Advances in Group Processes : 30th Anniversary Edition [30 ed.] 9781781909775, 9781781909768

Citation preview

ADVANCES IN GROUP PROCESSES: THIRTIETH ANNIVERSARY EDITION

ADVANCES IN GROUP PROCESSES Series Editors: Edward J. Lawler and Shane R. Thye Recent Volumes: Volumes 1–17:

Edited by Edward J. Lawler

Volume 18:

Edited by Edward J. Lawler and Shane R. Thye

Volume 19:

Group Cohesion, Trust and Solidarity – Edited by Edward J. Lawler and Shane R. Thye

Volume 20:

Power and Status – Edited by Edward J. Lawler and Shane R. Thye

Volume 21:

Theory and Research on Human Emotions – Edited by Jonathan H. Turner

Volume 22:

Social Identification in Groups – Edited by Shane R. Thye and Edward J. Lawler

Volume 23:

Social Psychology of the Workplace – Edited by Shane R. Thye and Edward J. Lawler

Volume 24:

Social Psychology of Gender – Edited by Shelley J. Correll

Volume 25:

Justice – Edited by Karen A. Hegtvedt and Jody Clay-Warner

Volume 26:

Altruism and Prosocial Behavior in Groups – Edited by Shane R. Thye and Edward J. Lawler

Volume 27:

Edited by Shane R. Thye and Edward J. Lawler

Volume 28:

Edited by Shane R. Thye and Edward J. Lawler

Volume 29:

Edited by Will Kalkhoff, Shane R. Thye and Edward J. Lawler

ADVANCES IN GROUP PROCESSES

VOLUME 30

ADVANCES IN GROUP PROCESSES: THIRTIETH ANNIVERSARY EDITION EDITED BY

SHANE R. THYE Department of Sociology, University of South Carolina, USA

EDWARD J. LAWLER School of Industrial and Labor Relations, and Department of Sociology, Cornell University, USA

United Kingdom – North America – Japan India – Malaysia – China

Emerald Group Publishing Limited Howard House, Wagon Lane, Bingley BD16 1WA, UK First edition 2013 Copyright r 2013 Emerald Group Publishing Limited Reprints and permission service Contact: [email protected] No part of this book may be reproduced, stored in a retrieval system, transmitted in any form or by any means electronic, mechanical, photocopying, recording or otherwise without either the prior written permission of the publisher or a licence permitting restricted copying issued in the UK by The Copyright Licensing Agency and in the USA by The Copyright Clearance Center. Any opinions expressed in the chapters are those of the authors. Whilst Emerald makes every effort to ensure the quality and accuracy of its content, Emerald makes no representation implied or otherwise, as to the chapters’ suitability and application and disclaims any warranties, express or implied, to their use. British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library ISBN: 978-1-78190-976-8 ISSN: 0882-6145 (Series)

ISOQAR certified Management System, awarded to Emerald for adherence to Environmental standard ISO 14001:2004. Certificate Number 1985 ISO 14001

CONTENTS LIST OF CONTRIBUTORS

vii

PREFACE

ix

THIRTY YEARS OF ADVANCES IN GROUP PROCESSES: A REVIEW ESSAY Morris Zelditch STILL COLOR-BLIND? THE TREATMENT OF RACE, ETHNICITY, INTERSECTIONALITY, AND SEXUALITY IN SOCIOLOGICAL SOCIAL PSYCHOLOGY Matthew O. Hunt, Pamela Braboy Jackson, Samuel H. Kye, Brian Powell and Lala Carr Steelman RACE AND ETHNIC COMPOSITION OF GROUPS: EXPERIMENTAL INVESTIGATIONS Carla Goar, Jane Sell, Bianca Manago, Calixto Melero and Bobbi Reidinger THE SPREAD OF STATUS VALUE: A THEORETICAL EXTENSION Joseph Berger and M. Hamit Fis- ek THE ROLE OF UNCERTAINTY IN SOCIAL INFLUENCE Scott V. Savage, David Melamed and Aaron Vincent

v

1

21

47

77

109

vi

CONTENTS

THE PROFESSIONAL CREDENTIALS OF IMMIGRANTS: A STATUS-AND-EXPECTATIONS APPROACH Martha Foschi

131

SELECTION AND INFLUENCE IN THE ASSIMILATION PROCESS OF IMMIGRANTS Arnout van de Rijt

157

USING SIMULATED INTERACTIONS TO EXPLORE EMOTIONAL PROCESSES AND STATUS ORGANIZING PROCESSES: A JOINT APPLICATION OF EXPECTATION STATES THEORY AND AFFECT CONTROL THEORY Joseph Dippong

195

AGE AND THE EXPERIENCE AND MANAGEMENT OF EMOTION Kathryn J. Lively

231

A GENERALIZED THEORY OF CONFLICTING LEADER IDENTITY ON GROUP PERFORMANCE Christabel L. Rogalin

267

THE IVY AND THE TRELLIS: AGENCY, BIOLOGY, AND SOCIALIZATION David Peterson

293

LIST OF CONTRIBUTORS Joseph Berger

Department of Sociology, Stanford University, CA, USA

Joseph Dippong

Department of Sociology, University of North Carolina-Charlotte, NC, USA

M. Hamit Fis- ek

Department of Psychology, Bog˘azic- i University, Turkey

Martha Foschi

Department of Sociology, University of British Columbia, BC, Canada

Carla Goar

Department of Sociology, Kent State University, OH, USA

Matthew O. Hunt

Department of Sociology, Northeastern University, MA, USA

Pamela Braboy Jackson

Department of Sociology, Indiana University, IN, USA

Samuel H. Kye

Department of Sociology, Indiana University, IN, USA

Kathryn J. Lively

Department of Sociology, Dartmouth College, NH, USA

Bianca Manago

Department of Sociology, Texas A&M, TX, USA

David Melamed

Department of Sociology, University of South Carolina, SC, USA

Calixto Melero

Department of Sociology, Texas A&M, TX, USA

David Peterson

Department of Sociology, Northwestern University, IL, USA

vii

viii

LIST OF CONTRIBUTORS

Brian Powell

Department of Sociology, Indiana University, IN, USA

Bobbi Reidinger

Department of Sociology, Kent State University, OH, USA

Christabel L. Rogalin

Department of Social Sciences, Purdue University, North Central, IN, USA

Scott V. Savage

Department of Sociology, University of California, Riverside, CA, USA

Jane Sell

Department of Sociology, Texas A&M, TX, USA

Lala Carr Steelman

Department of Sociology, University of South Carolina, SC, USA

Arnout van de Rijt

Department of Sociology and the Institute for Advanced Computational Science, Stony Brook University, NY, USA

Aaron Vincent

Department of Sociology, University of South Carolina, SC, USA

Morris Zelditch

Department of Sociology, Stanford University, Stanford, CA, USA

PREFACE Advances in Group Processes publishes theoretical analyses, reviews, and theory-based empirical chapters on group phenomena. The series adopts a broad conception of ‘‘group processes.’’ This includes work on groups ranging from the very small to the very large, and on classic and contemporary topics such as status, power, trust, justice, influence, decision-making, intergroup relations, and social networks. Previous contributors have included scholars from diverse fields including sociology, psychology, political science, business, philosophy, computer science, mathematics, and organizational behavior. Volume 30 represents a very important milestone for the series – marking 30 years of Advances in Group Processes. Over this time the series has seen a number of publishers (JAI Press, Elsevier Science, and now Emerald), benefited from the specialized-expertise of an occasional guest editor, varied formats between open topic volumes and those organized around a particular theme, and proudly published theoretical and empirical work relating to a wide assortment of group processes. Edward J. Lawler initiated the series 30 years ago in 1984 and Shane R. Thye joined as a series coeditor 16 years later in 2000. As editors it has been very gratifying to see the series flourish, its impact on the field grow, and the flexibility the series has provided authors over the years. Virtually all of the publishers have echoed these sentiments. The current volume opens with somewhat unique chapters for the series. In ‘‘Thirty Years of Advances in Group Processes: A Review Essay,’’ Morris Zelditch reflects on the evolution of group processes research over the past decades and comments on the role that the series has played in that progression. And while we did not realize this when he was invited, as series editors, it was affirmed for us that ‘‘Buzz’’ was the perfect person to write a review piece when we inadvertently discovered that he had published papers in volume 1 of Advances in Group Processes in 1984 (Zelditch and Henry Walker’s Legitimacy and the Stability of Authority), again in the 20th anniversary volume in 2003 (Zelditch and Walker’s The Legitimacy of Regimes), and now in the 30th anniversary volume. Buzz has had the benefit

ix

x

PREFACE

of seeing the discipline change over the past 30 years as one of the most creative, diverse, widely-read, and productive among group processes scholars. In some ways it is befitting that he should comment on 30 years of Advances in Group Processes given that he has occupied a front row seat to the show, and had a major hand in determining the program. We are very pleased to have his insights open this 30th anniversary volume. Students of group processes will learn much from his analysis. The next two chapters review and summarize the state of group processes research as it pertains to race. In ‘‘Still Color-Blind? The Treatment of Race, Ethnicity, Intersectionality, and Sexuality in Sociological Social Psychology,’’ Matthew O. Hunt, Pamela Braboy Jackson, Samuel H. Kye, Brian Powell, and Lala Carr Steelman update an earlier analysis regarding the treatment of race in social psychological scholarship. In this chapter they conduct a content analysis of articles published between 2000 and 2012 in Social Psychology Quarterly to determine the extent to which such articles refer to or seriously engage the topics of race/ethnicity and sexuality. They find a substantial increase in articles that incorporate race or ethnicity relative to the prior decade, yet, articles addressing sexuality in some important manner remain rare. The following chapter examines how race/ethnicity can be examined using experimental methods. In ‘‘Race and Ethnic Composition of Groups: Experimental Investigations,’’ Carla Goar, Jane Sell, Bianca Manago, Calixto Melero, and Bobbi Reidinger review the current experimental literature that focuses on race/ethnicity. They report data from a new experiment that examines interaction among (i) Mexican American and white participants in Texas and, (ii) African-American and white participants in Ohio. Like the previous authors, they argue that more could be done to promote the understanding of race/ethnicity among those who investigate group processes. Both chapters make salient the importance and often neglected role of race in group processes investigations. The following four chapters address some aspect of status or social influence in groups. Joseph Berger and M. Hamit Fis- ek formally theorize how it is that status valued elements come into existence via their relations to other status valued elements. In ‘‘The Spread of Status Value: A Theoretical Extension,’’ Berger and Fis- ek build on several strands of work related to the creation and spread of status value. Most important for those interested in status characteristics theory is that Berger and Fis- ek represent this phenomenon using the graph theoretical procedures of that theory. The effect should be to more fully integrate and represent the spread of status value with existing theoretical structures. Next, Scott V. Savage, David Melamed and Aaron Vincent work to integrate faction size with status

Preface

xi

characteristics theory in ‘‘The Role of Uncertainty in Social Influence.’’ The basic premise is that status matters most when the distribution of opinions regarding the correctness of a choice are evenly divided (i.e., group members equally select different options). Under such conditions, the theory predicts that a given decision maker will experience uncertainty, and thus rely on other information such as status to guide a choice. Data from a new laboratory study are generally supportive. In the next chapter Martha Foschi investigates whether being (i) native to an area or an immigrant, and (ii) the country of credentials serve as status characteristics in ‘‘The Professional Credentials of Immigrants: A Status-and-Expectations Approach.’’ Foschi reviews Canadian evidence from surveys, evaluations, and employment statistics to determine if these factors operate as status characteristics, and overall, the evidence suggests that they do. Based on these findings, she proposes a program of research for future investigations. Overall, these three chapters use the established core of status characteristics theory to forge new connections with related phenomena and all should stimulate new theoretical research. Finally, Arnout van de Rijt examines various dimensions of assimilation among Canadian immigrants in ‘‘Selection and Influence in the Assimilation Process of Immigrants.’’ Whereas previous theory and research has focused on the exogenous factors that alter the rate of assimilation, Arnout van de Rijt offers a new model that suggests variation in immigration outcomes can emerge endogenously, through cognitive dissonance processes. This new model is tested with data from the Longitudinal Survey of Immigrants to Canada. The result is a cutting edge analysis how the number of assimilated dimensions affects homophily. The next two chapters are aimed at the intersection of emotion and identity. Joseph Dippong offers new experimental simulations in ‘‘Using Simulated Interactions to Explore Emotional Processes and Status Organizing Processes: A Joint Application of Expectation States Theory and Affect Control Theory.’’ Dippong uses affect control theory’s Interact program to investigate the emotional and identity processes that undergird status structures. This represents one of the most thorough and useful joint investigations of how status processes interact with affect and identity processes to date, and will interest both senior and junior scholars interested in these and related phenomena. Next, Kathryn J. Lively explores the correlational structure of emotion in ‘‘Age and the Experience and Management of Emotion.’’ Using data from the 1996 General Social Survey’s emotion module, Lively examines the correlational structure of positive and negative emotions for individuals over (and under) the age of 60. Notable variation in emotion structure are found for older and younger individuals,

xii

PREFACE

and Lively draws out the implications for how these affect the psychology of aging and the process of emotion management. Both chapters are at the cutting edge of theory and research on emotion. The volume closes with two chapters related to leadership and socialization. Christabel L. Rogalin offers new theoretical insights in ‘‘A Generalized Theory of Conflicting Identity on Group Performance.’’ In this contribution she brings together elements from identity control theory and role congruity theory to explain the identity-related consequences for female leaders of groups both large and small. This theory builds on a long tradition of gender work in the status and identity traditions, and importantly, offers new solutions to improve group performance, and guidelines for women in leadership positions. A key theoretical contribution is that this work builds cumulatively on gender scholarship that traverses many facets of the group processes arena. Finally, David Peterson offers a new theory of socialization in ‘‘The Ivy and the Trellis: Agency, Biology, and Socialization.’’ This work synthesizes and brings together work in modern neuroscience, developmental psychology, and sociological theory. One key contribution is that this work suggests new mechanisms to explain and answer questions regarding cultural transmission. Overall, this chapter poses fresh questions and offers new solutions for contemporary theories of socialization.

THIRTY YEARS OF ADVANCES IN GROUP PROCESSES: A REVIEW ESSAY Morris Zelditch ABSTRACT Purpose – This chapter reviews 30 years of Advances in Group Processes. Its primary purpose is to study the part the series has played in the advances in the study of group processes that have taken place between 1984 and 2014. Design/methodology/approach – This chapter places the 30 years of Advances in Group Processes in the context of the changes that took place between small groups research in the 1950s and group processes research in the 1980s and beyond. Findings – Analyzing the policies of Advances in Group Processes and its contents, this chapter reflects on its role in the advances in group processes that have taken place since the 1980s. Between 1950 and 1980, small group research reinvented, reconceptualized, and reinvigorated itself as group process research. Between the two periods, small group research, its applied research, and its research programs became increasingly theory-driven and its concept of the group and its levels increasingly analytic. As a consequence of these changes, the concept of the field itself

Advances in Group Processes: Thirtieth Anniversary Edition Advances in Group Processes, Volume 30, 1–19 Copyright r 2013 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 0882-6145/doi:10.1108/S0882-6145(2013)0000030004

1

2

MORRIS ZELDITCH

became increasingly analytic. The changes between the two periods in its theory, research, application, programs, and in its concept of the group and the way the field was conceptualized led to marked advances in group process research in the 90s and beyond – to more theory, more impact of it on application, and more, and more cumulative, growth of it. Advances in Group Processes was at once a reflection of the changes that took place between the two periods and a driving force in the advances in group processes research that have taken place ever since. Originality/value – Advances in Group Processes is a fundamental resource for the development of theory and research on small groups and group processes. This chapter provides an overview of its contributions and places them in the context of the development of the field as a whole. Keywords: Small groups; group processes; programs of theoretical research; theory and research; theory and application; experimental methods

INTRODUCTION Between 1950 and 1980 small group research had changed so much that many were asking ‘‘whatever happened to small group research?’’ (See, e.g., Hare, 1976, appendix 2; Steiner, 1974, 1983, 1986). Some were even answering that the light had failed (Mullins, 1973). Despite the fact that well into the 80s Davis and Stasson (1988) provided ample evidence that the disappearance of small group research was largely an illusion, some continued to ask the question for another dozen years (e.g., Harrington & Fine, 2000; Manson, 1993) and answer it for another decade after that (e.g., Harrod, Welch, & Kushkowski, 2009; Moreland, Hogg, & Hains, 1994; Sanna & Parks, 1997). While theory and research on small groups was waning, theory and research on group processes was waxing. The fifties saw the rapid and prolific growth of research on small groups. The seventies saw what many thought was its decline, even, possibly, its disappearance. But the eighties saw the rapid and prolific growth of research on group processes, growth that accelerated in the nineties. I do not mean to suggest that small group research simply acquired an alias in the seventies. I do mean to suggest that what happened to it was not exactly its disappearance. It merely climbed a substantial number of rungs

Thirty Years of Advances in Group Processes

3

up the ladder of abstraction. Its research, its applications, and its programs became more theory-driven. Its concept of the group and its levels became more analytic. More theory-driven research, applications, programs, and a more analytic conception of the group transformed the way the entire field was conceptualized: Small group research didn’t just acquire an alias, it reconceptualized, reinvented, and reinvigorated itself as the study of group processes. As a consequence, there was more rapid, prolific growth of its theory, its research, its applications, and its programs. The 30 years (so far) of Advances in Group Processes were at once a reflection of the changes that took place in small group research between the fifties and the eighties and a driving force in the subsequent advances in our understanding of group processes that were its consequence – at once evidence that the field was in fact flourishing and a key factor in the advances that have since taken place in the field. The present introduction to this anniversary volume is, first, simply a comment on the changes in small group research that took place between 1950 and 1980; second, it comments also on their consequences for the subsequent growth of group processes research; and, finally, it comments on the part played in the growth of group processes research by the first 30 years of Advances in Group Processes.

WHATEVER HAPPENED TO SMALL GROUP RESEARCH? Disambiguating ‘‘decline’’ and ‘‘disappearance’’ is difficult, but some, for example Manson (1993) and Harrington and Fine (2000), thought that the research on group processes that flourished in the eighties and nineties was in itself evidence of the decline of theory and research on ‘‘the group in its own right.’’ While Davis and Stasson (1988) were certainly right that the disappearance of the small group was in fact illusory, because all that had happened to it was its speciation, Manson, Harrington, and Fine were also right that research on ‘‘the’’ group did in some sense disappear. But it was less a disappearance of the group than a marked transformation of the field. The research of the eighties, its applications, and the programs that emerged from them were all more theory-driven, the group and its levels were more analytic than they had been in the fifties. What had happened to small groups research between the fifties and the eighties was, in a word, theory.

4

MORRIS ZELDITCH

Research Small group research up to and into the fifties was a field of many methods, from ethnography (e.g., Whyte, 1943) to survey research (e.g., Stouffer et al., 1949), to field experiment (e.g., Roethlisberger & Dickson, 1939), to laboratory observation (e.g., Bales, 1950) and experiment (e.g., Asch, 1951). Little of the ethnography, survey research, or field experiment was theorydriven, but the signature accomplishment of the fifties was the rapid, prolific growth of laboratory observation and experiment. But not all laboratory observation and experiment is theory-driven: While the laboratory observation and experiment of the eighties was largely theory-oriented, testing, refining, and/or extending a theory of a process, such classic laboratory experiments of the fifties as Asch (1951) or Lewin (1947) and such classic laboratory observations of the fifties as Bales, Strodtbeck, Mills, and Roseborough (1951) or Bales and Slater (1955) simply demonstrated an effect rather than testing a theory. Not that there was no theory. Lewin had field theory (e.g., Lewin, 1951); the Bales of the Working Papers had functionalism (e.g., Bales, 1953); Goffman (1956) had interactionism crossed with functionalism. But, with some exceptions (e.g., Festinger, 1950; Goffman, 1956; Heider, 1958) the fifties were notable mostly for the gap between its theory and research. Its theory was more orienting strategy than testable theory, its research more demonstrations of an effect than tests of a theory.

Application By ‘‘applied,’’ the applied research of the eighties meant the use of a theory to explain a concrete, particular effect (e.g., Cohen, 1972, 1993; Driskell & King, 2007; Driskell & Radtke, 2003). Its social engineering derived interventions in social problems from the theory it applied to explain them, such as the learning disabilities of minorities (Cohen, Katz, & Lohman, 1976; Cohen & Roper, 1972) or the communication disabilities of computermediated communication (Driskell & Radtke, 2003). But the applied research of the fifties was as little driven by theory as its basic research. Despite Lewin’s maxim that there was nothing as practical as a good theory (Lewin, 1951, p. 169), applied research up to and into the fifties (e.g., Roethlisberger & Dickson, 1939; Stouffer et al., 1949) was largely atheoretical. By ‘‘applied’’ the fifties merely meant research oriented to social, economic, or political problems. It did not in fact apply anything.

Thirty Years of Advances in Group Processes

5

Its social engineering consisted of ad hoc interventions, of which the illumination experiments at Western Electric’s Hawthorne plant were merely the most famous example (Roethlisberger & Dickson, 1939).

Programs The differences between the two periods in the nature and functions of their basic and applied research led to marked differences in the nature and functions of the programs that grew out of them. The eighties were largely dominated by theoretical research programs, programs concerned with advancing the confirmation status, precision, and rigor of a theory of a process and extending its scope and domain (cf Berger & Zelditch, 1997). The fifties were largely dominated by effect research programs, that is, programs concerned with demonstrating the regularity of some effect of a group and explaining it (cf Zelditch, 2007). What looks like a subtle difference in the language I have just used made some not so subtle differences in the nature and functions of the two kinds of programs. Theoretical research programs are concerned with theories of a process, for example, the theory of relational cohesion with group formation (e.g., Thye, Yoon, & Lawler, 2002), identity control theory with identity processes (e.g., Stets & Burke, 2005), network exchange theory with power processes (e.g., Willer & Skvoretz, 1997), and group value theory with procedural justice (e.g., Tyler, Smith, & Ho, 1996). By contrast, the effects of the fifties, though often fundamental, powerful, and important, were often underanalyzed and complex. They were complex in the sense that they often involved more than one process. Asch, for example, confounded the normative effect of group pressures toward conformity with the informational effect of others as sense evidence (Deutsch & Gerard, 1955); Bales confounded the emergence of interaction hierarchies with the legitimacy problems of their emergence (Burke, 1967, 1968). They were under-analyzed in the sense that demonstration of their effects often did not explain them and conceptualization of them often did not distinguish one process from another. That the effects that drove them were under-analyzed had some undesirable consequences for the growth of theory in the fifties. In the first instance, demonstration of an effect was often followed by its replication in order to demonstrate its empirical regularity. Strodtbeck, for example, ‘‘validated’’ Bales’ interaction hierarchy in cross-cultural studies of family interaction (Strodtbeck, 1951), and his role-differentiation in juries

6

MORRIS ZELDITCH

(Strodtbeck & Mann, 1954). Replication was important in establishing empirical regularities but did nothing to explain them. In some cases an effect was not so much replicated as rediscovered. For example, an empirical regularity found in one concrete, particular setting, such as the effect of military rank on the interaction hierarchy of an air crew or of race in the classroom, was rediscovered, over and over, in many other concrete, particular settings, from juries to hospitals to universities (Cohen, Berger, & Zelditch, 1972). Unintended rediscoveries were also important in establishing empirical regularities but, absent a theory to systematically relate one concrete, particular result to another, left them fragmented, often compartmentalized, among different specialty areas (Davis & Stasson, 1988). But the programs to which effect-oriented research often gave rise were also often concerned with analyzing what, exactly, it was that had had the effect. Explanations depend on both initial conditions and laws, but much of this subsequent research was concerned with an effect’s initial conditions. Because the effect was under-analyzed, its initial conditions tended to be ad hoc, to multiply ad infinitum, and, because the effect was often complex, to be unrelated to each other. Hence, effect research programs tended often to be voluminous but incoherent. The Asch program, for example, multiplied procedural factors such as stimulus ambiguity, situational factors such as the size of the majority or status differences, and personality factors such as dependence, achievement orientation, and authoritarianism (Allen, 1965, 1977; Hare, 1976, p. 394). The Bales program multiplied size of the group, seating position of participants, consensus of participants, and, again, a host of personality factors (Hare, 1976, p. 393, 1989).1 But not all effect programs were concerned only with initial conditions. Some were concerned with exactly what processes had had the effect. But to the extent that an effect was not only under-analyzed but also complex, the effect of two or more processes, it was also irregular. (A process is regular if the same causes always have the same effects. If a process is complex, a change in conditions has different effects on its different processes. Whether an effect is or is not complex is an empirical question; that, if it is complex, it is irregular is true by definition.) Because the Asch effect was complex, for example, the same causes did not always have the same effects. Normative influence depended on the presence of a group and observability of choices, but informational influence depended only on the presence of an aggregate and occurred even if choices were anonymous (Deutsch & Gerard, 1955). Role-differentiation (Bales & Slater, 1955) depended on the legitimation problems of emergent hierarchies in homogeneous groups (Burke, 1968), a process not found in heterogeneous groups (Burke, 1967).

Thirty Years of Advances in Group Processes

7

The irregularity of an effect led to explication of the processes underlying it, not to complicating its initial conditions: What Deutsch and Gerard and Burke were searching for were processes the causes of which always had the same effects or, in other words, unit processes. (Again, whether a process is or is not a unit process is an empirical question; that, if it is a unit process, it is regular is true by definition.) Programs focused on unit processes – in the case of small group research, ‘‘group’’ processes – began to spin off of effect programs, as power-dependence theory did from the Asch program (Emerson, 1964)2 and expectation states theory did from the Bales program (Berger, 1958), morphing from effect programs into theoretical research programs.

The Group and its Levels The concept of a group in the fifties was as under-analyzed and complex as its effects. It is difficult to climb the ladder of abstraction any higher than field theory, functionalism, or, Blumer aside, interactionism. But functionalism, field theory, and interactionism were strategies, not theories. They were not, in themselves, directly testable. Strategies can be tested indirectly but only if they are embodied in more directly testable theories. There were, in fact, some directly testable theories in the fifties (e.g., Festinger, 1950; Goffman, 1956; Heider, 1958), but it was a period more noted for the gap between its theory and research than for the testability of its theories. The gap between its theory and its research meant that up to and including the fifties its studies of naturally occurring small groups were studies of concrete particulars (e.g., Roethlisberger & Dickson, 1939), its experiments, even when their method was analytic (e.g., when their groups were ad hoc), were conceptualized in terms of ad hoc, common-sense categories of the natural language rather than theories (e.g., Asch, 1951). Its concept of a group was itself nothing more than common usage: It meant by a ‘‘group’’ any aggregate of individuals whether taken individually or collectively. The elasticity of its concept of a group stretched from the simple presence of an other to encounters, to gatherings, or any other form of social interaction; to weak ties as well as strong ties, or any other form of relation; to indirect ties as well as direct ties, hence any form of network – or, in other words, to non-groups as well as groups; and, finally, to any form of cooperative problem-solving, decision-making, product-producing, or any other form of collectively acting group. It was a field held together largely by the looseness with which it used the term ‘‘group.’’ But all of its strategies, in one way or

8

MORRIS ZELDITCH

another, implied that there was ‘‘a’’ theory of ‘‘the’’ group, either of ‘‘a’’ generic social process or of ‘‘the’’ confluence of the fundamental processes of any social system. Cartwright and Zander (1953), it was true, eschewed any single theory (see their preface); nevertheless three of the six parts of Group Dynamics were a theory of the formation of ‘‘the’’ group and its pressures toward uniformity and group locomotion. Interactionism, of course, was not a theory of ‘‘the’’ group at all. But it was, with the notable exception of Goffman3, a theory of ‘‘the’’ process of all social interaction. The concept of a group in the more theory-driven eighties was just as elastic as it had been in the fifties (e.g., the small, task-oriented groups of expectation-states theory, the networks of behavioral exchange and network exchange theory, the referential other of justice theories). Nevertheless, between the fifties and the eighties it became more analytic in two ways. On the one hand, it became more abstract and general. On the other hand, it became less complex. The theory-driven eighties decomposed ‘‘the’’ generic process of ‘‘the’’ group, or ‘‘the’’ confluence of the fundamental processes of ‘‘any’’ group, into dissoluble unit processes (e.g., group formation, as in Thye et al., 2002, identity as in Stets & Burke, 2005, power as in Willer & Skvoretz, 1997, procedural justice as in Tyler et al., 1996). It also differentiated their scope conditions, the conditions under which their theories applied, for example, their task and interaction conditions and their time horizons – that is, whether or not there was a past, whether interaction was or was not repeated, whether there was or was not a future. Scope conditions are variables that condition the conditionals that are the laws of a process. One effect of the explication of a complex process is that such conditions become variables in the laws of unit processes – of ‘‘group’’ processes in the case of small groups research. Because they are more analytically conceived, concrete entities such as task-oriented groups, exchange relations, markets, or firms become abstract variables: variables such as the task-orientation or repeated interaction of the actors in them or the simultaneous, decentralized vs sequential, centralized decision-making of the actors in them. One by-product was that in the eighties, although just as elastic as it had been before, the use of the term ‘‘group’’ was more precise. The differences between the many things small groups research had always meant by a ‘‘group’’ became the scope conditions of the different theories of its different processes. In the process, ‘‘the’’ group disappeared, displaced by scope-conditioned group processes. The theory-driven eighties had not only a more analytic concept of the group than the fifties but also a more analytic concept of its levels. Abstract, general, inter-actor theories are as capable of macro as micro interpretations.

Thirty Years of Advances in Group Processes

9

There are marked differences between them, but their differences are differences in the elements, variables, or interpretations of the theory. That well into the eighties ‘‘micro’’ and ‘‘macro’’ meant at least four different things – small versus large, actor versus system, structure versus action, individual actors versus group actors – did not matter. Abstract, general, interactor theories are as capable of macro as micro interpretations no matter which meaning their terms are given (cf Berger, Eyre, & Zelditch, 1989). Nevertheless, to some, particularly in the sixties and seventies when small groups research was thought to be waning, application of an abstract, general inter-actor theory to macro-sociology was thought of as reductionism. The sociology of the period was caught up in a raging conflict over four unresolved antinomies: small versus large, actor versus system, structure versus action, individual versus group. One of this conflict’s by-products was the rejection by ‘‘macro-sociology’’ (e.g., by Blau, 1977; Mayhew, 1980) of the sort of reduction to micro-sociology implied by the argument by some small group research that micro-sociology was merely miniaturized macrosociology (Hare, 1976; Hare, Borgatta, & Bales, 1955, 1965) or, of course, the actual reduction of macro-sociology to micro-sociology by Homans (1958, 1964). If levels are antinomies there are only three solutions: independence – that is, different theories for different levels – reduction, or interrelation. The sixties and seventies were more caught up in either reducing or dichotomizing than interrelating levels. But the more analytically oriented approach that was emerging was dissatisfied with either solution. Rejecting reduction on the one hand, dichotomy on the other, the eighties instead found linkage between micro and macro-sociology, duality of structure and action, variability of scale within rather than between theories, and whether actors were individuals or groups simply different interpretations of one rather than two different theories (see, e.g., Alexander, Giesen, Muench, & Smelser, 1987; Huber, 1991).

The Field A more analytic concept of the group and its levels transformed the entire field of small group research. Davis and Stasson (1988) are right that its disappearance was illusory, but less because much of it was still going on in specialty areas than because it did not so much disappear as reinvent itself. It was, in fact, flourishing in the eighties. But the increase in its level of abstraction and decrease in complexity of its processes transformed

10

MORRIS ZELDITCH

small group research into group process research. It can even be said that small group research, rather than disappearing, played a fundamental role in the group process research of the eighties because group process research did not so much forget it as subsume it. Its effects, its regularities, its explications in fact were the foundations on the shoulders of which group process research was built. But in the process, its reinvention, its reconceptualization revitalized, reinvigorated it. The seeds of all this change were actually already sown in the fifties. The fifties were already closing the gap between theory and research, already constructing testable theories (e.g., Festinger, 1950; Goffman, 1956; Heider, 1958) already testing them (e.g., Back, 1951; Schachter, 1951), already explicating its complex effects (e.g., Cartwright & Harary, 1956; Deutsch & Gerard, 1955; Festinger, 1953), and theoretical research programs were already morphing out of the explication of them, for example, Berger (1958) out of Bales, and Cohen (1958) and Emerson (1964) out of Asch. On the one hand, it is true that the fifties were the apogee of small group research. But they were also the watershed of its transformation into the theory-driven research, application, and programs of the eighties.

ADVANCES IN GROUP PROCESSES The small group research of the fifties was path-breaking, foundational, but growth of its theory was limited. The eighties were not only more theorydriven, but had, as a consequence, more theory, more impact of theory on applied research and applied research on theory, more, and more cumulative, growth of theory, and, as a consequence of its more analytic approach to the group and its levels of analysis, also a more sociological social psychology than the fifties. By more theory I mean that, not only were research, application, and programs of research more theory-driven, but that there was more sheer theoretical activity than in the fifties. There were more theories being constructed, more of them were empirically grounded, testable theories, there was more theoretical innovation and more competition between theories than in the fifties. That there was more competition between theories that were more directly testable than orienting strategies was important in view of the period’s more sophisticated understanding of what it meant to refute a theory, in the sense that an anomaly was not in itself sufficient to refute a theory if there were no alternative superior to it. More competition, therefore, also meant more advances in the growth of theory.

Thirty Years of Advances in Group Processes

11

But there were also more ways that theory grew. In the fifties growth was explosive, but was often growth more of facts, effects, or empirical regularities than theory. Growth in the number of facts, effects, or empirical regularities is not necessarily growth of theory, that is, growth of what we know in some more general way about how something works, about the nature, causes, conditions, and consequences of a process. But the concept of an ‘‘advance’’ was itself reconceptualized between the fifties and the eighties. An ‘‘advance’’ in the eighties meant the modification of a theory of a process. A theory of a process advanced if its confirmation, precision, or rigor increased, competing theories were integrated, or its scope or domain were extended, extending its level of generality or its explanatory power. The research programs of the eighties grew more, in this sense, and more of their growth was cumulative, than in the fifties. For example, the confirmation status of its theories advanced both by testing untested implications of them and by explaining anomalies in the existing support for them. For example, Willer and Szmatka (1993) tested and confirmed Willer’s resistence theory of power. Michener (1992), on the other hand, found a way to explain the anomalies plaguing tests of cooperative game theories of coalition formation. Advances in precision included refinements by Markovsky, Willer, and Patton (1988) of the determinants of the power of a position in a network which made prediction of the effects of the power of a position more precise. Expectation-states theory was refined by Shelley (1998) who graded binary expectation-states, making predictions of influence more precise, an advance extended to status characteristics by Melamed (2011). The rigor of theories advanced by axiomatizing or formalizing them. For example, Barnum (1997) axiomatized situated identity theory; Fisek and Wagner (2003) formalized the expectation-states theory of reward allocation. Competing theories were integrated, for example by Lawler (1986), who unified deterrence and conflict spiral theories by subsuming them in one theory of intergroup conflict capable of differentiating the conditions under which they occurred. Extending the scope of a theory is a matter of generalizing the conditions under which it applies. For example, Molm (1996) generalized the behavioral exchange theory of power to punishment and coercion processes. Willer and Skvoretz (1997) generalized the scope of network exchange theory to any mix of network connections no matter what the structure of the network.

12

MORRIS ZELDITCH

Extending the domain of a theory, on the other hand, is a matter of extending the number of different phenomena it explains. For example, Berger, Webster, Ridgeway, and Rosenholtz (1986) extended the domain of expectation states theory to nonverbal and paraverbal social interaction; Stets and Burke (2005) extended identity control theory to the sociology of emotions. It was not simply that what it meant for research to be theory-driven implied that what grew was theory, it was also that it was theoretical relevance that drove it. It was focused on the unsolved problems of a theory, problems the solution of which promised greater theoretical payoff than the problems solved by effect programs. Both effect and theory programs are driven by the problems they have not solved, but the focus of a program on a theory has the obvious advantage of solving problems more relevant to and more promising for a theory. The more differentiated understanding of the levels of theory that emerged between the fifties and the eighties (cf. Berger & Zelditch, 1997) also had an impact on the growth of theory. Conflict between strategies tended often, throughout the sixties and seventies, to be irreconcilable, constituting a ‘‘crisis’’ rather than an ‘‘advance’’ of theory. But, as more theory-driven research emerged, more conflict was competition between testable theories and, especially if the competition was between variants of theories sharing common standards of assessment, led to actual resolution of theoretical disputes, hence actual theory growth. Thus, Ridgeway (1984) resolved a raging debate between status based on performance, the approach of expectation-states theory, and status based on assertive nonverbal and paraverbal behavior, the approach of behavioral dominance theory, in the process opening up new areas of research on nonverbal and paraverbal behavior in expectation-states theory (Berger et al., 1986).4 The fact that applied research also became more theory-driven between the fifties and the eighties also meant that social engineering became more powerful (Driskell & King, 2007) and not only did theory have a greater impact on application but application also had a greater impact on the growth of theory (e.g., Cohen, 1993). There was also more theory-oriented research on modeling the application of a theory to concrete, particular, instances of a process. Concrete phenomena are seldom unit processes. But, by definition, the conjuncture of two or more unit processes is irregular; that is, one unit process may conjoin with another some of the time but not all of the time – otherwise, they would (by definition) be one unit process. Status, for example, often combines with power, rewards, and/or sentiments. Empirically modeling any particular application of a theory of a unit

Thirty Years of Advances in Group Processes

13

process is therefore often a problem of inter-relating two or more processes, hence two or more theories. An important task of applied research is therefore interrelating, in one way or another, the two or more theories of the two or more unit processes found in the concrete instance to which a theory is applied. (For good examples of this sort of research see the interrelation of performance and reward expectations by Berger, Fisek, Norman, & Wagner, 1985; the interrelation of performance expectations and sentiment processes by Shelley, 1993, & Fisek & Berger, 1998; and the interrelation of multiple justice processes by Markovsky et al., 2008). Finally, the more analytic concept of levels of analysis emerging between the fifties and the eighties led to a more sociological conception of the social psychology of group processes. The inter-disciplinary reach of the group process research of the eighties was just as broad as that of small group research in the fifties but group process research not only reconceptualized small group research it also reconceptualized what it meant for a social psychology to be sociological rather than psychological. The more analytic approach to the micro–macro problem that emerged between the fifties and the eighties abstracted ‘‘group’’ processes from individual differences in pregiven dispositions. In their stead, it focused on more situationally conditioned motives, more situationally conditioned inter-actor processes, and more mutually contingent relations between the micro and macro levels of a system (e.g., Cohen, 1958; Lawler, Ridgeway, & Markovsky, 1993). For example, it rejected a form of reduction, found even among sociologists in the fifties, that constructed group phenomena by composition of individual differences in pre-given dispositions. (For example, see the reduction by Borgatta and Bales, 1953, of Bales’ interaction hierarchies to individual differences in pre-given rates of initiating performances.)

ADVANCES IN GROUP PROCESSES Advances in Group Processes5 is at once a reflection of the changes that took place between the fifties and the eighties and a driving force in the advances in group processes that have taken place ever since. The goal of its editorial policy, virtually unchanged for the last 30 years, has been to advance the field of group processes by generating more theoretical work. Its understanding of what that means has been marked by its breadth: On the one hand, it has continued to embrace meta-theory (e.g., Turner, 2000); on the other, it has done much to bridge the gap between theory and research. By encouraging more theory-driven empirical research

14

MORRIS ZELDITCH

it has done as much to ground theory in empirical research as it has to ground empirical research in theory. It has also understood theory as continually in progress, evolving into sustained programs testing, refining, articulating, integrating, and extending it. Consequently, not only its understanding of theory, but also its understanding of ‘‘advances’’ has been similarly marked by its breadth – not only conventional advances in confirmation status, but also in precision, rigor, integration, scope and domain. (All my earlier examples were from Advances in Group Processes.) Because of its understanding of programs, its publication policy has invited not only more theoretical, but also longer, more integrative papers than the piecemeal publication policy of the standard journals, facilitating a more coherent grasp of the current state of the art (e.g., Hare, 1989; Molm, 1996; Tyler et al., 1996; Thye et al., 2002).6 Finally, its editorial policy made explicit its understanding of a more analytic approach to the group and its levels, rejecting as false any dichotomy between micro and macro levels of analysis, inviting both and integrating them (e.g., Knottnerus, 2010; McPherson & Smith-Lovin, 2002; Turner, 2000). Its interdisciplinary outreach was also as broad as small group research had ever been but its more analytic concept of the group and its levels encouraged a more sociological, more situational, social psychology (e.g., Serpe & Stryker, 1993). Its tables of contents and its introductions to them, both ways of implementing its editorial policy, have also been instructive. For one thing, its understanding of theories in progress meant not only more emphasis on programs but also more emphasis on innovative theory construction, on new, untested, theory (e.g., Heckathorn & Rosenstein, 2002; Macy, 1993; Willer, 2009; Yamagishi, 1986). Furthermore, it has been more willing than the standard journals to publish them, if empirically grounded, before as well as after they have been tested, with data (e.g., Heckathorn & Rosenstein, 2002) or without (e.g., Macy, 1993; Willer, 2009; Yamagishi, 1986). It also strengthened the link between theory and application by its openness to papers positing, justifying, or demonstrating the applicability of theories to particular concrete instances, frequently repeating papers like Wagner’s application of status characteristics theory to the labeling of mental illness (Wagner, 1984); by its openness to papers like Dovidio, Gaertner, and Kafati’s deduction of interventions reducing prejudice and discrimination from their common inter-group identity model (Dovidio, Gaertner, & Kafati, 2000); and its openness to papers interrelating unit processes for applied purposes like Shelley (1993), Fisek and Berger (1998), and Markovsky et al. (2008). Finally, its introductions to its contents have

Thirty Years of Advances in Group Processes

15

been equally instructive, consistently underlining its editorial policy by focusing almost always on their theoretical significance. All this has made Advances in Group Processes a driving force in the advances in group processes that have taken place in the last 30 years: To the fact that there is more theory, that more of it has been testable, that more empirical research has been driven by it, that more of it has integrated macro with micro sociology, that there has been more application of it, and more and more cumulative growth of it.

NOTES 1. By contrast, although extension of the domain of a theory-driven program multiplies its effects, proliferation of which multiplies variants of the theory, the one thing it does not multiply is its processes and its variants remain interrelated by a core system of concepts and principles. 2. I am not trying to rewrite history. I do not mean that power-dependence theory was originally conceived of simply as an explanation of the Asch effect. But Emerson’s ‘‘two experiments’’ (1964) were written as spin offs of it. 3. Goffman began early to differentiate between the many forms of social interaction and progressively refined his analysis almost continually thereafter (see, e.g., Goffman, 1961, 1967). 4. I do not mean to imply that every theory is fruitful. Many are not. And if not, not every theory program grows. But if a theory is fruitful, the potential for growth of a theory-driven program is greater than the potential for growth of an effectdriven program (Zelditch, 2007). It is in this sense that the increase in the number of theory-driven programs drove more growth of theory. 5. That is, the series as a whole. 6. Whether the transformation of small group research into group process research was evolution, reform, or revolution probably doesn’t matter but in one respect Advances in Group Processes was not simply a reflection of how the field had changed. Its publication policy was more akin to Frickel and Gross’s social movement theory of scientific change (Frickel & Gross, 2005) with its editors as the cadre of the movement.

REFERENCES Alexander, J. C., Giesen, B., Muench, R., & Smelser, N. (Eds.). (1987). The micro–macro link. Berkeley, CA: University of California Press. Allen, V. I. (1965). Situational factors in conformity. Advances in Experimental Social Psychology, 2, 133–175. Allen, V. I. (1977). Social support for nonconformity. Advances in Experimental Social Psychology, 8, 1–43.

16

MORRIS ZELDITCH

Asch, S. E. (1951). Effects of group pressure on the modification and distortion of judgments. In H. Guetzkow (Ed.), Groups, leadership, and men (pp. 177–190). Pittsburgh, PA: Carnegie University Press. Back, K. W. (1951). Influence through social communication. Journal of Abnormal and Social Psychology, 46, 9–23. Bales, R. F. (1950). Interaction process analysis: A method for the study of small groups. Cambridge, MA: Addison-Wesley. Bales, R. F. (1953). The equilibrium problem in small groups. In T. Parsons, R. F. Bales & E. A. Shils (Eds.), Working papers in the theory of action (pp. 111–161). Glencoe, IL: Free Press. Bales, R. F., & Slater, P. E. (1955). Role differentiation. In T. Parsons & R. F. Bales (Eds.), Family, socialization, and interaction process (pp. 259–306). Glencoe, IL: Free Press. Bales, R. F., Strodtbeck, F. L., Mills, T. M., & Roseborough, M. E. (1951). Channels of communication in small groups. American Sociological Review, 16, 461–468. Barnum, C. (1997). A reformulated situated identity theory. Advances in Group Processes, 14, 29–57. Berger, J. (1958). Relations between performance, rewards, and action-opportunities in small groups. Ph. D. Dissertation, Harvard University, Cambridge, MA. Berger, J., Eyre, D. P., & Zelditch, M. (1989). Theoretical structures and the micro-macro problem. In J. Berger, M. Zelditch & B. Anderson (Eds.), Sociological theories in progress: New formulations (pp. 11–32). Newbury Park, CA: Sage. Berger, J., Fisek, M. H., Norman, R. Z., & Wagner, D. G. (1985). The formation of reward expectations in status situations. In J. Berger & M. Zelditch (Eds.), Status, rewards, and influence: How expectations organize behavior (pp. 215–261). San Francisco, CA: Jossey-Bass. Berger, J., Webster, M., Ridgeway, C., & Rosenholtz, S. (1986). Status cues, expectations, and behavior. Advances in Group Processes, 3, 1–22. Berger, J., & Zelditch, M. (1997). Theoretical research programs: A reformulation. In J. Szmatka, J. Skvoretz & J. Berger (Eds.), Status, network, and structure: Theory development in Group Processes. Stanford, CA: Stanford University Press. Blau, P. (1977). Inequality and heterogeneity: A primitive theory of social structure. New York, NY: Free Press. Borgatta, E. F., & Bales, R. F. (1953). Interaction of individuals in reconstituted groups. Sociometry, 16, 302–320. Burke, P. J. (1967). The development of task and social–emotional role differentiation. Sociometry, 30, 379–392. Burke, P. J. (1968). Role differentiation and the legitimation of task activity. Sociometry, 31, 404–411. Cartwright, D., & Harary, F. (1956). Structural balance: A generalization of Heider’s theory. Psychological Review, 63, 277–293. Cartwright, D., & Zander, A. (Eds.). (1953). Group dynamics: Research and theory. New York, NY: Harper & Row. Cohen, B. P. (1958). A probability model for conformity. Sociometry, 21, 69–81. Cohen, B. P., Berger, J., & Zelditch, M. (1972). Status characteristics and social interaction: A case study of the problem of developing cumulative knowledge. In C. McClintock (Ed.), Experimental social psychology (pp. 449–483). New York, NY: Holt, Rinehart, & Winston.

Thirty Years of Advances in Group Processes

17

Cohen, E. G. (1972). Interracial interaction disability. Human Relations, 25, 9–24. Cohen, E. G. (1993). From theory to practice: The development of an applied research program. In J. Berger & M. Zelditch (Eds.), Theoretical research programs: Studies in the growth of theory (pp. 385–415). Stanford, CA: Stanford University Press. Cohen, E. G., Katz, M., & Lohman, M. (1976). The center for interracial cooperation: A field experiment. Sociology of Education, 49, 47–58. Cohen, E. G., & Roper, S. S. (1972). Modification of interracial interaction disability: An application of status characteristic theory. American Sociological Review, 37, 643–657. Davis, J. H., & Stasson, M. F. (1988). Small group performance: Past and present research trends. Advances in Group Processes, 5, 245–277. Deutsch, M., & Gerard, H. B. (1955). A study of normative and informational social influences upon individual judgment. Journal of Abnormal and Social Psychology, 51, 629–636. Dovidio, J. F., Gaertner, S. L., & Kafati, G. (2000). Group identity and intergroup relations: The common in-group identity model. Advances in Group Processes, 17, 1–35. Driskell, J. E., & King, J. (2007). Conducting applied experimental research. In M. Webster & J. Sell (Eds.), Laboratory experiments in the social sciences (pp. 329–351). New York, NY: Elsevier. Driskell, J. E., & Radtke, P. H. (2003). The effect of gesture on speach production and comprehension. Human Factors, 45, 445–454. Emerson, R. M. (1964). Power-dependence relations: Two experiments. Sociometry, 27, 282–298. Festinger, L. (1950). Informal social communication. Psychological Review, 57, 271–292. Festinger, L. (1953). An analysis of compliant behavior. In M. Sherif & M. O. Wilson (Eds.), Group relations at the crossroads (pp. 232–256). New York, NY: Harper. Fisek, H. M., & Berger, J. (1998). Sentiment and performance expectations. Advances in Group Processes, 15, 23–40. Fisek, H. M., & Wagner, D. G. (2003). Reward expectations and allocative behaviors: A mathematical model. Advances in Group Processes, 20, 133–148. Frickel, S., & Gross, N. (2005). A general theory of scientific/intellectual movements. American Sociological Review, 70, 204–232. Goffman, E. (1956). Presentation of self in everyday life. Edinburgh: University of Edinburgh Press. Goffman, E. (1961). Encounters: Two studies in the sociology of interaction. Indianapolis, IN: Bobbs-Merrill. Goffman, E. (1967). Interaction Ritual: Essays on face-to-face behavior. New York, NY: Doubleday. Hare, A. P. (Ed.). (1976). Handbook of small group research. (2nd ed.). New York, NY: Free Press. Hare, A. P. (1989). New field theory: Symlog research, 1960–1988. Advances in Group Processes, 6, 229–257. Hare, A. P., Borgatta, E. F., & Bales, R. F. (Eds.). (1955). Small groups: Studies in social interaction. (1st ed.). New York, NY: Knopf. Hare, A. P., Borgatta, E. F., & Bales, R. F. (Eds.). (1965). Small groups: Studies in social interaction. (Revised ed.). New York, NY: Knopf. Harrington, B., & Fine, G. A. (2000). Opening the ‘‘black box’’: Small groups and twenty-first century sociology. Social Psychology Quarterly, 63, 312–323.

18

MORRIS ZELDITCH

Harrod, W. J., Welch, B. K., & Kushkowski, J. (2009). Thirty-one years of group research in Social Psychology Quarterly (1975–2005). Current Research in Social Psychology, 14, 75–103. Heckathorn, D. D., & Rosenstein, J. E. (2002). Group solidarity as the product of collective action: Creation of solidarity in a population of injection drug users. Advances in Group Processes, 19, 37–66. Heider, F. (1958). The psychology of interpersonal relations. New York, NY: Wiley. Homans, G. C. (1958). Social behavior as exchange. American Journal of Sociology, 63, 597–606. Homans, G. C. (1964). Bringing men back. American Sociological Review, 29, 809–818. Huber, J. (Ed.). (1991). Macro–micro linkages in sociology. Newbury Park: Sage. Knottnerus, J. D. (2010). Collective events, rituals, and emotions. Advances in Group Processes, 27, 39–61. Lawler, E. J. (1986). Bilateral deterrence and conflict spiral: A theoretical analysis. Advances in Group Processes, 3, 107–130. Lawler, E. J., Ridgeway, C., & Markovsky, B. (1993). Structural social psychology and the micro-macro problem. Sociological Theory, 11, 268–290. Lewin, K. (1947). Group decision and social change. In T. Newcomb & E. L. Hartley (Eds.), Readings in social psychology (pp. 330–344). New York, NY: Holt. Lewin, K. (1951). Field theory in social science. New York, NY: Harper & Row. Macy, M. W. (1993). Social learning and the structure of collective action. Advances in Group Processes, 10, 1–35. Manson, P. (1993). What is a group? A multi-level analysis. Advances in Group Processes, 10, 253–281. Markovsky, B., Dilks, L., Koch, P., McDonagh, S., Triplett, J., & Velasquez, L. (2008). Modularizing and integrating theories of justice. Advances in Group Processes, 25, 211–237. Markovsky, B., Willer, D., & Patton, T. (1988). Power relations in exchange networks. American Sociological Review, 53, 220–236. Mayhew, B. (1980). Structuralism and individualism: Part I, shadow boxing in the dark. Social Forces, 59, 335–375. McPherson, M., & Smith-Lovin, L. (2002). Cohesion and membership duration: Linking groups, relations, and individuals in an ecology of affiliation. Advances in Group Processes, 19, 1–36. Melamed, D. (2011). Graded status characteristics and expectation-states. Advances in Group Processes, 28, 1–31. Michener, H. A. (1992). Coalition anomalies in light of the central–union theory. Advances in Group Processes, 9, 65–88. Molm, L. (1996). Punishment and coercion in social exchange. Advances in Group Processes, 13, 149–188. Moreland, R. L., Hogg, M. A., & Hains, S. C. (1994). Back to the future: Social Psychological research on groups. Journal of Experimental Social Psychology, 30, 527–555. Mullins, N. (1973). Theories and theory groups in contemporary American sociology. New York, NY: Harper and Row. Ridgeway, C. (1984). Dominance, performance, and status in groups: A theoretical analysis. Advances in group processes, 1, 59–93. Roethlisberger, F. J., & Dickson, W. J. (1939). Management and the worker. Cambridge, MA: Harvard University Press.

Thirty Years of Advances in Group Processes

19

Sanna, L. J., & Parks, C. D. (1997). Group research trends in social and organizational psychology: Whatever happened to intragroup research? Psychological Science, 8, 261–267. Schachter, S. (1951). Deviation, rejection, and communication. Journal of Abnormal and Social Psychology, 46, 190–207. Serpe, R. T., & Stryker, S. (1993). Prior social ties and movement into new relationships. Advances in Group Processes, 10, 283–304. Shelley, R. K. (1993). How sentiments organize interaction. Advances in Group Processes, 10, 113–132. Shelley, R. K. (1998). Some developments in expectation-states theory: Graduated expectations. Advances in Group Processes, 15, 41–57. Steiner, I. D. (1974). Whatever happened to the group in social psychology? Journal of Experimental Social Psychology, 10, 94–108. Steiner, I. D. (1983). Whatever happened to the touted revival of the group? In H. H. Blumberg, A. P. Hare, V. Kent & M. Davies (Eds.), Small groups and social interaction (Vol. 2). Chichester, England: Wiley. Steiner, I. D. (1986). Paradigms and groups. Advances in Experimental Social Psychology, 19, 251–289. Stets, J., & Burke, P. J. (2005). New directions in identity control theory. Advances in group processes, 22, 43–64. Stouffer, S. et al. (Eds). (1949). The American soldier: Studies in social psychology in world war II. Princeton, NJ: Princeton University Press. Strodtbeck, F. L. (1951). Husband–wife interaction over revealed differences. American Sociological Review, 16, 468–473. Strodtbeck, F. L., & Mann, R. D. (1954). Sex role differentiation in jury deliberations. Sociometry, 19, 3–11. Thye, S., Yoon, J., & Lawler, E. J. (2002). The theory of relational cohesion: Review of a research program. Advances in Group Processes, 19, 139–166. Turner, J. H. (2000). A theory of embedded encounters. Advances in Group Processes, 17, 285–322. Tyler, T. R., Smith, H. J., & Ho, Y. J. (1996). Membership diversity and leadership effectiveness: Procedural justice, social identity, and group dynamics. Advances in Group Processes, 13, 33–66. Wagner, D. G. (1984). The labeling of mental illness as a status organizing process. Advances in Group Processes, 1, 127–144. Whyte, W. F. (1943). Street corner society: The social structure of an Italian slum. Chicago, IL: University of Chicago Press. Willer, D., & Skvoretz, J. (1997). Network connections and exchange ratios: Theory, predictions, and experimental tests. Advances in Group Processes, 14, 199–234. Willer, D., & Szmatka, J. (1993). Cross-national investigations of elementary theory: Implications for the generality of the theory and the autonomy of social structure. Advances in Group Processes, 10, 37–81. Willer, R. (2009). A status theory of collective action. Advances in Group Processes, 26, 133–163. Yamagishi, T. (1986). The structural, goal-expectation theory of cooperation in social dil emmas. Advances in Group Processes, 3, 51–87. Zelditch, M. (2007). Laboratory experiments in sociology. In M. Webster & J. Sell (Eds.), Laboratory experiments in the social sciences (pp. 517–531). New York, NY: Elsevier.

STILL COLOR-BLIND? THE TREATMENT OF RACE, ETHNICITY, INTERSECTIONALITY, AND SEXUALITY IN SOCIOLOGICAL SOCIAL PSYCHOLOGY$ Matthew O. Hunt, Pamela Braboy Jackson, Samuel H. Kye, Brian Powell and Lala Carr Steelman ABSTRACT Purpose – The chapter revisits and updates an earlier analysis to assess the extent to which sociological social psychology’s treatment of race has changed since 2000, and evaluates the degree to which issues of intersectionality and sexuality are engaged in social psychological scholarship. Design/methodology/approach – The chapter provides a content analysis of articles published between 2000 and 2012 in Social Psychology Quarterly, a leading journal in sociological social psychology, and of chapters published in two influential handbooks in social psychology. $

Note: Authors’ contributions are equal.

Advances in Group Processes: Thirtieth Anniversary Edition Advances in Group Processes, Volume 30, 21–45 Copyright r 2013 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 0882-6145/doi:10.1108/S0882-6145(2013)0000030005

21

22

MATTHEW O. HUNT ET AL.

Findings – It documents a notable increase in the percentage of articles in Social Psychology Quarterly in which race/ethnicity is referred to, included in the analysis, or seriously engaged. Patterns vary by methodological approach used in these articles. Social psychological attention to intersectionality and sexuality, as measured by the percentage of articles that broach these topics, is minimal. Research limitations/implications – This chapter restricts its analysis to the leading journal in sociological social psychology, but still demonstrates that there is potential for greater movement toward the incorporation of race/ethnicity, intersectionality, and sexuality in social psychological scholarship. Originality/value – It calls attention to core topics in sociology that would benefit from greater scholarly engagement by social psychologists. Keywords: Race/ethnicity; intersectionality; sexuality; gender; social psychology

INTRODUCTION More than a decade ago, we assessed social psychology’s treatment of race and ethnicity over the course of the final four decades of the 20th century (Hunt, Jackson, Powell, & Steelman, 2000).1 We found that social psychologists’ consideration of race increased over that time frame, but it did so in relatively trivial ways (e.g., by reporting the racial composition of a sample, rather than by actually using race in analyses). Indeed, as recently as the late 1990s, only about one in ten articles published in leading social psychology journals – such as Social Psychology Quarterly – analyzed race in what might reasonably considered a serious way (e.g., by employing race in analyses beyond its use as a control variable). These patterns prompted us to encourage social psychologists to more fully incorporate race and ethnicity into their research designs. For survey researchers, for example, this might mean greater utilization of large, publically available, omnibus data sources such as the General Social Surveys that include a sufficiently large sample of different racial/ethnic groups (Hunt, 2007). Similarly, for experimentalists this might mean including race as an experimental factor – much in the same way that

Treatment of Race, Ethnicity, Intersectionality, and Sexuality

23

gender has been studied in such work (Ridgeway, 2011; Ridgeway & SmithLovin, 1996). While we recognized the challenges to the broader incorporation of race and ethnicity in social psychological scholarship (e.g., especially in universities in which it would be difficult to find large enough minority pools for experiments), we also contended that the intellectual payoff was well worth it. In particular, we noted that greater attention to race and ethnicity would enable social psychologists to assess the extent to which general social psychological principles and processes are applicable across an increasingly racially and ethnically diverse population. In this chapter, we revisit and update our earlier analyses to assess the extent to which sociological social psychology’s treatment of race has changed over the ensuing years. Further, in the wake of growing sociological interest in issues of intersectionality (McCall, 2005) and sexuality (Loftus, 2001; Powell, Bolzendahl, Geist & Steelman, 2010), we expand our analyses to consider the extent to which (1) sociological social psychological scholarship that engages race does so in relation to issues of gender and social class, and (2) sexuality has become a focus of social psychological research. To accomplish these goals – and in line with our original work – we report patterns from a content analysis of articles from Social Psychology Quarterly, as well as of two important sourcebooks in sociological social psychology published over the past decade: Delamater’s Handbook of Social Psychology (2003) and Burke’s Contemporary Social Psychological Theories (2006).

BACKGROUND Sociologically oriented social psychologists commonly lament their parent discipline’s tendency to underappreciate the cognitive and interpersonal foundations of much of what sociologists study. As House (1981) documents, this tendency has deep roots, as sociologists have long ignored or downplayed the ‘‘social psychologies’’ embedded in the work of the classical social theorists – all of whom were concerned with basic questions regarding how dominant and/or emergent features of societies influenced the typical characteristics of persons.2 Alluding to the boundary work at play in concerns with establishing the legitimacy of a distinctly sociological approach to human social behavior and arrangements, House (1981) writes: Auguste Comte, Karl Marx, Max Weber, and Emile Durkheim were all centrally concerned with social psychology, especially problems of social structure and

24

MATTHEW O. HUNT ET AL. personality. This has not been sufficiently appreciated because of the tendencies of these writers, often accentuated by later interpreters, to stress the differences between their sociological approach and that of psychologists of their day. (p. 528, emphasis added)

Collins’ (1981) work on the ‘‘micro-foundations of macro-sociology,’’ Giddens’ (1984) structuration theory, and Sewell’s (1992) theorization of structure/agency dualism all represent attempts to overcome such tendencies by specifying relevant links between micro- and macro-level phenomena (or by dispensing with that distinction altogether). More recently, Ridgeway’s (2011) work on the social psychological bases of gender inequality represents an important contribution in this vein, and a notable exception to the failure of many sociological subfields to recognize and/or highlight the social psychological nature of their work.3 In contrast, Lareau’s (2003) widely lauded work on social class differences in cultural capital and parental investments in children is commonly identified as research on family, culture, and education, but rarely as a social psychologically informed scholarship, despite its intimate intellectual ties with the social structure and personality framework generally (McLeod & Lively, 2003), and with Kohn and associates’ (Kohn, 1969; Kohn & Schooler, 1983) research program on class, socialization, and values more specifically. Similarly, research on ‘‘doing gender’’ (West & Zimmerman, 1987) readily identifies as ‘‘cultural analysis,’’ while typically failing to acknowledge its close affinity with symbolic interactionism (Blumer, 1969; Stryker, 1980), and, in particular, with ethnomethodology (Garfinkel, 1967). Still further, social movements research utilizing resource mobilization and related ‘‘organizational’’ theories systematically under-theorizes social psychological (e.g., affective and identity-based) dimensions of movement participation and outcomes – despite the great potential utility in integrating insights from social movements and social psychological scholarship (Stryker, Owens, & White, 2000). These are just a few examples of sociology’s general failure to acknowledge and capitalize on social psychological scholarship. That being said, we also believe that social psychologists are complicit in some ways with this arrangement. To the extent that social psychologists fail to incorporate key concepts that have long held interest in sociological community, as well as concepts that are emerging as central to the discipline, we contribute – perhaps inadvertently – to the marginalization or isolation of social psychological work. Stryker (1980) made this general point several decades ago in his plea for social psychologists to take social structure – the principal subject matter of sociology – more seriously in their theory development and research. House (1977, 1981) and, more recently, McLeod and Lively (2003)

Treatment of Race, Ethnicity, Intersectionality, and Sexuality

25

make a similar point in contending that any truly multilevel social psychology must take seriously how various ‘‘components’’ of society (i.e., larger structural arrangements) shape more proximal social processes, and ultimately the person.

THE ORIGINAL STUDY Along these same lines, over a decade ago, we forwarded a critique of social psychology for its neglect of one key aspect of social structure (or ‘‘component’’ of society): race and ethnicity (Hunt et al., 2000). This study, which appeared in the millennial celebration edition of Social Psychology Quarterly had three primary objectives: (1) to make the case for increasing incorporation of race and ethnicity in social psychological scholarship; (2) to document the extent to which social psychology actually engaged issues of race and ethnicity between 1970 and 1999; and (3) to discuss strategies by which race might be given more attention – especially given the expanding availability of appropriate data sources and methodological advances in the field. As we noted in our original chapter (Hunt et al., 2000), race and ethnicity represent important bases of social identification and key axes around which groups form. That said, at the time of the publication of the article, race and ethnicity were also routinely ignored in social psychological research, a puzzling tendency in light of clear evidence that race matters for so many other sociological phenomena and the increasing attention social psychologists had given to gender – another ascribed, highly visible, and consequential status – over the last few decades of the 20th century (Cook, Fine, & House, 1995; Ridgeway & Smith-Lovin, 1996). In our original work, we were careful to note key exceptions that showcased how social psychological scholarship could further contribute to our understanding of race and to theory development. For example, research in the social structure and personality tradition has demonstrated the relevance of race for the self-concept (Hughes & Demo, 1989), stratification ideology (Kluegel & Smith, 1986; Tuch & Martin, 1997), and other important aspects of consciousness and/or mental functioning. However, at the time of our original article, other key social psychological topics such as distributive justice, expectation states theory, power, and leadership were largely ‘‘color-blind’’ in their neglect of race and ethnicity. While we considered various factors that might underlie race’s relative neglect,4 a fundamental reason – in our view – was an ‘‘assumption of

26

MATTHEW O. HUNT ET AL.

race/ethnic similarity’’: that is, the ‘‘generally untested proposition that basic social psychological processes and theories apply equally well to various racial and ethnic groups’’ (pp. 352–353). We cited a range of evidence challenging this assumption5 and concluded that, ‘‘this work not only documents racial differences in social psychological processes and the determinants of beliefs and attitudes, but also cautions that failure to attend to race may result in a social psychological knowledge that reflects a primarily ‘white’ experience of the world’’ (p. 354). To bring data to bear on the extent to which social psychologists had engaged issues of race and ethnicity, we analyzed the 954 articles published in Social Psychology Quarterly between 1970 and 1999, as well as the two comprehensive overviews of the field published in that time-period: Rosenberg and Turner’s Social Psychology: Sociological Perspectives (1981) and Cook, Fine, and House’s Sociological Perspectives on Social Psychology (1995). For the analysis of Social Psychology Quarterly, we employed a three-tiered measure to gauge each article’s consideration of race and ethnicity: (1) whether race was mentioned in the chapter (e.g., in the data and methods section, results, or discussion), regardless of the extent to which it was actually incorporated into the analysis (if at all); (2) whether race was used in the analysis, regardless of how centrally; (3) whether race was seriously considered as a pivotal focus in the analysis. Using this typology, we observed increases in the treatment of race over time in all three categories, although even in the 1995–1999 time period, the attention to race was, in our view, discouragingly low. Specifically, between the 1970–1974 and 1995–1999 time periods, the percentage of articles mentioning race increased from 8.0 to 41.3, the percentage using race in the analysis increased from 6.9 to 22.0, and the percentage giving serious consideration to race increased from 6.3 to 10.1.6 Moreover, our examination of the sourcebooks by Rosenberg and Turner (1981) and Cook et al. (1995) showed that neither devoted meaningful space to issues of race and ethnicity. To offer more of a vantage point to assess the percentages, we also analyzed shifts in Social Psychology Quarterly’s inclusion of gender. The analysis offered a stark comparison of the treatment of gender and race in social psychology, with the attention to gender being dramatically higher: in the 1995–1999 periods, over three-fourths of the articles mentioned gender (78.0 percent, an increase from 57.7 percent in the 1970–1974 period), over half included gender in the analysis (54.1 percent, an increase from 22.3 percent in the 1970–1974 period), and approximately two-fifths percent seriously considered gender (41.3 percent, an increase from 11.4 percent in the 1970–1974). Similarly, our examination of the sourcebooks by

Treatment of Race, Ethnicity, Intersectionality, and Sexuality

27

Rosenberg and Turner (1981) and Cook et al. (1995) showed strong upsurge in attention to gender between the 1981 and 1995 volumes. The increasing visibility in scholarly engagement of gender during this period certainly was welcome, but it was disappointing that consideration of race trailed so far behind. To assess the extent to which the relative inattention to race in Social Psychology Quarterly was a function of sociology as a discipline or the subfield of social psychology, we also examined changes in the treatment of race in two other specialty journals in sociology – the American Sociological Association (ASA)-sponsored Sociology of Education and Journal of Health and Social Behavior – and in a top specialty journal in psychological social psychology – the American Psychological Association-sponsored Journal of Personality and Social Psychology. The analysis of Sociology of Education and the Journal of Health and Social Behavior challenged the possibility that social psychologists’ inattention to race was endemic to the discipline of sociology (or of specialty journals). In fact, during the 1995–1999 period, in which only one-tenth (10.1 percent) of the articles in Social Psychology Quarterly seriously considered race, approximately one-third (31.1 percent) of the articles in the Journal of Health and Social Behavior and over one-half (55.1 percent) of the articles in Sociology of Education did so. In contrast, the patterns regarding the treatment of race in Social Psychology Quarterly more closely paralleled the patterns in the Journal of Personality and Social Psychology, in which slightly less than one-tenth (9.0%) of the articles seriously engaged race. These analyses offered persuasive evidence that when it comes to the question of race, sociological social psychologists are more distant from sociologists overall than they are from psychological social psychologists. Indeed, the theoretical paradigms and methodologies shared by sociological and psychology social psychologists may be a key factor in both groups’ relative inattention to race. We concluded our 2000 article with some thoughts regarding how social psychologists might increase their consideration of race. For example, regarding survey research, we cited the growing public availability of largescale data sets containing measures of interest to social psychologists. These include the National Educational Longitudinal Study (which contains a wealth of measures on expectations, aspirations, and self-concept), the National Survey of Families and Household (which boasts a wide range of behavioral and attitudinal measure), the Early Childhood Longitudinal Study: Kindergarten Class of 1998–1999 (which includes data from children, parents, and teachers on socialization, interaction, and attitudes), and the General Social Surveys (which contain a wide range of attitudinal items, as

28

MATTHEW O. HUNT ET AL.

well as experimentally based vignettes). We also pointed to several data sets that focused on the attitudes and experiences of racial and ethnic minorities – among them, the National Survey of Black Americans and the Chinese American Psychiatric Epidemiology Study. In addition, we also spotlighted technological advances in the form of computer simulation programs (Heise, 1997) and experimental software initiatives (Willer and Rutstrom, 1998, 1999) that could greatly facilitate the analysis of race in particular, as well as assessments of the external validity of basic social psychological principles more broadly (especially with the increased availability of such software on the internet).

GOING BEYOND RACE: INTERSECTIONALITY AND SEXUALITY A key goal of the current study, then, is to assess the extent to which the challenge we posed to social psychologists has been met in the ensuing years. During that same time-period, however, the discipline has evolved and other challenges and complexities have arisen. Among these are other issues of diversity – in particular, questions regarding intersectionality and sexuality. That is, scholars studying race increasingly appreciate the difficulty of disentangling race from other statuses, and – as a result – increasingly advocate the study of disadvantages (and privileges) as intersecting axes. For example, in highlighting the double-disadvantages faced by black women, Collins (1990) emphasized the importance of further interrogating the phenomenon of intersectionality – the combined, interactive effects of an individual’s (or group’s) identities and statuses (most typically gender, race, and class) that shape persons’ experiences. Although we did not address the topic of intersectionality in the original chapter, we do so in this chapter, as intersectionality has gathered momentum in studies of race (and gender) (Grollman, 2013; Harnois & Ifatunji, 2011; McCall, 2005). Indeed, the growth of interest in intersectionality has been impressive. Although the Race, Gender, and Class (RGC) section of the ASA was founded as recently as 1996 (in contrast to the Social Psychology section, which has been in existence for over 50 years), the section now ranks among the top five of the over fifty sections in number of members (by comparison, the Social Psychology section ranks fifteenth).7 The high percentage of student members in the RGC section – in fact, the highest of any section in the ASA (the Social Psychology section also has

Treatment of Race, Ethnicity, Intersectionality, and Sexuality

29

the distinction of ranking in the top five) – bodes well for the continued vitality of this section, interest in intersectionality, and future RGC scholarship. As attention to intersectionality (and gender) has intensified, so too has scholarly engagement in sexuality, as indicated by the increased visibility of research on this topic (Loftus, 2001; Pfeffer, 2010; Powell et al., 2010). In addition, as with the RGC section, the Sexualities section of the ASA has a relatively high percentage of student members. Given the myriad ways in which social psychology can inform scholarship on sexuality, this topic offers another important test-case for assessing the extent to which social psychological concepts – for example, self and identity, socialization, stigma, and status (Powell et al., 2010; Webster, Hysom, & Fullmer, 1998) – are being brought to bear on sexualities scholarship. As such, in addition to race and intersectionality, we also assess the incorporation of sexuality in recent articles in Social Psychology Quarterly (as discussed further below).

THE CURRENT STUDY Analytic Strategy Has the past inattention to race continued? Have social psychologists engaged issues of intersectionality and sexuality in their scholarship? The primary aim of the current study is to assess whether social psychologists have – knowingly or otherwise – heeded our call for greater consideration of race and ethnicity in their scholarship. To accomplish this task, we use the methodology from our prior study to examine the past 13 years of Social Psychology Quarterly (2000–2012). We selected this journal in the original study and in this study because of its distinction as the leading specialty journal in sociological social psychology and, accordingly, as a site of cutting-edge scholarship in the subfield. To assess social psychological contributions beyond those in Social Psychology Quarterly, we also examined the content of two important sourcebooks in sociological social psychology published in the past decade that reach a wide audience of social psychologists: Delamater’s Handbook of Social Psychology (2003), and Burke’s Contemporary Social Psychological Theories (2006). For comparative purposes, we also analyzed the treatment of gender in Social Psychology Quarterly for the 2000–2012 period and in the Delamater and Burke volumes. Finally, we considered the extent to which the topics of

30

MATTHEW O. HUNT ET AL.

intersectionality and sexuality were incorporated in Social Psychology Quarterly for the same time period and in the two sourcebooks.

Measures One of three coders read every article in Social Psychology Quarterly that was published between 2000 and 2012.8 Given the increase in the number of review pieces and other nonempirical comments published in this journal during this time period, the coders also analyzed these written pieces, along with Cooley-Mead Award addresses. This enables us to assess whether putatively broader written work in social psychology, which is less constrained by research protocol, has incorporated race/ethnicity – as well as intersectionality and sexuality – more readily than have research articles. We assessed the scholarly attention to race in each written piece in Social Psychology Quarterly in three ways. First, we examined whether the article made any reference to race/ethnicity, even if it was not used in any analysis. This category would include, for example, articles in which the author identified the racial composition of the sample but went no further than that in the analysis. Second, we considered whether the article included race and/or ethnicity in the analysis, even if it was not central to the analysis. This category would include, for example, articles in which race was used only as a control variable. Third, we noted whether the article seriously considered race/ethnicity in the analysis. Our interpretation of ‘‘seriously considered’’ is consistent with how we operationalized this measure in our previous study. We included an article in this category if it mentioned race in its title, used race as a central independent variable, analyzed a race-related dependent variable (e.g., racial attitudes), or conducted an analysis that tested, as its primary or secondary focus, the aforementioned ‘‘assumption of similarity’’ – that is, whether patterns are applicable across racial groups. Admittedly, this is a more expansive interpretation of ‘‘seriously considered’’ than some might prefer, which means that the estimates presented here may overestimate the extent to which race/ethnicity is genuinely engaged in social psychological scholarship. Also of note, the figures include analyses of international data, which more often than not emphasized ethnic patterns instead of racial patterns. One of two coders also read every chapter in the Delamater and Burke volumes in social psychology. To assess the coverage of race/ethnicity,

31

Treatment of Race, Ethnicity, Intersectionality, and Sexuality

intersectionality, and sexuality – as well as gender, as a comparison – in these volumes, coders first checked the index to identify references to each of these topics. Because indices may not be fully comprehensive, coders examined each chapter to see whether each topic was mentioned and, if so, how frequently.

Results Racial Patterns over Time Table 1 presents the percentage of empirical articles and research notes published in Social Psychology Quarterly between 2000 and 2012 that took race into account – as operationalized above. This table provides similar information regarding the treatment of sex/gender, intersectionality, and sexuality. From this table, we can see that approximately two-thirds (65.3 percent) of the articles referred to race and/or ethnicity, less than half of the articles (46.0 percent) included race in one way or another in the data analysis, and nearly one-fourth of the articles (24.2 percent) seriously engaged race in the analysis. These figures might seem low to some readers, but they do represent a notable increase in the incorporation of race over time. As demonstrated in Table 2, the percentage of articles that referred to race between 2000 and 2012 is over eight times as large as the percentage of articles that mentioned race (8.0 percent) between 1970 and 1974 – the very first time period studied in our previous analysis (Hunt et al., 2000). In our earlier study, we

Table 1. Number and Percentage of Empirical Articlesa in Social Psychology Quarterly That Engage Topic of Interest, 2000–2012 (N=248). Articles That Make Any Reference to Topic Race/Ethnicity Sex/Gender Intersectionality Sexuality a

162 219 17 13

(65.3%) (88.3%) (6.9%) (5.2%)

Articles That Include Topic in Analysis 114 162 7 8

(46.0%) (65.3%) (2.8%) (3.2%)

Articles that Seriously Considerb Topic 60 90 6 6

(24.2%) (36.3%) (2.4%) (2.4%)

These articles do not include theoretical articles, reviews, commentaries, and Cooley-Mead Award addresses. b See main text for operationalization of ‘‘seriously consider.’’

32

MATTHEW O. HUNT ET AL.

Table 2. Number and Percentage of Empirical Articles in Social Psychology Quarterly/Sociometry/Social Psychology That Mention Race/ Ethnicity, 1970–2012 (N=1202). Total Articles That Make Any Articles That Include Articles That Number of Reference to Race/ Race/Ethnicity in Seriously Consider Race/Ethnicityb Articlesa Ethnicity Analysis 1970–1974 1975–1979 1980–1984 1985–1989 1990–1994 1995–1999 2000–2012

175 206 176 163 125 109 248

14 26 27 36 35 45 162

(8.0%) (12.6%) (15.3%) (22.1%) (28.0%) (41.3%) (65.3%)

12 21 21 19 20 24 114

(6.9%) (10.2%) (10.2%) (11.7%) (16.0%) (22.0%) (46.0%)

11 17 15 14 12 11 60

(6.3%) (8.3%) (8.5%) (8.6%) (9.6%) (10.1%) (24.2%)

a

These articles do not include theoretical articles, reviews, commentaries, and Cooley-Mead Award addresses. b See main text for operationalization of ‘‘seriously consider.’’

documented an acceleration in the increase in the percentage of articles that mentioned race – with an average annual increase of less than one percentage point in the earlier time periods (e.g., from 1970 to 1984) and an average annual increase of nearly two percentage points in the later time periods (e.g., from 1990 to 1999). Since then, the increase has been unsteady. That is, there was a precipitous jump in the number of articles mentioning race in the early 2000s, after which the percentage of articles mentioning race plateaued – and, in fact, slightly decreased.9,10 Although this might suggest a ceiling effect in which race is mentioned, it is undeniable that there was a marked increase in the percentage of articles that refer to race from the early 1970s (8.0 percent) to the late 1990s (41.3 percent) to the 2000–2012 period (65.3 percent). The upswing in the percentage of articles that go beyond a mere mention of race is equally notable – perhaps more so. Of the articles published between 1970 and 1974, only a very small number (6.9 percent) included race in their analyses – even as a background variable. This figure slowly increased to 22.0 percent in the 1995–1999 period. Since then, however, this figure has more than doubled to 46.0 percent. The increase in the percentage of articles that seriously considered race in their analyses was even more glacial in the three decades covered in our previous study – from 6.3 percent in 1970–1974 to 10.1 percent in 1995–1999. This figure rose steeply – again more than doubling – to 24.2 percent in the 2000–2012 period.

33

Treatment of Race, Ethnicity, Intersectionality, and Sexuality

Gender Patterns over Time As a point of comparison, we also examined the scholarly attention to gender in Social Psychology articles published between 2000 and 2012. As shown in Table 1, the attention to gender – whether it is measured by references to gender (88.3 percent), inclusion of gender in analyses (65.3 percent), or seriously engagement of gender (36.3 percent) – continues to be greater than the consideration of race. That nearly nine out of ten articles now refer to gender – an increase of approximately 10 percentage points since the late 1990s, as shown in Table 3 – suggests that we are reaching or have reached a saturation point in which it is nearly obligatory to mention gender. In contrast, Table 3 offers evidence of a decrease since the late 1990s in the percentage of articles that seriously considered gender (from 41.3 percent to 36.3 percent).11 In other words, while gender continues to be more frequently discussed than race in articles, the difference has winnowed in the 2000s as a result of a larger increase in the representation of race than of gender. Methods and Theory in Studies of Race Some scholars have posited that the emphasis on empirical evidence in social psychological – and, more broadly, sociological – writings may be in part responsible for the inattention to less ‘‘researchable’’ topics such as race and ethnicity (Collins, 1998; Hunter, 2002). Extending this logic, one might expect that the ‘‘nonempirical’’ written work, by virtue of being unencumbered by constraints and challenged faced by empirical research (e.g., data limitations), will more frequently incorporate issues of race and ethnicity.

Table 3. Number and Percentage of Empirical Articles in Social Psychology Quarterly/Sociometry That Mention Gender/Sex/Sex Category, 1970–1974, 1995–1999, and 2000–2012 (N=532).

1970–1974 1995–1999 2000–2012 a

Total Number of Articlesa

Articles That Make Any Reference to Gender

Articles That Include Gender in Analysis

Articles That Seriously Consider Genderb

175 109 248

101 (57.7%) 85 (78.0%) 219 (88.3%)

39 (22.3%) 59 (54.1%) 162 (65.3%)

20 (11.4%) 45 (41.3%) 90 (36.3%)

These articles do not include theoretical articles, reviews, commentaries, and Cooley-Mead Award addresses. b See main text for operationalization of ‘‘seriously consider.’’

34

MATTHEW O. HUNT ET AL.

This expectation, however, is not supported by the evidence. As seen in Table 4, the percentage of nonempirical pieces – that is, theoretical articles, reviews, commentaries, and Cooley-Mead addresses – published in Social Psychology Quarterly between 2000 and 2012 that referred to race/ethnicity (47.9 percent), included race/ethnicity in their analyses (14.1 percent), or seriously considered race/ethnicity (9.7 percent) is significantly (po.01) lower than the aforementioned figures for empirical articles and research notes.12 In other words, if anything, a more empirical or research-oriented approach in sociological social psychology is linked to greater, not less, attention to race and ethnicity. This, however, does not necessarily mean that different research methodologies equally emphasize race. In our previous study, we documented that articles using survey methods were the most likely to consider race seriously (17.5 percent) and articles using experimental methods were the least likely to do so (2.8 percent). Interestingly, despite claims that qualitative scholarship is more sensitive to marginalized groups – for example, racial minorities, women, sexual minorities – our previous study found little support for this assertion: only 9.5 percent of the articles using qualitative methods took race into account in a serious way – a figure that was nearly identical to the percentage of all articles that seriously engaged race. As seen in Table 5, attention to race and ethnicity increased during the 2000–2012 period, regardless of methodology used – surveys, qualitative studies (interviews, ethnographies, and conversational analysis), and experiments. Still, articles that rely primarily on nonexperimental surveys continued to incorporate race most frequently – 79.7 percent refer to race,

Table 4. Number and Percentage of Nonempirical Articlesa in Social Psychology Quarterly That Engage Topic of Interest, 2000–2012 (N=71). Articles That Make Any Reference to Topic Race/Ethnicity Sex/Gender Intersectionality Sexuality a

34 36 11 13

(47.9%) (50.7%) (15.5%) (5.2%)

Articles That Include Topic in Analysis 10 13 1 1

(14.1%) (18.3%) (1.4%) (1.4%)

Articles that Seriously Consider Topicb 7 6 1 0

(9.7%) (8.5%) (1.4%) (0.0%)

These articles include theoretical articles, reviews, commentaries, and Cooley-Mead Award addresses and do not include empirical articles or research notes. b See main text for operationalization of ‘‘seriously consider.’’

35

Treatment of Race, Ethnicity, Intersectionality, and Sexuality

Table 5. Number and Percentage of Empirical Articlesa in Social Psychology Quarterly That Engage Topic of Interest by Methodology Used, 2000–2012 (N=248).

A. Race/ethnicity Surveys Interviews Ethnography/ Participant Observation Conversation Analysis/ Content Analysis Experimental B. Gender Surveys Interviews Ethnography/ Participant Observation Conversation Analysis/ Content Analysis Experimental C. Intersectionality Surveys Interviews Ethnography/ Participant Observation Conversation Analysis/ Content Analysis Experimental D. Sexuality Surveys Interviews Ethnography/ Participant Observation Conversation Analysis/ Content Analysis Experimental a

Total Number of Articles Using Specified Method

Articles That Make Any Reference to Topic

Articles That Include Topic in Analysis

Articles that Seriously Consider Topicb

138 18 14

110 (79.7%) 16 (88.9%) 10 (71.4%)

88 (63.8%) 9 (50.0%) 3 (21.4%)

43 (31.2%) 7 (38.9%) 2 (14.3%)

17

8 (47.1%)

5 (29.4%)

3 (17.6%)

61

18 (29.5%)

9 (14.8%)

5 (8.2%)

138 18 14

132 (95.7%) 18 (100%) 11 (78.6%)

118 (85.5%) 9 (50.0%) 5 (35.7%)

62 (44.9%) 5 (27.8%) 4 (28.6%)

17

8 (47.1%)

5 (29.4%)

3 (17.6%)

61

50 (82.0%)

25 (41.0%)

16 (26.2%)

138 18 14

9 (6.5%) 2 (11.1%) 0 (0.0%)

8 (5.8%) 1 (5.6%) 0 (0.0%)

5 (3.6%) 1 (5.6%) 0 (0.0%)

17

2 (11.8%)

1 (5.9%)

0 (0.0%)

61

0 (0.0%)

0 (0.0%)

0 (0.0%)

138 18 14

7 (5.1%) 4 (22.2%) 1 (7.1%)

4 (2.9%) 1 (5.6%) 0 (0.0%)

3 (2.2%) 1 (5.6%) 0 (0.0%)

17

3 (17.6%)

1 (5.9%)

1 (5.9%)

61

2 (3.3%)

1 (1.6%)

1 (1.6%)

Does not include theoretical articles, reviews, commentaries, and Cooley-Mead Award Addresses. b See main text for operationalization of ‘‘seriously consider.’’

36

MATTHEW O. HUNT ET AL.

63.8 percent include race in the analysis, and 31.2 percent use race as its focal point. Experimental work that addresses, studies, or engages race (29.5 percent, 14.8 percent, and 8.2 percent) continued to lag behind surveys, however – despite some concerted efforts to identify ways to more inclusively and aggressively study race in experimental scholarship (Sell & Goar, 2007).13,14 Intersectionality and Sexuality In an effort to expand our prior analysis beyond the issues of race and gender, we examined Social Psychology Quarterly articles’ treatment of intersectionality and sexuality between 2000 and 2012. Table 1 reports these results and shows that neither received much attention. Despite the intensifying interest in intersectionality in sociology – indicated by, as noted earlier, the large size and remarkable growth of the ASA section on RGC – only 6.9 percent of the empirical articles in Social Psychology Quarterly even mentioned the concept, while even more miniscule percentages either included or seriously engaged it (2.8 percent and 2.4 percent, respectively). Given the theory-generating possibilities that could arise from the question of intersectionality, it is telling that nonempirical pieces (Table 4) referred to intersectionality at a higher rate (15.5 each), but their inclusion or serious consideration of intersectionality was almost nonexistent (1.4 percent each).15 The figures regarding sexuality mostly echo those regarding intersectionality. Only one out of twenty research articles (5.2 percent) mentioned sexuality, approximately one out of thirty (3.2 percent) included sexuality in the analysis, and about one out of forty (2.4 percent) seriously engaged the topic. These numbers are stunningly low, although they still are slightly higher than the numbers among the nonempirical articles. Patterns from Influential Sourcebooks on Social Psychology It is possible that our emphasis on Social Psychology Quarterly, despite its status as the top specialty sociology journal in the area of social psychology, may misrepresent – by underestimating or perhaps overestimating – social psychologists’ engagement of race, gender, intersectionality, and sexuality. Certainly, social psychologists have written about these topics in different journals (e.g., Campbell & Troyer, 2007, 2011; Cheng & Powell, 2011; Goar & Sell, 2005; Loftus, 2001; Simpson, McGrimmon, & Irvin, 2007; Stepanikova, Triplett, & Simpson, 2011). To assess more broadly social psychology’s incorporation of these topics, we also explored the Handbook of Social Psychology (Delamater, 2003) and Contemporary Social Psychological Theories (Burke, 2006) – considered by many as the most

Treatment of Race, Ethnicity, Intersectionality, and Sexuality

37

comprehensive efforts in the 2000s to describe the many contributions of the subfield of social psychology. According to the Handbook of Social Psychology index (Delamater, 2003), race and ethnicity were referred to on 16 of the over 500 pages of text and in seven of the 21 chapters. A more detailed examination of the handbook, however, shows that race and ethnicity were mentioned in 62 pages. Five chapters dedicated sections or extensive examples regarding race (e.g., the chapter on expectations states theory [Correll & Ridgeway, 2003]), while two chapters called for future researchers to bring race and ethnicity closer to the foreground of social psychological scholarship (e.g., the chapter on social cognition [Howard & Renfrow, 2003]). In comparison, the question of gender was seriously engaged in nine chapters and was at least referred to in eleven additional chapters. In fact, only one chapter ignored gender completely. If the Handbook of Social Psychology (Delamater, 2003) effectively reflects the current state of social psychological knowledge or interest, then sexuality and especially intersectionality remain mostly outside the boundaries of contemporary social psychological scholarship. Sexuality was mentioned in seven pages and in seven chapters – more often than not, used briefly as an example of sexually marginalized groups (e.g., a collective identity, social movement, or identity). In contrast, intersectionality was neither seriously engaged nor even mentioned in the book. Race, gender, intersectionality, and sexuality were not referred to in the index of Contemporary Social Psychological Theories (Burke, 2006); however, with exception of intersectionality – which was overlooked in this volume – these topics were mentioned in selected chapters. That being said, attention to race was low, with fleeting references in six of the fifteen chapters. Gender also was mentioned in less than half of the chapters, but the chapters that included gender typically gave serious consideration to the topic – among these, Berger and Webster’s chapter on expectations, status, and behavior (2006), Stets’s chapter on identity theory (2006), Robinson and Smith-Lovin’s chapter on affect control theory (2006), and Ridgeway’s chapter on status construction theory (2006). These latter two chapters also have the distinction of being the only ones to seriously engage – or even make a reference to – the topic of sexuality. Overall, the analyses of these volumes – both of which offer remarkably comprehensive overviews of the field of sociological social psychology – discount the possibility that our examination of articles from Social Psychology Quarterly underestimates the extent to which race, intersectionality, and sexuality are addressed in social psychological scholarship.

38

MATTHEW O. HUNT ET AL.

CONCLUSION The primary goal of this chapter was to take stock of the social psychological treatment of the very important topic of race and ethnicity, as well as of intersectionality and sexuality – two additional topics that have garnered increased attention in the sociological community. Regarding race, we documented notable changes in the percentages of articles in Social Psychology Quarterly that referred to race and ethnicity, included race and ethnicity in their analyses, and seriously engaged race and ethnicity. These increases are sizable; for example, the percentage of articles that included race, or seriously considered race, in their analysis more than doubled between the late 1990s and the 2000–2012 period. These changes offer reason for optimism regarding the ways race/ethnicity is being considered in sociological social psychology. Some observers’ optimism may be tempered, however, by the fact that over half of all articles in Social Psychology Quarterly between 2000 and 2012 did not include race in their analysis, while over three fourths did not seriously engage the topic. These figures are even higher among articles relying on experimental approaches – a methodology that has made great advances in our understanding of key social psychological topics such as distributive justice, status processes, and power, and that could thus seemingly benefit from a more racially attuned analysis. Methodological challenges may account partially for this gap, though we also show that less empirically driven articles in Social Psychology Quarterly (e.g., theoretical articles, review essays, and Cooley-Mead addresses) – which should be less burdened by data constraints – devote even less attention to race and ethnicity than their empirically driven counterparts. As we noted in our original chapter, we do not claim to know the threshold at which race and ethnicity would receive its ‘‘fair’’ share of attention from the social psychological community. That said, it is clear that the treatment of race continues to lag noticeably behind that of gender in social psychology – even with the recent and anomalous reduction in the percentage of articles in Social Psychology Quarterly that seriously considered gender in their analysis. Moreover, compared to some other subfields in sociology – notably, sociology of education and medical sociology – social psychologists continue to pay less attention to race. As noteworthy as the increasing percentage of articles in Social Psychology Quarterly between 2000 and 2012 that incorporated race is, these percentages are still lower than the corresponding percentage of articles in Sociology of Education and the Journal of Health and Social Behavior from over a decade ago (i.e., 1995–1999).

Treatment of Race, Ethnicity, Intersectionality, and Sexuality

39

Further, and ironically, several recent articles in those journals that do deal centrally with race/ethnicity, also touch on or centrally engage social psychological concepts and theory (Cheng & Starks, 2002; Grollman, 2013). In addition, if Delamater’s Handbook of Social Psychology (2003) and Burke’s Contemporary Social Psychological Theories (2006) are reasonable barometers of the central issues addressed by social psychologists, race would appear to remain relatively peripheral to the central concerns of social psychologists. While the picture regarding the incorporation of race and ethnicity in social psychology is mixed, there is little ambiguity regarding the treatment of intersectionality and sexuality, at least as measured by their presence (or absence) in articles in Social Psychology Quarterly and in Delamater’s Handbook of Social Psychology (2003) and Burke’s Contemporary Social Psychological Theories (2006). As in the case of race, these topics – which have enjoyed increasing prominence in sociology as a discipline, and which many scholars consider integral to the future of sociology – have largely been ignored (and/or deemed unimportant) in recent social psychological scholarship. One potential danger of this path is the continued lack of disciplinary recognition of social psychology, which could further fortify what some have described as a wall between sociological social psychologists and other scholars (Steelman & Powell, 1996). Fortunately, there is reason for optimism on this front, as young scholars in particular are combining the best of methodological and theoretical advances in social psychology to study intersectionality and sexuality. Grollman (2013), for example, integrates insights from stress theory (Pearlin, 1989; Thoits, 2010) and intersectionality to consider the socioemotional consequences of multiple forms of discrimination. Similarly, Harnois and Ifatunji (2011) reassess attitudinal and perceptional indicators of discrimination from an intersectional framework. And, regarding sexuality, Doan, Loehr, and Miller (2012) resourcefully use experimental survey methodology to explore whether Americans differentiate between formal rights and informal privileges for lesbian and gay couples, and to assess the relative utility of group position theory and social identity theory. They do so by capitalizing on recently available Time-sharing Experiments for the Social Sciences (TESS; http://www.tessexperiments.org) – an initiative that enables scholars to collect experimental data from a nationally representative samples, as well as from oversamples of groups of interest (in Doan and associates’ case, lesbians and gay men). In our view, TESS offers an unprecedented opportunity for social psychologists interested in assessing the applicability of insights gained from an experimental lab setting to

40

MATTHEW O. HUNT ET AL.

the population-at-large, and to key societal subgroups of interest (e.g., racial groups, sexuality minorities). That said, laboratory experiments also have the potential to make great strides in the incorporation of sexuality (as well as intersectionality and race) in social psychological thought. Indeed, it is especially promising that the first recipient of the ASA Social Psychology Section’s inaugural graduate student investigator award is using a clever lab experimental design to parse out the effects of sexual orientation and gender on assessments of status and leadership (Mize, 2013). These advances among young scholars offer a glimpse into a possible future of the social psychological examination of intersectionality and sexuality – a future that could parallel the exciting developments and gains in social psychological insights into gender that occurred over the past few decades. At the same time, we also see possibilities for continued growth in the incorporation of race and ethnicity in social psychology. Toward that end, both up-and-coming social psychologists, and their more established counterparts, have been collaborating to identify challenges faced in the study of race and ethnicity, and to seek solutions to these problems. For example, a National Science Foundation/ASA-sponsored workshop of small groups researchers held at Texas A&M thoughtfully and productively made efforts: (1) to identify perceived incentives and barriers to studying aspects of race/ethnicity experimentally; (2) to map out a set of topics, recognized as important by both micro and macro sociologists, that can be investigated experimentally; and (3) to foster collaborations between established experimenters and new experimenters to assist in developing experimental designs for experimental studies of topics in race and ethnicity. (Sell & Goar, 2007)

The conference was fairly recent, so the fruits of the collaborative efforts have yet to be fully realized. Still, this workshop appears to have increased enthusiasm for research engaging race/ethnicity; if so, the results will likely pay intellectual dividends for social psychologists, scholars of race, and sociologists more generally in the years to come. Most readers of this volume likely do not need to be persuaded of the importance of social psychology. The same cannot necessarily be said for the sociological community at large. In our view, social psychologists too often cede important and relevant intellectual ground to other sociologists who, in turn, gloss over the contributions of social psychology. As a corrective, social psychologists should seek to highlight the places where our specialty area is particularly relevant to, and implicated in, the assumptions

Treatment of Race, Ethnicity, Intersectionality, and Sexuality

41

and theorizing of scholarship that is purportedly not social psychological. That is, we should more fully engage areas of expanding scholarly interest, such as race, intersectionality and sexual orientation without abandoning the principles and strengths of the social psychological approach to the study sociological phenomena. Failure to do so runs the risk of appearing disinterested in the most noteworthy disciplinary – and societal – changes of our times. By closely scrutinizing social psychologists’ consideration of race, intersectionality, and sexuality, we have attempted to contribute to an important conversation about the overlooked importance of social psychology: where social psychologists with their skills and perspectives can make needed contributions, and in so doing enhance the visibility, relevancy, and respect that social psychology so rightfully merits.

NOTES 1. We use the terms race, race/ethnicity, and race and ethnicity interchangeably in this chapter, while recognizing that race and ethnicity can be viewed as distinct or overlapping bases of identification (Cornell & Hartman, 1998). 2. For instance, Marx links the social and economic organization of capitalist societies to workers’ alienation and class consciousness. Weber was similarly concerned with specifying links between dominant modes of social organization and individuals’ subjectivities (beliefs, values), demonstrating, for instance, how bureaucratization leads to depersonalization, as well as how religious ideas and orientations represent a causal force in history. And, Durkheim’s notion of anomie relates the looseness of persons’ ties to religious institutions to maladaptive individual states of mind (e.g., lack of internal moral norms, etc.). 3. As an attempt to display the centrality of social psychology in sociological thought, many members of the Social Psychology section of the American Sociological Association (ASA) donned pins reading ‘‘Social Psychology: It’s Actually Everywhere’’ at the 2012 ASA meetings in Denver. 4. These include: race’s political sensitivity as a topic (a point we generally rejected, given that gender has also, historically, represented ideologically contested terrain), the demography of the field of social psychology (i.e., women saw much greater gains in the academy than did race/ethnic minorities over the latter part of the 20th century), and methodological challenges faced in studying race (e.g., experiments’ heavy reliance on college undergraduate samples, that make it easier to obtain a large numbers of male and female study participants and much more difficult to obtain a sufficient large number of participants for different racial/ethnic groups). 5. Examples include: studies of race/ethnic differences in the determinants of various beliefs and attitudes (Hunt, 1996; Steelman & Powell, 1993; Schnittker, Freese, & Powell, 2000), as well research from subfields such as medical sociology showing, for instance, that some commonly held views about role-related effects on mental health are less applicable to various race/ethnic minority populations than to non-Hispanic whites (Jackson, 1997).

42

MATTHEW O. HUNT ET AL.

6. Most of the increase in ‘‘mentions’’ of race stemmed from increases in the tendency of authors to identify the racial composition of the sample (or noting that the sample is limited to whites). The modest increase in the ‘‘used’’ category was generally attributable to employing race as a control variable (without more serious consideration). Most striking to us was the very modest (four percentage point) increase in the ‘‘seriously considers’’ category. 7. Information regarding section membership was provided by the ASA (personal communication, February 19, 2013). 8. A check for inter-coder reliability showed a very high degree of agreement – over 90% – in coding. In cases of disagreement, coding decisions were determined by a third coder. 9. These analyses, and other figures not displayed in the tables, are available upon request from the authors. 10. The 2000–2004 period was different from other time periods because it included three special editions – most notably, a special issue on ‘‘Race, Racism, and Discrimination,’’ in which, as would expected, every article referred to, analyzed, and seriously considered race. Supplementary analyses confirm that the slight decline in subsequent years is attributable to the prominence of race in the special issues. 11. Closer examination, however, shows that the real decrease – indeed a precipitous drop – in the percentage of articles that seriously considered gender occurred in the second half of the 2000–2012 period. In fact, of all time periods studied in the current and previous study, the early 2000s represents the period in which gender was the most seriously engaged. 12. Table 4 also confirms that the nonempirical articles also placed much less emphasis on gender than did the empirical ones. For example, gender was seriously engaged in only 8.5 percent of the nonempirical articles (vs. 36.3 percent of the empirical articles). 13. Compared to previous years, the 2000–2012 period can also be characterized by a shift in methodological approaches employed. Indeed, there was a notable increase in the number of Social Psychology Quarterly articles that used surveys (from approximately 40 percent in 1995–1999 to 55 percent in 2000–2012) and a corresponding decrease in articles that used experiments (from 37 percent to 25 percent). It is unclear whether this is a sign of changing submission patterns, changing preferences in the reviewing process, or the changing nature of sociological social psychology. In a supplementary analysis that decomposed the overall fluctuation in the treatment of race over time, we found that a small portion of this change can be attributed to the changing distribution of methodological approaches in Social Psychology Quarterly. 14. Articles using survey designs also are more likely than other articles to refer to, analyze, and seriously engage gender, although the differences are less stark. For example, 44.9 percent of survey-based articles seriously consider gender, in contrast to 24.5 percent of qualitative articles and 26.2 percent of experimental articles. 15. Given the very small numbers involved, one should be cautious in extrapolating too much about the role of methodological approach in the likelihood of intersectionality being mentioned. That said, 6.5 percent of survey-based articles, 8.2 percent of qualitative articles (using interviews, ethnography, or conversational analysis), and 0 percent for experimental articles even referred to intersectionality.

Treatment of Race, Ethnicity, Intersectionality, and Sexuality

43

ACKNOWLEDGMENTS This study was supported by the Center for Research on Race and Ethnicity in Society at Indiana University. We thank Elizabeth Zack for her assistance in collecting the data analyzed in this chapter.

REFERENCES Berger, J., & Webster, M., Jr. (2006). Expectations, status, and behavior. In P. Burke (Ed.), Contemporary sociological theories (pp. 268–301). Stanford, CA: Stanford University Press. Blumer, H. (1969). Symbolic interactionism: perspective and method. Berkeley, CA: University of California Press. Burke, P. (Ed.). (2006). Contemporary social psychological theories. Stanford, CA: Stanford University Press. Campbell, M. E., & Troyer, L. (2007). The implications of racial misclassification by observers. American Sociological Review, 72(5), 750–765. Campbell, M. E., & Troyer, L. (2011). Further data on misclassification: A reply to Cheng and Powell. American Sociological Review, 76(2), 356–364. Cheng, S., & Powell, B. (2011). Misclassification by whom? A comment on Campbell and Troyer (2007). American Sociological Review, 76(2), 347–355. Cheng, S., & Starks, B. (2002). Racial differences in the effects of significant others on students’ educational expectations. Sociology of Education, 75(4), 306–327. Collins, P. H. (1990). Black feminist thought: Knowledge, consciousness and the politics of empowerment. New York, NY: Routledge. Collins, P. H. (1998). On fighting words: Black women and the search for justice. Minneapolis, MN: University of Minnesota Press. Collins, R. (1981). On the microfoundations of macrosociology. American Journal of Sociology, 86(5), 984–1014. Cook, K. S., Fine, G. A., & House, J. S. (Eds.). (1995). Sociological perspectives on social psychology. Boston, MA: Allyn and Bacon. Cornell, S., & Hartman, D. (1998). Ethnicity and race: Making identities in a changing world. Thousand Oaks, CA: Pine Forge. Correll, S. J., & Ridgeway, C. L. (2003). Expectation states theory. In J. Delamater (Ed.), Handbook of social psychology (pp. 29–52). New York, NY: Kluwer Academic/Plenum Publishers. Delamater, J. (Ed.). (2003). Handbook of social psychology. New York, NY: Kluwer Academic/ Plenum Publishers. Doan, L., Loehr, A., & Miller, L. (2012). Attitudes toward formal rights and informal privileges for gay and lesbian couples: Evidence from a National Experimental Survey. Presented at the 62nd Annual Meeting of the Society for the Study of Social Problems. Denver, CO. Garfinkel, H. (1967). Studies in ethnomethodology. Englewood Cliffs, NJ: Prentice Hall. Giddens, A. (1984). The constitution of society: Outline of the theory of structuration. Berkeley, CA: University of California Press.

44

MATTHEW O. HUNT ET AL.

Goar, C., & Sell, J. (2005). Using task definition to modify racial inequality within task groups. Sociological Quarterly, 46(3), 525–543. Grollman, E. A. (2013). Multiple forms of perceived discrimination and health among adolescents and young adults. Journal of Health and Social Behavior, 53(2), 199–214. Harnois, C. E., & Ifatunji, M. A. (2011). Gendered measures, gendered models: Toward an intersectional analysis of interpersonal racial discrimination. Ethnic and Racial Studies, 34(6), 1005–1028. Heise, D. (1997). Interact. Retrieved from http://www.indiana.edu/Bsocpsy/ACT/interact.htm. Accessed on May 5, 2013. House, J. (1977). The three faces of social psychology. Sociometry, 40(2), 61–177. House, J. (1981). Social structure and personality. In M. Rosenberg & R. Turner (Eds.), Social psychology: Sociological perspectives. New York, NY: Basic Books. Howard, J. A., & Renfrow, D. G. (2003). Social cognition. In J. Delamater (Ed.), Handbook of social psychology (pp. 259–282). New York, NY: Kluwer Academic/Plenum Publishers. Hughes, M., & Demo, D. H. (1989). Self-perceptions of Black Americans: Self-esteem and personal efficacy. American Journal of Sociology, 95(1), 139–159. Hunt, M. O. (1996). The individual, society, or both? A comparison of Black, Latino, and White beliefs about the causes of poverty. Social Forces, 75(1), 293–332. Hunt, M. O. (2007). African American, Hispanic, and white beliefs about black/white inequality, 1977–2007. American Sociological Review, 72(3), 390–415. Hunt, M. O., Jackson, P. B., Powell, B., & Steelman, L. C. (2000). Color-blind: The treatment of race and ethnicity in social psychology. Social Psychology Quarterly, 63(4), 352–364. Hunter, M. (2002). Rethinking epistemology, methodology, and racism: Or is White sociology dead? Race & Society, 5(2), 119–138. Jackson, P. B. (1997). Role occupancy and minority mental health. Journal of Health and Social Behavior, 38(3), 237–255. Kluegel, J. R., & Smith, E. R. (1986). Beliefs about inequality. New York, NY: Aldine De Gruyter. Kohn, M. L. (1969). Class and conformity: A study in values (2nd ed.). Chicago, IL: University of Chicago Press. Kohn, M. L., & Schooler, C. (Eds.). (1983). Work and personality. Norwood, NJ: Ablex. Lareau, A. (2003). Unequal childhoods: Class, race, and family life. Berkeley, CA: University of California Press. Loftus, J. (2001). America’s liberalization in attitudes toward homosexuality: 1973–1998. American Sociological Review, 66(5), 762–782. McCall, L. (2005). The complexity of intersectionality. Signs: Journal of Women in Culture and Society, 30(3), 1771–1800. McLeod, J. D., & Lively, K. J. (2003). Social structure and personality. In J. Delamater (Ed.), Handbook of social psychology (pp. 77–102). New York, NY: Kluwer Academic/Plenum Publishers. Mize, T. (2013). Status disadvantages for gay men and lesbian leaders: Orientation or masculinity? American Sociological Association, Social Psychology Section Young Investigator Grant. Washington, DC. Pearlin, L. I. (1989). The sociological study of stress. Journal of Health and Social Behavior, 30(3), 241–256. Pfeffer, C. A. (2010). ‘‘Women’s work’’? Women partners of transgender men doing housework and emotion work. Journal of Marriage and Family, 72(1), 165–183.

Treatment of Race, Ethnicity, Intersectionality, and Sexuality

45

Powell, B., Bolzendahl, C., Geist, C., & Steelman, L. C. (2010). Counted out: Same-sex relations and Americans’ definitions of family. New York, NY: Russell Sage. Ridgeway, C. L. (2006). Status construction theory. In P. Burke (Ed.), Contemporary sociological theories (pp. 301–323). Stanford, CA: Stanford University Press. Ridgeway, C. L. (2011). Framed by gender: How gender inequality persists in the modern world. New York, NY: Oxford University Press. Ridgeway, C. L., & Smith-Lovin, L. (1996). Gender and social interaction. Social Psychology Quarterly, 59(3), 173–175. Robinson, D. T., & Smith-Lovin, L. (2006). Affect control theory. In P. Burke (Ed.), Contemporary sociological theories (pp. 137–164). Stanford, CA: Stanford University Press. Rosenberg, M., & Turner, R. H. (Eds.). (1981). Social psychology: Sociological perspectives. New York, NY: Basic Books. Schnittker, J., Freese, J., & Powell, B. (2000). Nature, nurture, neither, nor: Black-White differences in beliefs about the causes and appropriate treatment of mental illness. Social Forces, 78(3), 1101–1132. Sell, J., & Goar, C. (2007). Expanding experimental investigations of race in sociology. American Sociological Association’s Fund for the Advancement of the Discipline (ASA/NSF). Sewell, W. F. (1992). A theory of structure: Duality, agency, and transformation. American Journal of Sociology, 98(1), 1–29. Simpson, B., McGrimmon, T., & Irvin, K. (2007). Are Blacks really less trusting than Whites? Revisiting the race and trust question. Social Forces, 86(2), 525–552. Steelman, L. C., & Powell, B. (1993). Doing the right thing: Race and parental locus of responsibility for funding college. Sociology of Education, 66(4), 223–244. Steelman, L. C., & Powell, B. (1996). The family devalued: The treatment of the family in small groups literature. In B. Markvosky, M. J. Lovaglia & R. Simon (Eds.), Advances in groups processes (Vol. 13, pp. 213–238). Greenwich, CT: JAI Press. Stepanikova, I., Triplett, J., & Simpson, B. (2011). Implicit racial bias and prosocial behavior. Social Science Research, 40(4), 1186–1195. Stets, J. (2006). Identity theory. In P. Burke (Ed.), Contemporary sociological theories (pp. 88–110). Stanford, CA: Stanford University Press. Stryker, S. (1980). Symbolic interactionism: A social-structural version. Menlo Park, CA: Benjamin/Cummings. Stryker, S., Owens, T. J., & White, R. W. (Eds.). (2000). Self, identity, and social movements. Minneapolis, MN: University of Minnesota Press. Thoits, P. (2010). Stress and health: Major findings and policy implications. Journal of Health and Social Behavior, 51(1), S41–53. Tuch, S. A., & Martin, J. K. (Eds.). (1997). Racial attitudes in the 1990s: Continuity and change. Westport, CT: Praeger. Webster, M., Jr., Hysom, S. J., & Fullmer, E. (1998). Sexual orientation and occupation as status. In J. Skvoretz & J. Szmatka (Eds.), Advances in group processes (Vol. 15, pp. 1–21). New York, NY: JAI Press. West, C., & Zimmerman, D. H. (1987). Doing gender. Gender & Society, 1(2), 125–151. Willer, D., & Rutstrom, L. (1998). Ex Net III: A multifunctional instrument for social science experimentation. National Science Foundation Grant. Willer, D., & Rutstrom, L. (1999). A software and data library for experiments, simulations, and archiving. National Science Foundation Grant.

RACE AND ETHNIC COMPOSITION OF GROUPS: EXPERIMENTAL INVESTIGATIONS$ Carla Goar, Jane Sell, Bianca Manago, Calixto Melero and Bobbi Reidinger ABSTRACT Purpose – This chapter considers whether and how race/ethnicity can be examined using experimental methods. Design/methodology/approach – We begin by discussing the highly contextual nature of race/ethnicity and reviewing the properties of experiments. After examining existing experimental literature that focuses on race/ethnicity, we turn to our current study that uses the incompatible complexity condition to examine the multilevel interactions of diverse racial/ethnic groups composed of Mexican American and White participants in Texas and Black and White participants in Ohio. Findings – We argue that experiments, when guided by formal theoretical approaches that allow for general inquiries of theoretical principles, are especially suitable for studying interactional characteristics such as race/ ethnicity. $

The first two authors are equal coauthors.

Advances in Group Processes: Thirtieth Anniversary Edition Advances in Group Processes, Volume 30, 47–75 Copyright r 2013 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 0882-6145/doi:10.1108/S0882-6145(2013)0000030006

47

48

CARLA GOAR ET AL.

Originality/value – We suggest a particular approach that emphasizes interactional aspects of race/ethnicity and how these aspects can be utilized to diminish inequality in group processes. Keywords: Race/ethnicity; inequality; expectation states; stereotyping

INTRODUCTION In this chapter we consider the issue of race and how group processes are affected by race. No social scientist would claim that the study of race is unimportant; yet there is relatively little research on race and group processes and even less within the experimental paradigm (Goar & Sell, 2009). Why is this the case? The issue is complex, most directly because what race is and what it means is complex. We discuss the meaning of race, how this meaning affects how race can be conceptualized in group processes, and how such conceptualizations affect methodologies. In particular, we will argue, contrary to some critiques, that experimental methodology can enable penetrating analysis of race.

The Problem: What is Race? Discussion, controversy, and political issues surround fundamental understandings of what constitutes race and ethnicity. To address defining race/ ethnicity1, we focus upon the role of theory and the purpose of particular kinds of investigations. Biological-based definitions of race were historically used for domination, both socially and ‘‘scientifically.’’ This history has not been erased and continues to affect how we think about research questions related to race/ethnicity. Perhaps understandably, anthropologists sometimes chastise sociologists and other social scientists for an unreflective use of race. In the statement of the American Anthropological Association in 1998, it was asserted: ‘‘In the United States both scholars and the general public have been conditioned to viewing human races as natural and separate divisions within the human species based on visible physical differences. With the vast expansion of scientific knowledge in this century, however, it has become clear that human populations are not unambiguous, clearly demarcated, biologically

Race and Ethnic Composition of Groups

49

distinct groups.’’ The statement further articulated that the very concept of race was and is an instrument for domination. The domination becomes supported by racial beliefs, which constitute myths about diversity. ‘‘The myths fused behavior and physical features together in the public mind, impeding our comprehension of both biological variations and cultural behavior, implying that both are genetically determined. Racial myths bear no relationship to the reality of human capabilities or behavior. Scientists today find that reliance on such folk beliefs about human differences in research has led to countless errors.’’ For its part, the American Sociological Association (ASA) in 2003 asserted the importance of collecting information on race, while acknowledging that what race means relates predominantly to its social construction. In part this pronouncement was a reaction to discussions both from scholars in fields such as physical anthropology and political actors that race is not a biological concept and that its use seems to solidify differences. In its response, ASA acknowledged the potential dangers of presentation of race as somehow biologically based, but maintained that ‘‘However, a large body of social science research documents the role and consequences of race in primary social institutions and environments, including the criminal justice, education and health systems, job markets, and where people live. These studies illustrate how racial hierarchies are embedded in daily lifey.’’ In particular, ASA argued that collecting information about race allowed assessment of issues related to social justice. In a nuanced examination of the issue of race, especially as it relates to medical studies, Gravlee (2009, p. 54) asserts that critiques of the use of race need to be further refined to address three considerations: that research in genetics do not justify any return to racial thinking, that there is, however, a biocultural conceptualization of race (and especially race and health) that should be further advanced, and finally that the very concept of race should be more carefully assessed as it may mean different things in different contexts. We begin with Gravelee’s point. His illustration of his own research highlights the importance of the context as defining what race means. Using both an ethnography and a survey in Puerto Rico, he found how the local, cultural definitions of color form a coherent, cultural definition of color and hierarchy. He then tested how blood pressure varied by the local definition of color versus color as measured by reflectometry. His findings indicate that socially ascribed color, but not measured skin pigmentation, was associated with blood pressure (Gravlee & Dressler, 2005; Gravlee, Dressler, & Bernard, 2005, as cited in Gravlee, 2009).

50

CARLA GOAR ET AL.

Context is critical to defining the meaning of race. To study group processes and race is to fundamentally commit to an idea of race as interactional, not a characteristic that is an individual-level, stable characteristic. Of course for some questions, race/ethnicity might be such a stable, categorization characteristic. But these studies are of issues such as segregation or voting behavior, etc., which can be conducted based on examining how people categorized in a certain way at a certain point in time. (Historical studies of the census, indeed, clearly illustrate this. The classification of individuals has changed over time based upon the governmental classifications and uses of the categories.) However, to analyze group process, we are interested in examining how race/ethnicity is read by self and others and then how initial readings develop differently over time. As such, it is an issue of how race is viewed as identity. In our work, this conception of race and ethnicity is key to understanding what questions can be asked and what kinds of questions can’t be asked within group process context. For example, the question ‘‘Are Latinos less influential than Anglos in groups?’’ might be a descriptive question. But it is certainly not a group process question. From a group process question, we would ask about group composition, the task, and the degree of prior history within the group. So, for example, we would get a different response if we ask, ‘‘Are Latinos less influential than Anglos in groups in which Latinos are the majority both within the group and within the institutional context in which the group operates?’’ than the question, ‘‘Are Latinos less influential than Anglos in groups in which Latinos are in the minority both within the group and within the institutional context in which the group operates?’’ Further, these group process questions can be assessed within an experimental context in a way that can develop theories about how race/ethnicity organizes interaction and subsequent theories about the ways negative effects of race/ethnicity can be dismantled.

THE EXPERIMENTAL METHOD AND ITS PURPOSE In sociological research methods textbooks, the experimental method is almost always presented first. Often, it is explained that experiments are able to show causality because they enable random assignment and/or controlled conditions, the establishment of treatments, and then the comparison of conditions to assess whether or not change occurs. Because of the

Race and Ethnic Composition of Groups

51

experiment’s ability to rule out alternative factors that results in outcomes, they are said to have high internal validity, or the ability to demonstrate casual connections. So, in the matter of internal validity, experiments are often characterized as ‘‘ideal.’’ However, there is the well-known ‘‘but’’ that is usually added to this ideal method: ‘‘The greatest weakness of laboratory experiments lies in their artificiality’’ (Babbie, 1975, p. 254). This artificiality weakness is often cited and cited within the context that such artificiality creates problems in external validity. And what is external validity? Campbell and Stanley (1967) in their oftencited book on research design define external validity in terms of generalizability. ‘‘To what populations, settings, treatment variable, and measurement variables can their effect be generalized’’ (p. 5)? As Mook (1983) points out, the problem enters when Campbell and Stanley continue by saying that ‘‘the selection of designs strong in both types of validity is obviously our ideal’’ (Campbell & Stanley, 1967, p. 5). Mook as well as others argue that this view of the ideal misrepresents the goal of different kinds of research methods. The issue basically distills into the population to which one is generalizing. It is possible that researchers may want to generalize to characteristics of an actual population defined within a particular time and space, for example, the attitudes of Black Americans toward immigration issues in 2013. Such generalization is usually termed ‘‘statistical generalizations,’’ and as discussed by Meeker and Leik, concerns whether ‘‘we can use these observations to describe the existing characteristics of people or situations that were not observed’’ (1995, p. 639). But, this is only one kind of generalization. As Lucas (2003) points out, generalization is always relative to the questions being asked. One cannot make a decision about whether the method itself allows generalizability; rather one must assess generalizability relative to the particular question being asked and the method applied. If research seeks to elaborate the processes that occur at a particular point in time, it is seeking to generalize from a sample to a larger, time-defined population. Martin and Sell (1979) describe such research as asking descriptive theoretical questions. They are descriptive because they ask about the state of things at a particular point in time and space. However, not all populations are defined at particular points in time. When we ask questions such as ‘‘Under what conditions is stereotyping decreased?’’ or ‘‘What kinds of learning techniques are most efficient for

52

CARLA GOAR ET AL.

increasing creativity?’’ we are not asking a question about a defined time and place. Instead, such questions seek to test general, theoretical principles that are meant to apply, not only to a specific point in time, but to many different places and points in time. These theories are termed ‘‘formal theoretical’’ by Martin and Sell (1979) because they accept the formal requirements of deduction in the developments. Abiding by the requirements or stipulations of deduction to some extent determines the structure of resulting theories.

Artificiality and Experimental Research The universal form of deductive arguments and the specification of scope conditions of formal theoretical statements mean that such theories themselves are artificial. (For discussion of scope conditions see Berger, Cohen, & Zelditch, 1972; Berger & Webster, 2006; Cohen, 1980, 1993; Foschi, 1997; Freese, 1980; Henshel, 1980a; Markovsky, 1994; Walker & Cohen, 1985; Webster & Sell, 2007; Webster & Kervin, 1971). They cannot reflect all the complexities of the world or the situations within them. Theories are meant to express the most theoretically meaningful variables, not all the variables of a specific context. It is in this sense that theories are artificial; it is in this sense that experiments are advantageous in theorytesting precisely because they are artificial. Formal theories are composed of deductive statements with specified scope conditions; the populations of interest are not defined by a particular time and place. Because this is the case, experiments are particularly advantageous for the testing of formal theories because they allow the creation or instantiation of these theoretically specified relationships and the control of other conditions. Experiments are meant to create artificial worlds to test artificial theories (Cook & Campbell, 1979; Freese, 1980; Henshel, 1980b; Lucas, 2003; Martin & Sell, 1979; Sell & Martin, 1983; Webster & Sell, 2007; Willer, 1987; Willer & Walker, 2007; Zelditch, 1969). For generalization to populations of theoretically defined outcomes, experiments have a decided advantage because of their basic defining characteristics: random assignment and/or controlled conditions, the establishment of treatments, and then the comparison of conditions to assess whether or not change occurs. So, the high internal validity of experiments allows high external validity for generalizing to populations that are NOT defined by time and space. But how does this translate into knowledge about society today? How can we use such theories if our interest is in phenomena like race/ethnicity?

Race and Ethnic Composition of Groups

53

How Can Experimental Designs Address Issues Relevant to Society? Although the artificiality of experiments enables rigorous tests of formal theories, it seems that there is a missing link to society. If both the theories and the experiments are artificial, how can they address the important, relevant issues of our surroundings? Can such theories have important implications for issues about race/ethnicity? We argue that, indeed, experiments are invaluable for addressing issues related to race. Because formal theories are highly abstract, they do not describe specific settings like the streets of Detroit or a group of first graders in Texas. But they certainly can apply to such settings. The key to such application is the use of initial conditions (see Cohen, 1980, 1989, 1994a, 1994b, 2003; Sell & Kuipers, 2009; Sell & Martin, 1983 for discussion of initial conditions). Initial conditions serve to provide instances of or instantiate the theoretical concepts. Suppose, for example, that a formal theory describes the process by which characteristics of individuals are generalized and also specifies a set of circumstances under which the process might be decreased. The next step in either a test or an application of the theory is to link the theory to a specific setting. This step involves an informed understanding of the particular cultural context with respect to the theory being tested. Remembering that the theory specifies only the most relevant of characteristics or factors, it becomes important to be intimately familiar with a specific context and to have empirical evidence of that specific context. As an example, suppose the particular theory being invoked uses the concept of a diffuse status characteristic. As defined within status characteristics theory (a research area within the expectations states research program), a diffuse status characteristic is a characteristic possessed by an individual that has at least two states of differential evaluation. Connected with each state there are associated specific performance expectations as well as associated general performance expectations ‘‘without limit as to scope’’ (Webster & Foschi, 1988). That is, not only are those who possess the high state of this diffuse status characteristic societally defined as better at a particular task, but those individuals are defined as ‘‘generally better than’’ those who have the low state of the diffuse status characteristic. The definition does not have any time or space components, that is, it does not define the characteristic in ways that reference a historical context, like slavery in the United States or Islamic religion in Sudan. In this way, matching the definition with the specific culture and time of interest identifies an instance of a diffuse status characteristic. In the case of race/ethnicity, it can be seen that while many

54

CARLA GOAR ET AL.

times being Black is a diffuse status characteristic, it is not always a diffuse status characteristic. Further it does not carry the same meaning in different times and places. Being Black in Brazil has a different meaning than being Black in South St. Louis; being Black in the United States in 1800 has a different meaning than being Black in the United States in 2013. Further complicating this picture is the idea that meaning changes by time and place and interactants. But this is the complex nature of the ‘‘real’’ world and when testing or applying theory, the researcher must be extremely familiar with the context.

How Experimental Design Relates to the Formal Theory Because the ‘‘real world’’ is so context-driven, well-designed experiments testing formal theories incorporate measurement of the contexts themselves to provide an initial condition or baseline. These baseline or control conditions are relative to the theoretical issues being addressed. So, for example, if a theory posits a mechanism to interrupt the process of status generalization, there must be a baseline condition that assesses the degree to which the process organizes interaction in a specific context, absent interruption. Such baselines capture the context. This context may be the interaction of Black and White college-aged women working on a SAT-like test in Austin, Texas, or the interaction of Black and White elderly women working on a group weaving in Charleston, South Carolina. The point is that baselines are necessary measurements to gauge the ‘‘beginning points’’ of the investigation. And without such baselines, experiments cannot reach their powerful potential as tests of formal theories. As defined earlier, experiments are defined by their ability to control or eliminate plausible alternative reasons for obtaining the results of the experiment. Baselines are an important way to accomplish this. Random assignment is another critical technique. The power of randomization is the power assured by probability theory: if errors are distributed randomly, they sum to zero. When treatments are randomly assigned, we can be assured that differences among treatments are not due to the characteristics of the individuals. But what about characteristics that are possessed by individuals and cannot be separated? We cannot randomly assign race, class, and gender. We cannot randomly assign personality characteristics. Sometimes in discussions about experimental design, variables that cause ‘‘noise’’ in interpretation, often because they cannot be randomly assigned, are termed ‘‘nuisance’’ variables. The term arises because the error caused

Race and Ethnic Composition of Groups

55

by these variables impedes interpretation of treatment effects. However, we agree wholeheartedly with Sue’s statement that ‘‘Ethnicity should not be treated as a nuisance variable’’ (Sue, 1999, p. 1076). We also agree with Sue’s (and others) arguments about the paucity of social psychological (especially experimental) research devoted to race/ethnicity (see Graham, 1992; Iwamasa & Smith, 1996; Padilla and Lindholm (1995) for reports on the very low percentage articles focused on ethnic minority groups in psychology; see Hunt, Jackson, Powell, & Steelman, 2000 for reports on social psychology). However, we do not agree with Sue’s argument that the one of the biggest problems inhibiting research about ethnicity is the overconcern with internal validity. Sue argues that his position: ‘‘is not that psychological principles or theories cannot be generalized from one populations to another. Many principles can be applied to different populations. Problems occur when the assumption of generalizability is made. Generality is a phenomenon that should be empirically tested’’ (Sue, 1999, p. 1074). We would argue that Sue is actually discussing the kind of generalization that experiments are weakest at: statistical generalization or descriptive theoretical generalization. Indeed, if one is interested in questions that Sue mentions such as ethnic differences in sexual aggression, generalizability to specific populations defined in terms of space and time is paramount. Further, experimental research is fundamentally incapable of answering these sorts of questions precisely because it is artificial. In light of these kinds of questions, Sue’s plea to lessen attention to internal validity and focus on generalization to different ethnic groups at this point in time make sense. But, we would argue that this is certainly not the only kind of research question that is important for race/ethnicity. The kinds of questions that can be addressed by experiments involve testing of formal theories. These formal theories can enable diffuse status characteristics such as race/ethnicity to be central to the formulation. In the test of these theories, we cannot randomly assign a person to a given race/ethnicity. But we can assure that we make race/ethnicity a focus to both the theory and the experiment. And, although we cannot randomly assign race/ethnicity, we can randomly assign by race/ ethnicity or by race/ethnic group composition. If, for example, we are testing a theory about the interruption of the burden of proof process, we might examine how the ethnic composition of the group affects the distribution of action rates and opportunities. We can randomly assign individuals of a given ethnicity to groups with differing compositions. Important to the interpretation of any test of this nature are the baseline

56

CARLA GOAR ET AL.

conditions. Baselines, in this case of composition, are most likely the group composed of all one race/ethnicity to provide an estimate of the particular cultural context under investigation. And, importantly, as mentioned by Sue (1999) the baselines are not simply assumed to be White control groups. Whiteness is not the default category or the category that is always the reference point.

Challenges from Critical Race Theory Critical race theory (CRT) and related perspectives often critique the scientific method and, in particular, experiments. CRT began as a critique of the U.S. legal system as being developed through systemic racism that facilitates the advancement of some racial groups at the expense of others (Delgado & Stefancic, 2001; Ladson-Billings, 1998a). CRT scholars maintain that this system has normalized racism such that even policies or methods of investigation support a status quo privileging Whites (Crenshaw, 1988; Gordon, 1990; Ladson-Billings, 1998a, 1998b). As described by Adams and Salter (2011), Critical Race Realism is a perspective that draws from social science research, which demonstrates, for example, that all actors, including, judges, lawyers, etc., are subjective and share the same proclivities for attributional biases and consequently cannot be ‘‘color-blind perceivers,’’ as is sometimes claimed. But, Adams and Salter also argue that psychology, in particular, needs a Critical methodology that corresponds more closely with the broader CRT framework. Specifically, they argue that there should be more methodological critique of the scientific method, a method that is permeated by White subjectivities. The authors note that within the discipline of social psychology there is ‘‘an almost fetishistic reliance on laboratory experiments as a direct route to scientific truth. This near exclusive reliance on laboratory-experimental methods is a problem to the extent that it reflects and reproduces ideologies of conceptual and methodological individualism’’ (Adams & Salter, 2011, pp. 1363–1364). This could be especially dangerous, argue the authors, when racism is attributed to individuals rather than systemic contexts, and as a result, scientists ‘‘may unwittingly reproduce racist outcomes when they deploy a racism-denying atomistic conception as individual prejudice’’ (Adams & Salter, 2011, p. 1376). We would disagree with the basic tenet that experimental methods automatically reproduce issues of methodological individualism. Experiments are not necessarily about individuals at all; a very well-known article

Race and Ethnic Composition of Groups

57

that addresses this issue is entitled, ‘‘Can You Study an Army in a Laboratory?’’ (Zelditch, 1969). Zelditch argues that experiments are designed to study the theoretically relevant factors of phenomena, not the actual phenomena in all its complexity. So while one cannot study an actual army in a laboratory, one can study certain organizational aspects, such as authority structure, in a laboratory. We further argue that experiments can actively test how embedded racial hierarchies might operate. In particular, we would argue that taking a structural social psychological view (Lawler, Ridgeway, & Markovsky, 1993) could specifically help avoid some of the pitfalls of methodological individualism. This framework is multilevel. Analysis at any point in time can focus on the actors themselves, their encounters or interactions, microstructures which include constraints such as norms involved in group interaction, and macrostructures which include larger networks such as institutions. This structural view focuses researchers’ attention, and we would maintain, most especially in experimental studies where race/ethnicity, gender, or other such diffuse characteristics forces close consideration of the different levels involved in any group interaction. We will be using this structural social psychological framework in later discussions.

EXPERIMENTS ANALYZING RACE/ETHNICITY WITHIN GROUP PROCESSES Within the group process research, there have been comparatively few studies of race/ethnicity, especially relative to, for example, research examining gender. But there certainly is a history of experimental research on groups, particularly related to Blacks and Whites. Much of this research is related to expectation states framework, a natural fit since much of the framework is devoted to status generalization processes. Some of the earliest experimental studies, which considered interactions in mixed race groups were conducted by Katz and associates. Over a series of experiments, Katz created groups composed of Black and White men who worked on different tasks. In Katz, Goldston, and Benjamin (1958) and Katz and Benjamin (1960), groups composed of two Black and two White male college students worked together. In both studies, men were told that they would receive extra pay if they worked together as a team and that their particular team had higher ability than other teams. Blacks made fewer suggestions than Whites and more readily accepted Whites suggestions. This

58

CARLA GOAR ET AL.

occurred even in the Katz and Benjamin (1960) experiment in which Black and White teammates were matched on the basis of intelligence tests. In the Katz and Cohen (1962) study, groups were composed of only one White and one Black man. Clues were manipulated so that one person had clues that definitely helped solve problems more so than the other partner. Further, half of the Black participants were given ‘‘Assertion Training,’’ training which demonstrated their high performance ability on similar tasks. Results did demonstrate that some Blacks were more influential in the Assertion Training than they were in the baseline condition; however Whites reported negative feelings about Blacks when they were assertive. Webster and Driskell (1978) discuss the above studies in terms of expectations states formulations. The theoretical importance of these studies stems from recognition that race is a diffuse status characteristic. In the case of the first two experiments, the burden of proof process is clearly demonstrated: While there was no rational reason Whites should have more influence on the tasks, there was no intervention and consequently the diffuse status characteristics order the interaction. The ‘‘Assertion Training’’ was a method to insert information about the performance characteristics or specific status characteristics of Blacks that contradicted the societally defined low state of the diffuse status characteristic. As Webster and Driskell (1978) note, the fact that Whites expectations were not changed, however, led to status struggle. Perhaps the most ambitious program for investigation of group processes and race/ethnicity was the decades long research of Elizabeth Cohen and her associates. Beginning in the 1970s, Elizabeth Cohen conducted research designed to eliminate racial/ethnic inequalities in school settings. This work explicitly built upon the expectation states theoretical foundation. In the 1972 Cohen and Roper article, the researchers report their classroom interventions designed to decrease inequality between Black and White children. They devised a way to measure influence by creating a board game, which involved different strategies. Interaction was videotaped and ‘‘units of interaction’’ or uninterrupted speech related to the task were coded as initiations; whereas the person most responsible for the group decision was a measure of influence. In the research reported in Cohen and Roper, baseline conditions with no interventions indicated that great inequality existed, and that being Black was a low state of the diffuse status characteristic of race while being White was the high state. In the experimental or intervention conditions, Cohen and Roper taught the Black students how to put together a two-transistor radio. In one condition of the study, Black students (trained by Black

Race and Ethnic Composition of Groups

59

trainers) were taught to put together the radio and their successful experiences were videotaped and then shown to them (to reinforce the process). Then after this successful acquisition of performance skills (or in the language of expectation status, specific status characteristics), the Black and White students were brought together. This intervention had little effect. However, if the Blacks taught the Whites how to put together the two-transistor radio (a process that Cohen called ‘‘expectation training’’), then inequality in the board game decreased. Importantly, the White children were always told that if they had experiences or training, they would also understand how to put together the radio. The researchers believed that the high performance demonstration (a demonstration of a high status of the specific task characteristic for Black students), the cooperative setting, and the expectation modification or training of BOTH Blacks and Whites were important to the process of decreasing inequality or intervening in the burden of proof process. Further, little hostility occurred. Riordan and Ruggiero (1980) replicated some of the Cohen and Roper (1972) findings; Cook (1974) replicated the interventions and their success with Indians and Whites in British Columbia, and Robbin (1977) replicated the interventions with Mexican Americans and Whites in California. One failure in replication led to the discovery of the importance of trainers or referent actors for the performance skills task. It was determined that the results could be effective only if the trainer for those with the low state of the status characteristics also had the low state of that characteristic (see discussion in Cohen, 1982). This prompted the development of the Reference Actor principle, the idea that the model is an important component of the interaction and intervention. Cohen describes the development of another intervention, the Multiability definition, as an important intervention, one that developed serendipitously. Teachers and researchers found that if a task was defined such that it had multiple abilities and the children were told that nobody was expected to be good on all the abilities but that everyone would be good on at least one of the abilities, task inequality decreased (Stulac, 1975). Rosenholtz and Wilson (1980) and Rosenholtz and Simpson (1984) further developed this notion and began developing curricula around the idea (see further discussion in Cohen, 1994b; Cohen & Lotan, 1997; Lotan, 2006). An important note in this applied curriculum relates to the importance of treating expectations of both those who possess the low state of the diffuse status characteristic as well as those who possess the high state of the diffuse status characteristics. This is a reminder to theorists and applied researchers as well that the stereotyping of issues and the inequality generated on its

60

CARLA GOAR ET AL.

basis is an interactional issue; the vast literature on stereotype threat demonstrates that even those disadvantaged by the stereotypes are affected by them in real ways affecting performance (see Lovaglia, Lucas, & Thye, 1998; Steele, 1997); the literature on expectation states training emphasizes the importance of modifying expectations of both those who possess the high state and those who possess the low state of the diffuse status characteristic. Additional theoretical work by Webster and Driskell (1978) seemed to support the applied finding by Cohen and associates. They tested three models of status generalization by examining White subject’s reactions to a partner about whom they know the following: the partner’s race, Black; the partner’s race, Black and high performance compared to their own; or the partner’s high performance only. The experiment seemed to indicate that participants used both information about race and information about performance in decision-making. That is, the high performances of the Black participants decreased (but did not eliminate) the status generalization associated with the diffuse status characteristics of themselves and their partner. In 2005, Goar and Sell published a theoretical expansion and test of some ideas inspired by Cohen’s applied work and Hamit Fisek’s theoretical extensions (Fisek, 1991). Goar and Sell reasoned that if the task was represented as one which is very complex and consists of many different skills that do not necessarily relate to one another, then this complexity will decrease the status generalization associated with race. (In keeping with Fisek’s terminology, this was termed Inconsistent Complexity.) This would be the case because some of the specific status characteristics associated with doing well on the task are positive but some are negative. Consequently, a person possessing the low state of the diffuse status characteristic would also possess some positive states of the specific status characteristic and the person possessing a high state of the diffuse status characteristics would also possess some low states of the specific status characteristics. In this way, the effects of the diffuse status characteristic are weakened. To test their theory, they employed two conditions. In the baseline condition participants were told that while groups did better than individuals at the task, ‘‘some people do better and some people do worse.’’ In the experimental condition, participants were told that the task was very complex, and had many different abilities associated with it, ‘‘everybody will have some ability to contribute to the task.’’ Their study differed in important ways from the applied work of Cohen and associates. In applied settings, the tasks themselves were different

Race and Ethnic Composition of Groups

61

(so that some involved different kinds of skills that had to be combined), teacher reinforcement for certain strategies was used, modeling was employed, and referent actors were introduced. In the Goar and Sell study, all extraneous factors were controlled, and further, the group tasks themselves were exactly the same in both conditions; only the way they were introduced differed. The baseline condition indicated significant inequality between Black and White participants with Blacks talking less and participating less. In the experimental condition, inequality decreased, at least in terms of time talked and in terms of procedural operations (acts such as writing and organizing). Further, and importantly, when inequality was lessened, task efficiency in terms of correct responses was not affected. Simpson, McGrimmon, and Irwin (2007), is related to race and group processes but is not in the expectation states tradition. The theoretical focus of this research relates to a question we discussed earlier: what is race? How might it change by context? Simpson et al. (2007) begin with the observation that many studies have demonstrated that Blacks are less trusting than Whites, even after controlling for different correlates of trust, including income and education. But they ask, what does this mean, given that the standard measure of trust used in these studies is, ‘‘Generally speaking, would you say that most people can be trusted or that you can’t be too careful in dealing with people?’’ Their major theoretical point is that researchers need to consider what respondents consider ‘‘most people?’’ Simpson et al. (2007) note that it is reasonable to think that both Black and White respondents are thinking about Whites, given the demographics of the United States. As such, they point out that the category of ‘‘White’’ might be triggering a response based on category-based trust. To test their ideas, the researchers brought in Black and White participants. In their responses to the general trust question, the researchers duplicated the usual results: Blacks responded as less trusting than Whites. The researchers then asked participants to interact in a trust game. These games are used as a behavioral measure of trust. In these settings, participants receive money and are given the choice to either keep the money or to pass the money to another person. The amount passed to another person is greatly increased (usually tripled in most studies) and then the second person has the choice of either keeping all the money or passing money back to the first person. The first person’s choice is usually regarded as a behavior measure of trust; the second person’s choice is usually regarded as a measure of trustworthiness. Simpson et al. (2007) found that both Blacks and Whites were more trusting when their partner was in the same race category. However, Blacks were more trusting than were Whites.

62

CARLA GOAR ET AL.

These results are important for many reasons including methodological reasons, but also because they demonstrate that race is not an individual characteristic that exists outside of context. And in this specific case, the context implicates group processes. In the next section we highlight our own research, which is based to a large extent on the expectation states group process experiments we discussed above. In this section, we try to be mindful of the critiques upon which we have focused. Our purpose is to experimentally test theories directed at dismantling inequality generated by race/ethnicity. To avoid the criticisms directed at such experimental investigations, we propose to investigate general, theoretical principles. Our studies are not aimed at describing how race/ethnicity affects peoples’ ordinary interactions, nor are they descriptive of how ‘‘most people’’ act or interact. To avoid the CRT criticism of methodological individualism, we incorporate a structural social psychological approach. We examine the macrostructure in terms of the importation of societal stereotypes and institutional power relations. In terms of the microstructure, we pay particular attention to the creation of norms for the group. The group interaction is our focus, because the interaction itself reproduces or fails to reproduce existing power structures. At the same time, the repeated interaction includes what Lawler, Thye, and Yoon (2009) term productive exchange, the particular kind of exchange that is particularly powerful at developing emotion toward the group itself. Finally, we consider the individual actors, most especially how they conceive of their own and others’ ethnic identities and contributions to the group.

RACIAL/ETHNIC INEQUALITIES: EXAMINING GROUP PROCESS EFFECTS OVER TIME Our new project is an extension of the Goar and Sell (2005) research. The research project utilizes the same inconsistent complexity definition of the task, but examines groups composed of Mexican American and Whites (or Anglos) in Texas and Blacks and Whites in Ohio. In addition to the two different racial/ethnic group compositions, we greatly expanded the time in which groups interact: the time for each group session was increased and groups also met once a week for three weeks. In this way, we can test whether the effects of the initial intervention last over time. Further, we included many other measures of attitudes and feelings toward the group to assess how different models of prejudice might relate to a series of

Race and Ethnic Composition of Groups

63

behavioral measures. Below we detail the experimental designs themselves as manifestations of the structural social psychological approaches. We then briefly discuss some preliminary results.

The Macrostructure Because we are considering diffuse status characteristics, we are considering ‘‘states’’ of the characteristics, meaning that comparisons are always relative to one another. We can consider how the stereotypes related to the states of the characteristics are imported into the group processes. As we all know, stereotypes are not true depictions, but they are powerful because they unconsciously or consciously affect behavior. In groups composed of White and Blacks and groups composed of Whites (or Anglos) and Mexican Americans, we can examine the negative stereotypes about Mexican Americans and Blacks and compare them to the less negative stereotypes about Whites. African Americans and Stereotypes The pervasiveness of negative stereotypes associated with racial and ethnic minorities has been historically documented. Lemons (1977) traces Black stereotypes in popular culture back to 1880 and discusses representations which are still recognizable today such as Old Uncle Tom, Aunt Jemima, and Ol’ Mammy. These negative representations are rooted in racism, the efforts of a dominating group to deny status and power to a dominated group. One of these efforts is the perpetuation of stereotypes (Hall, 2001). Definitions of what constitutes a stereotype are varied however, citing the work of both Pettigrew (1979) and Hamilton (1981), Sigelman & Tuch (1997) state, ‘‘Supported by cultural and psychological, as well as cognitive inputs, stereotypes are simplistic, resist disconfirming evidence, and create selffulfilling prophecies when mutually stereotyping groups interact’’ (p. 87). Hall (2001) cites the work of Lombardo (1978) in discussing two distinctive stereotypes, the brute and the sambo. The brute stereotype describes African Americans as ‘‘yprimitive, temperamental, violent, and sexually powerfuly’’ whereas the sambo stereotype describes them as childlike (p. 106). These stereotypes have permeated various social institutions such as education and the justice system. Stereotypes are not just perpetuated by a dominating group; an invidious aspect is that they also affect the dominated group. Maddox and Gray (2002) found that stereotypes associated with African Americans can be seen not only in prejudices expressed by Whites, but by other African Americans as well.

64

CARLA GOAR ET AL.

It is recognized that there are many negative stereotypes associated with African Americans, such as criminal (Gilliam, Iyengar, Simon, & Wright, 1996; Hurwitz & Peffley, 1997; Steffensmeier, Ulmer, & Kramer, 1998; Welch, 2007), lazy or lacking work ethic (Bullock, Fraser Wyche, & Williams, 2001; Fiske, Xu, Cuddy, & Glick, 1999; Lipset, 1990; Peffley, Hurwitz, & Sniderman, 1997), violent (Gilliam et al., 1996; Hurwitz & Peffley, 1997; Sigelman & Tuch, 1997), and sexually aggressive (Baker, 2005; Maddox & Gray, 2002; Orbe, 2009; West & Rose, 2000). For the purposes of this piece, focus will be placed on the ‘‘dumb Black’’ stereotype and the role of stereotype threat. The stereotype of African Americans being less intelligent is rooted in historical racism and perpetuated by scholarly work. Arthur Jensen’s published work adheres to genetic suppositions of intelligence, discounts social and economic disadvantages encountered by African Americans, and argues that government funding of compensatory education is wasteful (Hall, 2001). Later the publication of The Bell Curve by Herrnstein and Murray (1994) reinforced Jensen’s work and added the argument of a genetic basis for intelligence and reaffirming the ‘‘dumb Black’’ and sambo stereotypes. Not only are African Americans stereotyped as less intelligent, but various research, as cited by Chang & Demyan (2007), notes that teachers hold lower expectations regarding academic performance, higher expectations of disruptive behavior, provide higher rates of negative verbal feedback, and report them more often than their White peers to disciplinary and special education services (p. 92). Students are aware of these stereotypes and often report feeling invisible, self-doubting, and pressured to change or drop classes (Solorzano, Ceja, & Yosso, 2000). The role of stereotype threat during social interactions is significant. Steele & Aronson (1995) discuss stereotype threat as the threat of judgment and resulting treatment based on affirmation of negative stereotypes associated with one’s group. Notably, an individual need not believe the negative stereotype associated with their group. They simply need to be aware of its existence. Steele and Aronson (1995) write, ‘‘This threat can befall anyone with a group identity about which some negative stereotype exists, and for the person to be threatened in this way, he need not even believe the stereotype. He need only know that it stands as a hypothesis about him in situations where the stereotype is relevant’’ (p. 798). In their experimental study of test performance, it was found that when the test is explained as a measure of individual ability African Americans experience significantly greater cognitive activation of stereotypes of African Americans, their own ability, and unwillingness to have their test results linked to their racial identity (Steele & Aronson, 1995). The significance of stereotype threat was

Race and Ethnic Composition of Groups

65

reinforced by the experimental work of Stone, Lynch, Sjomeling, and Darley (1999) regarding athletic performance and intelligence. It was determined that African Americans performed considerably worse on a golf task when the task was framed as a measure of athletic intelligence (Stone et al., 1999). These results are consistent with the findings of Steele and Aronson (1995), indicating that the salience of negative stereotypes of African Americans have an impact on performance when the stereotype is relevant to the task. Further work by Aronson, Fried, and Good (2002) indicates the impact of stereotype threat can be reduced by framing the object of the stereotype as flexible rather than a fixed characteristic. Encouraging African American students to view intelligence as malleable reduced stereotype threat and resulted in greater enjoyment of the academic process, greater engagement, and higher academic performance (Aronson et al., 2002). Mexican Americans and Stereotypes As we have pointed out, and as Dovidio, Ellerx, and Hewstonex (2011) also emphasize, most literature on stereotyping and prejudice is based on Black/ White distinctions. But, it is important to address other distinctions. While the theoretical principles of the status generalization process and interventions in that process are the same, regardless of the specific instance, the testing and application must consider the specific cases. In the case of diffuse status characteristics, we need to be aware of the specific stereotypes existing at this point in time and in this culture. We have ample evidence that Latinos are discriminated against in the United States. There is much statistical evidence related to education and income and occupation. There is also clear experimental evidence. In studies in which trained experimenters who are either Latino or Anglo (and matched on physical characteristics and who use the exact same script) confront store employees asking for gift certificates, Latinos were asked for more information and not given the same service as Anglos (Dovidio et al., 2011). In studies of employment, when Latinos and Whites have the same credentials, Whites more often receive call backs for jobs or offers (Pager, Western, & Bonikowski, 2009). As pointed out by Dovidio et al. (2011) and Devos and Banaji (2005), there is an identification of ‘‘American’’ that is associated with Whites or Anglos, and also with Blacks, but considerably less with Latinos. That is, there is a ‘‘foreignness’’ ascribed to Latinos. Gonzalez notes that there has been a tradition of dehumanization, especially for Latinas, with labels going back to the 1800s in which they were labeled witches and whores (Gonzalez, 2003). Using two different methodologies, one

66

CARLA GOAR ET AL.

using checklists of adjectives and another using open-ended responses, Niemann, Jennings, Rozelle, Baxter, and Sullivan (1994) note the top 15 responses contrasting Anglo American males and females with African American males and females and Mexican American males and females. Perhaps the most discriminating characteristics were ‘‘criminal’’ applied much for often for minority males than Anglo males, ‘‘lower-class’’ mentioned especially for Mexican Americans, ‘‘hard workers’’ for men, and ‘‘family oriented’’ and ‘‘baby makers’’ for Mexican American females. Status was clearly demonstrated by the fact that ‘‘intelligent’’ was often mentioned for Anglos, but rarely mentioned for African Americans or Mexican Americans. In sum, we anticipate that the societal stereotypes associated with the diffuse status of ethnicity will be incorporated into the group processes that we develop. We first test whether these stereotypes become instantiated in the experimental setting.

Microstructure The microstructure relates to the norms established for the group. These norms are important for the establishment of the scope conditions for the expectation states framework. Many of these norms are introduced to the participants as an important aspect of their task. So, participants are told that groups do better when they work together than when they work alone. Further they are provided incentives to do well – a bonus is provided for the more correct responses.

Interaction The interaction in the group consists of all group members participating in solving the tasks at hand. We use the expectation states tradition to help us define observable power and prestige, and use as a set of indicators about the equality within a group. One of the simplest measures is the influence of a given person relative to others in the group. In our case, the tasks we use all have correct responses (as designed by experts in the areas). The tasks included a NASA task (what is most important for survival in space), a SEA task (what is most important while at sea) and a DESERT task (what is most important in a desert). Individuals first decide on their own ranking and then the group develops a group ranking. The measure of influence is the difference between the individual response at time one and the group

Race and Ethnic Composition of Groups

67

response at time two. Other measures correspond to measures of action rates (how often a particular group members speaks and whether they use directives), action opportunities (who is spoken to or asked questions of), and influence and compliance (agreements or disagreements). Most of the observable power and prestige components must be assessed through a coding that involves developing intercoder reliability among coders who represent the diversity of the participants (Compton, Love, & Sell, 2012).

Actors The actors in our study are the individual participants for some predictions but the groups are the actors in other predictions. For example, we predict differences in group efficiency or success depending upon the condition or treatment. But, we also have predictions about how individual group members feel about their interaction in the group. The fact that the actor or the unit of analysis changes based on the theory illustrates an earlier point about the theoretical nature of structural social psychology. Procedures For our experiments (Black–White Experiment at Northern Illinois; Mexican American–non-Hispanic White Experiment at Texas A&M), participants were recruited for studies that met for one hour each week for over three weeks. At the time of recruitment, they self-identified their race/ethnicity. Each group had one minority group member and two majority group members. For each time period, they first worked alone on a relatively complex task that had correct and incorrect responses (as judged by experts in the particular task area). After each person had written her own responses, the group met, discussed the problem for 20 minutes and came up with a group response. Groups received payment for each of the time periods and were told that they could also receive a bonus if their group performed at a high level. To control for status effects other than ethnicity, all participants were female undergraduates. Independent Variable. There were two conditions: Inconsistent Complexity and the Baseline. In the Inconsistency condition, the participants were told that the tasks were complex, involving many different abilities. In the Baseline condition, participants were told that some people do better and

68

CARLA GOAR ET AL.

some people do worse on these tasks. This manipulation was only delivered at the first group meeting and was not repeated at the second or third meeting.

Initial Results The first question that we test relates to the power of race/ethnicity. There are several ways to test this, but one way is to determine if, in the baseline condition, those who possess the lower state of the diffuse status characteristic (African American or Mexican American) have less influence than those who have the higher state of the diffuse status characteristic. Based on the individuals’ initial rankings relative to the final group rankings, both African Americans and Mexicans Americans have less influence that Whites (Anglos) in the Baseline. Based on this, we can conclude that Race/Ethnicity is a diffuse status characteristic for the contexts we consider. A second question relates to task success. Does the experimental condition lead to any differences in task correctness or efficiency? This is an especially important question for applications. It is easy to see that companies, schools, and other institutions would be far more likely to implement an intervention that would lead to less inequality in participation IF such an intervention did not affect task success. In the analysis of this question, we find that the task apparently makes some difference in the answer to this question. Our participants did quite well in Task 1 and Task 3 which corresponded to their first week of interaction and their third week. However, the second week involved the ‘‘Lost at Sea,’’ a task that involved the group deciding on the importance of items to save a crew stranded because their ship is on fire. All participants (and consequently groups) did much worse on this task than the other two. For this task, there was no difference between those in the Experimental Condition and those in the Baseline condition. They all did quite poorly. However, for the other two tasks, the Experimental Groups did better than did the Baseline Groups. So, importantly, the intervention never negatively affected the success of the group, and apparently sometimes increased task success.

SUMMARY We have argued that experiments can play an important role in the investigation of race/ethnicity. However, experiments cannot tell us about

Race and Ethnic Composition of Groups

69

descriptive aspects of race/ethnicity. We cannot ask questions about the experience of being Mexican American in a predominantly Anglo Institution in 2013 by conducting an experiment. We cannot directly generalize from our experimental groups to elementary school groups, research and development groups or teams in hospitals. However, we can ask questions such as, ‘‘under what conditions can stereotyping be decreased?’’ These kinds questions can be addressed by experiments because they are formal, theoretical questions. They are not meant to address an empirical population defined at a particular point in time. As we are have argued, these kinds of questions are the precursors to both tests and eventual application. Such questions are conditional, that is, they require scope conditions, conditions specifying when the theories apply. As an example, our illustrations concern task groups as an important scope condition. When we test the theory, we instantiate the scope conditions or develop initial conditions by creating groups in which group members work on specific activities such as survival strategies related to space. To apply the theory, we ask whether a group might meet the scope conditions: does this classroom activity of measuring a dinosaur meet the requirements for a task group? Theoretical questions about diffuse status characteristics can be fruitfully examined through experimental methods. Indeed, we argue that experiments are especially advantageous for answering these kinds of questions. The complementarity of formal approaches and experimental methods lead us, not toward asking, ‘‘what happens in these real world settings such as classroom and research and development groups?’’ but rather, ‘‘what might be possible?’’ ‘‘what kinds of conditions in task groups lead to equality in participation, decreases in stereotyping, and success?’’ We believe these questions to be exceptionally important: ones that can serve to countermand the reproduction of discrimination.

NOTE 1. We use the term ‘‘Race/Ethnicity’’ to refer to the sociological categorization of people. We realize that this is sometimes a controversial term, but feel that it fits the goals of our research program, because it emphasizes social construction.

ACKNOWLEDGMENTS This work was supported in part by National Science Foundation Grant 0961940 awarded to the first two authors. An American Sociological

70

CARLA GOAR ET AL.

Association/National Science Foundation, Future of the Discipline (FAD) grant was also awarded to the first two authors; the conference supported by this grant was critical for our thinking about experiments and race/ethnicity. We are grateful to all the participants. We also thank the departments of sociology at Kent State University and Texas A&M University for their support for this project, and are especially appreciative of all the undergraduate research assistants.

REFERENCES Anonymous. (2008). AAA Statement on Race. American Anthropologist, 100(3), 712–713. doi:10.1525/aa.1998.100.3.712 Adams, G., & Salter, P. (2011). A critical race psychology is not yet born. Connecticut Law Review, 43(5). Retrieved from http://heinonlinebackup.com/hol-cgi-bin/get_pdf.cgi? handle=hein.journals/conlr43§ion=47 American Sociological Association. (2003). The importance of collecting data and doing social scientific research on race. Washington, DC: American Sociological Association. Aronson, J., Fried, C., & Good, C. (2002). Reducing the effects of stereotype threat on african american college students by shaping theories of intelligence. Journal of Experimental Social Psychology, 38, 113–125. Babbie, E. R. (1975). Practice of social research. Belmont, CA: Wadsworth. Baker, C. N. (2005). Images of womens sexuality in advertisements: A content analysis of blackand white-oriented women’s and men’s magazines. Sex Roles, 52(1-2), 13–27. Berger, J., Cohen, B. P., & Zelditch, M. J. (1972). Status characteristics and social interaction. American Sociological Review, 37(3), 241–255. Retrieved from http://www.jstor.org/ stable/10.2307/2093465 Berger, J., & Webster, M. (2006). Expectations, status, and behavior. In P. J. Burke (Ed.), Contemporary social psychological theories (pp. 269–300). Stanford, CA: Stanford University Press. Bullock, H. E., Fraser Wyche, K., & Williams, W. R. (2001). Media images of the poor. Journal of Social Issues, 57(2), 229–246. doi:10.1111/0022-4537.00210 Campbell, D. T., & Stanley, J. C. (1967). Experimental and quasi-experimental designs for research. Chicago, IL: Rand McNally. Chang, D. F. D., & Demyan, A. A. L. (2007). Teachers’ stereotypes of Asian, Black, and White students. School Psychology Quarterly, 22(2), 91. doi:10.1037/1045-3830.22.2.91 Cohen, B. P. (1980). The conditional nature of scientific knowledge. In L. Freese (Ed.), Theoretical methods in sociology: Seven essays. Pittsburgh, PA: University of Pittsburgh Press. Cohen, B. P. (1989). Developing sociological knowledge: Theory and method (2nd ed.). Chicago, IL: Nelson-Hall. Cohen, B. P. (2003). Creating, testing, and applying social psychological theories. Social Psychology Quarterly, 66(1), 5–16. Retrieved from http://www.jstor.org/stable/10.2307/ 3090137

Race and Ethnic Composition of Groups

71

Cohen, E., & Lotan, R. (1997). Working for equity in heterogeneous classrooms: Sociological theory in practice. Sociology of Education Series. New York, NY: Teachers College Press. Cohen, E. G. (1982). Expectation states and interracial interaction in school settings. Annual Review of Sociology, 8(1982), 209–235. Retrieved from http://www.jstor.org/stable/ 10.2307/2945994 Cohen, E. G. (1993). From theory to practice: The development of an applied research program. In J. Berger, (Ed.) Theoretical research programs: Studies in the growth of theory (pp. 385–415). Stanford, CA: Stanford University Press. Cohen, E. G. (1994a). Restructuring the classroom: Conditions for productive small groups. Review of Educational Research, 64(1), 1–35. doi:10.3102/00346543064001001 Cohen, E. G. (1994b). Designing groupwork: Strategies for heterogeneous classrooms. New York, NY: Teachers College Press. Cohen, E. G., & Roper, S. S. (1972). Modification of interracial interaction disability: An application of status characteristic theory. American Sociological Review, 37(6), 643–657. Compton, D., Love, T., & Sell, J. (2012). Developing and assessing intercoder reliability in studies of group interaction. Sociological Methodology, 42(1), 348–364. doi:10.1177/ 0081175012444860 Cook, T. D. (1974). Producing equal status interaction between Indian and white boys in British Columbia: An application of expectation trainin. Stanford, CA: Stanford Univeristy. Cook, T. D., & Campbell, D. T. (1979). Quasi-experimentation: Design and analysis for field settings. Chicago, IL: Rand McNally. Crenshaw, K. W. (1988). Race, reform, and retrenchment: Transformation and legitimation in antidiscrimination law. In L. Back & J. Solomos (Eds.), Harvard Law Review, 101(7), 1331–1387. doi:10.2307/1341398 Delgado, R., & Stefancic, J. (2001). Critical race theory: An introduction. In R. Delgado & J. Stefancic (Eds.) Race (pp. 60–110). New York, NY: New York University Press. Retrieved from http://uolibraries.worldcat.org/title/critical-race-theory-an-introduction/ oclc/45387241&referer=brief_results Dovidio, J. F., Ellerx, A., & Hewstonex, M. (2011). Improving intergroup relations through direct, extended and other forms of indirect contact. Group Processes & Intergroup Relations, 14(2), 147–160. doi:10.1177/1368430210390555 Devos, T., & Banaji, M. R. (2005). American=White? Journal of Personality and Social Psychology, 88(3), 447–466. doi:10.1037/0022-3514.88.3.447 Fisek, M. H. (1991). Complex task structures and power and prestige orders. Advances in Group Processes, 8, 115–134. Fiske, S. T., Xu, J., Cuddy, A. C., & Glick, P. (1999). (Dis)respecting versus (Dis)liking: Status and interdependence predict ambivalent stereotypes of competence and warmth. Journal of Social Issues, 55(3), 473–489. doi:10.1111/0022-4537.00128 Foschi, M. (1997). On scope conditions. Small Group Research, 28(4), 535–555. doi:10.1177/ 1046496497284004 Freese, L. (1980). The problem of cumulative knowledge. Theoretical methods in sociology: Seven essays (pp. 13–69). Pittsburgh, PA: University of Pittsburgh Press. Gilliam, F. D., Iyengar, S., Simon, A., & Wright, O. (1996). Crime in black and white: The violent, scary world of local news. Harvard International Journal of PressPolitics, 1(3), 6–23. doi:10.1177/1081180X96001003003 Goar, C., & Sell, J. (2005). Using task definition to modify racial inequality within task groups. The Sociological Quarterly, 46(815), 525–543. Retrieved from http://onlinelibrary. wiley.com/doi/10.1111/j.1533-8525.2005.00025.x/full

72

CARLA GOAR ET AL.

Goar, C., & Sell, J. (2009). The use of fad funds: Expanding experimental investigations of race/ ethnicity in sociology. ASA Footnotes. Retrieved from http://www.asanet.org/footnotes/ mar09/fad_0309.html Gonzalez, D. (2003). ‘Lupe’s song’: On the origins of mexican-women-hating in the United States. In Velvet barriers: Popular culture and Chicano/a sexualities (pp. 251–264). New York, NY: Palgrave Macmillan. Gordon, R. (1990). New developments in legal theory. In D. Kairya (Ed.), The politics of law: A progressive critique (pp. 413–425). New York, NY: Pantheon Books. Graham, S. (1992). Most of the subjects were white and middle class: Trends in published research on African Americans in selected APA journals. American Psychologist, 47(5), 629. Gravlee, C. C. (2009). How race becomes biology: Embodiment of social inequality. American journal of physical anthropology, 139(1), 47–57. doi:10.1002/ajpa.20983 Gravlee, C. C., & Dressler, W. W. (2005). Skin pigmentation, self-perceived color, and arterial blood pressure in Puerto Rico. American Journal of Human Biology the Official Journal of the Human Biology Council, 17(2), 195–206. Retrieved from http://www.ncbi.nlm. nih.gov/pubmed/15736179 Gravlee, C. C., Dressler, W. W., & Bernard, H. R. (2005). Skin color, social classification, and blood pressure in southeastern Puerto Rico. American Journal of Public Health, 95(12), 2191–2197. Retrieved from http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid= 1449506&tool=pmcentrez&rendertype=abstract Hall, R. E. (2001). The ball curve: Calculated racism and the American men of African stereotype. Journal of Black Studies, 32(1), 104–119. doi:10.1177/002193470103200106 Hamilton, D. (1981). Cognitive processes in stereotyping and intergroup behavior. Hillsdale, NJ: Lawrence Erlbaum Associates. Retrieved from http://scholar.google.com.lib-ezproxy. tamu.edu:2048/scholar?q=hamilton+1981+stereotype&btnG=&hl=en&as_sdt=0% 2C44#0 Henshel, R. (1980a). Seeking inoperative laws: Toward the deliberate use of unnatural experimentation. Theoretical methods in sociology: Seven essays (pp. 175–199). Pittsburgh, PA: University of Pittsburgh Press. Henshel, R. L. (1980b). The purposes of laboratory experimentation and the virtues of deliberate artificiality. Journal of Experimental Social Psychology, 16(5), 466–478. Retrieved from http://www.sciencedirect.com/science/article/pii/0022103180900529 Herrnstein, R., & Murray, C. A. (1994). The Bell Curve: The reshaping of American life by differences in intelligence. In F. Bess, (Ed.). New York, NY: Free Press. Hunt, M., Jackson, P., Powell, B., & Steelman, L. (2000). Color-blind: The treatment of race and ethnicity in social psychology. Social Psychology Quarterly, 63(4), 352–364. Retrieved from http://www.jstor.org/stable/10.2307/2695845 Hurwitz, J., & Peffley, M. (1997). Public perceptions of race and crime: The role of racial stereotypes. American Journal of Political Science, 41(2), 375–401. Iwamasa, G. Y., & Smith, S. K. (1996). Ethnic diversity in behavioral psychology: A review of the literature. Behavior Modification, 20(1), 45–59. Katz, I., & Benjamin, L. (1960). Effects of white authoritarianism in biracial work groups. The Journal of Abnormal and Social Psychology, 61(3), 448. Katz, I., & Cohen, M. (1962). The effects of training Negroes upon cooperative problem solving in biracial teams. The Journal of Abnormal and Social Psychology, 64(5), 319. Katz, I., Goldston, J., & Benjamin, L. (1958). Behavior and productivity in biracial work groups. Human Relations, 11, 123–141.

Race and Ethnic Composition of Groups

73

Ladson-Billings, G. (1998a). Just what is Critical Race Theory and what’s it doing in a nice field like education? International Journal of Qualitative Studies in Education, 11(1), 7–24. Ladson-Billings, G. (1998b). Teaching in dangerous times: Culturally relevant approaches to teacher assessment. The Journal of Negro Education, 67(3), 255. doi:10.2307/2668194 Lawler, E., Ridgeway, C., & Markovsky, B. (1993). Structural social psychology and the micromacro problem. Sociological Theory, 11(3), 268–290. Retrieved from http://cat.inist.fr/?a Modele=afficheN&cpsidt=4281789 Lawler, E., Thye, S., & Yoon, J. (2009). Social commitments in a depersonalized world. New York, NY: Russell Sage Foundation Publications. Lemons, S. (1977). Black stereotypes as reflected in popular culture 1880–1920. American Quarterly, 29(1), 102–116. Lipset, S. M. (1990). The continental divide: The values and institutions of the United States and Canada. London: Routledge. Lombardo, B. (1978). The Harlem Globetrotters and the perception of the black stereotype. The Physical Educator, 35(2), 60–63. Retrieved from http://scholar.google.com. lib-ezproxy.tamu.edu:2048/scholar?q=lombardo+1978+stereotypes&btnG=&hl= en&as_sdt=0%2C44#0 Lotan, R. (2006). Managing groupwork. In C. Everston & C. Weinstein (Eds.), Handbook of Classroom management: Research, practice and contemporary issues (pp. 525–539). Mahwah, NJ: Lawrence Erlbaum Associates. Lovaglia, M., Lucas, J., & Thye, S. (1998). Status processes and mental ability test scores. American Journal of Sociology, 104(1), 195–228. Retrieved from http://www.jstor.org/ stable/10.1086/210006 Lucas, J. W. (2003). Theory-testing, generalization, and the problem of external validity. Sociological Theory, 21(3), 236–253. doi:10.1111/1467-9558.00187 Maddox, K. B., & Gray, S. A. (2002). Cognitive representations of Black Americans: Reexploring the role of skin tone. Personality and Social Psychology Bulletin, 28(2), 250–259. doi:10.1177/0146167202282010 Markovsky, B. (1994). The structure of theories. In M. Foschi & E. J. Lawler (Eds.), Group processes: Sociological analysis (pp. 3–24). Chicago, IL: Nelson-Hall. Martin, M., & Sell, J. (1979). The role of the experiment in the social sciences. The Sociological Quarterly, 20(4), 581–590. Retrieved from http://onlinelibrary.wiley.com/doi/10.1111/ j.1533-8525.1979.tb01237.x/abstract Meeker, B. F., & Leik, R. (1995). Experimentation in sociological social psychology. In K. S. Cook, G. A. Fine & J. S. House (Eds.), Sociological perspectives on social psychology (pp. 630–649). Allyn and Bacon. Mook, D. G. (1983). In defense of external invalidity. American Psychologist, 38(4), 379–387. doi:10.1037/0003-066X.38.4.379 Niemann, Y. F., Jennings, L., Rozelle, R. M., Baxter, J. C., & Sullivan, E. (1994). Use of free responses and cluster analysis to determine stereotypes of eight groups. Personality and Social Psychology Bulletin, 20(4), 379–390. doi:10.1177/0146167294204005 Orbe, M. P. (2009). Construction of reality on MTV’s ‘‘The Real World’’: An analysis of the restrictive coding of masculinity. Southern Communication Journal, 64(1), 32–47. Padilla, A., & Lindholm, K. (1995). Quantitative educational research with ethnic minorities. Handbook of Research on Multicultural Education, 97–113. Pager, D., Western, B., & Bonikowski, B. (2009). Discrimination in a low-wage labor market: A field experiment. American Sociological Review, 74(5), 777–799. Retrieved from http:// asr.sagepub.com.lib-ezproxy.tamu.edu:2048/content/74/5/777.short

74

CARLA GOAR ET AL.

Peffley, M., Hurwitz, J., & Sniderman, P. M. (1997). Racial stereotypes and Whites’ political views of Blacks in the context of welfare and crime. American Journal of Political Science, 41(1), 30–60. doi:10.2307/2111708 Pettigrew, T. F. (1979). The ultimate attribution error: Extending Allport’s cognitive analysis of prejudice. Personality and Social Psychology Bulletin, 5(4), 461–476. doi:10.1177/01461 6727900500407 Riordan, C., & Ruggiero, J. (1980). Producing equal-status interracial interaction: A replication. Social Psychology Quarterly, 43(1), 131–136. Robbin, A. (1977). Fostering equal status interaction through the establishment of consistent staff behavior and appropriate situated norms. Paper presented at the American Education Research Association. New York, NY. Rosenholtz, S. J., & Simpson, C. (1984). The formation of ability conceptions: Developmental trend or social construction? Review of Educational Research, 54(1), 31–63. doi:10.3102/ 00346543054001031 Rosenholtz, S. J., & Wilson, B. (1980). The effect of classroom structure on shared perceptions of ability. American Educational Research Journal, 17(1), 75–82. Retrieved from http:// aer.sagepub.com/content/17/1/75.short Sell, J., & Kuipers, K. J. (2009). A structural social psychological view of gender differences in cooperation. Sex Roles, 61(5-6), 317–324. doi:10.1007/s11199-009-9597-5 Sell, J., & Martin, M. W. (1983). An acultural perspective on experimental social psychology. Personality and Social Psychology Bulletin, 9(3), 345–350. doi:10.1177/01461672830 93003 Sigelman, L., & Tuch, S. A. (1997). Metastereotypes: Blacks’ perceptions of Whites’ stereotypes of Blacks. Public Opinion Quarterly, 61(1), 87–101. doi:10.1086/297788 Simpson, B., McGrimmon, T., & Irwin, K. (2007). Are Blacks really less trusting than Whites? Revisiting the race and trust question. Social Forces, 86(2), 525–552. Retrieved from http://sf.oxfordjournals.org/content/86/2/525.short Solorzano, D., Ceja, M., & Yosso, T. (2000). Critical race theory, racial microaggressions, and campus racial climate: The experiences of African American College students. The Journal of Negro Education, 69(1/2), 60–73. doi:10.1080/095183998236926 Steele, C. M. (1997). A threat in the air: How stereotypes shape intellectual identity and performance. American Psychologist, 52(6), 613. Steele, C. M., & Aronson, J. (1995). Stereotype threat and the intellectual test performance of African Americans. Journal of Personality and Social Psychology, 69(5), 797–811. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/7473032 Steffensmeier, D., Ulmer, J., & Kramer, J. (1998). The interaction of race, gender, and age in criminal sentencing: The punishment cost of being young, black, and male. Criminology, 36(4), 763–798. doi:10.1111/j.1745-9125.1998.tb01265.x Stone, J., Lynch, C. I., Sjomeling, M., & Darley, J. M. (1999). Stereotype threat effects on Black and White athletic performance. Journal of Personality and Social Psychology, 77(6), 1213–1227. doi:10.1037/0022-3514.77.6.1213 Stulac, J. T. (1975). The self-fulfilling prophecy: Modifying the effects of a unidimensional perception of academic competence in task-oriented groups. Stanford, CA: Stanford University. Sue, S. (1999). Science, ethnicity, and bias. 54(12), 1070–1077. Walker, H. A., & Cohen, B. P. (1985). Scope statements: Imperatives for evaluating theory. American Sociological Review, 50(3), 288. doi:10.2307/2095540

Race and Ethnic Composition of Groups

75

Webster, M., & Driskell, J. E. (1978). Status generalization: A review and some new data. American Sociological Review, 43, 220–336. Webster, M., & Foschi, M. (1988). Status generalization: New theory and research. Stanford, CA: Stanford University Press. Webster, M., & Kervin, J. B. (1971). Artificiality in experimental sociology. Canadian Review of Sociology and Anthropology, 8, 263–272. Webster, M., & Sell, J. (2007). Laboratory experiments in the social sciences. Academic Press. Welch, K. (2007). Black criminal stereotypes and racial profiling. Journal Of Contemporary Criminal Justice, 23(3), 276–288. doi:10.1177/1043986207306870 West, C. M., & Rose, S. (2000). Dating aggression among low income African American youth: An examination of gender differences and antagonistic beliefs. Violence Against Women. doi:10.1177/10778010022181985 Willer, D., & Walker, H. (2007). Building experiments: Testing social theory. Stanford, CA: Stanford University Press. Willer, D. (1987). Theory and the experimental investigation of social structures. Routledge. Zelditch, M. (1969). Can you really study an army in the laboratory? In A. Etzioni (Ed.), A sociological reader on complex organizations (2nd ed., pp. 528–539). New York, NY: Holt.

THE SPREAD OF STATUS VALUE: A THEORETICAL EXTENSION Joseph Berger and M. Hamit Fis- ek ABSTRACT Purpose – The Spread of Status Value theory describes how new diffuse status characteristics can arise out of the association of initially nonvalued characteristics to existing status characteristics that are already well-established in a society. Our objective is to extend this theory so that it describes how still other status elements, which have become of interest to researchers such as ‘‘status objects’’ (Thye, 2000) and ‘‘valued roles’’ (Fis- ek, Berger, & Norman, 1995), can also be socially created. Design/methodology/approach – Our approach involves reviewing research that is relevant to the Spread of Status Value theory, and in introducing concepts and assumptions that are applicable to status objects and valued roles. Findings – Our major results are an elaborated theory that describes the construction of status objects and valued roles, a graphic representation of one set of conditions in which this creation process is predicted to occur, and a design for a further empirical test of the Spread of Status Value theory. This extension has social implications. It opens up the possibility of creating social interventions that involve status objects and valued roles to ameliorate dysfunctional social situations.

Advances in Group Processes: Thirtieth Anniversary Edition Advances in Group Processes, Volume 30, 77–107 Copyright r 2013 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 0882-6145/doi:10.1108/S0882-6145(2013)0000030007

77

78

JOSEPH BERGER AND M. HAMIT FIS-EK

Originality/value – Our elaborated theory enables us to understand for the first time how different types of status valued elements can, under appropriate conditions, be socially created or socially modified as a result of the operation of what are fundamentally similar processes. Keywords: Spread of status value; status objects; valued role; graphic representation; status characteristics theory; social interventions

INTRODUCTION While the role of diffuse status characteristics in determining behavior in goal-oriented social situations has long been the concern of Status Characteristics Theory (Berger, Cohen, & Zelditch, 1966; Berger, Fis- ek, Norman, & Zelditch, 1977), it was the research of Ridgeway (1991) that first addressed the issue of how such diffuse status characteristics came to be established to begin with. Over a 20-year period, she and her colleagues have developed a theory of Status Construction that has been extensively tested and elaborated through a body of ongoing experimental and theoretical research. Recently, Berger and Fis- ek (2006) have formulated a second theory, the Spread of Status Value, which like Ridgeway’s work is concerned with how diffuse status characteristics come to be constructed, but employs a different basic mechanism – that involves the spread of status value. Our objective in this chapter is to extend that theory so that it will cover other objects in which status value plays an important role, namely, status objects and social roles. In the next section we briefly review the research of Ridgeway and her colleagues, focusing on the key mechanism for status construction developed in their work. We then examine some of the research on the spread of status value, in particular that of Berger, Zelditch, Anderson, and Cohen (1972) and that of Thye (2000) that has served as the background to our work. In section three we present revised assumptions for the extended theory of Spread of Status Value, followed by studies that are relevant to spread of status value processes. In section four we consider the problem of testing the extended theory, and in section five, we consider some of the more general implications of our results.

79

Spread of Status Value

RELEVANT THEORETICAL BACKGROUND Ridgeway’s Program Recent research in the creation of status categories has been stimulated by the work of Cecilia Ridgeway and colleagues. Ridgeway (1991) formulated a theory describing processes by which status characteristics can come to be constructed. She envisions situations in which individuals are consistently discriminated on initially non-valued states of a characteristic and high and low levels of resources. She argues that possessing high or low resources can lead actors to achieve, respectively, high and low positions in the group’s interaction or influence hierarchy. These high and low positions, in turn, become the bases of the evaluation of the actor’s status level and performance capacity (or competence). Assuming that such situations are repeated and consistent – the same initially non-valued states of the characteristic are associated with high or low levels of resources – these status and competence evaluations will become status beliefs that are associated with the initially non-valued states of the characteristic. In short, the fundamental underlying mechanism in Ridgeway’s account of how new status characteristics are formed involves actors achieving and enacting high or low positions in the group’s interaction hierarchy on the basis of the level of resources they possess. The basic arguments in Ridgeway’s theory have been subjected to extensive experimental investigation (Ridgeway, Boyle, Kuipers, & Robinson, 1988; Ridgeway & Ericson, 2000; Ridgeway & Correll, 2006; Ridgeway, Li, Erickson, Backor, & Tinkler, 2009). One of the most interesting findings from this experimental research is that the possession of high or low levels of resources is not a logically necessary condition for the emergence of these new status beliefs. Status beliefs will be created and associated with the initially non-valued states provided that actors occupy and enact high or low positions in the interaction hierarchy given whatever means by which such positions are attained. This result was predicted by Webster and Hysom (1998) and experimentally confirmed by Ridgeway and Ericson (2000). Ridgeway’s original theory (1991) has been elaborated further by Ridgeway et al. (2009); Ridgeway and Correll (2006); Webster and Hysom (1998); and Berger, Ridgeway, and Zelditch (2002). We build upon Ridgeway’s research in applying a theory on the spread status value to the creation of new status characteristics.

80

JOSEPH BERGER AND M. HAMIT FIS-EK

Spread of Status Value While the idea of spread of status value is an old one in sociology (Veblen, 1899), there have been few attempts to develop abstract and general theories that embody this idea. In terms of the background to our work two bodies of research have been particularly important, Berger et al.’s theoretical research on a ‘‘Structural Aspects of Distributive Justice: A Status Value Formulation’’ (1972) and Shane Thye’s research on ‘‘A Status Value Theory of Power in Exchange Networks’’ (2000). In the distributive justice chapter, Berger and his colleagues argued for the important role of status value processes in defining justice situations. They reasoned that the status value of elements in an immediate local setting (the situation of action) and the embedded shared moral expectations in such situations are conditioned by referential structures that are part of the social or cultural framework that encompass these local settings. Further, they argued that one of the ways by which the status value of established elements in the cultural framework operates to define the status value of elements in a local situation is through a process where status value spreads from established status valued elements to elements that initially do not have status value. More specifically they argued that if a set of status valued elements share the same status value and are related to a nonstatus valued element in a local situation, status value will spread to the non-valued element. If the valued elements differ in their status value there will be no transfer of status value. They called this principle the Spread of Status Value. Building on these ideas on the spread of status value as conceptualized above, and the idea in Status Characteristics theory (Berger, Fis- ek, Norman, & Zelditch, 1977) that different status values are embedded in status characteristics, as well as ideas on power use as developed in Emerson’s (1972a, 1972b) power-dependence theory and in Willer’s (1981) elementary theory, Shane Thye, in a major chapter, presented a theory on the Status Value of Power in Exchange Relations (2000). The theory consists of three assumptions. The first argues that if A has higher status than B then A’s resources will be more highly valued than B from both the perspective of A and B. The status value of characteristics possessed by high status actors will be higher than that possessed by low status actors and will spread to the resources these actors possess. Assumption 2 argues that if the resources of A are more highly valued by A and B than the resources of B, then A will have power over B, and assumption three

Spread of Status Value

81

argues that if A has power over B, A will acquire greater resources than B in an exchange with B. To test these theoretical arguments, Thye developed an effective experimental procedure to study the spread of status value. In the first phase of the experiment, subjects were randomly assigned to high and low status conditions on the basis of purported differences in the diffuse and specific status characteristics they possessed. In the second phase of the experiment, subjects were engaged in a standard exchange bargaining task that involved using poker chips (Lovaglia, Skvoretz, Markovsky, & Willer, 1995), with purple chips presumably being possessed by a subject’s high status partner and orange chips presumably being possessed by a subject’s low status partner. Subjects were motivated to obtain as many resources (poker chips) as possible from the bargaining exchanges with their partners. In general, the results supported Thye’s predictions: high status subjects acquired more resources than low status subjects, and subjects reported that it was more important to acquire the chips held by high status partners than those possessed by low status partners, although they were told that all chips had equal monetary value for them. With this experiment, Thye had shown how to move research on the spread of status value from the purely theoretical level to a level involving both theory and relevant experimental research. Before turning to the concepts and assumptions of the Status Value theory, we note that our principal concern in this chapter is with the core processes of status creation as they occur in a situation of action. Following Ridgeway and others, we separate as a distinct research issue these core processes from those that describe the diffusion of a newly constructed status distinction through a particular social population. (For relevant chapters on this diffusion issue, see, in addition to Ridgeway’s original chapter, Ridgeway and Balkwell (1997) and Mark, Smith-Lovin, and Ridgeway (2009).)

THE INITIAL FORMULATION OF THE STATUS VALUE THEORY The primary objective of our formulation of ‘‘Diffuse Status Characteristics and the Spread of Status Value’’ (Berger & Fis- ek, 2006) was to develop a theory that describes a process by which new diffuse status characteristics

82

JOSEPH BERGER AND M. HAMIT FIS-EK

come to be constructed. On the most general level, the theory describes how new status characteristics can emerge out of their particular association with status characteristics that exist and already are well-established within a society or relevant social body. This theory, like others in the expectation states program, rests upon the basic concepts and assumptions of the graph version of the theory of status characteristics and expectation states (Berger, Fis- ek, Norman, & Zelditch, 1977). A key concept in the status characteristics theory is that of a diffuse status characteristics, which is seen to involve four major features: (1) a socially significant characteristic such as gender, race, and occupational position; (2) states of the characteristic such as male–female, white–black, higher– lower occupational positions, which can partition a relevant population; (3) different status evaluations of these states relative to each other in terms of honor, esteem, prestige, respect and general social worth; and (4) high and low conceptions of the generalized performance capacities of the individuals who possess these different states, where these high and low conceptions are consistent with the status evaluation of the states of the characteristic. Thus gender may be a diffuse status characteristic for a population at a given time if the members of that population hold differential status evaluations for male and female, with, say, males being more highly esteemed, honored, and in general more socially valued than women, and with members of the population commonly assuming that men are in general superior and more capable than women on a wide (and typically unspecified) range of valued tasks and activities. As should be clear from the above, a diffuse status characteristic implies the existence of two dimensions of inequality between individuals in advantaged and disadvantaged categories. There is the inequality in performance capacities with the individual in the advantaged category assumed, in general, to be more capable than the individual in the disadvantaged category. In addition, there is the inequality in status valuations with the individual in the advantaged category assumed to be more socially valuable and more socially worthy than an individual in the disadvantaged category. Therefore, a theory that explains how new diffuse status characteristics come to be created is also a theory that describes the social origin and bases of prejudices and invidious expectations and evaluations of disadvantaged status groups. To describe the process by which a new diffuse status characteristic comes to be constructed, the Spread of Status Value theory (Berger & Fis- ek, 2006) presented three major arguments. First, it argued that if actors possessed the states of established diffuse status characteristics and other established

Spread of Status Value

83

status elements that already have status value in society and these actors also possessed an initially non-valued but significant characteristic then status value would spread from the established status elements to the states of the non-valued characteristic. Second, it argued that as status value spreads to the states of the initially non-valued characteristic high and low performance expectations would come to be connected to these initially non-valued states consistent with the status value they had acquired. Further, it argued that as the states of the initially non-valued characteristic build up status value through their associations with established status characteristics and other valued status elements, the connections between the states of the new characteristic and the high and low states of generalized performance expectations become increasingly stronger until they approach a maximum strength. Thus as the initially non-valued characteristic accumulates status value it becomes increasingly effective in determining the status behaviors of individuals. But third, we argued that more is required for such a characteristic to become a new stable status characteristic. We claimed that the characteristic must become an object in the public domain, that it must be externalized, and that it must become part of a social reality. One way by which these changes can be achieved is through a process of social validation. This involves enacting, and consensually supporting the meanings and behaviors implied by the new status characteristic.

THE EXTENDED FORMULATION As already indicated, our primary objective in this chapter is to extend our theory so that it will also apply to social objects and to social roles. With respect to social objects, we build on Thye’s work, which provides evidence for the creation of status objects by the transfer of positive and negative status values to social objects. Our aim here is to explain what is involved in the spread of relevance process by representing it with graphs from the Status Characteristics theory (Berger et al., 1977) and using the assumptions of the extended Spread of Status value theory to investigate the consequences of the transfer process. With respect to social roles, we start with the observation that social roles often have positive or negative status values embedded in them and these values can play a major part in determining the behavior of the occupant of these roles. A study that deals with this issue is research by Jackson, Esses, and Burris (2001) who in a set of studies tested and in

84

JOSEPH BERGER AND M. HAMIT FIS-EK

general found support for the argument that the greater respect normally accorded men as a compared to women acts as a mediating variable in hiring biases favoring men in situations that involve high status jobs. In our terms the general difference in respect accorded to males and females is an expression of the difference in status value, which is a component of gender as a diffuse status characteristic in our society. What the research of Jackson and associates suggest is that a mechanism involved in gender-job discrimination may be the matching of the sign of the status value of states of status characteristics with the sign of the states value associated with specific roles. More generally, our goal in applying the Spread of Status Value theory to roles is to account for a process by which roles can acquire differential status values.

Concepts from the Graph Version of the Status Characteristics Theory We shall first describe the basic concepts of the graph formulation of the status characteristics theory (Berger, Fis- ek, Norman, & Zelditch, 1977) that were used to develop the original status value formulation (Berger & Fis- ek, 2006), which we are extending in this chapter. Since the status value formulation builds on the graph formulation of the status characteristics and the reward expectations theories (Berger, Fisek, Norman, & Wagner, 1985), formal statements of both these theories are presented in the appendix.1 The status characteristics and the reward expectations theories describe how actors use status information to form expectations for task performances and for rewards levels for themselves and the other actors with whom they are interacting. The theories describe how status characteristics become admitted as usable items of information in a task situation (salience); how they become connected with task and reward outcome states (Burden of Proof and Activation of Referential structures); and how the actors’ expectations undergo changes with the admission of new actors, new status distinctions, and new tasks in the task situation (Sequencing). The theories are concerned with situations where actors are collectively oriented toward solving a task, or a sequence of distinct tasks, where each task has the valued outcomes of success and failure. A directed graph is constructed to represent a given situation, and makes it possible to compute task performance and reward expectation values for the actors in the situation. The success and failure outcomes of the group’s task, represented by T(+) and T(), and the states of the performance characteristic, C(+)

Spread of Status Value

85

and C(), which are instrumental to the success and failure outcomes of the group’s task, are points in the graph of the situation. The typical situation contains actors, p, o1, o2,yon. Points in the graph represent actors. The actors possess states of status characteristics which are also represented as points in the graph, the possession relation being represented by a line connecting the actor point to the characteristic state point; possession lines are positively signed. An actor may possess the positive or negative states of diffuse or status specific characteristics. Oppositely signed states of the same characteristic, when possessed by actors, are connected by negatively signed lines, called dimensionality lines that represent the mutual exclusion of the other of the two states of the same characteristic. For simplicity only negative signs are displayed in status graphs. Status characteristics are either (a) initially connected to task outcomes by relevance relations, or (b) discriminate between the actors and become connected by relevance relations through processes described by the status characteristics or the reward expectations theories. Paths starting from an actor and connected to a task outcome state are called actor paths of relevance. The length of a path is simply the number of relations in the path. The sign of any such path is the product of the signs of the relations on the path and the sign of the outcome state to which the path is connected. Paths of different lengths contribute different amounts to the expectations of an actor – specifically, shorter paths contribute more to expectations then longer paths. The theory assumes that the strength of a path is given by a function, f(i), of its length, i. This function is an inverse function, that is, the longer the path the less its strength, paths can take on values in the interval (0,1). Examples of different kinds of paths are shown in the Appendix. How their lengths and signs are determined, as well as how their strengths are computed also is demonstrated in the Appendix. If differential rewards for performance are distributed in the situation, then there are additional points in the graph of the situation, representing high and low reward levels, R(+/), and these points are respectively, relevant to positively and negatively evaluated goal objects, GO(+/). The notion of actor paths of relevance also applies to connections of actors to goal-object outcomes. In any given situation actors may be confronted with multiple salient status characteristics, with additional status information specific to the actors, and with information on behaviors that have occurred in their immediate situation. If such information is salient it generates multiple actor

86

JOSEPH BERGER AND M. HAMIT FIS-EK

paths of relevance to outcome states. It is assumed that actors combine these paths in forming their aggregated expectation states in the immediate situation subject to a process we call the principle of organized subsets. The paths are combined within like signed subsets according to the function: f ði [ jÞ ¼ f ðiÞ þ f ðjÞf ðiÞf ðjÞ and given the sign of the paths in the subset. The values of the two subsets are algebraically summed to yield the expectations for the actor. Expectations for actors can be combined in different ways to yield different expectation indices to predict particular kinds of behaviors, like the expectation advantage or the expectation standing.

Concepts from the Status Value Theory2 Status value theory introduces a new type of path of relevance called a ‘‘characteristic path of relevance.’’ A characteristics path of relevance starts with the state of a characteristic, N(a/b), that is non-valued and possessed by an actor who also possesses a number of valued status elements, for example, states of status characteristics or states of task abilities. Depending upon the situation, these paths may be initially connected to task and/ or reward outcomes or they may come to be connected by processes described in the status characteristics or the reward expectation states theories. Status value theory directly applies the concepts that have been previously developed for the sign, the length, and the strength of actor paths of relevance to characteristic paths of relevance. As is true with possessed valued status elements, the states of an initially non-valued characteristic become salient if they are a basis of discrimination or are directly connected to the task outcome states in the situation. The theory then introduces the idea of a process by which the states of N acquire status value because of their association with the valued states of status elements possessed by the actors in the situation.3 By the status value of characteristic N, the theory refers to the relatively positive or negative or high–low evaluations of states of N in terms of honor, esteem, prestige, or more generally, social worthiness. Considering one state, say N(a), of a nonvalued characteristic and asking what combination of paths will increase the positive status value of this state while simultaneously increasing

87

Spread of Status Value

the negative status value of the contrasting state, N(b), the theory asserts that all paths are paths of positive significance which are either: 1. positive paths connecting N(a) to task and reward outcome states, or 2. negative paths connecting N(b) to task and reward outcome states. Similarly paths increasing the negative status value of N(a) and the positive status value of N(b) are paths of negative significance, which are either: 1. positive paths connecting N(b) to task and reward outcome states, or 2. negative paths connecting N(a) to task and reward outcome states. The theory assumes that characteristic paths of relevance will be combined within subsets of similar significance, subject to an attenuation effect, and that the strength of these subsets will be algebraically summed to determine the status value acquired by the non-valued characteristic through association.4

Status Value Assumptions to cover Status Objects and Valued Roles Assumption: Formation of Aggregated Status Value Association. Given that states of an initially non-valued discriminating characteristic, N, or non-valued states of object, O, or the non-valued states of a social role, VR, are connected to outcome states by a set of paths of positive significance of strengths f(i),y, f(n) or a set of paths of negative significance of strengths f(iu),y, f(nu). These paths will first be combined within subsets of similar significance to yield a positive association value, v+, and a negative association value, v, via the following functions: vþ ¼ ½1ð1f ðiÞÞ . . . ð1f ðnÞÞ v ¼ ½1ð1f ði0 ÞÞ . . . ð1f ðn0 ÞÞ

(1)

The aggregated status value acquired by an initially non-valued characteristic, a non-valued state of object, or the states of a social role through association is given by v ¼ vþ þ v

(2)

Assumption: Evaluated states, generalized expectations, and reward levels. Given that the states of an initially non-valued discriminating characteristic N, or the states a non-valued object O, or the non-valued states of

88

JOSEPH BERGER AND M. HAMIT FIS-EK

a social role VR, have become differentially evaluated. Then these states, N(+/), O(+/), VR(+/) will become connected in a consistent manner by lines to the states of generalized expectations if the valued element is from a characteristic, or to states of reward levels in a consistent manner if the valued element is from an object or social role. These lines will be of variable length, and their length is given by a function: l ¼ 1=va

(3)

Assumption: Construction of a diffuse status characteristic, status object, or valued social role. Given a characteristic, N, an object, O, or a social role, VR, whose status value association, v, comes within epsilon of 1. Assume an actor in the group behaves in accord with the expectations generated by this characteristic, object, or social role. If the actor’s behavior is strongly validated by a specific other there is a probability that the element, if it is a characteristic becomes a stable diffuse status characteristic, if it is an object becomes a stable status object, or if it is a role becomes a stable valued social role for the actors involved. The greater the number of specific others validating the actor’s behavior, the greater this probability.

The Creation of Status Objects and Status Valued Roles As noted previously, a major objective in extending the theory of Spread of Status Value to social objects is to be able to produce a transfer process for creating a status object. In addition, we argue that such a transfer process also is applicable to creating a status-valued role. In this context a status object is a social object with states that embody positive or negative status values, and similarly, a status-valued role is a social role with embedded positive or negative status values. Using our extended theoretical assumptions, how can we describe the process by which these transformations can occur? In Fig. 1, we present a graph that is a realization of a status transfer process that is applicable both to the creation of status objects and status-valued roles.5 In Fig. 1, p and o represent the actors in a two-person task group with p possessing O(a), the initially unevaluated a-state of the object or social role and o possessing O(b), the initially unevaluated b-state. In this example, p and o also possess different states of three diffuse status characteristics, D1, D2, D3, with p possessing the highly evaluated states and o the low evaluated states of these characteristics. We assume that each of these states have become relevant in a consistent manner to the positively and negatively

89

Spread of Status Value R(+)

GO(+)

O(a) Γ3(+)

D3(+)

Γ2(+)

D2(+)

P

D1(+)

Γ1(+)

C*(+)

T(+)

o

D1(–)

Γ1(–)

C*(–)

T(–)

Γ2(–)

D2(–)

D3(+)

Γ3(–)

O(b) R(–)

Fig. 1.

GO(–)

The Creation of a Status Object and a Status Valued Role.

evaluated states of generalized performance expectations, G (+/–), through the burden of proof process. G (+/–) become relevant in turn to the high and low states of the task ability, C(+/–) which are connected to the success and failure outcomes on the group task, T (+/–). We further assume that high and low states of the task ability have become relevant in a consistent manner to the high and low reward levels, R (+/–), and these in turn, are relevant to the specific types of goal objects (rewards), GO (+/–), that are distributed in this situation (see Appendix). By the spread of status value processes described in our assumptions, we argue that the high and low status values that are embodied in the diffuse status states possessed, respectively, by p and o will spread to the particular states of the object or role that they possess. As these states accumulate status value they become relevant in a consistent manner to the high and low levels of rewards in the situation. The dashed lines between these states and reward levels represent the fact that these relevance lines are of variable strength. The strength of the relevance lines is a function of the type of

90

JOSEPH BERGER AND M. HAMIT FIS-EK

status state (a low or highly evaluated state), the number, the degree of consistency, and the degree of task relevance of the status characteristics possessed by the actors. As noted, Fig. 1 represents one particular realization of the creation process. For example, the status transfer process may involve many actors with some contributing to a positive evaluation of one state of the object or role while others contributing to a negative evaluation of the second state, the actors may possess different status characteristics, the status characteristics may differ in their degree of relevance, and so forth.6 Nevertheless, the structure described above, with appropriate modifications, is applicable to the creation of goal-objects (specific types of rewards), awards, prizes, and valued social roles of many different types. In summary, Fig. 1 illustrates three core features of the process of creating a status object or status valued role. First, the status value that comes to be attributed to the states of the object or role is a function of the established status elements possessed by individuals possessing that state of the object or role. Second, as the state of the object or role accumulates status value its connections (or linkages) with reward and task outcomes are strengthened. Third, once an object has become a status object or a role has become a status valued role, depending upon the status sign of the state, its possession when salient will either increase or decrease an actor’s task and reward expectations in a situations of action.7

EMPIRICAL ASSESSMENT At present there is recent research that is relevant to the theory that we shall review. In addition we shall consider the question of how to design further tests of the theory. These are the issues we turn to next.

Recent Research Two newly reported studies provide information that bears on the Spread of Status Value theory. The first of these by Shane Thye (2010) studied twoperson same gendered groups whose members interacted with each other in a bargaining exchange situation. Using experimental procedures that are similar to those employed in his earlier study (Thye, 2000), he first created high and low status conditions, and associated different colored chips with different status conditions. In the exchange phase of the study subjects were

Spread of Status Value

91

motivated to gain as many chips as possible in their exchanges. In one set of conditions, subjects were differentiated on a single status characteristic and in a contrasting condition on multiple status characteristics. Given our theoretical arguments above, we would expect to find significant status value effects in the single and multiple status conditions with those in the multiple conditions being stronger than those in the single characteristic conditions. Among other results, Thye found that significantly high status values were attributed by high and low status subjects to the chips held by high status actors in both the single and multiple status conditions. Further, higher status values were attributed by high status subjects to chips of high status actors in multiple status conditions as compared with those in the single status conditions. However this latter result did not hold for low status subjects. In general, this experiment provides partial support of our theoretical expectations. The second study by Walker, Webster, and Bianchi (2011) constitutes an independent test of the initial version of the Status Value theory. In a fourcondition experiment, they studied two-person task groups involving female subjects. All subjects were led to believe that they possessed an unevaluated state, either N(a) or N(b), of a newly identified characteristic. In the first two conditions, subjects learned of the task achievements of referent actors. A referent actor is distinct from the subject but is an actor who in this situation possessed the same or opposite states of the unevaluated characteristic the subject possessed. Subjects were informed that referent actors possessing one type of unevaluated state were uniformly successful while referent actors possessing the second type of an unevaluated state were uniformly unsuccessful at a highly valued task that was independent of the task they were working on. In the second two conditions, subjects were led to believe that referent individuals who possessed one of the unevaluated states of N had attained the highly evaluated states of a number of already established status characteristics and/or reward outcomes while those referent individuals who possessed the second unevaluated state of N had attained the low evaluated states of these established status characteristics and/or reward outcomes. In accord with the Spread of Status Value theory, Walker et al. (2011) reasoned that in each set of conditions, high or low status values will spread to the initially unevaluated states of N depending upon the status elements with which they are associated. Further they reasoned that high and low states of generalized performance expectations would become connected in a consistent manner to these newly evaluated states. They found, as predicted, that differential status beliefs did become associated with the states of N.

92

JOSEPH BERGER AND M. HAMIT FIS-EK

Subjects believed that most people, as was also true of themselves, attribute greater status and leadership and greater competence and knowledge to individuals possessing states of N that had been associated through referent actors with established high valued status elements as compared to individuals possessing states of N that had been associated through referent actors with low valued elements.8 They also found, as predicted, that behavioral differences were associated with these states of N. Individuals possessing states of N that had been associated with high valued status elements or high achievements on the independent task exercised greater influence over individuals possessing states of N that had been associated with low valued elements or low achievements than did the latter over the former. Overall, the experiment provides empirical support for the Spread of State Value theory.9

Further Tests of the Spread of Status Value Theory While the Walker and associates research is significant in providing direct empirical support for the Spread of Status Value theory, nevertheless, at present we are at a very early stage in the empirical assessment of this theory. In particular we are in need of studies that will provide tests of some of the highly precise predictions of the status creation process. We now consider one such test where the objective is to create a very strong linkage between states of N and those of generalized performance expectations. Design for an Experimental Test According to the Spread of Status Value theory, the strength of the status value acquired by the states of an initially unevaluated characteristic, N, is a function of the status sign (positive or negative), the number, the consistency and the degree of task and/or reward relevance of established status elements possessed by the actors. Our theory enables us to determine in a precise manner of how these factors are related to the strength of status value acquired by states of N and its effect on behavior. Assume that in a given situation differential status evaluations of the states of N occur. The theory argues that as this takes place states of high and low generalized performance expectations, G(+/), become connected in a consistent manner with these states of N. Further, as differential status values of the states of N increase, the lines connecting these states with those of generalized performance expectations decrease until they approach a line of length 1, which is the length of the line connecting generalized

93

Spread of Status Value

performance expectations with the states of already established diffuse status characteristics. The theory enables us to determine the relation of the status values acquired by the states of N and the strength of the line connecting these states to states of generalized performance expectations. Our objective in designing this experiment is to create relations between states of N and established status elements that will generate a strong linkage between states of N and generalized performance expectations. In order to test for our precise predictions of this creation process, we need a strongly differentiated status structure from which differential and established status values will spread to initially unevaluated states of N with the resulting line connecting these states with high and low generalized performance expectations being close to one. The design in Fig. 2 is an example of one such structure. In the design represented in Fig. 2, p and o possess the positively and negatively evaluated states of D1 and D2, respectively, which are directly relevant in a consistent manner to the positively and negatively evaluated states of the task characteristic, C(+/). In addition p possesses the success, T(+), and o the failure, T(), outcome states on this group task. As a result p and o’s status positions are based on three directly relevant status elements. Further, these status elements are fully consistent, with p o2

N(a)

D2(+)

D1(+) p

C*(+) T(+)

T(–) o1

C*(–) D1(–)

o3

Fig. 2.

N(b)

D2(–)

The Creation of a Diffuse Status Characteristic.

JOSEPH BERGER AND M. HAMIT FIS-EK

94 O2

N(+)

Γ(+)

C*(+)

T(+)

O3

N(–)

Γ(–)

C*(–)

T(–)

Fig. 3.

Testing for the Status Creation of a Diffuse Status Characteristic.

possessing the positive elements and o possessing the negative elements, and these status elements are strongly relevant to the task outcomes. Our design includes actors, o2 and o3, who also initially possess the different unevaluated states of N. In the first phase of the experiment, these actors simply observe the interaction and outcomes of the behavior of p and o in the task situation. We assume that the observers share the spread of status value process from the status elements possessed by p and o to the states of N, and the states of N acquire the same status values for o2 and o3 that they do for p and o. If so, these observing actors can interact with each other in a second phase of the experiment without D1 and D2 being salient, and without any task to task transfer of expectations since the second phase of the experiment will actually be their first task from the observer’s point of view (see Fig. 3). Since o2 and o3 do not possess the established status elements that are involved in the spread of status value process these established status elements will not affect their behavior on that subsequent task. Consequently their behavior on the subsequent task can be used to test the theory’s precise predictions for this status creation process.10

SUMMARY AND CONCLUDING COMMENTS To briefly recapitulate the theoretical background of our formulation, we start out with the theory of status characteristics and expectation states (Berger et al., 1966), which describes how diffuse status characteristics, possessed by actors in a task situation, determine the power and prestige order of the group and how, in turn, the power and prestige order determines the influence processes within the group. This theory has been elaborated, and developed in a number of different ways as we have described in our introductory statement of theoretical background. A crucial development has been the formalization of the theory

Spread of Status Value

95

(Berger et al., 1977), which has provided the framework for still later developments. One such later development arises out of the question posed by Cecilia Ridgeway and her colleagues (1988) on how status characteristics actually come into existence. Briefly, the answer given by Ridgeway and colleagues is that if actors in a group task situation possess initially nonevaluated states of a discriminating characteristic and high or low levels of valued resources in a consistent manner, they will come to occupy high and low positions in the group’s interaction hierarchy as a function of the resources they possess. Over time the assessment of the behavior of the actors in these high and low hierarchy positions will lead to the association of differential status values and differential status beliefs with the initially nonevaluated states of the discriminating characteristic that the actors possess. As previously noted, Status Construction theory, has received considerable empirical support. Berger and Fis- ek (2006) have formulated a second theory of the creation of status characteristics, which is not an alternative to the Ridgeway’s Status Construction theory, but which invokes mechanisms other than the group’s interaction hierarchy in the status creation process. The basic idea is that when actors are in a group task situation and possess established status valued characteristics and also possess the non-valued states of a discriminating characteristic, status value can transfer from any of the established status valued characteristics to the initially non-valued states of the discriminating characteristic. And as these status values accumulate, high and low generalized performance expectations are predicted to become more strongly connected in a consistent manner to the states of a discriminating characteristic. This gives rise to a new status characteristic that is developed out of its particular associations with already well-established status distinctions that exist in the society or the relevant social group. More recent elaborations of the Status Characteristics theory deal with status elements other then status characteristics that enter the expectation formation process, such as ‘‘valued roles’’ (Fis¸ek et al., 1995) and ‘‘valued objects’’ (Thye, 2000). These developments involving new status elements again raise the question that Ridgeway and colleagues put forth about status characteristics: ‘‘How is it that valued roles and valued objects come into being?’’ That is the question addressed in this extension of the Spread of Status Value theory. The response given in our formulation is that while allowing for differences in the elements involved, status value spreads in basically a similar manner from valued status elements to non-valued roles or non-valued objects as it does from valued status characteristics to nonvalued characteristics in creating new status characteristics.

96

JOSEPH BERGER AND M. HAMIT FIS-EK

Finally, we would like to observe that the initial status characteristics theory (Berger et al., 1966) has undergone considerable development from its original formulation. That theory dealt with two actors discriminated by a single status characteristic, and performing a single task operating within a social or cultural framework. Currently that theory has become a network of interrelated theories that can address situations involving any number of actors (within limits of practicality, of course) who may possess any number of status characteristics, prior task outcome states, goal objects, valued roles, and status elements that can be created in the same situation in which the actors are immediately interacting, e.g. behavior interchange patterns (Fis¸ek, Berger, & Norman, 1991). In addition, the network can address a situation where the actor’s status expectations can evolve across a sequence of tasks that extend through time (see Berger & Webster, 2006). And, as this chapter indicates, this network has begun to deal with questions of how status distinctions of very different types are created. We should add, however, that this observation has to be qualified by the fact that while there is empirical evidence that supports this family of theories, considerably more empirical evidence is needed to more firmly establish the empirical basis of the different theories in this expectation states research program.

NOTES 1. The Appendix is taken with revisions from Berger and Fis- ek (2006). 2. This section is largely a paraphrase of Berger and Fis- ek (2006, pp. 1042–1049). 3. While we focus on association relations in this presentation, our formulation can be applied in a straight forward fashion to other connections between status elements such as when the states of an initially non-valued characteristic or goalobject (see below) is relevant to a valued status element. 4. The general form of the combining principle we use here is similar to that introduced in Berger et al. (1977) to describe the way actors aggregate established status information. It has been used effectively in accounting for relevant empirical research and in developing theoretical extensions of the core status characteristics theory. 5. In what follows, we are not claiming to represent in detail the experimental situation that Thye has used in his research. What we do claim is that within the concepts and principles of our theory that describe the spread of status value, we can construct a version of the transfer process used by Thye to create status valued objects. 6. It should be noted that it may be the case that a status object or a status valued role has only a single evaluated state, and the second, less valued, state is ‘‘not possessed.’’ To represent this situation requires involves a modification of the

97

Spread of Status Value

transfer structure represented in Fig. 1. However, status valued objects and valued roles also may be defined so that differential status valued states are implied if not explicit, for example, leader–follower. 7. There is no assumption that objects and roles can only acquire differential status values by the spread of status value process. The claim being made is that under the appropriate conditions the spread of status value is a sufficient process, but not a logically necessary process. Other factors can lead to differential status values, for example in the case of roles, the nature of the tasks defined within the role or the position of the role in some formal or informal hierarchy in the relevant social unit. 8. Combining the results of Walker and associates with those previously established by Ridgeway on the attribution of status beliefs we can conclude the following: Given that one initially unevaluated category is advantaged over a second in terms of positions in the interaction hierarchy, or task achievements, or status associations, then high or low status beliefs will be assigned to these categories consistent with this relative advantage. 9. For an application of the spread of status value to a situation where relations between an immigrant and a native population are transformed to a situation involving status defined ethnic relations, see Berger and Fis¸ek, ‘‘Immigration groups and the emergence of status inequality: An application of the spread of status value theory’’ (2008). 10. We can imagine a situation where the observers 02, 03, who are the subjects, are given the information at the same time during the experiment on what states of Ds and what states of task success or task failure p and 01 each possess.

ACKNOWLEDGMENT We would like to acknowledge support for Joseph Berger’s work from the Hoover Institution.

REFERENCES Berger, J., Cohen, B. P., & Zelditch, M., Jr. (1966). Status characteristics and expectation states. In J. Berger, M. Zelditch, Jr. & B. Anderson (Eds.), Sociological Theories in Progress (Vol. 1, pp. 29–46). Boston, MA: Houghton Mifflin. Berger, J., & Fis- ek, M. H. (2006). Diffuse status characteristics and the spread of status value: A formal theory. American Journal of Sociology, 111(4), 1038–1079. Berger, J., & Fis¸ek, M. H. (2008). Immigration groups and the emergence of status inequality: An application of the spread of status value theory. A version of this paper was presented at the ISA First World Forum of Sociology in Barcelona, 5–9 September. Berger, J., Fis¸ek, M. H., Norman, R. Z., & Wagner, D. G. (1985). Formation of reward expectations in status situations. In J. Berger & M. Zelditch, Jr. (Eds.), Status, rewards, and influence (pp. 215–261). San Francisco, CA: Jossey-Bass. Berger, J., Fis- ek, M. H., Norman, R. Z., & Zelditch, M., Jr. (1977). Status characteristics and social interaction. New York, NY: Elsevier.

98

JOSEPH BERGER AND M. HAMIT FIS-EK

Berger, J., Ridgeway, C. L., & Zelditch, M. (2002). Construction of status and referential structures. Sociological Theory, 20(2), 157–179. Berger, J., & Webster, M., Jr. (2006). Expectations, status and behavior. In P. J. Burke (Ed.), Contemporary social psychological theories (pp. 268–300). Stanford, CA: Stanford University Press. Berger, J., Zelditch, M., Jr., Anderson, B., & Cohen, B. P. (1972). Structural aspects of distributive justice. In J. Berger, M. Zelditch, Jr. & B. Anderson (Eds.), Sociological theories in progress (Vol. 2, pp. 119–146). Boston, MA: Houghton Mifflin. Emerson, R. M. (1972a). Exchange theory part I: A psychological basis for social exchange. In J. Berger, M. Zelditch, Jr. & B. Anderson (Eds.), Sociological theories in progress (Vol. 2, pp. 38–57). Boston, MA: Houghton Mifflin. Emerson, R. M. (1972b). Exchange theory part II: Exchange relations and networks. In J. Berger, M. Zelditch, Jr. & B. Anderson (Eds.), Sociological theories in progress (Vol. 2, pp. 58–87). Boston, MA: Houghton Mifflin. Fis- ek, M. H., Berger, J., & Norman, R. Z. (1991). Participation in heterogeneous and homogeneous groups: A theoretical integration. American Journal of Sociology, 97, 721–746. Fis¸ek, M. H., Berger, J., & Norman, R. Z. (1995). Evaluations and the formation of expectations. American Journal of Sociology, 101(3), 721–746. Fisek, M. H., Norman, R. Z., & Nelson-Kilger, M. (1992). Status characteristics and expectation states theory: A priori model parameters and test. Journal of Mathematical Sociology, 16, 285–303. Jackson, L. M., Esses, V. M., & Burris, C. T. (2001). Contemporary sexism and discrimination: The importance of respect for men and women. Personality and Social Psychology Bulletin, 27, 48–61. Lovaglia, M. J., Skvoretz, J., Willer, D., & Markovsky, B. (1995). Negotiated outcomes in social exchange networks. Social Forces, 74, 123–155. Mark, N. P., Smith-Lovin, L., & Ridgeway, C. L. (2009). Why do nominal characteristics acquire status value? A minimal explanation for status construction. American Journal of Sociology, 115(3), 832–862. Ridgeway, C. L. (1991). The social construction of status value: Gender and other nominal characteristics. Social Forces, 70, 367–386. Ridgeway, C. L., & Balkwell, J. W. (1997). Group processes and the diffusion of status-value beliefs. Social Psychology Quarterly, 60, 14–31. Ridgeway, C. L., Boyle, E. H., Kuipers, K. J., & Robinson, D. T. (1988). How do status beliefs develop? The role of resources and interactional experience. American Sociological Review, 63(3), 331–350. Ridgeway, C. L., & Correll, S. J. (2006). Consensus and the creation of status beliefs. Social Forces, 85(1), 431–453. Ridgeway, C. L., & Ericson, K. G. (2000). Creating and spreading status beliefs. American Journal of Sociology, 106, 579–615. Ridgeway, C. L., Li, Y. E., Erickson, K. G., Backor, K., & Tinkler, J. E. (2009). How easily does a social difference become a status distinction? Gender matters. American Sociological Review, 74(1), 44–62. Thye, S. R. (2000). A status value theory of power in exchange relations. American Sociological Review, 65, 407–432.

Spread of Status Value

99

Thye, S. R. (2010). Status, power and mechanisms of micro stratification: Theory and new experimental evidence. Unpublished manuscript, University of South Carolina, South Carolina. Veblen, T. (1899). The theory of the leisure class. New York, NY: Macmillan. Walker, L. S., Webster, M., & Bianchi, A. J. (2011). Testing the spread of status value theory. Social Science Research, 40, 1652–1663. Webster, M., Jr., & Hysom, S. J. (1998). Creating status characteristics. American Sociological Review, 63, 351–378. Willer, D. (1981). The basic concepts of elementary theory. In D. Willer & B. Anderson (Eds.), Networks, exchange, and coercion (pp. 25–53). New York, NY: Elsevier.

100

JOSEPH BERGER AND M. HAMIT FIS-EK

APPENDIX: CORE THEORY AND EXTENSIONS In this appendix we first give a full statement of the assumptions of the core theory of status characteristics and expectation sates together with the auxiliary assumptions introduced by the reward expectations and the evolution of expectations extensions of the core theory. We then use the situational graph given in (Fig. A.1), and a further development of this graph given in Fig. A.2 to describe the applications of these assumptions. We also demonstrate the computation of expectation advantage of one actor over another and the status value association of an initially non-valued characteristic using the earlier formulations presented here together with the current developments presented in this chapter.

Assumption: Salience 1. Given existing paths connecting an interactant to outcome states of the group task the elements and the relations in these paths become salient in the task situation. 2. Given status elements that provide a basis for discrimination between the interactants the states of these characteristics become salient in the task situation.

Assumption: Burden of Proof Given that a salient status element possessed by an actor is not connected to a task outcome state, if the element is 1. a diffuse status characteristic, then the associated general expectation state will be activated, and it will become relevant to the similarly valued state of the instrumental ability relevant to the task, 2. a specific status characteristic, then its relevant task outcome state will be activated and will become relevant to the similarly evaluates state of abstract task ability, and the latter will become relevant to the similarly evaluated task outcome state, or 3. a behavior interchange pattern, then its relevant status typification states will be activated and will become relevant to similarly evaluated states of abstract task ability, and the latter will become relevant to similarly evaluated task outcome states.

101

Spread of Status Value GO(+)

R(+)

N(a)

p

o

D(+)

D(–)

N(b)

Γ+

C1*(+)

T1(+)

Γ–

C1*(–)

T1(–)

R(–)

GO(–)

Fig. A.1. Male with Preference for Klee and Female with Preference for Kandinsky Work Together to Prepare a Scenario for an Advertising TV Clip. Rewards to be Given on Completion of the Task.

Assumption: Sequence of Structure Completion A given structure will be developed for the interactants, through the processes of salience and burden of proof. If a noninteractant later becomes an interactant, or a new actor enters the situation, or a new task is

JOSEPH BERGER AND M. HAMIT FIS-EK

102

GO(+)

R(+)

Γ+

N(a)

C2*(+) p

D(+)

o

D(–)

Γ+

C1*(+)

C1*(–)

Γ+

T1(+)

T1(–) C2*(–)

N(b)

Γ–

T2(+)

T2(–)

R(–)

GO(–)

Fig. A.2. Male with Preference for Klee and Female with Preference for Kandinsky, after the Scenario Task, Work on a Set of Cartoons to be Used in advertising a Different Product. Rewards to be Given on Completion of the Task.

introduced, or a new behavior interchange pattern comes into existence, or new imputations of status characteristics are made, the structure will be further developed through the operation of the same processes. For any actor, those parts of the actor’s structure previously completed will remain while the actor is in the given situation.

Assumption: Formation of Aggregated Expectation States Given that an actor x is connected to states of an outcome a(7) (task or reward) by sets of positive paths of strengths f(i), y, f(n) and negative paths of strengths f(iu), y, f(nu). These paths will first be combined within

103

Spread of Status Value

like-signed subsets to yield a positive paths value, eaðþÞ x , and a negative path value, eaðÞ via the following function: x ¼ 1½1f ðiÞ . . . ½1f ðnÞ eaðþÞ x (A.1) eaðÞ ¼ ð1½1f ði0 Þ . . . ½1f ðn0 ÞÞ x The aggregated expectation state is then given by þ eaðÞ eax ¼ eaðxÞ x x

(A.2)

Assumption: Basic Expectation Assumption Given that p has formed aggregated expectation states for self and other, and that p’s power and prestige position relative to o will be a direct continuous function of p’s expectation advantage over o.

Auxiliary Assumption: Activation of Referential Structures Given status characteristics that are salient in situation S, if these characteristics are referentially associated with states of reward levels, relevance relations will come to be induced between states of the characteristics and similarly evaluated states of the reward levels in S.

Auxiliary Assumption: Relevance of Task Outcome States Given any successive group tasks that are not dissociated or inversely related to each other. If on completion of the first task, the actor engages in the second task, then relevance relations will be established in a consistent manner between the outcome states of the two tasks. Fig. A.1 gives the situational graph for a hypothetical situation – a male Klee fan and a female Kandinsky fan are writing a scenario together for an TV advertising clip, and rewards are to be given for task performance. We now want to describe how this situational graph is determined by our formulation. First any situational graph starts out with a minimal initial structure. The minimal structure consists of one point for each actor (at least

104

JOSEPH BERGER AND M. HAMIT FIS-EK

two of them), and a task outcome structure which consists of the task outcome states, T(+) and T(), standing for task success and failure respectively, and the states of the instrumental characteristic (i.e., the ability necessary to perform the task) C(+) and C(), standing for high and low ability states respectively. The states of the instrumental characteristic are relevant to the similarly signed states of task outcomes. If rewards are to be given in the situation the initial graph also includes the reward outcome structure which consists of the reward outcome states GO(+) and GO() standing for the more valued and less valued rewards respectively, and the states of ‘‘reward level’’ (an abstract conception of high and low rewards), R(+) and R(), standing for high and low rewards respectively. The states of reward levels are relevant to similarly signed states of reward outcomes. We also assume that in any situation where rewards are allocated, the ‘‘ability referential structure’’ will be automatically salient, so that states of the instrumental characteristic will be relevant to the similarly signed states of reward level. All the elements of this initial structure can be seen in Fig. A.1. Given the initial structure, the states of the diffuse status characteristic, gender, D(+) and D(), standing for male and female respectively, and the states of the non-valued characteristic, N(a) and N(b), standing for Klee preference and Kandinsky preference respectively, became salient through the action of the salience process because they discriminate the actors. The actors are connected to these states by possession lines. Since gender is not initially task relevant the burden-of-proof process comes into play the generalized expectations associated with gender are activated and burden of proof paths are established connecting the actors to the task outcomes. Furthermore we assume that gender is referentially connected to reward levels and therefore a categorical referential structure for gender will become salient and the states of D will be relevant to reward levels. The resulting situational graph is as given in Fig. A.1. We now need to determine the paths connecting the actors to the outcome states in order to determine their expectations according to the formation of aggregated expectations assumption. We first look at task expectations, and see that all paths connecting p to task outcome states, T(7), are positive. We see that p is task connected with two paths of lengths 4 and 5 going through D and C states and two other paths of the same lengths going through D and R states, so that etðþÞ ¼ 1½1f ð4Þ ½1f ð5Þ ½1f ð4Þ ½1f ð5Þ p

105

Spread of Status Value

We need actual numerical values for the strengths of paths of different lengths to compute the actual expectation value. Fis- ek, Norman, and Nelson-Kilger (1992) have shown that the function giving these values has a specific form and the parameters of the form can be determined by making theoretical assumptions about path lengths. We use the function proposed by them: f ðiÞ ¼ 1 expð2:618ð2iÞ Þ

(A.3)

According to this function: ¼ :332 f ð4Þ ¼ :1358 and f ð5Þ ¼ :0542 and therefore etðþÞ p Since p does not have any negative paths this is actually p’s task expectations, etp , and because the graph structure happens to be symmetric, o’s expectations have exactly the same magnitude but is negative in sign. Therefore, p’s expectation advantage over o is tðþÞ etpo ¼ etðþÞ ¼ :332ð:332Þ ¼ :664 p eo

Turning our attention to the reward expectation advantage, we see that the same general conditions hold, all paths connecting p to reward outcome states, GO(7), are positive and symmetric, that is o will have exactly the same expectation value as p except for sign. P is connected to reward outcome states with two paths of lengths 3 and 4 through R(7), and two paths of lengths 5 and 6 through C(7). Thus the reward expectations of p is given by erðþÞ ¼ 1½1f ð3Þ ½1f ð4Þ ½1f ð5Þ ½1f ð6Þ p We need the numerical values of paths of lengths of 3 and 6 for this computation: ¼ :454 f ð3Þ ¼ :3175 and f ð6Þ ¼ :0211 and therefore erðþÞ p By the same reasoning as in the case of task expectations, the reward expectation advantage of p over o is erpo ¼ :908 To compute the status value association of N(x) we need to note several points. The first is that all paths connecting N(x) to the outcome states (both task and reward) are of positive significance, that is all paths connecting N(a) to outcome states are positive and all paths connecting N(b) to the

106

JOSEPH BERGER AND M. HAMIT FIS-EK

outcome states are negative. Therefore all paths will be combined together according to our combining function. The second thing to note is that each path connecting an actor to an outcome state constitutes, a path connecting N(x) to an outcome state of one greater length. Thus there were two paths of length 4 and two paths of length 5 connecting p to T(7),constituting two paths of lengths 5 and two paths of length 6 connecting N(x) to T(7). Since o was task connected by paths of the same lengths which were of negative sign for task expectations, but are of positive significance for status value association, we have four paths of length 5 and four paths of length 6 connecting N(x) to task outcomes. By the same reasoning there are two paths of length 4 and two paths of length 5, and two paths of length 6 (the paths of length 6 connecting actors to reward outcomes drop out as they become paths of length 7) connecting N(x) to GO(7). In total there are six paths of length 5 and six paths of length 6 contributing to the status value association of N(x), vN: vN ¼ 1½1f ð4Þ2 ½1f ð5Þ6 ½1f ð6Þ6 ¼ :529 Fig. A.2 gives the situational structure after the actors move on to a new task. By the sequence of structure completion assumption all of the structure developed on the first task remains in the graph, and the salient ability referential structure establishes to connections of reward levels to the instrumental ability states of the new task. Furthermore the burden of proof process establishes the connection of gender to the new task and the relevance of task outcome states provides the relevance connection between the two task outcome states (Fig. A.2). The new part of the structure due to the current formulation is the connection of N(x) to the second task because it has acquired status value on the first task. The evaluated states and generalized expectations assumption states that when evaluate states are salient then generalized expectations will become salient and task-connect the evaluated states. However the line connecting the evaluated states to the generalized expectation states, will not be a standard length 1 line but a variable length line depending on the status value association of N(x). These variable length lines are represented by dashed lines in the graph. The length of the line is given by l ¼ 1=va

(A.4)

Spread of Status Value

107

The parameter a is to be empirically estimated, for our illustrative purposes to carry out the computations we will simply assume its value to be 1. Therefore: vN ¼ :529; a ¼ 1 and l ¼ 1=:529 ¼ 1:889 Therefore p is connected to the task outcome with two positive paths of lengths 4.889 and 5.889 and o with two negative lines of the same lengths through the states of N. The strengths of these paths can also be obtained from f(i) as given above; these values are .0601 and .0234. We will not repeat the exercise of tracing paths and computing expectations since no new concepts are involved. We give below, the values of the expectation advantages and status value association on the second task: etpo ¼ :864; erpo ¼ 1:012; vN ¼ :568 We should note that the paths through the evaluated states N(x) are included in the computations of the expectation advantages, but not in the computation of status value association, since inclusion would bring circularity into the system.

THE ROLE OF UNCERTAINTY IN SOCIAL INFLUENCE$ Scott V. Savage, David Melamed and Aaron Vincent ABSTRACT Purpose – This study examines how the distribution of opinions and social status combine in a collectively oriented task group to affect perceptions about the correctness of a final decision. Design/methodology/approach – We relied on data from a controlled laboratory experiment to test a series of theoretically derived hypotheses. Findings – The study shows that both the distribution of opinions and status affect perceptions of correctness. It also establishes that the effects of status on uncertainty are strongest when the group is initially evenly split about the correctness of an opinion, and that like the distribution of opinions, the effects of status on uncertainty are curvilinear. Research limitations/implications – Previous research shows that by integrating research on faction sizes with status characteristics theory (SCT), more accurate predictions of social influence are possible. Assumed therein is that people use information about the distribution of opinions and status to reduce uncertainty about correctness of a choice. $

Portions of this paper were presented at the 2013 Annual Meeting of the Southern Sociological Society.

Advances in Group Processes: Thirtieth Anniversary Edition Advances in Group Processes, Volume 30, 109–129 Copyright r 2013 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 0882-6145/doi:10.1108/S0882-6145(2013)0000030008

109

110

SCOTT V. SAVAGE ET AL.

The current study establishes this point empirically by examining the effects of the distribution of opinions and status in a four-person, collectively oriented task group. Future research should consider groups of different sizes and other moderating factors. Originality/value – This study advances and elaborates upon previous research on social influence that integrates research on faction sizes with SCT. Keywords: Social influence; status; faction size; uncertainty

What makes people more or less susceptible to the influence of others? Status characteristics theory (SCT) argues that one’s relative status affects whether a collectively and task-oriented actor is likely to succumb to the opinions of another (Berger, Fis- ek, Norman, & Zelditch, 1977; Berger & Webster, 2006; Correll & Ridgeway, 2003). Social impact theory offers an alternative explanation, contending that, in addition to other factors, different numerical distributions of opinions in groups larger than two exert varying amounts of influence on individuals (DiFonzo et al., 2013; Latane, 1981). Tests of these two theories confirm that both status and numbers affect social influence. Until recently, however, research on the influential effects of status and research on the influential effects of numbers have occurred independently of one another. Melamed and Savage (2013) sought to fill this void with a new theoretical model of social influence that examines the effects of status in collectively oriented task groups larger than two. The model states that the effects of status become increasingly important as the distribution of opinions about the correctness of a choice become evenly divided. When equal numbers of people hold differing opinions about the correctness of an opinion, numbers provide little insight into the appropriateness of an opinion, and additional information is necessary to reduce any uncertainty about its correctness. Status characteristics, or attributes of individuals that carry widely held cultural beliefs that attach greater value and competence to one category of the attribute than the other, serve this purpose. Critical to the theoretical model, then, is uncertainty, as it is the uncertainty signaled by an even numerical split that motivates actors to consider status when evaluating the correctness of an opinion. The initial test of the model, however, did not assess actual perceptions of uncertainty and instead relied on previous research

The Role of Uncertainty

111

(e.g., Chaiken & Stangor, 1987; Festinger, 1954; Moscovici, Lage, & Naffrenchoux, 1969) to assume its presence. Not evaluating perceptions of uncertainty left a number of questions unanswered: Do people actually feel more uncertain about the correctness of a final opinion when equal numbers of people hold differing initial opinions, does this moderate the effect of status, and if so, how? Finally, how does changing one’s mind affect one’s confidence in the correctness of a final choice? In this chapter, we rely on a laboratory experiment to answer these questions empirically. We also examine additional relationships about how the presence of others affects perceptions of uncertainty. Below we flesh out our predictions, describe the experiment designed to test our hypotheses, and summarize the results. We close with a discussion of what these findings mean for investigations of social influence and social life more generally.

UNCERTAINTY AND SOCIAL INFLUENCE Uncertainty is a powerful motivator, and much social scientific research shows this (see, for instance, Savage and Bergstrand, 2013). The new institutional theory of organizations, for example, posits that organizations model themselves after others in a seemingly rational attempt to respond to uncertainty (DiMaggio & Powell, 1983). Similarly, social identity theorists assert that people enact social identities in an effort to cope with uncertainty about self and others (Hogg, 2007), and Kollock (1994) contends that uncertainty is the force that drives cohesion and stability in exchange networks. Thus, even though people cannot achieve certainty (Pollock, 2003), they nonetheless rely on social cues to enhance perceptions of it (Dewey, 1929 [2005]). The motivation to reduce uncertainty fits well with Festinger’s claim that humans have a ‘‘drive for self-evaluation’’ (1954, p. 138) and this drive requires social comparisons (Festinger, 1954; Suls & Wheeler, 2000). In groups, when these comparisons produce discrepancies in opinions there is a strong motivation for individual group members to try to resolve these inconsistencies (Heider, 1958). This can happen in one of two ways: (1) an individual can change his or her position or (2) the individual can attempt to change the position of others (Festinger, 1954). Whether a group member remains recalcitrant depends in part on his or her confidence in that position, and absent clear confirmation of the position’s correctness, this confidence will come from social comparisons with group members. Two pieces of information that individuals use to compare themselves with others are status (Berger, Zelditch, & Cohen, 1966; Berger et al., 1977)

112

SCOTT V. SAVAGE ET AL.

and the distribution of opinions (Latane, 1981; Mullen, 1983; Tanford & Penrod, 1984; Tormala & DeSensi, 2009). SCT holds that in collectively oriented and task-focused groups, people will evaluate the relative competence of self and others using status characteristics or attributes with widely held cultural beliefs about the value and utility of the different categories that comprise those attributes. The theory does not assume that this will be a conscious process, but that behavior will unfold as if it were. Status-based evaluations, in turn, affect behavioral inequalities in small groups (Berger et al., 1966, 1977). Much research confirms the general logic of the theory (for a review, see Berger & Webster, 2006), with a number of studies showing that the theory can predict social influence in groups of two (e.g., Berger, Cohen, & Zelditch, 1972; Moore, 1968; Pugh & Wahrman, 1983; Webster & Driskell, 1978). A second piece of information that people use to evaluate their opinions with those around them is the numerical distribution of opinions. Psychological research reveals a clear relationship between faction size and an individual’s propensity to go along with the opinions of others (Asch, 1951, 1956; Latane, 1981; Moscovici, 1980; Mullen, 1983; Tanford & Penrod, 1984). As the numerical distribution of opinions shifts in favor of an opinion, individuals become increasingly likely to support that opinion. Although some recent research in psychology has examined how other factors interface with the distribution of opinions (e.g., Gordijn, de Vries, & de Dreu, 2002; Phillips, 2003), this research has generally ignored how faction sizes interface with status processes (c.f., Kalkhoff, Friedkin, & Johnsen, 2010). Thus, in an effort to reduce uncertainty, group members compare themselves to one another, and both relative status and the numerical distribution of opinions inform these comparisons. Depending on where one stands in the status hierarchy of the group and on the number of individuals who agree with him or her, an individual may feel more or less uncertainty about an opinion and correspondingly may be more or less susceptible to social influence or changes to an individual’s thoughts or behaviors as a result of social interaction (Rashotte, 2007). Despite this, very little experimental research has analyzed how status processes affect uncertainty and social influence in groups larger than two (Kalkhoff et al., 2010; Ridgeway, 1984). Two exceptions to this are recent works by Kalkhoff et al. (2010) and Melamed and Savage (2013). Kalkhoff, Friedkin, and Johnsen integrate SCT and social influence network theory to demonstrate how social status weights the relative importance of network contacts in terms of social

The Role of Uncertainty

113

influence. Melamed and Savage (2013) integrate SCT with the research on how the distribution of opinions shapes influence processes to generate novel predictions. A key difference between these two models is that the latter focuses on the relative importance of social status, given a distribution of opinions (i.e., it is nonlinear with respect to status). It is this second model that concerns us here.

UNCERTAINTY WITH RESPECT TO STATUS AND FACTIONS Melamed and Savage’s (2013) model of social influence draws on SCT and research on faction sizes to argue that both status and numbers affect social influence. Specifically, the theoretical model relies on the logic of SCT to explain how status processes affect influence in collectively oriented task groups larger than two. It also restricts itself to those situations where individuals know both the initial opinions and the status of other group members before making a final choice. An example of such a situation would be each iteration of a jury taking an open vote. When everyone agrees, there is no need for a second vote, but when there is disagreement, everyone will know the opinions of everyone else prior to any subsequent votes. In such situations, the unified model of social influence holds that people will rely on the opinion and status of others when determining the correct solution to a binary nonveridical task. Consistent with current research on social cognition (e.g., Fiske & Taylor, 1991), Melamed and Savage (2013) posited that people will process the distribution of opinions within the group first. When consensus holds, there is little uncertainty about the correct choice and no pressure to change one’s mind. In these situations, there is no need for actors to consider the status distribution of the group (i.e., consensus creates an absorbing state). When the distribution of opinions deviates from consensus, however, the correctness of an opinion becomes increasingly murky as the factions for and against a particular position become evenly divided. Uncertainty about the correctness of an opinion motivates actors to seek out additional information, such as status information, to evaluate the merits of the alternative choices. Research shows that people use status characteristics to evaluate the competence of self and others (Cuddy, Norton, & Fiske, 2005; Fiske, 2000; Ridgeway, 1997). As such, Melamed and Savage (2013) argued

114

SCOTT V. SAVAGE ET AL.

people will rely on the status distribution of those holding competing opinions to assess the correctness of a choice and that they will increasingly do this as the distribution of opinions becomes more even. At the heart of their model then is the notion that people in collectively oriented task groups attempt to minimize the uncertainty associated with a nonveridical binary choice and discriminatorily rely on information about numbers and status to achieve this end. The initial test of the model, however, did not establish empirically (i) that individuals use information about the distribution of opinions to evaluate how certain they are about the correctness of a final decision, (ii) that individuals feel more uncertain about a final choice when the initial distribution of opinions about that choice is evenly split, (iii) that individuals also use status to evaluate how certain they are about the correctness of a final choice, and (iv) that the effects of status on uncertainty will be strongest when the group is initially evenly split about the correctness of an opinion. For Melamed and Savage’s unified model of social influence to hold up, all of these points need to be supported. Thus, in line with the theoretical model, we hypothesize: Hypothesis 1. Absent information about the status of others, receiving information about the initial opinions of others will decrease confidence in one’s final choice if the group is evenly split over the correctness of an opinion. Hypothesis 2. Net of the initial opinions of others, status information will increase one’s confidence in a final choice if their faction is relatively high in status, and status information will decrease one’s confidence in a final choice if their faction is relatively low in status. Hypothesis 3. Information about the distribution of opinions will moderate the relationship between relative status and confidence in the correctness of a final opinion, with status mattering most when the distribution of opinions is evenly split. These hypotheses suggest that the effect of status will be strongest when the division of opinions in the group reaches a 50/50 split. But, like the distribution of numbers, the uncertainty reducing effects of status depend on the relative distribution of status across factions. When status becomes more evenly split, individuals cannot use it to evaluate the correctness of an opinion. Thus, like numbers, we posit a curvilinear relationship between status and perceptions of uncertainty. People will be more certain about a final choice if the relative status of their faction is either high or low.

115

The Role of Uncertainty

They will express less certainty the closer it is to the middle. Formally, we expect the following: Hypothesis 4. The effects of relative status on perceptions of uncertainty about the correctness of a final opinion will be curvilinear. These hypotheses document the expected relationships between relative status, the distribution of initial opinions, and perceptions of uncertainty about the correctness of a choice. Recall, however, that Melamed and Savage’s model is ultimately concerned with social influence, arguing that whether an individual succumbs to social influence depends on whether the distribution of initial opinions and relative status convince the individual that the alternative is in fact the proper choice. We therefore hypothesize: Hypothesis 5. Individuals will be more certain about their final opinion if it is consistent with their initial opinion.

METHOD We test our hypotheses with data from a controlled laboratory experiment. The experimental setting was a variant of the standard experimental setting for tests of SCT (Berger et al., 1977, pp. 43–48; Berger, 2007) and involved participants working on an experimental task to earn money. The experiment had two phases. In the first phase, participants individually completed 20 trials of a fictitious ability test. In the second phase, participants engaged in 25 trials of a modified version of the same test, with the amount of information they received about the initial opinions and relative status of others varying by condition.

Design and Participants To test the effects of numbers and status on perceptions of uncertainty, our experiment manipulated the amount of information participants had about the opinions and status of others. In all conditions, participants were led to believe they were working in groups of four, and that they would be paid for the group’s performance during phase two. The group size allows general comparisons with past research (Melamed & Savage, 2013). The experiment was a two-phase experiment. In the first, participants took an individualized ability test, and in the second, they completed a second ability test, with

116

SCOTT V. SAVAGE ET AL.

some conditions requiring that participants work together on the test. There were three experimental conditions. In the first condition, each participant was led to believe that he or she would work independently during phase two of the experiment. The second condition differed from the first, in that prior to making a final decision on each trial in the second phase of the experiment (see below) participants received information about how many of their partners agreed with their initial choice. The sequence of whether zero, one, two, or three partners agreed with the participant was randomly generated and consistent for every participant. The third condition was similar to the second except that in this condition participants also knew the relative status of self and others. Thus, the manipulations in this experiment are both between subjects (information about the opinions and status of others) and within subjects (the number in agreement). Participants were undergraduate students from a large public university. Sixty-four students participated in the experiment. We excluded data from four participants because these participants failed to meet three criteria: (i) if participants did not try to get each trial of the experimental task correct, (ii) if they did not know in condition 3 how many trials they or their group members got correct in phase one (see below), and (iii) if participants in conditions 2 and 3 did not consider their partners’ initial opinions when making their final decisions in phase two of the experiment, then the participants’ data were not used. As in SCT, we assume individuals in groups will be both task and collectively oriented. When these conditions were not met, the data from those participants was unusable and was replaced with new data. We replaced cases in a random order. This procedure resulted in usable data from 60 students, with 20 of these participants assigned to each of the three conditions. Twenty-four of the students were male and 36 were female. Sensitivity analyses showed no differences between men and women.

Procedures Upon arriving at the experimental laboratory, participants were directed to isolated rooms and informed that they would be participating in a two-phase, computer-mediated experiment with several partners. In reality, partners were simulated actors that behaved in predetermined ways. Multiple participants were scheduled at the same time, but they did not see each other to avoid nonmanipulated status characteristics from becoming salient. After a brief introduction to the study, the experimenter left the room and participants received computer-mediated instructions from that point forward.

117

The Role of Uncertainty

The instructions introduced participants to the experimental task: contrast sensitivity. Contrast sensitivity was described as a relatively new ability that is not associated with other abilities such as mathematical or verbal skills. Participants were shown a rectangular image, composed of black and white areas, and were asked to select which color constituted the majority of the image. They were told that there is a correct answer to each trial and that the test in phase one, which was designed for student populations, would measure the presence of contrast sensitivity ability. In reality, contrast sensitivity is a bogus ability and there is no correct answer (Moore, 1968). To motivate participants in these conditions to be both task and collectively oriented, we made earnings contingent on the subject’s performance and also the performance of their partners. Instructions indicated that the minimum an individual could earn was $6.00 but that each correct answer in phase two of the experiment would result in them earning more money. Specifically, participants were told that they could earn an additional $.125 for each trial they got correct in phase two and $.125 for each trial that each of their three partners got correct in phase two. This operationalization of valued outcomes fits with SCT’s assumption of actors working on a valued task in a manner consistent with past experimental work that operationalizes valued outcomes with monetary rewards (e.g., Markovsky, Willer, & Patton, 1988; Molm, Whitham, & Melamed, 2012; Ridgeway & Correll, 2006). Finally, the instructions for conditions 2 and 3 were used to create the initial conditions of our theoretical model. In condition 1, participants knew nothing about the choices their partners made. In conditions 2 and 3, however, participants were led to believe that in phase two they would be working with several partners over a computer network on the contrast sensitivity task. The instructions for conditions 2 and 3 described the team portion of the experiment as a ‘‘critical choice’’ situation in which taking others’ opinions into consideration leads to an increased likelihood of making a correct decision. This portion of the instructions was intended to increase participants’ collective orientation for these two conditions.

Manipulations The between subject manipulation in this experiment was the information subjects had about the opinions and status of others. In condition 1, participants were told that they would be participating in the experiment at the same time as three other students, but they received no information

118

SCOTT V. SAVAGE ET AL.

about the relative status or initial opinions of those other students. Thus, while participants believed that their earnings would depend on both how they as well as their partners did, they had no information about their partners. In condition 2, participants received information about the initial opinions of the three partners prior to making a final decision in phase two of the experiment. The number of partners who agreed with the participant’s initial opinion in phase two of the experiment was a within participants manipulation. No matter what the participant’s initial choice was, the program had some of the others agree with the participant in a systematic fashion. We randomly assigned zero, one, two, or all three of the others to agree with the participant’s initial opinion on each of the 25 trials. The pattern of agreement was constant across all subjects. Condition 3 was the same as condition 2, but it also used feedback from the phase one contrast sensitivity test to manipulate status between the participant and the three partners. Regardless of the condition, phase one of the experiment had participants respond to 20 contrast sensitivity questions, so scores ostensibly ranged from 0 to 20. All participants received a score of 12, one less than the national average. They were also told that scores between 0 and 10 indicate a poor performance, scores between 11 and 15 indicate an average performance, and scores between 16 and 20 indicate a superior ability on the test (see Berger & Fis- ek, 1970). In condition 3, we made participants aware of their partners’ performance on the test. The two high status partners were given scores of 17 and 19, respectively. The low status partner received a score of 7. Thus, as the number of partners who agreed with the participant changed, the relative status of the participant’s faction also changed.

MEASURES AND MODELING STRATEGY We designed the experiment to determine whether information about the numerical distribution of opinions and relative status affects perceptions of uncertainty with a final choice. The dependent variable for our analysis was certainty in the correctness of a final choice. We measured this with a threeitem scale. Questions asked participants to report how certain, sure, and confident they were in the final choice. Answers to each of these items ranged in value from 1 to 9, with higher scores indicating that participants were more certain, more sure, and more confident in the correctness of their final choice. We created a highly reliable certainty scale from these items

119

The Role of Uncertainty

(a=.97). To examine how the numerical distribution of opinions and relative status affect perceptions of certainty about a final choice, we considered two measures. We measured the numerical distribution of opinions by counting the number of individuals who held the participant’s initial opinion. This measure ranges in value from 1 to 4. Scores of 1 indicate that only the participant holds an opinion. Scores of 4 indicate consensus. To measure relative status in condition 3, we calculated the expectation standing (Fis- ek, Berger, & Norman, 1991) for each member of the fourperson group. At the trial level, we summed the expectation standings for the participant and the partners that agreed with the participant’s initial opinion, and then divided that value by the total number in agreement (e.g., if two others agreed with P then the sum of the expectation standing was divided by three) in order to isolate, as best as possible, the effect of status from the effect of numbers. Thus, the total expectation standing of P’s faction represents the proportion of the total expectations in the situation for that faction by placing the faction on the expectation dimension relative to the other faction (Fis- ek et al., 1991, p. 121). This measure captures the influential effect of status, net of numbers, by quantifying the amount of status that is consistent with the participant’s initial opinion and placing it in relation to the rest of the status in the situation. Although there are three methods for calculating expectation state values and expectation standings, we present results using the Balkwell (1991) estimates. Given that both the number of partners who agree and the status of those partners vary at the trial level, the expectation standings also vary at the trial. Table 1 presents the various possibilities for expectation standings in agreement with P. In addition to these two measures, we also consider participants’ perceptions of certainty about their initial opinion. Identical to the measure for perceptions of certainty regarding their final opinion, this measure was

Table 1.

Variation in Expectation Standings in Agreement with P’s Initial Opinion For Trials in Condition 3.

Number and Status of Os in Agreement 0 1 1 1 2

Expectation Standing

Trials

.03 .06 .5 .53 .97

6, 13, 15, 20 12, 24 4, 7, 8, 17, 19, 23, 25 1, 3, 11, 16, 18, 21, 24 10

Low Status High Status Low Status; 1 High Status High Status

Consensus occurred on trials 2, 5, 9, 14, and 22.

120

SCOTT V. SAVAGE ET AL.

created from three questions that asked participants how certain, sure, and confident they were about their initial choice. Responses could range in value from 1 to 9 with scores closer to 9 meaning the participant was more certain about the correctness of the choice. Averaging responses to these three questions resulted in a final certainty scale with an alpha reliability of .99. Finally, we also recorded whether the participant stayed with the initial choice (1) or changed it (0). Using these measures we tested our five hypotheses. First we relied on descriptive statistics to illustrate how perceptions of certainty vary across the three conditions that we study. Second, we used data from all three conditions for analysis of variance as well as difference in means t-tests to establish that the numerical distribution of opinions and status combine in predicted ways to influence perceptions of certainty. Finally, we relied on linear mixed modeling and data from condition 3 to examine these relationships more fully. Mixed models allow researchers to evaluate data that are nested in larger units (Raudenbush & Bryk, 2002; Snijders & Bosker, 1999), and thus are the appropriate method to account for the nesting of trials in participants. After excluding trials with consensus in condition 3, we have 20 instances of participant certainty in their final opinion nested in each of the participants.

RESULTS Melamed and Savage’s model of social influence assumes that both numbers and status affect how certain an individual is about the correctness of a final choice. The means presented in Table 2 and visually represented in Fig. 1 begin to illustrate the unique and combined effects of each. They also show that an individual’s confidence in an initial opinion depends on the additional information he or she expects to receive about the opinions of others. In particular, Fig. 1 illustrates that participants were least certain when the group was evenly divided (groups of 2 out of 4) and that status had an omnibus positive effect on certainty, especially when the group was evenly divided. Across the board, the participants in our experiment expressed more certainty in their final choices when they had information about both the status and the number of people who agree with them. A difference in means t-test provides statistical evidence: participants in condition 3 indicated on average significantly higher levels of certainty in their final choice than participants in condition 2 (po.001, one-tailed). Thus, there was a

121

The Role of Uncertainty

Table 2. Sample Sizes, Mean Initial, and Final Perceptions of Certainty by Condition and Numbers in Agreement (Standard Deviations). Number in Agreement

Condition 1 –

N Initial Certainty (I)

500 –

Final Certainty (F)

5.60 (1.56) –

F–I

Condition 2 1

2

3

Condition 3 4

1

2

3

4

80 160 160 100 80 160 160 100 4.81 4.94 4.91 4.86 5.35 5.40 5.44 5.44 (2.00) (1.96) (1.81) (1.99) (1.59) (1.67) (1.57) (1.49) 5.15 4.63 6.13 6.9 6.20 5.86 6.63 8.03 (2.01) (1.82) (1.68) (1.80) (1.60) (1.48) (1.53) (1.08) .34 .31 1.22 2.04 .85 .46 1.19 2.59

Note: Number in Agreement includes P.

9 8 7 No Status Information

6

Status Information

5 4 3

Fig. 1.

1

2

3

4

Perceptions of Certainty by Number in Agreement (Including self) and Whether Status Information Was Known.

significant increase in overall confidence in a final choice when individuals received information about both the status and the number of others who agreed with them. A similar result was not found when comparing conditions 1 and 2. Although having information about the opinions of others increased final perceptions of certainty above condition 1’s baseline average of 5.6 when two (average equal to 6.13) or three (average equal to 6.9) others (O) agreed with the participant, it actually decreased perceptions when zero (average equal to 5.15) or one (average equal to 4.63) O agreed with the participant. That average perceptions of certainty fell below condition 1’s average of 5.6

122

SCOTT V. SAVAGE ET AL.

when one O agreed with the participant (average equal to 4.63) makes sense, given the predictions of Melamed and Savage’s theoretical model. That this also occurred when no one agreed with the participant, however, was surprising, as the model suggests that having a unified block of individuals disagree with an individual should result in the individual becoming more confident in the correctness of that choice. While these data do not entirely support that position, they do not entirely discredit it either, as individuals in this scenario reported higher levels of certainty in their final choice than they did in their initial choice. Data from conditions 1 and 2 therefore show that informing people that they will receive information about the initial opinions of others decreases certainty about an initial opinion below what they express when they receive no information about the opinions of others. However, when they actually receive information about the opinions of others, people report final perceptions of certainty above their initial perceptions for all situations except for that situation where the distribution of opinions is evenly split. Our results then generally support hypothesis 1. So long as the group was not evenly split over the correctness of an opinion, condition 2 shows that participants were more certain about their final opinion than they were about their initial opinion. Difference in means t-tests reveals that the differences between initial certainty and final certainty were highly significant when two or three other people agreed with the participant (when two others agreed, the difference between final certainty and initial certainty was 1.22 or 6.13–4.91; when three others agreed, the difference was 2.04 or 6.9–4.86; po.001, one-tailed) and borderline significant when no one agreed with the participant (5.15–4.81=.34; po.07, one-tailed). When the group was evenly split, however, participants in condition 2 actually reported that they were significantly less confident in their final opinion (4.63–4.94=.31; po.001, one-tailed). Other than the one difference in means t-test barely missing the .05 significance level, these results bode well for the validity of Melamed and Savage’s theoretical model. The next step in testing the model involves examining whether the effects of status vary across the different numbers in agreement. The mean scores in Table 2 as well as the difference in means t-test comparing perceptions of certainty in final opinions across conditions 2 and 3 provide support for hypothesis 2: relative status affects perceptions about the correctness of a final opinion. This, however, does not mean that the effects of status are consistent across the different distributions of opinions. To test this, we compare the difference in means between the initial and final perceptions of certainty across conditions 2 and 3 for the different numbers in agreement.

The Role of Uncertainty

123

We compare the difference as opposed to the final perceptions of uncertainty in an effort to control for variation in initial perceptions of uncertainty across these two conditions. This difference is as follows: .85.34=.51, .46(.31)=.77, and 1.19–1.22=.03 when zero, one, and two people agree with the participant, respectively. Difference in means t-test show that this difference is highly significant when one O agreed with the participant (po.001, two-tailed test), borderline significant when no Os agreed with the participant (po.09, two-tailed test), and insignificant when two Os agreed with the participant (po.84, two-tailed test). Although we were not expecting status to have such a large effect when no one agreed with the participant, these results still partially support hypothesis 3 as information about the distribution of opinions seems to moderate the relationship between relative status and certainty about the correctness of a final opinion. For the most part, these analyses confirm our expectations, showing that both numbers and status affect perceptions of certainty about a final opinion, with the effect of status mattering more as the distribution of initial opinions in the group become evenly divided. To further test these ideas as well as hypotheses 4 and 5, we rely on data from condition 3 and a series of mixed models. Linear mixed models are appropriate for nested data when clustering is evident. The intraclass correlation coefficient (ICC) assesses this by measuring the proportion of variance that is between groups (Raudenbush & Bryk, 2002). The ICC for final perceptions of certainty about the correctness of an opinion when grouped by participant and with no independent variables included in the model is .53, suggesting a high degree of clustering and the need for a multilevel modeling approach. Table 3 presents results from a series of mixed models predicting how certain a participant was in his or her final opinion. All of these models control for the initial perceptions of certainty. Models 1, 2, and 3 once again illustrate that the numerical distribution of opinions affect perceptions of certainty about a final choice. Model 1 simply establishes a relationship between these two measures (b=.45, sb=.05, po.001). Adding the squaredterm in Model 2 demonstrates that this relationship is in fact curvilinear, with individuals expressing less certainty when the numerical distribution of opinions becomes evenly split. Calculating predicted values shows this. Assuming an initial perception of certainty score of 5, the predicted perception of certainty about a final choice is 5.64 when no one else expressed the same initial opinion as the participant, 5.17 when one other individual expresses the same initial opinion, and 6.28 when two others express the same initial opinion. Given this, we created a new variable that

124

SCOTT V. SAVAGE ET AL.

Table 3.

Linear Mixed Models Predicting Perceptions of Certainty about a Final Choice.

Intercept Initial Perceptions of Certainty Numerical Distribution of Opinions Numerical Distribution of Opinions Squared Even-Split

Model 1 B (S.E.)

Model 2 B (S.E.)

Model 3 B (S.E.)

Model 4 B (S.E.)

Model 5 B (S.E.)

3.04 (0.26) 0.34 (0.03) 0.45 (0.05)

5.99 (0.38) 0.34 (0.03) 2.84 (0.33) 0.79 (0.08)

4.35 (0.23) 0.34 (0.03)

4.41 (0.32) 0.29 (0.05)

4.96 (0.32) 0.23 (0.04)

0.61 (0.23)

0.08 (0.28) 2.48 (0.70) 1.34 (0.61)

0.91 (0.08)

Expectation Standing of P’s Faction EvenSplit  Expectation Interaction Expectation Standing of P’s Faction Squared Stay Response 2 Log-Likelihood BIC

2589.49 2622.91

2495.32 2535.42

2535.71 2569.13

1271.67 1301.62

3.14 (0.74) 0.79 (0.15) 1194.54 1248.47

po.05; po.01; po.001. Note: N is 400 participant-trials (20 trials nested in 20 participants). Consensus trials were excluded.

indicates whether the numerical distribution of opinions is evenly split (1) or not (0). Model 3 uses this new variable and also shows that an even-split decreases perceptions of certainty (b=.91, sb=.08, po.001). For ease of interpretation, we use the even-split variable when examining whether the effects of status are moderated by the number in agreement. We next examined the effects of status. Model 4 tests to see if the expectation standing of the participant’s faction affects perceptions of certainty. As expected, a significant relationship exists between the two (b=.61, sb=.23, po.01). Model 5 exists as our ‘‘full’’ model and tests to see if the effects of status depend on the numerical distribution of opinions, if the effects of status are curvilinear, and if people who stay with their

125

The Role of Uncertainty

initial opinion report higher levels of certainty in their final opinion. Both the even-split by expectation standing interaction term and the square of the expectation standing term are significant (b=1.34, sb=.61, po.05 and b=3.14, sb=.74, po.001, respectively), thereby supporting hypotheses 3 and 4. Like the numerical distribution of opinions, the effect of status on final perceptions of certainty is curvilinear and this relationship fluctuates depending on the number of people in agreement. That is to say individuals are more likely to rely on status information when the numerical distribution of opinions is evenly split, but that status information is only useful for reducing uncertainty if the factions differ in relative status. Because of the limited distribution of expectation standings in our experiment, we cannot explore these relationships in their entirety, but preliminary investigations indicate that in situations where the numerical distribution of opinions are evenly split, people turn to information about status to evaluate the correctness of an opinion. When that status information also provides little insight into correctness of an opinion, people report less certainty in their final opinion. This finding is important insofar as it refines Melamed and Savage’s initial argument about the relative effects of status in evenly distributed groups. Model 5 also examines the relationship between staying with an initial opinion and perceptions of certainty about a final decision. The results show that those individuals who stay with their initial opinion report higher levels of certainty about the correctness of their final response (b=.79, sb=.15, po.001). This finding confirms hypothesis 5.

DISCUSSION Melamed and Savage (2013) recently integrated research on faction sizes with SCT. In doing so, they showed that in collectively oriented task groups with more than two individuals, the effects of status on social influence depend on the distribution of opinions about the correctness of a final decision. This finding added to previous research on how a host of other factors combine with faction sizes (e.g., Erb, Bohner, Rank, & Einwiller, 2002; Gordijn et al., 2002; Kalkhoff & Barnum, 2000; Martin, Gardikiotis, & Hewstone, 2002; Tormala & DeSensi, 2009). The critical underlying mechanism for Melamed and Savage’s theory is uncertainty. The theory holds that the effects of status should be strongest when the distribution of opinions reduces the least amount of uncertainty (i.e., when the distribution of opinions was evenly divided). Up to this point,

126

SCOTT V. SAVAGE ET AL.

however, this assertion had not been tested empirically. In this chapter, we rectified this shortcoming and in the process, uncovered a number of additional insights that add nuance to Melamed and Savage’s initial statement. The results reported here generally confirm the theory’s logic. As predicted, having information about the distribution of opinions affected how certain participants were about the correctness of a final decision, with an even-split producing the most uncertainty. The results also show that people use status when evaluating how certain they are about the correctness of a final choice and that the effects of status on uncertainty are strongest when the distribution of opinions is evenly split. Although these findings strengthen our confidence in Melamed and Savage’s (2013) theory, there was one unexpected finding. In the four-actor groups that we studied, the uncertainty reducing effects of numbers were not the same for factions of one and factions of three, and as a result, the relative effects of status varied across these two scenarios. Specifically, in the four-person groups we found that absent status information, when no one agreed with the participant, participants felt only slightly more certain in their final choice but when two others agreed, they felt much more certain and that as a result, the uncertainty reducing effects of status was much more pronounced when no one agreed with the participant. We suspect that this difference may be the result of some form of self-bias, but without further research, we cannot conclude this definitively. What we can say definitively, however, is that like numbers, the uncertainty reducing effects of status are curvilinear. As the amount of status of those in agreement and disagreement become more evenly divided, actors cannot use this information to evaluate the correctness of a final decision and, thus, cannot use this information to enhance how certain they are about a nonveridical choice. In fleshing out the form of this relationship, the current research extends Melamed and Savage’s (2013) earlier statement. Another contribution of this research concerns the relationship between a stay response and confidence in a final decision. Melamed and Savage’s theory suggests that uncertainty affects social influence, but our experiment measured perceptions of certainty after an individual made their final choice. We view this inconsistency as a strength insofar as it let us examine whether the act of staying increased confidence in the appropriateness of a final choice. Our results confirm this, thereby highlighting the relationship between behavioral consistency and confidence in the action. Thus, this study offers empirical support for earlier assertions made by Melamed and Savage. As hypothesized, the distribution of opinions and

127

The Role of Uncertainty

social status combine to affect perceptions of uncertainty. It also generates new insights into social processes by uncovering the curvilinear relationship between status and certainty about the correctness of a final opinion. By empirically documenting how the distribution of opinions and status affect perceptions of certainty about a final choice, this research advances the science of social influence by illustrating how the structural conditions of the group can frame whether intragroup interactions create cognitive pressures for individuals to conform to the opinions of others.

REFERENCES Asch, S. (1951). The effects of group pressure upon the modification and distortion of judgments. In H. Guetzkow (Ed.), Groups, leadership, and men (pp. 177–190). Pittsburgh, PA: Carnegie Press. Asch, S. (1956). Studies of independence and conformity: A minority of one against a unanimous majority. Psychological Monographs, 70, Whole No. 416. Balkwell, J. W. (1991). Status characteristics and social interaction. In E. J. Lawler, B. Markovsky, H. A. Walker & C. Ridgeway (Eds.), Advances in group processes (Vol. 8, pp. 135–176). Greenwich, CT: JAI Press. Berger, J. (2007). The standardized experimental situation in expectation states research: Notes on history, uses, and special features. In M. Webster & J. Sell (Eds.), Laboratory experiments in the social sciences (pp. 353–378). New York, NY: Elsevier. Berger, J., Cohen, B., & Zelditch, M. (1972). Status characteristics and social interaction. American Sociological Review, 37, 241–255. Berger, J., & Fis- ek, M. H. (1970). Consistent and inconsistent status characteristics and the determinations of power and prestige orders. Sociometry, 33, 287–304. Berger, J., Fis- ek, M. H., Norman, R., & Zelditch, M., Jr. (1977). Status characteristics and social interaction: An expectations states approach. New York, NY: Elsevier. Berger, J., & Webster, M. (2006). Expectations, status, and behavior. In P. Burke (Ed.), Contemporary social psychological theories. Palo Alto, CA: Stanford University Press. Berger, J., Zelditch, M., Jr., & Cohen, B. (1966). Status characteristics and expectation states. In J. Berger, M. Zelditch & B. Anderson (Eds.), Sociological theories in progress (pp. 29–46). Boston, MA: Houghton Mifflin. Chaiken, S., & Stangor, C. (1987). Attitudes and attitude change. Annual Review of Psychology, 38, 575–630. Correll, S., & Ridgeway, C. (2003). Expectation states theory. In J. Delamater (Ed.), The handbook of social psychology (pp. 29–52). New York, NY: Kluwer Academic/Plenum. Cuddy, A., Norton, M., & Fiske, S. (2005). This old stereotype: The pervasiveness and persistence of the elderly stereotype. Journal of Social Issues, 61, 265–283. Dewey, J. (1929 [2005]). The quest for certainty: A study of the relation of knowledge and action. Whitefish, MT: Kessinger Publishing. DiFonzo, N., Bourgeois, M. J., Suls, J., Homan, C., Stupak, N., Brooks, B. P., y Bordia, P. (2013). Rumor clustering, consensus and polarization: dynamic social impact selforganization of hearsay. Journal of Experimental Social Psychology, 49, 378–399.

128

SCOTT V. SAVAGE ET AL.

DiMaggio, P., & Powell, W. W. (1983). The iron cage revisited: Institutional isomorphism and collective rationality in organizational fields. American Sociological Review, 48, 147–160. Erb, H., Bohner, G., Rank, S., & Einwiller, S. (2002). Processing minority and majority communications: The role of conflict with prior attitudes. Personality and Social Psychology Bulletin, 28, 1172–1182. Festinger, L. (1954). A theory of social comparison processes. Human Relations, 7, 117–140. Fis- ek, M. H., Berger, J., & Norman, R. (1991). Participation in heterogeneous and homogeneous groups: A theoretical integration. American Journal of Sociology, 97, 114–142. Fiske, S. T. (2000). Stereotyping, prejudice, and discrimination at the seam between the centuries: Evolution, culture, mind, and brain. European Journal of Social Psychology, 30, 299–322. Fiske, S. T., & Taylor, S. (1991). Social cognition. New York, NY: McGraw-Hill. Gordijn, E. H., de Vries, N. K., & de Dreu, C. K. W. (2002). Minority influence on focal and related attitudes: Changes in size, attributions, and information processes. Personality and Social Psychology Bulletin, 28, 1315–1326. Heider, F. (1958). The psychology of interpersonal relations. New York, NY: Wiley. Hogg, M. A. (2007). Uncertainty-identity theory. In M. P. Zanna (Ed.), Advances in experimental social psychology (pp. 69–126). San Diego, CA: Academic Press. Kalkhoff, W., & Barnum, C. (2000). The effects of status-organizing and social identity processes on patterns of social influence. Social Psychology Quarterly, 63, 95–115. Kalkhoff, W., Friedkin, N., & Johnsen, E. (2010). Status, networks and opinions: A modular integration of two theories. In S. R. Thye & E. Lawler (Eds.), Advances in group processes (pp. 1–38). New York, NY: JAI Press. Kollock, P. (1994). The emergence of exchange structures: An experimental study of uncertainty, commitment, and trust. American Journal of Sociology, 100, 313–345. Latane, B. (1981). The psychology of social impact. The American Psychologist, 36, 343–356. Markovsky, B., Willer, D., & Patton, T. (1988). Power relations in exchange networks. American Sociological Review, 53, 220–236. Martin, R., Gardikiotis, A., & Hewstone, M. (2002). Levels of consensus and majority and minority influence. European Journal of Social Psychology, 32, 645–665. Melamed, D., & Savage, S. V. (2013). Status, numbers and influence. Social Forces, 91, 1085–1104. Molm, L. D., Whitham, M., & Melamed, D. (2012). Forms of exchange and integrative bonds: Effects of history and embeddedness. American Sociological Review, 77, 141–165. Moore, J. C., Jr. (1968). Status and influence in small group interactions. Sociometry, 31, 47–63. Moscovici, S. (1980). Towards a theory of conversion behavior. In L. Berkowitz (Ed.), Advances in experimental social psychology (pp. 209–237). New York, NY: Academic Press. Moscovici, S., Lage, E., & Naffrenchoux, M. (1969). Influence of a consistent minority on the responses of a majority in a color perception task. Sociometry, 32, 365–380. Mullen, B. (1983). Operationalizing the effect of group on the individual: A self-attention perspective. Journal of Experimental Social Psychology, 19, 296–322. Phillips, K. W. (2003). The effects of categorically based expectations on minority influence: The importance of congruence. Personality and Social Psychology Bulletin, 29, 3–13. Pollock, H. N. (2003). Uncertain science y uncertain world. Cambridge: Cambridge University Press.

The Role of Uncertainty

129

Pugh, M. D., & Wahrman, R. (1983). Neutralizing sexism in mixed-sex groups: Do women have to be better than men? American Journal of Sociology, 88, 746–762. Rashotte, L. (2007). Social influence. In G. Ritzer (Ed.), The Blackwell encyclopedia of sociology (pp. 4426–4429). Hoboken, NJ: Blackwell. Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models: Applications and data analysis methods (Second Edition). Thousand Oaks, CA: Sage Publications. Ridgeway, C. (1984). Dominance, performance, and status in groups: A theoretical analysis. In E. Lawler (Ed.), Advance in group processes (pp. 59–93). New York, NY: JAI Press. Ridgeway, C. (1997). Sex differences in gender role attitudes. American Sociological Review, 62, 218–235. Ridgeway, C. L., & Correll, S. J. (2006). Consensus and the creation of status beliefs. Social Forces, 85, 431–453. Savage, S. V., & Bergstrand, K. (2013). Negotiating the unknown: The role of uncertainty in social exchange. Sociology Compass, 7, 315–327. Snijders, T. A. B., & Bosker, R. J. (1999). Multilevel analysis: An introduction to basic and advanced multilevel modeling. Thousand Oaks, CA: Sage. Suls, J., & Wheeler, L. (2000). Handbook of social comparison: Theory and research. New York, NY: Kluwer/Plenum. Tanford, S., & Penrod, S. (1984). Social influence model: A formal integration of research in majority and minority influences processes. Psychological Bulletin, 95, 189–225. Tormala, Z. L., & DeSensi, V. L. (2009). The effect of minority/majority source status on attitude certainty: A matching perspective. Personality and Social Psychology Bulletin, 35, 114–125. Webster, M., Jr., & Driskell, J., Jr. (1978). Status generalization: A review and some new data. American Sociological Review, 43, 220–236.

THE PROFESSIONAL CREDENTIALS OF IMMIGRANTS: A STATUS-AND-EXPECTATIONS APPROACH Martha Foschi ABSTRACT Purpose – (a) To examine ‘‘native-born/immigrant’’ (nativity) and ‘‘national/foreign professional credentials’’ (country of credentials) as status factors in terms of expectation states theory, and (b) to lay out a blueprint for a theory-based, experimental research agenda in this area. Design/methodology/approach – (for (b) above). I propose a research program based on three types of expectation states experimental designs: the open group-discussion, the rejection-of-influence standardized setting, and the application-files format. All three incorporate measures of either biased evaluations or double standards for competence, or of both. I illustrate how these designs can be adapted to assess, through the presence/absence of one or the other of those practices, the separate impacts of nativity, country of professional credentials and selected additional factors on the inference of task competence. The need for and the advantages of systematic, experimental work on this topic are highlighted. Advances in Group Processes: Thirtieth Anniversary Edition Advances in Group Processes, Volume 30, 131–155 Copyright r 2013 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 0882-6145/doi:10.1108/S0882-6145(2013)0000030009

131

132

MARTHA FOSCHI

Findings – (from (a) above). I review evidence of the status value of nativity and country of credentials through data on evaluations, employment, and earnings. My evidence originates in contemporary Canadian studies that present results from surveys, interviews, census records, and – to a lesser extent – experiments, and these findings support my claim. Practical/social implications – The proposed research will facilitate the development of interventions toward the standardized and unbiased assessment of immigrants’ foreign credentials. Originality/value – The agenda I put forth constitutes a novel approach to the study of nativity and country of credentials. The work will extend the expectation states program, and enhance immigration research both theoretically and methodologically. Keywords: Professional credentials; immigrants; Canada; status characteristics; performance expectations; experiments

INTRODUCTION Many immigrants have professional degrees from their original homelands. This chapter focuses on the assessment of those credentials. I examine the topic within the context of contemporary immigration to Canada and offer examples from the English-language Canadian research literature. The subject, however, is also of importance and concern in many other societies, and I submit that the ideas and proposals that I offer here are generally applicable beyond the Canadian context that I consider. Immigrants to Canada come from numerous parts of the world, hold degrees in a wide diversity of fields, and represent a significant addition to the Canadian talent pool. The country needs immigrants to make up for labor shortages in various professions, and the official policy is to encourage applications from that sector. Over the last several years, however, academic publications as well as official statistics and news-media releases have frequently called attention to the difficulties that immigrants with professional degrees often encounter in having their foreign credentials recognized in Canada. This, in turn, translates into barriers in finding jobs that are commensurate with their qualifications, as well as into underemployment and lower earnings relative to the native-born. Depending on the individual cases, several factors contribute to these difficulties. Those factors may

The Professional Credentials of Immigrants

133

involve, for example, the applicants’ lack of Canadian work experience, their limited language skills, the assessors’ unfamiliarity with degrees from abroad, restrictive rules governing professional practice in certain fields, and skill-demand fluctuations in the labor market (see, e.g., Li, 2003, Chapters 3 and 6; Reitz, 2001a, 2001b). My interest is in studying the extent to which those difficulties are linked to status factors and, particularly, in whether ‘‘native-born/immigrant’’ and ‘‘country of credentials’’ can be conceptualized as status characteristics in terms of the expectation states program.1 I present and discuss the theoretical and methodological advantages that this program brings to the study of immigration and accreditation issues. The topic I consider has repercussions at various levels, from consequences for the Canadian economy to patterns of interaction in task groups. I view it as a subject of special relevance to the study of diversity in work teams and other groups within organizations (Brief, 2008; Herring, 2009). In particular, in this chapter I use an expectation states approach to address the assessment of native-born applicants with Canadian credentials in relation to immigrant candidates with foreign credentials. A key question is: ceteris paribus, if the native-born has been favored, to what extent does this reflect either a bias in performance evaluation or a double standard for competence? My central objective here is to offer a blueprint for a theorybased, experimental research agenda in this area.

STATUS CHARACTERISTICS AND EXPECTATION STATES Expectation states theory is a comprehensive, widely recognized research program on the assignment of competence in task groups (for reviews, see Berger & Webster, 2006; Correll & Ridgeway, 2003; Wagner & Berger, 2002). Its core ideas concern the formation of performance expectations (anticipated outcomes) in groups where members have no prior history of interaction with each other, and are committed to doing well on a task that they consider valuable and must complete as a team. These latter clauses constitute the two main scope conditions of this tradition, namely, ‘‘task orientation’’ and ‘‘collective orientation.’’ According to the theory, members of those groups assign competence to each other and hold corresponding task-related expectations, on the basis of knowledge of either status characteristics or the quality of prior task performance, or of both. A substantial

134

MARTHA FOSCHI

literature, spanning several decades and focusing on the experimental test of hypotheses, supports these predictions. The expectation states program involves several interconnected unittheories. My discussion in this chapter centers on status characteristic theory – a major component of this program. A ‘‘status characteristic’’ is any valued attribute believed by many in a society to convey information about task competence. Such attributes consist of at least two levels (or states), one viewed as carrying more worth (prestige) than the other (e.g., high vs. low mechanical ability). The beliefs associated with them are social constructions (i.e., they do not have to have an objective validity) and are not necessarily conscious. These attributes also vary in the range of their perceived relevance to the task at hand: specific characteristics provide information about task competence in a narrowly defined area (e.g., the ability to solve crossword puzzles) while diffuse characteristics are seen as associated with a wide, indeterminate set of tasks. In many societies, gender, ethnicity/race, nationality, formal education, and socioeconomic class, for example, constitute status characteristics for a large number of individuals (for references to this evidence, see Berger & Webster, 2006, pp. 273–274). The dimension ‘‘diffuse-specific’’ covers of course many values, and status characteristics vary in their location on this dimension depending on the social system and individual beliefs under consideration. Here I propose ‘‘place of birth’’ and ‘‘country of credentials’’ as diffuse factors, each affecting how a person’s task competence is assessed and, in turn, how he/she fares in chances of professional employment and related outcomes. Note also that there are several other social distinctions that carry differential evaluations, such as moral/not moral, liked/not liked, and in-group/out-group; under some conditions, these too function as status factors (see, e.g., Bianchi, 2004). In this chapter I am concerned only with those attributes that primarily imply different levels of competence at a valuable task, that is, those defined as status attributes by this theory. Work on expectation states follows a situational approach. Thus, the core propositions are formulated from the point of one individual (‘‘self’’ or ‘‘focal actor’’) who performs a given task with a single partner (‘‘other’’). If there are two or more partners, the interaction is viewed as a series of dyadic links between self and each partner. As I discuss later in this chapter, the theory has also been extended to settings in which self is an assessor of one or more persons but not a performer of the task. Furthermore, expectations are conceptualized as relative to the particular situation – that is, they vary

The Professional Credentials of Immigrants

135

in their implied levels of competence as a function of characteristics of both the task and the other performers(s) under consideration.

PERFORMANCES AND THE STABILITY OF THE STATUS QUO Studies based on expectation states theory have investigated two key practices that contribute to the maintenance of a given status order in the assignment of competence: biased evaluations and double standards.2 A biased evaluation occurs when there are no objective and agreed-upon means to assess a performance, and information seen as related is used instead as an indicator of its value. Here I focus on performer’s status as one such substitute. As a result of that process, higher-status individuals tend to be evaluated more positively for their outputs than do their lower-status counterparts. Thus, a book that is presented as written by an expert in the field in question (e.g., urban transportation systems) may be judged, a priori, to be of higher quality than if the same book is reported as authored by a novice in the area. The notion of a ‘‘biased evaluation’’ is at the center of expectation states theory. It is the social mechanism through which performances are assessed in line with status and in turn help preserve it, as shown, for example, in two classic experiments in this tradition – Moore (1968) and Webster (1969). The practice is also demonstrated in the wellknown study by Goldberg (1968) and is often referred to as ‘‘the Goldberg paradigm.’’ (See Top, 1991 for a review of studies investigating this topic.) The notion of a double standard for competence, on the other hand, is a more recent introduction to this program. A double standard relates to situations in which individuals are seen in terms of two items of information: status characteristics and task results (see, e.g., Foddy & Smithson, 1999; Pugh & Wahrman, 1983; Wagner, Ford, & Ford, 1986; Zelditch, Lauderdale, & Stublarec, 1980). Of particular interest is the case of two performers who differ on status but exhibit objectively evaluated task outcomes that show that both have performed equally well. Are these outcomes sufficient to suppress status effects? In Foschi (1989, 2000) I have proposed that, under certain conditions, they are not, and that the means through which this occurs is the activation of double – and sometimes multiple – standards requiring more and better evidence of competence from the lower-status performer.3 This formulation has been tested for gender,

136

MARTHA FOSCHI

ethnicity/race, parental status, and academic rank, in all cases with supporting results. (See, e.g., Correll, Benard, & Paik, 2007; Foschi, 1996; Leahey, 2004; see also Biernat, 2005 for a related formulation and significant research program on ‘‘shifting standards.’’) I submit that the case of immigrants and their difficulties in obtaining employment commensurate with their foreign professional qualifications can be studied in relation to one or the other of these processes – depending on the extent and quality of the information the assessor has about the credentials. But first, I must address whether ‘‘native-born/immigrant’’ and ‘‘country of credentials’’ are established (i.e., socially validated) status variables in the Canadian context. To my knowledge, no previous research has investigated these two variables and their joint consequences in the terms I discuss here.

NATIVITY AND COUNTRY OF CREDENTIALS I propose that ‘‘native-born’’ and ‘‘Canadian credentials’’ are, respectively, often evaluated in Canada as worthier than ‘‘immigrant’’ and ‘‘foreign credentials.’’ This affects how much competence is inferred and, in turn, job prospects and related outcomes in general. I examine the evidence for my proposal by reviewing studies on nativity, credentials, employment, and earnings in the Canadian context. This literature is extensive, concerns a variety of professions, and provides evidence of a troubling social problem. The work contains surveys, interviews, census data, and, to a lesser extent, experiments.4 The fields commonly studied are engineering, health, and the natural sciences. For the most part, I focus on findings published in academic journals over the last 15 years. The studies I cite in this section are from that set and represent those that I consider of close relevance to this chapter. (For examples of outside-Canada findings in this area, see Kler, 2006 (Australia), Ward & Masgoret, 2007 (New Zealand), DiTomaso, Post, Smith, Farris, & Cordero, 2007, and Zeng & Xie, 2004 (United States); for a comparison between the United States and Canada, see Somerville & Walsworth, 2009; for a valuable review of experiments on immigration topics, and on the need for more such research, see McKenzie & Yang, 2012.) Performances and Evaluations Ferrer and Riddell (2008) report that immigrants’ years of schooling as well as professional experience acquired before arrival in Canada counted much

The Professional Credentials of Immigrants

137

less, regarding job offers and actual hiring, than Canadian qualifications obtained by the native-born. Wang and Lo (2005) present comparable findings about educational credentials. Several studies also report that, in the eyes of the immigrants themselves, their foreign degrees are devalued (e.g., Basran & Zong, 1998; Grant & Nadin, 2007). Note, however, that these findings do not provide direct evidence of the status value of those factors: while the findings associate both nativity and country of credentials with evaluations, it is also possible that the latter reflect considerations other than status, as I observe below.

Status and Rewards Expectation states studies have also investigated the relationship between these two variables. Findings not only support a direct link from status to rewards, but also a reverse association whereby rewards affect the assignment of status (Hysom, 2009; Stewart & Moore, 1992). The latter link is especially relevant to the present discussion, as it offers an additional indication of the status value of nativity and country of credentials. Let us examine hiring recommendations, actual hiring-rates, and earnings as instances of rewards. Evidence that the native-born and/or the holders of Canadian credentials are favored in one or more of these respects is presented in, for example, Boyd and Thomas (2002); Esses, Dietz, and Bhardwaj (2006); Ferrer and Riddell (2008); Fong and Cao (2009); Li (2001); Nakhaie (2007); Oreopoulos (2011); Pendakur and Pendakur (1998); Reitz (2001b); Wang and Lo (2005). See also Fuller and Vosko (2008) on the related topic of nativity and temporary employment. However, note that – as the authors of these works recognize – it is important to separate nativity effects from those of country of degree (particularly in terms of how these factors are perceived by potential employers) and to consider other key variables. These include whether the foreign and the Canadian credentials under review are indeed comparable, whether the employers have judged them to be comparable, and whether the immigrants had the Canadian experience and linguistic abilities needed for the Canadian jobs in question. Sex category, age at time of arrival in Canada, socioeconomic class, and ethnicity/race of the immigrants also need to be added.5 Moreover, one should pay attention to the interaction context and take into account characteristics of both the assessor(s) and the person(s) being assessed. As well, it is important to note that nativity and country of degree are more likely to become salient when they are seen as

138

MARTHA FOSCHI

relevant to the task at hand and involve contrasting – rather than equating – comparisons (on these points, see Berger & Webster, 2006, p. 272). Let us then keep in mind that, when persons who differ on either nativity or country of credentials are reviewed, there are often other factors at play and that, in perceptions of competence, the comparison does not always favor either the native-born, or the holder of a Canadian degree, or both. The result may depend, among other attributes, on the sex category, ethnicity/race, formal education, country of credentials, and social class of the differing groups. Field of degree and prestige of conferring academic institution should also be considered. Two examples from the Canadian context would be (a) a low-income, Canadian-born versus a wealthy immigrant and (b) a Canadian-born with limited formal education versus an immigrant with a professional degree from a university such as Oxford or Harvard. In addition, having been born in Canada could represent an advantage if a person belongs to a long lineage of either an English-speaking or a French-speaking group – but the advantage may be diminished if, for instance, his/her parents are recent immigrants from Italy. Finally, I should also note that the two central factors I discuss here may not only be linked to perceived competence but could as well be moderated by both affect and social identity – that is, nativity and country of credentials may generate reactions involving liking and/or solidarity (such as an employer’s sympathy based on shared characteristics). See also, for example, Esses, Dovidio, Semenya, and Jackson (2005) on the roles of patriotism and nationalism in attitudes toward immigrants. What then can one conclude about nativity and country of credentials? Within the limits mentioned above, research conducted so far offers considerable evidence (direct as well as indirect) that both are status factors affecting the evaluation of credentials and subsequent rewards. However, are they diffuse status characteristics? All status characteristics, either specific or diffuse, are based on cultural beliefs. As I indicate earlier in this chapter, statements about such characteristics are formulated from the point of view of a person who holds those beliefs, and reflect a comparison across two or more actors. It is often the case that not all in a given social system accept the status value of a certain attribute (or that they all accept it to the same extent). Furthermore, status characteristics are predicted to be a source of performance expectations and to affect interaction in groups only under the situational boundaries identified by the theory. The formal definition of a diffuse status characteristic is that ‘‘(1) it involves two or more states that are differentially valued, (2) associated with each state are distinct sets of specific

The Professional Credentials of Immigrants

139

expectations, each itself evaluated, and (3) also associated with each state is a similarly evaluated general expectation state’’ (Berger & Webster, 2006, p. 272). I propose that nativity and country of credentials meet this definition, for the following reasons. The evidence discussed earlier in this section (under Performances and Evaluations and Status and Rewards) associates each factor with two states, ‘‘Canadian-born/immigrant’’ and ‘‘national/foreign credentials,’’ respectively, and these states are assigned different values, as required by (1) above. That evidence also indicates that in Canada, for some people and depending on the comparison group, ‘‘immigrant’’ and ‘‘foreign credentials’’ signal a competence disadvantage in several specific respects (such as job offers, actual hiring, and earnings differentials). These findings line up with requirement (2) for each factor. Moreover, the views linked to nativity and country of credentials extend to comprise overall levels of worth and general expectation states (requirement (3): in the case of nativity, for immigrants that expectation is usually associated with inferior language skills (in either English or French, both written and oral) and insufficient understanding of ‘‘the Canadian ways of doing things’’). In particular, those persons tend to be perceived as bringing in values that are ‘‘foreign to Canada,’’ and ‘‘immigrant’’ is often assumed to be linked to ‘‘non-White.’’ For a more complete discussion of these points, see ‘‘the folk version of immigrants’’ presented in Li (2003, pp. 44–46) and Garnett, Guppy, and Veenstra (2008) on the notion of ‘‘cultural talent.’’ For evidence of overall expectations, see Palmer’s (1996) study of Canadian attitudes towards immigration, and Creese (2010) and Scassa (1994) on language standards, perceptions of immigrants, and employment. With respect to ‘‘country of foreign credentials,’’ the general expectation state includes the belief that, frequently, immigrants come from societies that are traditional (as opposed to ‘‘advanced’’), inefficient, rather disorganized, and not highly technological. As a result, educational institutions in those countries are seen as overall inferior in quality to their counterparts in Canada. For investigations on immigrants’ foreign degrees (either known directly or inferred from age), country of degree, and employment/earnings in Canada, see, for example, Boyd and Thomas (2002); Satzewich and Li (1987).6

RESEARCH QUESTIONS For the reasons presented in the previous section, I propose that it would be a fruitful strategy to carry out experimental work in this topic. This would

140

MARTHA FOSCHI

result in a more thorough appraisal of the status value of these two factors and would contribute to isolate their individual effects. To achieve these ends, I suggest creating simplified group-settings and studying a wide range of cases. Table 1 presents the key interrelated comparisons that I deem worth examining. It describes situations in which an assessor C reviews performers A and B in relation to each other. (In particular cases, self is both an assessor and a performer. In the first two types of studies I discuss below, C=A.) I assume that C treats ‘‘country of birth’’ and ‘‘country of credentials’’ as diffuse status factors, and Table 1 includes baseline Comparisons 1 and 2 to determine empirically whether and to what extent this is the case for each. Since, as I illustrated earlier, the answer will be affected by the particular countries considered, it would be best to vary them in different replications. Such replications will serve to ascertain how well established those two factors are, and can also investigate the degree to which participants view them as relevant to task ability (be it directly, indirectly, or conversely). In all the comparisons, A and B differ in either nativity or country of credentials, or in both. Depending on design, it may only be necessary to identify birthplace simply as either ‘‘Canadian-born’’ or ‘‘immigrant.’’ All other information that C has about A and B (from either written or visual cues, and subject to type of experiment) on, for example, sex category, age, formal education, appearance, work experience, professional employment, and language skills equate the two performers. Throughout Table 1, date of arrival in Canada would be similar for the immigrant persons.

Table 1.

Comparisons Across Country of Birth and Country of Credentials.

Comparison

1 2 3 4 5 6 7 8

Person A

Person B

Nativity

Credentials

Nativity

Credentials

Canadian-born – Canadian-born Canadian-born Canadian-born Immigrant Canadian-born Canadian-born

– Canadian Canadian Foreign Canadian Canadian Canadian Foreign

Immigrant – Immigrant Immigrant Canadian-born Immigrant Immigrant Immigrant

– Foreign Canadian Foreign Foreign Foreign Foreign Canadian

The Professional Credentials of Immigrants

141

Let us then assume an experimental project with ‘‘country of birth’’ and ‘‘country of credentials’’ as the independent variables, and inference of competence for the task at hand (e.g., solving a complex geometrical task; carrying out successfully the duties of a professional appointment) as the dependent variable. Given random assignment to conditions, Comparisons 1, 3, and 4 as a set will sort out further whether nativity carries status; 2, 5, and 6 will do so for country of credentials; and 7 and 8 will assess the relative impact of each of those two factors. The comparisons shown in Table 1 could either be implemented in a single experiment, or split into two interconnected studies (namely, one focusing on nativity and the other on country of credentials). Throughout the project, I would make the operational definitions of ‘‘credentials’’ as alike as possible. For example, I would use the same Canadian university (an accredited institution) for A’s and B’s credentials in Comparison 3, and the same foreign university for the two persons in Comparison 4. I would use those same two universities for Comparisons 2 and 5 through 8, and would also make the two institutions highly similar in academic prestige. In all cases, the credentials will be presented as equivalent in field of study, required courses, and year of degree (depending on type of design and amount of information provided about A and B – as discussed in the next section). In order to enhance the realism of the situation, there will be minor differences in peripheral details (e.g., optional courses, extracurricular activities). I would follow a corresponding methodological approach in operationalizing ‘‘immigrant’s country of birth.’’ The research program I outline in Table 1 is best approached by gradually adding replications with various types of assessors (e.g., in terms of sex category, ethnicity/race, formal education, employment status). In addition, this basic framework can be made more complex. For example, the performers’ sex category and age, as well as the assessors’ level of knowledge about the credentials could be varied. Another change could be in the overall prestige of the Canadian and the foreign degree and in the field of the credentials (e.g., engineering, business, health services). Comparisons 2, 5, and 6 are particularly important regarding the social mechanisms I examine. If the assessor has insufficient factual knowledge and low familiarity with the credentials, I submit that biased evaluations in line with status of country of degree will take place in those cases. On the other hand, if the situation is such that the credentials are seen by the assessor as being truly equivalent (e.g., either this person has sufficient knowledge to arrive at that decision by him/herself, or an accreditation board has reached that conclusion), then those comparisons would

142

MARTHA FOSCHI

investigate if double standards occur nevertheless through the devaluation of the foreign credentials. The scope of the proposed agenda may appear to be vast because of the number of possibly interacting variables. However, different teams of researchers could focus on separate aspects of the program, including the addition of other factors and operationalizations. As results accumulate, more will be learned about which ones are the key relevant variables. Those which are not can then be excluded from future replications.

THREE TYPES OF EXPECTATION STATES STUDIES The research questions I pose in this chapter have the benefit of being asked within the context of a theory. In particular, I submit that designs based on expectation states theory would be useful, complementary tools for the study of the present topic, as work in that theory has assembled a rich array of welltested experimental situations to investigate status effects. Several of these situations could be adapted to examine topics linked to nativity, country of credentials, biased evaluations, and double standards for competence. Below I discuss three designs that I have selected from that literature. Each can be modified to test whether a partiality emerges in appraising Canadian-born professionals relative to their immigrant counterparts when the two groups have overall comparable degrees from their respective birthplaces. All the experiments would incorporate control conditions as discussed in the previous section. Those taking part would be alike in several key respects (such as age) except of course for the factors to be studied as independent variables in each case; all would be paid volunteers. The following summarizes the main common features of several studies that have used one or the other of these designs, as well as the adaptations that I offer.7 In the three cases, the main objective is to investigate the practices that affect assignment of task competence to one person relative to another. For different ways of measuring biased evaluations and double standards see, for example, Foschi (2000, pp. 22–32). All three designs focus on first impressions, which are usually powerful determinants of subsequent interaction. The task at hand may be viewed by the participants as directly associated with one or both of the independent variables of interest here, or may be linked to them indirectly, through a process of ‘‘status generalization’’ (Wagner & Berger, 2002, p. 48). In all the studies, participants would be assured of their anonymity in the reporting of results.

The Professional Credentials of Immigrants

143

Discussion Groups A small group of persons (typically four to six) are asked to meet in an open (i.e., face-to-face) interaction context and discuss a human-relations problem that does not have a clear-cut answer. They are, nevertheless, urged to do their task well and to reach an agreement within a specific time limit – instructions that implicitly encourage them to assess each other’s contributions to the discussion and thus reveal status effects. A commonly used situation involves asking the participants to act as a jury and make recommendations about the guilt/innocence of the person undergoing trial (not one of the group members). Depending on the specific objectives of the study, those taking part are either similar or different in key background variables. The open-interaction setting facilitates the salience of factors such as sex category, skin color, accent, and age. (For a thorough exploration of these factors, it would of course be helpful to vary them one at a time.) An introductory ‘‘get-acquainted’’ session is often used for group members to communicate information about themselves. Trained observers positioned behind a one-way mirror record data on status effects, namely, each person’s chances to perform, actual performances, evaluations received, and exerted influence. Alternatively, audio and/or tape recording of the session could be used. The following are some of the changes that I would introduce to study the value attached to my two independent variables. Participants would have equivalent levels of formal education in the same professional field (e.g., a recently obtained B.S. in Mechanical Engineering). Initially, I would carry out this experiment with groups that are also homogeneous in sex category. Control conditions would include only native-born persons while in the experimental conditions, one or more of the participants would be confederates of the experimenter who will present themselves as immigrants and who will differ from the rest of the group in either place of birth or country of professional degree, or in both. In addition, depending on the study, the immigrant person(s) could either differ from or be similar to the others in the group on characteristics such as accent. Three examples of studies that use this interaction setting are Goar and Sell (2005), Shelly and Munroe (1999), and Shelly and Webster (1997). The design can also be modified so that selected information about those taking part is given through prerecorded audio or video taping (see, e.g., Tuzlak, 1989); moreover, some or all of the interaction may occur only online.

144

MARTHA FOSCHI

Computer-Mediated Interaction in Dyads A large portion of expectation states research has used a standardized design to investigate effects of status on influence behavior. The following describes the common elements of several computerized versions of that design. Such versions have been implemented in, for example, Bianchi (2004), Foddy and Smithson (1999), Foschi (1996), and Hysom (2009). Two previously unacquainted persons work together on the solution of a visual-perception task. The two are separated by a partition, and do not see each other before or during the session. The experimenter provides each with ‘‘demographic information’’ about the partner – a procedure that is used to create perceptions of that person on several status factors (sex category, ethnicity/race, formal education) as intended. Each person then views a series of abstract, two-colored patterns on the computer screen; in each case the instructions are to identify which color is predominant. Although the task does not have a correct solution, participants are told that it does, and that the ability to discriminate between the two colors is a valuable, recently discovered skill. To further ensure that results are not affected by other factors, this ‘‘ability’’ is often explicitly dissociated from variables such as intelligence and mathematical proficiency. For each pattern, each of the two persons makes an initial choice as to its predominant color and then forwards this choice electronically to the partner. In turn, the relayed response from that person is preset to result in a large proportion of disagreements between the two. Next, each is prompted to make a final choice. The difference between a person’s initial and final choices after each disagreement constitutes a measure of acceptance of influence from the partner and an indicator of the competence attributed to self and other. Several versions of this design, including the computerized setting I describe above, have been successfully used to capture the effects of biases originating in a variety of status factors. Examples of studies that have used one or another version to examine ethnicity/race are Foddy and Riches (2000), Foschi and Buchan (1990), Riches and Foddy (1989), Tuzlak and Moore (1984), and Webster and Driskell (1978). Some of these studies have also incorporated pretaped video images of the person said to be the partner – a procedure that effectively communicates characteristics such as skin color. Moreover, the ‘‘demographic information’’ can be easily adapted to present details about this person’s nativity and academic/professional credentials.

The Professional Credentials of Immigrants

145

Assessment of Job Applicants This design involves the examination of applications for professional positions. It implements ideas that expand the core of the theory from groups in which each member is both a performer and an assessor, to settings in which those two roles are occupied by different persons. Correll et al. (2007: Study 1) and Foschi and Valenzuela (2008) are two examples of work that uses this design. Their central common features are as follows. The objective of the session, as described to its participants, is to extend the hiring process to seek, on behalf of one or more business/technology companies, the views and choices of various types of respondents. Participation may be in either individual or group sessions; the data collected, however, are all individual-level responses. In order to highlight the status factors of interest, each person reviews the files of two applicants (fictitious but presented as real) who are different in those respects. The dependent variables include measures of job competence and suitability, choice of applicant, and suggested salary. Job responsibilities could also be added. This design can be readily modified to study nativity and country of credentials as well as associated variables such as sex category and ethnicity/ race, and to sort out their individual effects. Furthermore, the decision to be made need not be about hiring but could concern, for example, a place on a wait-list, a promotion, or a salary increase. The format could also be expanded to incorporate a group discussion at the end, with or without the requirement of achieving consensus on the various assessments the participants are asked to make. (On the related factorial-survey methodology and its application to immigration research, see Jasso, 1988.) For all three of the designs I describe above, instruments have been developed to measure unobtrusively any biased evaluations and/or double standards on the part of the respondents. In the three cases, the cover story about the nature of the session and the characteristics of both the task and the performers are removed during a postexperimental interview that is followed by a full debriefing. It is highly unlikely that respondents would disclose any of their own biases if asked about them directly during the experiment itself. Note also that in the first two types of studies, Comparisons 4, 6, and 8 in Table 1 would appear to be more difficult to implement than the other five, because of the attributes required of Person A as both a performer and an assessor. However, this could be overcome by, for example, recruiting

146

MARTHA FOSCHI

participants from a business organization with a large and diverse number of employees.

ADVANTAGES OF THESE EXPERIMENTAL DESIGNS Experiments, especially those requiring a laboratory setting, are not common in sociological research. Yet, they offer many advantages as a hypothesis-testing instrument. In particular, there are several key benefits to studying effects from nativity and country of credentials through the designs presented here. Some of these advantages pertain to experiments in general, others are linked to the expectation states approach in particular. I outline both types of advantages next. (For a more detailed presentation of some of those points, see McDermott, 2006; Webster & Sell, 2007.)  The factors of interest (be they constants or independent variables) do not emerge naturally in the situation but their levels are specifically set by the experimenter. For example, participants may be equated on sex category and formal education, or they may be asked to assess others who are either similar to or different from themselves on specific respects. These procedures result in control over the level of each factor being investigated and, in turn, facilitate the assessment of its effect(s) on the dependent variable(s). In addition, since more than one factor can thus be established, it is possible to efficiently determine the relative impact of each. In fact, because experiments typically deal with a few variables at a time, they are ideally suited for testing hypotheses that involve a small number of comparisons. Subsequent work can add other variables and thus build on the initial formulation, as discussed below.  Random assignment of a sufficiently large number of persons to different levels of the independent variable(s) ensures that the experimental and control conditions will have highly comparable distributions for those factors that are not directly manipulated. This, in turn, strengthens the inferences that one can draw from the results.  Control groups help assess with precision the effect of a factor or set of factors on the dependent variable(s). For example, let us consider Comparison 1, where A and B differ only in birthplace. As indicated earlier, control-group results on the dependent variable(s) would serve (a) to determine whether nativity carries status for the sample under study and (b) to provide an informative baseline against which to appraise the results from the other conditions. Additional control groups could be

The Professional Credentials of Immigrants

147

created, such as comparisons that investigate the status value, for those taking part, of sex category and field of degree. In all cases, the use of different samples of participants would help assess how well established that value is. Moreover, the value of nativity as status could be measured by dividing participants at random into two groups: one receives a questionnaire designed to obtain views on ‘‘native-born/immigrant’’ (as well as on associated factors and/or fillers), while the second group is assigned to the experiment proper. In this way, those views are obtained less obtrusively and can still be inferred to hold for the second group. For Canadian examples of survey instruments that could be adapted for this purpose, see Berry and Kalin (1995); Esses, Jackson, and Armstrong (1998); Tougas et al. (2004). Corresponding procedures could be used for ‘‘country of degree.’’ For a questionnaire explicitly developed to measure expectation states deriving from any status characteristic, see Zeller and Warnecke (1973); Webster and Driskell (1978) discuss the application of this tool to a case of Black/White interactions.  The studies I propose foster the understanding of evaluations and other decisions at the interactional (group-processes) level, in a way that is more thorough and less reactive than commonly possible through surveys, interviews, census questions, and other types of experiments.  The expectation states program has been characterized by the use of comparable experimental designs. In this way, studies have been planned to examine different aspects of the same topic while retaining common characteristics – a feature that advances cumulativeness of findings. For example, experiments may be constructed to be either more or less ‘‘realistic’’ (i.e., to correspond to various extents to the concrete properties of everyday situations). The three types of designs that I describe here illustrate various levels of that factor. Following this approach, the oftenexpressed concern about the artificiality of experiments can be reduced through a series of experiments that are similar in key respects but vary in the degree to which they resemble familiar situations.  Viewing ‘‘native-born/immigrant’’ and ‘‘country of degree’’ as status characteristics facilitates comparison, within a single theoretical context, with analogous attributes. For example, one can pose questions on whether nativity or sex category has a stronger impact on perceptions of competence. Furthermore, one may use the same experimental design to make such comparisons. For instance, the relative weight of two status factors may be assessed through different control and experimental

148

MARTHA FOSCHI

conditions of the discussion-groups design: some conditions may show variation in nativity but not in sex category while others offer the opposite combination.  Working within a research program also enables authors to build on each other’s projects in yet another way: the studies can be designed to proceed from the investigation of simple situations to more complex ones. Thus, as I have proposed here, one could begin with variations in only two factors – ‘‘native-born/immigrant’’ and ‘‘Canadian/foreign professional degree’’ – and gradually enlarge the focus to investigate, for example, ethnicity/race and language skills of the professionals involved. One could also make the designs more complex by examining, for example, the extent to which ‘‘solidarity with native-born Canadians’’ is a factor in employment recommendations. Such a systematic approach facilitates both theory building and the interpretation of results. In turn, a good theory is always a strong basis for policy development.  The rejection-of-influence design requires special equipment and is usually conducted in a laboratory setting. Similarly, discussion-group sessions have generally been carried out in a laboratory facility in which trained observers record the interaction from behind a one-way mirror. However, the equipment for the former design need not be located in a laboratory, and the latter may take place in a regular seminar-room. Moreover, discussion groups could, alternatively, include trained observers who are said to be doing a related task while, in fact, are recording unobtrusively some aspects of the interaction. Finally, the applicationfiles design does not require any specialized equipment. All three types of experiments may then be held at different locations and implemented with a variety of respondents (e.g., high-school students, residents of senior-citizens homes, business employees). That is, it is not necessary to carry out any of these experiments with undergraduates in a university setting. Often those groups are preferred for reasons of easy accessibility and convenience to the researchers. However, if one puts those reasons aside, the studies I describe can be easily transported to other settings and conducted with other samples. For instance, respondents in the application-files format could be either graduate students in professional schools, or immigration officers, or prospective employers. Note also the rejection-of-influence experiment reported in Berger, Cohen, and Zelditch (1972), in which the participants were US Air Force personnel.  Expectation states work has assembled a variety of techniques that make participants aware of each other’s status without making that

The Professional Credentials of Immigrants

149

information reactive. For example, depending on type of design, those taking part may be told that ‘‘the groups under study have been selected at random from volunteers within the university-student population,’’ or that ‘‘the purpose of the session is to collaborate with industry in the hiring of new graduates.’’ In the latter case, it is also a good idea to ask respondents to make decisions about two types of pairs of applicants, those of interest in the particular study (or ‘‘critical pair’’) and those who are included only to mask the real purpose of the session (e.g., control groups in which the two persons are highly comparable in all key respects). For a discussion of some of these procedures, see Foddy and Riches (2000, pp. 118–120), Goar and Sell (2005, pp. 531–534). Several may be readily adapted to study perceptions of competence based on either nativity or country of credentials, or on both. For a helpful overview of methodological topics in the study of stereotyping, see Schneider (2004, Chapter 2).

SUMMARY AND CONCLUSIONS Although experiments have many advantages, no single method or study can provide all the answers of interest on a given subject. Designing a study within an established research tradition enhances comparability with related work at both the theoretical and methodological levels, and thus facilitates cumulativeness. Using a combination of different methodological approaches and specific techniques to study the same topic helps cancel the limitations of any particular procedures and offers a larger, more complete picture. Expectation states theory provides a sound theoretical background and a wide variety of proven experimental designs, while experiments have been underutilized in immigration studies. In this chapter I present ideas from work on expectation states, biased evaluations, and competence double-standards, and examine ‘‘Canadianborn/immigrant’’ and ‘‘Canadian/foreign credentials’’ as special cases of status variables within that framework. I also review three types of experimental designs and outline the contributions that they can make to the study of those two factors and their consequences, both within and outside the Canadian context. My proposed agenda constitutes a novel approach to the topic. I conclude that this line of research holds much promise of productive results for both the expectation states and the immigration literatures.

150

MARTHA FOSCHI

NOTES 1. In this text I treat the terms ‘‘immigrant’’ and ‘‘foreign-born’’ as synonyms. Similarly, I use ‘‘native-born’’ and ‘‘Canadian-born’’ interchangeably. I also use ‘‘nativity,’’ ‘‘birthplace,’’ and ‘‘country of birth’’ to refer to the ‘‘native-born/ immigrant’’ distinction. I abbreviate ‘‘professional credentials (or degree)’’ to ‘‘credentials (or degree),’’ and refer to them in terms of either ‘‘national/foreign’’ or ‘‘country of credentials.’’ ‘‘Task ability,’’ ‘‘competence,’’ ‘‘talent,’’ and ‘‘skill’’ are treated as equivalent. The expression ‘‘job prospects’’ refers to work that is commensurate with a person’s professional qualifications. For simplicity, I focus on the clearly contrasting labels of ‘‘native-born’’ and ‘‘immigrant.’’ Thus, I exclude from my discussion those former immigrants who have now acquired Canadian citizenship. As my interest is on the attributes that are associated with the term ‘‘immigrant’’ alone, I do not distinguish among the different ways by which a person may be admitted as an immigrant (e.g., through the standard point-system, as an ‘‘economic-class’’ applicant, as a family-sponsored applicant, or as a refugee), although in all cases I assume that he/she is in Canada legally (see Li, 2003, pp. 39–44 for a description of the legal categories for entry into Canada). Since I am interested in ‘‘country of credentials’’ and ‘‘native-born/immigrant’’ as status variables, here they should be understood as abbreviations of ‘‘perceived prestige (worth) of either country of credentials or country of birth.’’ 2. My interest in the present work is on that maintenance and its consequences. On the precedent question of how a characteristic acquires status, see Ridgeway (1991, 2006)’s social construction theory and Berger and Fis- ek (2006)’s social value theory. About the latter, and of particular interest here, see Berger and Fis- ek (2008)’s on the acquisition of status value by ‘‘native-born/immigrant’’ in two historical cases from the last century: the immigration of Turks to Germany and that of Chinese to Malaysia. Also writing within the context of status acquisition, Walker, Webster, and Bianchi (2011, p. 1661) outline an interpretation of ‘‘native’’ and ‘‘immigrant’’ as status factors; for a useful clarification of the term ‘‘status,’’ see Jasso (2011, p. 1299). 3. If the status factor has more than two states (e.g., three groupings of skin color or four categories of socioeconomic class) and each activates a different standard, then the standard is multiple. For simplicity, however, here I use ‘‘double standards’’ as the generic term. The practice also occurs if the performances have been judged to be equally poor, and the higher-status person is treated less harshly in the assessment of incompetence. Note as well that there are situations in which the double standard is reversed, and the lower-status person is favored. These include contexts in which, for example, either a same-sex preference or an intention to redress past inequalities carries more weight than the assessment of competence. For a discussion of this topic and empirical results, see, for example, Foschi and Valenzuela (2008). 4. Experiments are often classified as either ‘‘laboratory’’ or ‘‘field.’’ A discussion of that classification is beyond the objectives of this chapter. It is important to emphasize, however, that regardless of location and type of facilities/equipment used, a study is an experiment only if random assignment to experimental and control conditions has taken place. Since respondents in a study cannot be assigned at random to levels of attributes such as sex category or ethnicity/race, these and similar factors (often called ‘‘organismic’’) are only quasiexperimental variables.

The Professional Credentials of Immigrants

151

Note, however, that random assignment is possible regarding the presentation of those attributes in others. 5. In particular, since immigrants to Canada come from many countries and a broad variety of ethnic/racial backgrounds, indirect evidence of the status value of ‘‘nativity’’ is also provided by studies on indicators of ethnicity/race (particularly skin color, foreign accent, first and last name, and country of birth) and the relationship between those factors and employment, income, and, in general, attributed competence (see, e.g., Creese, 2010; Foschi & Buchan, 1990; Henry & Ginzberg, 1985; Kalin & Rayko, 1978; Oreopoulos, 2011; Pendakur & Pendakur, 2002; Satzewich & Li, 1987; Tuzlak & Moore, 1984). Notice also that (a) the nativeborn as well differ in ethnicity/race, and (b) in many cases, immigrants who are labeled as ‘‘ethnics’’ in Canada were not identified as such in their countries of origin (e.g., because of skin-color homogeneity in those countries). As a result, once in Canada such immigrants find themselves having to cope with an imposed, new social identity. 6. To place this chapter in its larger context, I should also mention that several measures have been taken (and others are currently being developed) to standardize the accreditation of national and foreign professional credentials. Thus, (a) as part of the point system of admission to Canada, foreign degrees are assessed by consular offices at the location where prospective immigrants make their applications, (b) once in Canada, in some cases those credentials are also evaluated by professional associations, and (c) several offices at various levels of government provide assistance to newly arrived immigrants regarding such evaluation (e.g., the Canadian Government’s new Foreign Credentials Referral Office; Alberta’s Immigrant Access fund; the Toronto Region Immigrant Employment Council Programs). The effectiveness of these measures in influencing the views of the Canadian population at large and those of prospective employers in particular has yet to be determined. 7. I should note that there are also designs with comparable features to the ones I describe in this text, but that those designs originate in other traditions. That is, there are other types of studies that use discussion-groups, dyadic-interaction, and assessment-of-applicants formats. My focus here is only on those designs that have been put forward within the context of expectation states theory.

ACKNOWLEDGMENTS I thank Emma Dewit, Neil Guppy, Michael Kehl, and Maria Zeldis for their valuable comments and suggestions on an earlier text of this chapter.

REFERENCES Basran, G. S., & Zong, L. (1998). Devaluation of foreign credentials as perceived by visible minority professional immigrants. Canadian Ethnic Studies, 30, 6–23.

152

MARTHA FOSCHI

Berger, J., & Fis- ek, M. H. (2006). Diffuse status characteristics and the spread of status value: A formal theory. American Journal of Sociology, 111, 1038–1079. Berger, J., & Fis- ek, M. H. (2008, September). Applying the spread of status value theory. Paper presented at the First World Forum of Sociology of the International Sociological Association. Barcelona, Spain. Berger, J., Cohen, B. P., & Zelditch, M., Jr. (1972). Status characteristics and social interaction. American Sociological Review, 37, 241–255. Berger, J., & Webster, M., Jr. (2006). Expectations, status, and behavior. In P. J. Burke (Ed.), Contemporary social psychological theories (pp. 268–300). Stanford, CA: Stanford University Press. Berry, J. W., & Kalin, R. (1995). Multicultural and ethnic attitudes in Canada: An overview of the 1991 National Survey. Canadian Journal of Behavioural Science, 27, 301–320. Bianchi, A. J. (2004). Rejecting others’ influence: Negative sentiment and status in task groups. Sociological Perspectives, 47, 339–355. Biernat, M. (2005). Standards and expectancies: Contrast and assimilation in judgments of self and others. New York, NY: Psychology Press. Boyd, M., & Thomas, D. (2002). Skilled immigrant labour: Country of origin and the occupational locations of male engineers. Canadian Studies in Population, 29, 71–99. Brief, A. P. (Ed.). (2008). Diversity at work. New York, NY: Cambridge University Press. Correll, S., Benard, S., & Paik, I. (2007). Getting a job: Is there a motherhood penalty? American Journal of Sociology, 112, 1297–1338. Correll, S., & Ridgeway, C. L. (2003). Expectation states theory. In J. Delamater (Ed.), Handbook of social psychology (pp. 29–51). New York, NY: Kluwer Academic/Plenum. Creese, G. (2010). Erasing English language competency: African migrants in Vancouver, Canada. International Migration & Integration, 11, 295–313. DiTomaso, N., Post, C., Smith, D. R., Farris, G. F., & Cordero, R. (2007). Effects of structural position on allocation and evaluation decisions for scientists and engineers in industrial R&D. Administrative Science Quarterly, 52, 175–207. Esses, V. M., Dietz, J., & Bhardwaj, A. (2006). The role of prejudice in the discounting of immigrant skills. In R. Mahalingam (Ed.), Cultural psychology of immigrants (pp. 113–130). Mahwah, NJ: Lawrence Erlbaum. Esses, V. M., Dovidio, J. F., Semenya, A. H., & Jackson, L. M. (2005). Attitudes toward immigrants and immigration: The role of national and international identity. In D. Abrams, M. A. Hogg & J. M. Marques (Eds.), The social psychology of inclusion and exclusion (pp. 317–337). New York, NY: Psychology Press. Esses, V. M., Jackson, L. M., & Armstrong, T. L. (1998). Intergroup competition and attitudes toward immigrants and immigration: An instrumental model of group conflict. Journal of Social Issues, 54, 699–724. Ferrer, A., & Riddell, W. C. (2008). Education, credentials, and immigrant earnings. Canadian Journal of Economics, 41, 186–216. Foddy, M., & Riches, P. (2000). The impact of task and categorical cues on social influence: Fluency and ethnic accent as cues to competence in task groups. Advances in Group Processes: A Research Annual, 17, 103–130. Foddy, M., & Smithson, M. (1999). Can gender inequalities be eliminated? Social Psychology Quarterly, 62, 307–324. Fong, E., & Cao, X. (2009). Effects of foreign education on immigrant earnings. Canadian Studies in Population, 36, 87–110.

The Professional Credentials of Immigrants

153

Foschi, M. (1989). Status characteristics, standards, and attributions. In J. Berger, M. Zelditch, Jr. & B. Anderson (Eds.), Sociological theories in progress: New formulations (pp. 58–72). Newbury Park, CA: Sage. Foschi, M. (1996). Double standards in the evaluation of men and women. Social Psychology Quarterly, 59, 237–254. Foschi, M. (2000). Double standards for competence: Theory and research. Annual Review of Sociology, 26, 21–42. Foschi, M., & Buchan, S. (1990). Ethnicity, gender, and perceptions of task competence. Canadian Journal of Sociology, 15, 1–18. Foschi, M., & Valenzuela, J. (2008). Selecting job applicants: Effects from gender, selfpresentation, and decision type. Social Science Research, 37, 1022–1038. Fuller, S., & Vosko, L. F. (2008). Temporary employment and social inequality in Canada: Exploring intersections of gender, race and immigrant status. Social Indicators Research, 88, 31–50. Garnett, B., Guppy, N., & Veenstra, G. (2008). Careers open to talent: Educational credentials, cultural talent, and skilled employment. Sociological Forum, 23, 144–164. Goar, C., & Sell, J. (2005). Using task definition to modify racial inequality within task groups. The Sociological Quarterly, 46, 525–543. Goldberg, P. (1968). Are women prejudiced against women? Transaction, 5, 28–30. Grant, P. R., & Nadin, S. (2007). The credentialing problems of foreign trained personnel from Asia and Africa intending to make their home in Canada: A social psychological perspective. International Migration & Integration, 8, 141–162. Henry, F., & Ginzberg, E. (1985). Who gets the work: A test of racial discrimination in employment. Toronto, ON: The Urban Alliance on Race Relations and The Social Planning Council of Metropolitan Toronto. Herring, C. (2009). Does diversity pay? Race, gender, and the business case for diversity. American Sociological Review, 74, 208–224. Hysom, S. J. (2009). Status valued goal objects and performance expectations. Social Forces, 87, 1623–1648. Jasso, G. (1988). Whom shall we welcome? Elite judgments of the criteria for the selection of immigrants. American Sociological Review, 53, 919–932. Jasso, G. (2011). Migration and stratification. Social Science Research, 40, 1292–1336. Kalin, R., & Rayko, D. S. (1978). Discrimination in evaluative judgments against foreignaccented job candidates. Psychological Reports, 43, 1203–1209. Kler, P. (2006). Graduate overeducation and its effects among recently arrived immigrants to Australia: A longitudinal survey. International Migration, 44, 93–128. Leahey, E. (2004). The role of status in evaluating research: The case of data editing. Social Science Research, 33, 521–537. Li, P. S. (2001). The market worth of immigrants’ educational credentials. Canadian Public Policy, 27, 23–38. Li, P. S. (2003). Destination Canada: Immigration debates and issues. Don Mills, ON: Oxford University Press. McDermott, R. (2006). Guest editor’s introduction: Special issue on Experiments in political psychology. Political Psychology, 27, 347–358. McKenzie, D., & Yang, D. (2012). Experimental approaches in migration studies. In C. VargasSilva (Ed.), Handbook of research methods in migration (pp. 249–269). Cheltenham, UK: Edward Elgar.

154

MARTHA FOSCHI

Moore, J. C., Jr. (1968). Status and influence in small group interactions. Sociometry, 31, 47–63. Nakhaie, M. R. (2007). Ethnoracial origins, social capital, and earnings. International Migration & Integration, 8, 307–325. Oreopoulos, P. (2011). Why do skilled immigrants struggle in the labor market? A field experiment with thirteen thousand re´sume´s. American Economic Journal: Economic Policy, 3, 148–171. Palmer, D. L. (1996). Determinants of Canadian attitudes toward immigration: More than just racism? Canadian Journal of Behavioural Science, 28, 180–192. Pendakur, K., & Pendakur, R. (1998). The colour of money: Earning differentials among ethnic groups in Canada. Canadian Journal of Economics, 31, 518–548. Pendakur, K., & Pendakur, R. (2002). Colour my world: Have earnings gaps for Canadianborn ethnic minorities changed over time? Canadian Public Policy, 28, 489–512. Pugh, M. D., & Wahrman, R. (1983). Neutralizing sexism in mixed-sex groups: Do women have to be better than men? American Journal of Sociology, 88, 746–762. Reitz, J. G. (2001a). Immigrant success in the knowledge economy: Institutional change and the immigrant experience in Canada, 1970–1995. Journal of Social Issues, 57, 579–613. Reitz, J. G. (2001b). Immigrant skill utilization in the Canadian labour market: Implications of human capital research. Journal of International Migration and Integration, 2, 347–378. Riches, P., & Foddy, M. (1989). Ethnic accent as a status cue. Social Psychology Quarterly, 52, 197–206. Ridgeway, C. L. (1991). The social construction of status value: Gender and other nominal characteristics. Social Forces, 70, 367–386. Ridgeway, C. L. (2006). Status construction theory. In P. J. Burke (Ed.), Contemporary social psychological theories (pp. 301–323). Stanford, CA: Stanford University Press. Satzewich, V., & Li, P. S. (1987). Immigrant labor in Canada: The cost and benefit of ethnic origin in the job market. Canadian Journal of Sociology, 12, 229–241. Scassa, T. (1994). Language standards, ethnicity and discrimination. Canadian Ethnic Studies, 26, 105–121. Schneider, D. J. (2004). The psychology of stereotyping. New York, NY: The Guilford Press. Shelly, R. K., & Munroe, P. T. (1999). Do women engage in less task behavior than men? Sociological Perspectives, 42, 49–67. Shelly, R. K., & Webster, M., Jr. (1997). How formal status, liking, and ability status structure interaction: Three theoretical principles and a test. Sociological Perspectives, 40, 81–107. Somerville, K., & Walsworth, S. (2009). Vulnerabilities of highly skilled immigrants in Canada and the United States. American Review of Canadian Studies, 147–161. Stewart, P. A., & Moore, J. C., Jr. (1992). Wage disparities and performance expectations. Social Psychology Quarterly, 55, 78–85. Top, T. J. (1991). Sex bias in the evaluation of performance in the scientific, artistic, and literary professions: A review. Sex Roles, 24, 73–106. Tougas, F., Desruisseaux, J. C., Desrochers, A., St-Pierre, L., Perrino, A., & de la Sablonnie`re, R. (2004). Two forms of racism and their related outcomes: The bad and the ugly. Canadian Journal of Behavioural Science, 36, 177–189. Tuzlak, A., & Moore, J. C., Jr. (1984). Status, demeanor and influence: An empirical reassessment. Social Psychology Quarterly, 47, 178–183. Tuzlak, A. S. (1989). Joint effects of race and confidence on perceptions and influence: Implications for Blacks in decision-making positions. Canadian Ethnic Studies, 21, 103–119.

The Professional Credentials of Immigrants

155

Wagner, D. G., & Berger, J. (2002). Expectation states theory: An evolving research program. In J. Berger & M. Zelditch, Jr. (Eds.), New directions in contemporary sociological theory (pp. 41–76). Lanham, MD: Rowman & Littlefield. Wagner, D. G., Ford, R. S., & Ford, T. W. (1986). Can gender inequalities be reduced? American Sociological Review, 51, 47–61. Walker, L. S., Webster, M., Jr., & Bianchi, A. J. (2011). Testing the spread of status value theory. Social Science Research, 40, 1652–1663. Wang, S., & Lo, L. (2005). Chinese immigrants in Canada: Their changing composition and economic performance. International Migration, 43, 35–71. Ward, C., & Masgoret, A. M. (2007). Immigrant entry into the workforce: A research note from New Zealand. International Journal of Intercultural Relations, 31, 525–530. Webster, M., Jr. (1969). Source of evaluations and expectations for performance. Sociometry, 32, 243–258. Webster, M., Jr., & Driskell, J. E., Jr. (1978). Status generalization: A review and some new data. American Sociological Review, 43, 220–236. Webster, M., Jr., & Sell, J. (2007). Why do experiments? In M. Webster, Jr. & J. Sell (Eds.), Laboratory experiments in the social sciences (pp. 5–23). Burlington, MA: Elsevier. Zelditch, M., Jr., Lauderdale, P., & Stublarec, S. (1980). How are inconsistencies between status and ability resolved? Social Forces, 58, 1025–1043. Zeller, R. A., & Warnecke, R. B. (1973). The utility of intervening constructs in experiments. Sociological Methods & Research, 2, 85–110. Zeng, Z., & Xie, Y. (2004). Asian-Americans earning disadvantage reexamined: The role of place of education. American Journal of Sociology, 109, 1075–1108.

SELECTION AND INFLUENCE IN THE ASSIMILATION PROCESS OF IMMIGRANTS Arnout van de Rijt ABSTRACT Purpose – Empirical studies show substantial variation across immigrants in the rate and direction of assimilation along various dimensions (e.g., cross-ethnic contact, language, identity). To explain this variation, past research has focused on identifying exogenous factors, such as discrimination, human capital, and settlement intention. In this chapter we argue that variation in immigrant outcomes emerges endogenously through positive interaction effects between dimensions of assimilation. We propose a new assimilation model in which processes of social influence and selection into congruent social environments give rise to multiple long-term equilibria. In this model, migrants who are already assimilated along many dimensions tend to also adapt along other dimensions, while less assimilated migrants become more strongly embedded in their ethnic group. Design/methodology/approach – To test the assimilation model, we derive a number of hypotheses, which we evaluate using trend analysis and dynamic panel regression on data from the Longitudinal Survey of Immigrants to Canada. Advances in Group Processes: Thirtieth Anniversary Edition Advances in Group Processes, Volume 30, 157–193 Copyright r 2013 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 0882-6145/doi:10.1108/S0882-6145(2013)0000030010

157

158

ARNOUT VAN DE RIJT

Findings – The data mostly confirm the hypotheses, providing overall support for the assimilation model. Research implications – Our theory and findings suggest that immigrants would follow divergent assimilation trajectories even in the absence of a priori population heterogeneity in external factors. Social implications – The positive interaction effects between cultural and structural dimensions of assimilation suggest that mixed policies that promote integration while seeking to prevent loss of identity go against the natural tendency for cultural and structural assimilation to go hand in hand. Originality/value – The present chapter proposes a novel model of immigrant assimilation and an empirical test. Keywords: Assimilation; dissonance; selection; influence; homophily

INTRODUCTION A classic thesis in the immigration literature is that immigrants over time adapt to the cultural mainstream of the host country (Gans, 1973; Gordon, 1964; Park, 1950; Park & Burgess, 1921; Sandberg, 1973; Shibutani & Kwan, 1965; Warner & Srole, 1945). ‘‘Assimilation,’’ sometimes referred to as ‘‘incorporation,’’ ‘‘integration,’’ or ‘‘adaptation,’’ may involve a shift away from exclusive use of the mother tongue, decreasing ethnic identification, abandonment of homeland practices, and the diminishing of coethnic work, family, and friendship relations. While assimilation can take place over multiple generations (Alba & Nee, 2003; Brown & Bean, 2006; Perlmann & Waldinger, 1997; Portes & Zhou, 1993; Rumbaut & Portes, 2001), the focus of this chapter is on the dynamics of individual assimilation, whereby individual migrants incrementally adapt to the host culture along social and cultural dimensions over the course of their stay in the host country. In contrast to the uniform trajectory of increasing assimilation from ethnic group to mainstream posited by ‘‘straight-line’’ immigration theory, empirical evidence suggests that immigrant cohorts exhibit great heterogeneity in the pace and even the direction of adaptive change. Large variation has been observed in official language proficiency

Assimilation Process of Immigrants

159

(Van Tubergen & Kalmijn, 2009a, 2009b) and job transitions from ethnic economies into multiethnic labor markets (Nee, Sanders, & Sernau, 1994; Sanders, Nee, & Sernau, 2002). Surprising levels of continuity have been found for ethnic church attendance (Hurh & Kim, 1990), civic participation (Fong & Ooka, 2006), and residential segregation of various ethnic communities (Alba, Logan, Stults, Marzan, & Zhang, 1999; White, Biddlecom, & Guo, 1993; Zhou & Logan, 1991). Some studies even find reverse trends, such as increasing lifetime fertility rates (Rose, 1942), growing levels of transnational engagement (Guarnizo, Portes, & Haller, 2003; Itzigsohn & Saucedo, 2002; Portes, Guarnizo, & Haller, 2002), and strengthening ethnic self-identification (Golash-Boza, 2006; Zhou, 2001, pp. 201–202). While the majority of empirical work on immigrant assimilation attends to the movement of immigrants along distinct dimensions (language, contact, employment, identification) independently, the very factors that promote or prevent adaptive changes along one dimension are often themselves forms of assimilation along other dimensions. For example, cross-ethnic contact has been shown to facilitate language acquisition (Van Tubergen & Kalmijn, 2009b), and language skills in turn enable cross-ethnic contact (Kalmijn & Van Tubergen, 2010; Martinovic, Van Tubergen, & Maas, 2011). Language skills also promote national self-identification (De Vroome, Coenders, Van Tubergen, & Verkuyten, 2011), which stimulates cross-ethnic contact through shared traditions (Martinovic et al., 2011). This suggests that the combined whole of adaptive change is more than just the sum of its parts. Consequently, the interaction between dimensions of assimilation gives rise to important systemic features of the overall process that are missed when analyzing one component at a time. We argue that these interaction effects may be particularly important for understanding the empirical diversity in adaptation trajectories that lead some immigrants to integration in the majority group while others remain in ethnic enclaves (Sanders & Nee, 1987; Wilson & Portes, 1980) or blend with a different minority group. While exogenous factors such as contexts of reception (Golash-Boza, 2006; Portes & Rumbaut, 1996; Rumbaut & Portes, 2001), forms of capital (Alba & Nee, 2003; Nee & Sanders, 2001), and settlement intention (Paul, 2011; Van Tubergen & Kalmijn, 2009b) have been demonstrated to affect immigrant outcomes, we suggest that the interaction across dimensions of immigrant assimilation can also, endogenously, steer these dynamics. In what follows, we propose a model of assimilation grounded in the notion that assimilation along one dimension reinforces assimilation along other dimensions. The model shows how heterogeneity in trajectories and outcomes across immigrants can emerge

160

ARNOUT VAN DE RIJT

from such endogenous features of the assimilation process, even when exogenous factors are held constant. To do this, we study immigrant adaptation as a multidimensional process. Dimensions of assimilation, cultural and structural, tend to interact following principles of ‘‘selection’’ and ‘‘influence,’’ with important implications for the overall dynamics of immigrant assimilation. We test predictions derived from our model with data from the Longitudinal Survey of Immigrants to Canada (LSIC) on six parallel dimensions of adaptive change during the early settlement years of a recent immigrant cohort. The multidimensional model we propose complements previous efforts at understanding the variation in levels of assimilation among immigrant populations. We show that while exogenous factors can also be very important for understanding this process, assimilation outcomes can be highly varied even in the absence of such a priori structural heterogeneity in immigrant cohorts.

LITERATURE Intergenerational versus Individual Assimilation One way by which adaptive changes may occur is from one generation to the next, where the children of immigrants are more assimilated than their parents and less than their own children. Critics of intergenerational assimilation have argued that for many immigrant groups in the United States, the road to integration and adaptation remains blocked (e.g., Glazer & Moynihan, 1970), that more recent immigrants to the United States from non-European origin countries do not fit the traditional assimilation model (Portes & Zhou, 1993; Zhou, 1997) and that often ethnic identities remains strong in later generations even as workplace boundaries are penetrated and social networks become more ethnically diverse (Brown & Bean, 2006). Yet others argue using quantitative data on multiple generations that the overall trend nevertheless appears monotonic and progressive for many countries of origin (e.g., Alba & Nee, 2003; Pagnini & Morgan, 1990) and destination (e.g., Lee & Boyd, 2008). The focus in this chapter is on a second form of assimilation, individual assimilation, where individual migrants over the course of their stay in the host country bit-by-bit adapt along behavioral dimensions and establish new relationships.

Assimilation Process of Immigrants

161

Mixed Evidence for Individual Assimilation The body of evidence for individual assimilation is rather mixed. The magnitude and direction of assimilation varies heavily across and within studies. Some find support for classic predictions of forward movement. Massey, Alarco´n, Durand, and Gonza´lez (1987) and Martinovic, Van Tubergen, and Maas (2009a) report an average trend of increasingly crossethnic friendship networks among immigrants to respectively the United States and the Netherlands. Support has also been found for job transitions out of the ethnic economy (Nee et al., 1994; Roth, Seidel, Ma, & Lo, 2012), increased national attachment (Waldinger, 2007), and civic participation (Stoll & Wong, 2007) in the Los Angeles area. Trilla, Esteve, and Domingo (2008) found changing intermarriage rates of immigrants to Spain in accordance with classic assimilation theory. However, while proficiency in the official language generally increases, rates vary greatly within and across ethnic groups and destination countries (Akresh, 2007; Bean & Stevens, 2003; Carliner, 2000; Da´vila & Mora, 2000; Espenshade & Fu, 1997; Massey & Espinosa, 1997; Grenier, 1984; Massey et al., 1987; McConnell & LeClere, 2002; Medvedeva, 2012; Stevens, 1992, 1999; Van Tubergen & Kalmijn, 2005, 2009a, 2009b). Furthermore, findings regarding weakening identification with various origin countries are equivocal, with many studies failing to find any average trend (Ajrouch & Jamal, 2007; Berry, 2006; Diehl & Schnell, 2006; Ersanilli & Saharso, 2011; Matute-Bianchi, 1986; Sears et al., 2003; Zimmerman, Zimmermann, & Constant, 2007). Also, employment trajectories in Canada have been found to vary greatly within migrant cohorts, with some quickly integrating into cross-ethnic workplaces while others remain excluded (Fuller & Martin, 2012). And while some immigrant groups show an average trend toward residential integration, members of other groups appear to remain spatially segregated (Alba et al., 1999; White et al., 1993; Zhou & Logan, 1991). Hurh and Kim (1990) found persistent attendance of the Korean ethnic church among Korean immigrants long after arrival. While Anderson and Milligan (2010) find a positive duration effect for voluntary association membership, Fong and Ooka (2006) find no effect of duration on civic participation in Canada. Some scholars even find reverse tendencies. Rose (1942) observed climbing, not falling lifetime fertility rates among Italian immigrants in the United States – away from the national average. Colombian, Dominican, and Salvadorian immigrants to the United States have been observed to increase, not reduce their engagement in political, social, and business

162

ARNOUT VAN DE RIJT

activities involving the country of origin (Guarnizo et al., 2003; Itzigsohn & Saucedo, 2002; Portes et al., 2002). And discriminatory experiences have been found to strengthen ethnic identity (Golash-Boza, 2006). Finally, some find diverging trends even within relatively homogeneous groups. Esser (1987) observes polarization in the friendship networks of Yugoslavian and Turkish immigrants to Germany, with some integrating while others segregate. Zhou (2001) finds a similar divergence in the ethnic selfidentification of Vietnamese immigrants, with mixed identities becoming less common over time (pp. 201–202).

Antecedents of Assimilation This high degree of variation in the magnitude and direction of individual assimilation within and across groups has led scholars to search for initial conditions under which adaptive changes occur and factors that inhibit assimilation or even lead to movement away from integration and acculturation. Among these antecedents are contexts of reception and family structure (Portes & Rumbaut, 1996) as well as forms of capital (Nee & Sanders, 2001) which ‘‘shape the trajectory of their incorporation into the host society’’ (p. 386).1 Discrimination differentiates ethnic groups in terms of their job market success and as a consequence leads some to exhibit strengthened ethnic self-identification (De Vroome et al., 2011; GolashBoza, 2006; Portes & Rumbaut, 1996; Zhou, 2001). Family ties and acquaintances (social capital) and foreign education (human capital) condition workplace integration (Nee & Sanders, 2001; Nee et al., 1994; Portes & Sensenbrenner, 1993; Roth et al., 2012) while at the same time increasing political and economic involvement in the country of origin (Guarnizo et al., 2003; Portes et al., 2002). Other a priori determinants of assimilation are migration motive and settlement intention, both of which affect language skills and use (Van Tubergen & Kalmijn, 2009a, 2009b), and immigrant group size which is inversely related to intermarriage (Blau & Schwartz, 1984; Kalmijn & Van Tubergen, 2010). These theoretical strategies provide exogenous explanations for how alternative conditions differentiate immigrant cohorts into distinct assimilation trajectories and outcomes. However, many of the conditions that studies have identified as exogenous precursors of assimilation are themselves forms of assimilation. For example, ‘‘ethnicity of spouse’’ and ‘‘ethnic composition of social networks’’ are both assimilation dimensions as well as important determinants of adaptive changes in language use

Assimilation Process of Immigrants

163

and proficiency (Van Tubergen & Kalmijn, 2009a, 2009b). Language skills and practice, in turn, fuel cross-ethnic contact (Kalmijn & Van Tubergen, 2010; Martinovic, Van Tubergen, & Maas, 2009b; Roth et al., 2012) and national self-identification (De Vroome et al., 2011). First- and second-generation immigrants are more likely to marry someone outside their ethnic group if they are linguistically more proximate to other groups (Kalmijn & Van Tubergen, 2010). And migrants more readily form interethnic friendships if they already have an interethnic partner and embrace the traditions of the host country (Martinovic et al., 2011). Each of the above examples represents a positive cross-dimensional effect; one type of adaptive change reinforces another, while lack of change in one dimension inhibits assimilation along another dimension. Below we develop a model of how these reinforcing factors may lead to diverging trajectories of either increasing or decreasing levels of overall assimilation. (For a simple formalization of this model see the appendix).

A MODEL OF INDIVIDUAL ASSIMILATION The various positive cross-dimension effects, we argue, can be brought back to two governing processes in the dynamics of social networks: ‘‘selection’’ and ‘‘influence’’ (Aral, Muchnik, & Sundararajan, 2009; Christakis & Fowler, 2007; Crandall, Cosley, Huttenlocher, Kleinberg, & Suri, 2008; Kandel, 1978; Lazarsfeld & Merton, 1954; McPherson, Smith-Lovin, & Cook, 2001; Steglich, Snijders, & Pearson, 2010; Wasserman & Faust, 1994). Selection occurs when an individual has a tendency to disproportionately initiate and strengthen relationships with those who are similar to them in the cultural beliefs they hold and the behavior they display, while limiting or eliminating contact with dissimilar others. Influence occurs when an individual comes to hold beliefs and exhibit behavior that is similar to the beliefs and behavior present in their immediate social environment. A variety of basic social–psychological mechanisms underpin these dual processes, such as the tendency to reduce discomfort stemming from conflicting cultural beliefs and behavioral expectations (Festinger, 1957; Heider, 1946; Tazelaar, 1980), the tendency to adopt the identity and behaviors from one’s immediate social environment (Akers, 1985; Bandura, 1977), and the tendency to seek interaction with members of one’s in-group (Alba, 1990; Huston & Levinger, 1978; Tajfel & Turner, 1979). Selection and influence are further driven by exposure effects from initially foreign beliefs, words, and behaviors in interaction with dissimilar others and the

164

ARNOUT VAN DE RIJT

greater practicality of contact between individuals who share a language and belief system. The end result of selection and influence processes is homophily, the disproportionate similarity of individuals to their social contacts (Centola, 2011; Lazarsfeld & Merton, 1954; McPherson et al., 2001; Newman, 2002; Watts, Dodds, & Newman, 2002). If for the sake of theoretical parsimony we assume a simplistic binary reality with only two groups2,3 – the migrant’s ethnic group and a relatively homogeneous mainstream – then homophily in assimilation is maximal in two circumstances. Homophily is maximal when interacting with members of the mainstream while simultaneously adhering to mainstream values and culturally prescribed behaviors. But maximal homophily also occurs at the other extreme, when the migrant is immersed in coethnic neighborhoods and workplaces and at the same time identifies, speaks the language of, holds a system of beliefs, and behaves all in accordance with the ethnic group of origin. In mixed situations between these two extremes, incongruities exist between the migrant’s own beliefs and behaviors and those prevalent in the migrant’s immediate social environment.4 Two types of changes can resolve these incongruities. First, migrants can seek changes in their social environment that produce a better match with their own behavior and beliefs (selection). Alternatively, migrants may adapt their behavior and beliefs to those of their social environment (influence). Following a common convention in the immigration literature we will distinguish between changes to the social environment of the migrant and changes to her own actions, skills, and mindset as (respectively) structural and cultural assimilation (Berry, 1997, 2006; Berry et al., 1987; Gans, 1997; Gordon, 1954, 1964; Rosenthal, 1960). Structural and cultural assimilation can occur either in ‘‘forward’’ direction away from the migrant’s ethnic group, following a traditional straight-line understanding of assimilation, or ‘‘in reverse,’’ toward the migrant’s ethnic group, with both migrant and her contacts becoming further embedded in the ethnic community of origin. This application of social network processes to immigration provides an account of empirical variance in assimilation outcomes. Namely, assimilation toward the mainstream is one way by which homophily is accomplished; immigrants surround themselves with members of another culture and adapt their beliefs and behavior to match those of their new friends, family, and colleagues. But it is not the only way. Often an easier route to congruence is maintenance of ethnic beliefs and traditions and immersion in neighborhoods and workplaces where these beliefs and traditions are shared. The striving for social accordance thus provides an

165

Assimilation Process of Immigrants

explanation for diverse trajectories and outcomes. Which path – toward or away from the ethnic group – is of lesser resistance depends on situational conditions of the migrant. Migrants with a reasonable command of the host language, some familiarity with mainstream customs, and a job in the main economy may proceed to assimilate. Those with poor language skills, an already strong ethnic identity, and largely co-ethnic job and friendship networks more naturally remain in or recede back into the ethnic community. This theory of homophilous assimilation can then be summarized as follows: Selection and influence form a ‘dual engine’ of assimilation, resulting in a bi-directional process. Moderately assimilated migrants move toward the mainstream, while less assimilated migrants remain with or recede back to their ethnic group. Fig. 1 illustrates how the direction in which homophily drives assimilation depends on the number of dimensions along which a migrant is currently assimilated. A total of six distinct dimensions of assimilation are assumed (corresponding with six empirical measures that we introduce later): ethnic identification, attachment to ethnic values and traditions, command of the host language, co-ethnic friendship network, co-ethnic workplace, and co-ethnic household members. Three dimensions – identity, values, language skills – pertain to the migrant herself and are cultural dimensions of assimilation, while the remaining three measures – friendship, workplace, and household integration – concern her social environment and are therefore structural dimensions of assimilation. Homophily is maximal for a migrant in two extreme cases, namely either if assimilated along none of the dimensions (outer left) or if assimilated along all six of them (outer right). When migrants identify strongly with their ethnic group of origin, speak their first language, and wish to carry on the values and traditions of their homeland while at the same time maintaining many co-ethnic friendship, work, and family relations (# assimilated dimensions ¼ 0), there is congruence between self and other. Yet there is congruence also in a situation where communication is done in the host language, when homophily

homophily 0

1

2

3

4

5

6

# assimilated dimensions

Fig. 1. Homophily as a Function of Assimilation. Shown is homophily as a function of the number of dimensions along which a migrant is assimilated. Arrows indicate whether reverse (pointing left) or forward (pointing right) assimilation leads to greater homophily.

166

ARNOUT VAN DE RIJT

simultaneously ethnic identification and cultural attachment are weak and when at the same time household, friendship, and work ties are largely with the mainstream (# assimilated dimensions ¼ 6).5

Hypotheses From this model of homophilous assimilation, several hypotheses can be derived. First, the model suggests that an immigrant will typically move from an initial ‘‘distressed’’ state of incongruence between self and social environment to one of two homophilous equilibria, where she is likely to remain. It follows that as more and more immigrants reach a state of homophily, with time we should see fewer and fewer changes occur in the networks and behaviors of an immigrant cohort. Hypothesis 1. Assimilation exhibits ‘‘deceleration’’: As time goes by, fewer and fewer changes occur along structural and cultural dimensions. Also, as Fig. 1 illustrates, the direction of assimilation along any one structural or cultural dimension depends on the number of dimensions a migrant is already assimilated along, because dimensions of assimilation reinforce one another: Hypothesis 2. Assimilation exhibits ‘‘reinforcement’’: The more assimilated an immigrant is along any one dimension, the more (less) likely s/he will assimilate forward (in reverse) along other dimensions. An additional consequence of the tendency for homophily is that migrants during the adjustment process will tend to spend relatively little time in partially assimilated states – which are nonhomophilous – and a lot of time in either a state of nonassimilation or full assimilation along all dimensions. Hypothesis 3. Assimilation exhibits ‘‘parallel change’’: A migrant will at any point during her adjustment process tend to be assimilated either along none or along all of the dimensions, but not partially assimilated. In the long run the bidirectional movement that the tendency for homophily sets in motion should converge on one of the two homophilous states. As time goes by more migrants reach such a state so that less and less change should occur. If this increasing stability is indeed the result of convergence on homophilous states, then the probability of remaining in a

Assimilation Process of Immigrants

167

homophilous state would have to be greater than the probability of remaining in a nonhomophilous state. Migrants who are unassimilated along all dimensions or assimilated along all dimensions should be less subject to change than migrants who are partially assimilated. In other words, unassimilated migrants and fully assimilated migrants more likely remain in that state than partially assimilated migrants. Hypothesis 4. Assimilation exhibits ‘‘stability of extremes’’: Migrants who are unassimilated to the mainstream as well as fully assimilated migrants will show greater stability than partially assimilated migrants. Moreover, when near one state of homophily while far away from the other, we would expect migrants to move to the first. If assimilation follows the arrows in Fig. 1, then migrants move toward the nearest extreme. Hypothesis 5. Assimilation exhibits ‘‘movement toward extremes’’: When a migrant is near an extreme state, s/he will tend to move toward that state. The closer s/he is to his/her ethnic or cultural group (the farther away from the mainstream) the more likely s/he will assimilate in reverse toward this group (and the less likely forward toward the mainstream). At the population level this will then result in a polarized distribution of assimilation levels with many immigrants concentrating near the extremes. As individual migrants move toward the most proximate of the two maximally homophilous states of Fig. 1, for any cohort of migrants we should see a tendency for a majority to be assimilated along either few or many dimensions, while only a minority is partially assimilated. Hypothesis 6. Assimilation exhibits ‘‘polarity’’: Assimilation levels in any immigrant cohort are distributed in a polarized fashion, with unassimilated and assimilated migrants being overrepresented and partially assimilated migrants underrepresented. As it takes immigrants time to find their way in the host society, this polarized state may not be obtained immediately. Rather, over the course of years, a cohort of migrants will gravitate toward a more and more polarized population-wide assimilation pattern. Hypothesis 7. Assimilation exhibits ‘‘polarization’’: The tendency for polarity increases with time.

168

ARNOUT VAN DE RIJT

Data Traditional data on immigrant assimilation has been of a cross-sectional nature and scholars have long called for longitudinal data that permits analysis of lifetime changes: ‘‘The lack of y life-course information for immigrant cohorts hinders understanding of the integration of immigrants.’’ (Jasso, Massey, Rosenzweig, & Smith, 2000, p. 127); ‘‘Actual longitudinal data that track individual immigrants over time y have largely been lacking’’ (Bean, Brown, & Rumbaut, 2006, p. 309). Here we draw on data from the LSIC (Statistics Canada, 2005a). A key feature of immigration policy in Canada makes longitudinal data on this particular immigration country attractive for a test of the theory developed. Namely, Canada’s official multicultural policy is aimed at stimulating structural assimilation while discouraging most forms of cultural assimilation (cf. Bloemraad, 2006), thus counteracting the hypothesized positive interaction effects between structural and cultural assimilation. These data thus allow a conservative test of the theory. The target population of the LSIC is all immigrants aged 15+ who entered Canada between October 1, 2000 and September 30, 2001 from abroad with legal ‘‘landed immigrant’’ status and who were still in Canada after four years. The survey conducted face-to-face or telephone interviews in one of 15 languages with 14,356 immigrants six months after entry. From these interviewees 9,332 were reinterviewed after two years, and of those second-wave respondents 7,713 were interviewed a third time, four years after landing. If respondents were not reinterviewed that was either because they could not be located at the next wave of the survey or because they chose not to participate again. Attrition analysis shows that the probability of exit from the sample was unrelated to satisfaction with life in Canada and various sociodemographic variables, with the exception of old and familyclass immigrants for whom exit was somewhat more likely (Houle & Schellenberg, 2010). A comparison of the wave 1 frequency distributions of the assimilation variables of interest for respondents who only participated in wave 1 with those for wave 2 and wave 3 respondents yielded no significant differences. Weights constructed by Statistics Canada, who collected the data, were designed to make the sample of immigrants who completed all three interviews representative of the target population (Statistics Canada, 2005b, ch. 12) and these weights are used throughout the analysis. We consider six measures of assimilation. As a measure of strength of group identity, scholars often record perceived closeness to the in-group

Assimilation Process of Immigrants

169

(e.g., Bosson, Johnson, Niederhoffer, & Swann, 2006; Otten & Epstude, 2006; Staerkle´, Sidanius, Green, & Molina, 2005). We follow this convention, using the question ‘‘When you think of others in Canada from the same ethnic or cultural group as yourself, how close would you say you feel to that group as a whole?’’ as an indicator of ethnic identification. Answers ranged from ‘‘very close,’’ to ‘‘close,’’ to ‘‘not very close,’’ to ‘‘not close at all.’’ Attachment to cultural traditions and values is measured by the following question: ‘‘How important is it for you to carry on the values and traditions of your ethnic or cultural group or your homeland?’’ Answer categories are ‘‘very important,’’ ‘‘important,’’ ‘‘not very important,’’ and ‘‘not important at all.’’ The LSIC provides a self-assessment of language skills: ‘‘How well can you speak English/French?’’ Answer categories are ‘‘cannot speak this language,’’ ‘‘poorly,’’ ‘‘fairly well,’’ ‘‘well,’’ and ‘‘very well.’’ In the analysis reported here we used the measure of French language skills for residents of Quebec and English otherwise. Alternative treatment of the language variables such as models with French only, English only, or both languages does not lead to substantively different results. Brown (2006) has argued that structural assimilation variables that directly measure the ethnic composition of immigrants’ networks are preferable to measures of neighborhood composition, which ‘‘y rely on the assumption that relationships develop among racially and ethnically dissimilar people who live near one another’’ (p. 85). The LSIC provides such measures of network ethnic composition. For friendship relations, we use the question ‘‘How many of your new friends are of the same ethnic or cultural group as you?’’ Possible answers were ‘‘all of them,’’ ‘‘most of them,’’ ‘‘half of them,’’ ‘‘some of them,’’ and ‘‘none of them.’’ New friends were defined as friends made since landing (wave 1) or since the previous wave (waves 2 and 3). For work relations, we use the question ‘‘How many of your co-workers are of the same ethnic or cultural group as you?’’ Possible answers were ‘‘all of them,’’ ‘‘most of them,’’ ‘‘some of them,’’ and ‘‘none of them.’’ For household relations we employ a measure of the proportion of household members who speak English or French (Quebec) – excluding the respondent. Missing values reduce the sample from 7,713 to 7,031 cases. We report on analyses of this smaller sample of list-wise deleted cases. Analyses with imputed missing values are consistent. Table 1 displays means and standard deviations of all six assimilation variables. The fraction of household members speaking the official language

Descriptive Statistics (N=7,031).

Table 1. Time-Varying Variables

Wave 1 (6 months)

Wave 2 (2 years)

Wave 3 (4 years)

Mean

St. Dev.

Mean

St. Dev.

Mean

St. Dev.

English/French speaking skills Does not feel close to cultural group Finds unimportant to carry on values/traditions Proportion cross-cultural friends Proportion cross-cultural colleagues Proportion household speaking English/French

.676 .351

.323 .262

.743 .343

.299 .246

.752 .343

.293 .253

.277

.245

.261

.232

.265

.242

.314 .319

.314 .384

.395 .470

.295 .395

.422 .514

.314 .387

.675

.249

.849

.242

.864

.226

Log annual individual income Age Proportion neighbors not born in country of origin

5.36 35.0 .961

4.18 12.0 .054

7.08 36.7 .962

4.29 12.0 .052

7.70 38.6 .964

4.16 12.0 .052

Percentage

Percentage

Percentage

23.5% 14.9% 1.3% 48.2%

21.7% 16.2% 3.7% 26.0%

20.2% 15.4% 5.8% 20.1%

Marital status (1=single) Attends religious services Owns own business Unemployed Time-Constant Variables

Percentage

Gender (1=female) Holds college degree from foreign country Immigrant class ‘‘Family’’ Immigrant class ‘‘Independent’’ Immigrant class ‘‘Business’’ Immigrant class ‘‘Refugee/Other’’ Country of Origin China India Philippines Pakistan Afghanistan South Korea Iran Romania Russia Sri Lanka U.K.

Percentage 16.7% 14.5% 6.6% 4.5% 4.0% 3.7% 3.4% 3.1% 2.1% 1.9% 1.9%

50.3% 64.8% 26.3% 67.0% 6.1% 0.5% Country of Origin Bosnia Algeria Morocco Iraq France Ukraine Yugoslavia Bangladesh Columbia Sudan South Africa

Percentage 1.8% 1.7% 1.7% 1.7% 1.5% 1.5% 1.3% 1.3% 1.2% 1.1% 1.1%

n.a.=not applicable. Note: Statistics Canada does not permit reporting of minimum and maximum values or crosstabulations with cell counts below 10 (Statistics Canada, 2005b: ch. 12).

Assimilation Process of Immigrants

171

is of cardinal measurement level. For the remaining five assimilation variables, the data in Table 1 are based on a cardinalization where the unit interval is divided up into equidistant fractions. For example, for language skills we have 0=cannot speak this language, 14=poorly, 12=fairly well, 3 4=well, and 1=very well/fluently. Such cardinal level variables are used for estimation of the dynamic panel models, as elaborated in the following section. Alternative coding decisions yield no qualitative differences in results. Table 1 additionally shows various time-varying control variables. These are income (logged with base 10 so that a unit increase corresponds to an extra digit), age (in years), marital status (dummy; 1=single), church attendance (dummy; 1=attends), neighborhood ethnic composition (proportion migrants from same country of origin in same three-digit postal code area), business ownership (dummy; 1=owner), and unemployment (dummy; 1=unemployed). Table 1 also shows some time-constant variables that give the reader a sense of the demographic composition of the sample. For example, the most strongly represented nationalities in the 2000/2001 cohort were Chinese and Indians, together accounting for 31% of all immigrants.

Analysis Strategy Hypothesis 1 (‘‘deceleration’’) can be tested by comparing migrant-level changes in the six assimilation variables between waves 1 and 2 with changes between waves 2 and 3. The hypothesis would be confirmed if fewer migrants exhibit change in the latter period than in the former period. To test hypothesis 2 (‘‘reinforcement’’), we must verify if changes in the six assimilation variables are positively affected by other assimilation variables. We do so by estimating six regression equations, one for each dimension d=1, 2, y , 6, of the following form: xdit ¼ ad þ

X

bed xeit þ cd xdit1 þ ydi þ dit

(1)

ead

Equation (1) is the standard dynamic panel model (Finkel, 1995; Halaby, 2004, p. 540), where i indexes the N migrants, t indexes time, xdit is the level of assimilation along dimension d, e=1, 2, y , 6 indexes dimensions other than d, a is a vector of constants for each assimilation dimension, b the sixby-six matrix of cross-dimension coefficients of interest, yi a migrant-specific vector of constants and e error. We employ the Arellano–Bond GMM1 estimator (Arellano & Bond, 1991), which provides fixed-effect estimates and uses lags of the dependent variables as instruments for the cd

172

ARNOUT VAN DE RIJT

coefficients. As controls we include the time-varying control variables shown in Table 1 as well as lags of all predictors. Positive entries in b, this would support the hypothesis. A test of hypothesis 3 (‘‘parallel change’’) must involve comparisons within migrants across time, which the panel data allow us to make. We use fixed effects linear regression (Allison, 2009) and estimate the multivariate effects of the level of assimilation in each source dimension on the level of assimilation in each target dimension. Positive effects would support the hypothesis. Note that straight-line theory would also predict these correlations, namely as a byproduct of contemporaneous change in forward direction. To capture such cross-dimensional correlations that are merely due to simultaneous unidirectional change, we add dummy variables for wave 2 and wave 3. We also add the aforementioned control variables to the regression model. To test hypothesis 4 (‘‘stability of extremes’’) we predict the probability of extreme and non-extreme assimilation levels remaining unchanged from wave to wave. We consider a space capturing all combinations of the six assimilation variables at each wave. As to keep the dimensionality of this space manageable, we first dichotomize the six assimilation variables, considering migrants either assimilated or not assimilated. We split variables at the wave 1 median, consistently used these same median thresholds across waves, and verified that the particular choice of threshold is inconsequential for the results. On each assimilation dimension, a migrant can at any wave thus either be assimilated or not, giving rise to 26=64 states. Two of these 64 states, namely nonassimilation in all dimensions and assimilation in all dimensions, are homophilous and predicted to be more stable than the others. We estimate six logistic regression models, one for each of the six assimilation variables, with prior levels of assimilation in all six variables as predictors. These same logistic regression models can be used to test hypothesis 5 (‘‘movement toward extremes’’). We assess the differences in probabilities of forward and reverse assimilation between those who are nearly assimilated to the mainstream with those who are hardly assimilated. We estimate the probability that a migrant who is assimilated along one dimension but unassimilated along all other dimensions will have reverse-assimilated in that dimension by the next wave. We compare this with the probability that a migrant who is unassimilated along that dimension as well as all other dimensions assimilates along that dimension. We similarly compute probabilities of forward and reverse assimilation along each dimension conditional upon assimilation along all other dimensions.

173

Assimilation Process of Immigrants

For hypothesis 6 (‘‘polarity’’) we consider the same state space of dichotomous assimilation variables, but now at all waves. This yields 3 waves  26 states=192 possible assimilation histories that any one migrant can have experienced. We estimate a Negative Binomial regression model for cell frequencies in the corresponding seven-dimensional matrix with 192 cells: X X lnðf c Þ ¼ j0 þ jd xcd þ gg zcg þ dc (2) .

d

g

Here, fc is the frequency of cell c, f0 a constant, fd a coefficient for the effect of a cell involving assimilation along dimension d, xcd taking on value 1 if cell c involves assimilation along dimension d and 0 otherwise, gg with g=0, 1, 2, 3 a coefficient for the effect of homophily zcg, where higher values represent P positions closer to the two homophilous states. Variable zcg equals 1 if j d xcd 3j ¼ g and 0 otherwise. dc is an error term.6 Note that g equals 3 both at maximal assimilation (g=j63j=3) and at minimal assimilation (g=j03j=3), while intermediate levels of assimilation yield smaller values (e.g., g=j33j=0). The pattern of polarity predicted by hypothesis 6 would be visible if gg were increasing in g. Finally, hypothesis 7 (‘‘polarization’’) can be tested by evaluating if the degree to which gg increases with g (i.e., polarity) becomes stronger with each subsequent wave. Hypothesis 7 would also be supported if we find a pattern of increasing correlation between dimensions of assimilation. To this end we calculate pairwise correlations for all pairs of assimilation variables at each wave.

RESULTS Table 2 shows lifetime changes in the six measures of assimilation for both the 6 months–2 years and the 2 years–4 years period. The table shows that for all measures and all periods the direction of adaptive change strongly varies across immigrants, unlike what classic assimilation theory predicts. While many immigrants after longer residence are indeed less connected to their ethnic group and are less attached to its values and traditions, the table also shows that many other immigrants move in opposite direction, which does not support a unidirectional perspective. For two measures, namely, values & traditions and group identity, such reverse assimilation is even more common than forward assimilation in both periods (po.05 using a w2 test). These longitudinal data on Canada thus exhibit heterogeneity in

174

Table 2.

ARNOUT VAN DE RIJT

Lifetime Changes in Six Measures of Immigrant Assimilation between 6 Months and 4 Years after Entry. Period

Group identity

1 2–2

years of residence

2–4 years of residence Values & traditions

1 2–2

years of residence

2–4 years of residence Language proficiency

1 2–2

years of residence

2–4 years of residence Friendship integration

1 2–2

years of residence

2–4 years of residence Workplace integration

1 2–2

years of residence

2–4 years of residence Household integration

1 2–2

years of residence

2–4 years of residence

Reverse

No change

Forward

Total

2,061 (29%) 1,939 (28%) 1,938 (28%) 1,737 (25%) 979 (14%) 1,371 (19%) 1,300 (18%) 1,531 (22%) 790 (11%) 1,178 (17%) 844 (12%) 1,192 (17%)

3,035 (43%) 3,251 (46%) 3,385 (48%) 3,577 (51%) 3,834 (55%) 4,113 (58%) 3,098 (44%) 3,530 (50%) 4,012 (57%) 4,228 (60%) 1,666 (24%) 4,370 (62%)

1,935 (28%) 1,841 (26%) 1,708 (24%) 1,717 (24%) 2,218 (32%) 1,547 (22%) 2,633 (37%) 1,970 (28%) 2,229 (32%) 1,625 (23%) 4,521 (64%) 1,469 (21%)

7,031 (100%) 7,031 (100%) 7,031 (100%) 7,031 (100%) 7,031 (100%) 7,031 (100%) 7,031 (100%) 7,031 (100%) 7,031 (100%) 7,031 (100%) 7,031 (100%) 7,031 (100%)

Note: For all measures, more change occurs during the first period of one-and-a-half years than during the second period of two years, confirming Hypothesis 1 (‘‘deceleration’’).

the rate and direction of assimilation, as has been noted before in the literature on immigration to Western countries. Table 2 further shows that for all measures there is less change in the second period than in the first, suggesting that the biggest adjustments in an immigrant’s life are made at the very beginning, after which a process of equilibration is set in motion. This finding confirms hypothesis 1 (‘‘deceleration’’). Table 3 shows results for the six dynamic panel models regressing temporal changes in each of the six assimilation variables on levels of assimilation in each of the others. Hypothesis 2 (‘‘reinforcement’’) predicts positive effects of assimilation along one dimension on assimilation along other dimensions. With six assimilation dimensions, there are 30 such

175

Assimilation Process of Immigrants

Table 3. Estimates of Six Dynamic Panel Models Regressing Dimensions of Assimilation on One Another.a Identity coef (z)

Values coef (z)

Language coef (z)

Friendship coef (z)

Workplace coef (z)

Household coef (z)

Identity

–

Values

.287 (20.35) .007 (.38) .078 (5.91) .027 (1.81) .014 (.79)

.234 (19.79) –

.010 (.96) .006 (.52) –

.072 (5.46) .025 (1.73) .076 (4.11) –

.028 (2.07) .003 (.21) .035 (1.84) .077 (5.78) –

.018 (1.69) .012 (1.06) .269 (18.77) .017 (1.60) .000 (.04) –

Language Friendship Workplace Household w2 (df) # migrants

556.75 (25) 7,031

.003 (.17) .016 (1.33) .004 (.30) .011 (.68)

.031 (3.18) .027 (2.42) .254 (20.04)

552.20 (25) 7,031

515.12 (25) 7,031

.100 (6.78) .020 (1.17) 157.99 (25) 7,031

.018 (1.01) 5045.62 (25) 7,031

549.64 (25) 7,031

Note: With many cross-dimension effects being insignificant and all significant effects positive, there is weak overall support for Hypothesis 2 (‘‘reinforcement’’). po.05; po.01; po.001 a Effects of the following control variables are included but not shown: Income (log), age (in years), marital status (dummy; 1=single), church attendance (dummy; 1=attends), proportion migrants from same country of origin in neighborhood, business ownership (dummy; 1=owner), and unemployment (dummy; 1=unemployed). Additionally included but not shown are time lags of the dependent, independent and control variables.

cross-dimension effects. Of these 30 effects none have z values below –1.96, 18 have z values between –1.96 and +1.96, while 12 have z values above +1.96. As all statistically significant effects are in the predicted direction, these results are consistent with the hypothesis. However, with only a minority of the predicted positive cross-dimension effects being statistically significant, empirical support is weak overall. Looking more closely at the various effects, we can see that the directionality of cultural assimilation – forward or reverse – indeed depends on how structurally assimilated a migrant is. Cross-ethnic friendships and workplace relations speed up acquisition of the official language, and so does contact with household members who speak the official language. The findings support our theoretical propositions and are also consistent with the intuition that exposure to a language eases an immigrant’s acquisition

176

ARNOUT VAN DE RIJT

of that language (e.g., Van Tubergen & Kalmijn, 2005). Friendship effects on identification are in the expected direction as well, which is also consistent with the notion of ‘‘influence.’’ There is no evidence that workplace segregation or interaction with household members who do not speak the official language reinforces identification with one’s original ethnic group, nor is there any evidence for structural effects on cultural values. The directionality of structural assimilation is also associated in the expected direction with how culturally assimilated a migrant is. A better command of the official language and less identification with one’s own ethnic or cultural group are found to promote cross-ethnic friendship contact, consistent with the notion of ‘‘selection.’’ The effect of language skills is in line with the present theory and also with the intuitive understanding of language skills as allowing conversations one could otherwise not have (Chiswick & Miller, 1995; Espinoza & Massey, 1997). In a similar vein, language skills are expected to increase the odds of transitioning from an ethnic to a mainstream work environment (e.g., Nee et al., 1994), but this effect is not significant in the dynamic panel models. Value orientation does not seem to affect the directionality of change in the ethnic composition of migrants’ friendship or work networks. As for friendship relations, for work relations an identity effect is also observed. Change in the proportion of household members speaking the official language is strongly affected by one’s own command of that language, but not by identification or attachment with the own ethnic/cultural group. Degrees of assimilation in cultural dimensions do not significantly affect the directionality of change in other cultural dimensions, with one exception; attachment to cultural values of the own group reinforces identification with that group and vice versa. This mutual relationship can be explained as a combination of selection and influence whereby the effect of identity on values and the reverse effect are both mediated by some form of structural assimilation; stronger ethnic identification leads to increased co-ethnic contact which in turn reinforces ethnic values and vice versa. Analogously, influence and selection can in combination produce effects between dimensions of structural assimilation. Indeed, Table 3 shows that friendship relations with the own group lead to more work relations of that same type and this effect also runs in the opposite direction. These effects are thus consistent with the theory of selection and influence and also with the argument that friends from one cultural group refer one to jobs with employers and co-workers of that same cultural group (Bashi, 2007; Hagan, 1998; Menjı´ var, 1999; Portes & Rumbaut, 1996; Portes & Sensenbrenner,

177

Assimilation Process of Immigrants

1993; Sanders et al., 2002). No evidence was found for any of the remaining effects between dimensions of structural assimilation. Table 4 displays the results of the six static fixed-effects panel models, including the estimated cross-dimension coefficients with associated t values, the F statistics with the associated numbers of degrees of freedom (all are significant), and the number of migrants. Note that the matrix of t values is symmetric because each model includes the same variables; for example, the t value for the effect of friendship integration on language skills is the same as the t value for the effect of language skills on friendship integration. Of the 15 cross-dimension correlations, 7 are significantly positive, 7 are not statistically significant, while 1 is significantly negative. The exception involves the apparent negative relationship between household relations and values, which the present theory cannot account for. All in all, these results provide modest support for hypothesis 3 (‘‘parallel change’’): degrees of assimilation are mostly positively correlated with one another over the life course of a migrant. Note that this tendency is not a mere consequence of

Table 4. OLS Estimates for Six Models with Migrant-Level Fixed Effects, Regressing Levels of Assimilations on Any One Dimension on Levels of Assimilation Along the Five Other Dimensions.a Identity coef (t)

Values coef (t)

Language coef (t)

Friendship coef (t)

Workplace coef (t)

Household coef (t)

Identity

–

Values

.259 (28.75) .007 (.61) .066 (8.12) .016 (4.56) .019 (1.85)

.215 (28.75) –

.004 (.61) .004 (.67) –

.071 (8.12) .035 (3.67) .062 (5.06) –

.028 (2.55) .003 (.34) .056 (3.65) .103 (9.75) –

.013 (1.85) .018 (2.39) .366 (40.30) .006 (.89) .023 (4.48) –

Language Friendship Workplace Household F (df) # migrants

438 (7) 7,031

.007 (.67) .027 (3.67) .002 (.34) .023 (2.39)

.029 (5.06) .029 (3.65) .283 (40.30)

137 (7) 7,031

130 (7) 7,031

.065 (9.75) .010 (.89)

.061 (4.48)

155 (7) 7,031

333 (7) 7,031

1073 (7) 7,031

Note: All but one of the significant cross-dimension effects are positive, providing weak overall support for Hypothesis 3 (‘‘parallel change’’). po.05; po.01; po.001 a Effects of constants and wave 2 and wave 3 dummy variables are not shown.

178

ARNOUT VAN DE RIJT

contemporaneous increases across waves, as a unidirectional perspective would predict. The model contains dummy variables for individual waves, thus controlling confounding effects of time. Fig. 2 graphs the replication probabilities of assimilation levels remaining unchanged from one wave to another as a function of the number of dimensions along which a migrant is assimilated. The probabilities were computed from logistic regression models, as detailed in the previous section. The light grey bars in Fig. 2 closely follow the theoretical pattern depicted in Fig. 1. They show that migrants who are completely unassimilated are most likely to remain where they are from one wave to the next. They also demonstrate that in accordance with our theoretical propositions, the two most stable states in the empirical analysis are the ones that maintain homophily. All forms of partial assimilation are less stable, with the most mixed forms being the least stable, again in accordance with the theory. There is strong evidence that the two homophilous states are

Replication probability

0.3

0.2

0.1

0 0 homophily

1

2 3 4 # assimilated dimensions

5

6 homophily

Fig. 2. Stability by Degree of Assimilation. The ‘‘replication probability’’ shown is the conditional probability that a migrant is again found in a state with the given number of assimilated dimensions in the current wave provided that s/he exhibited no change along any of the six dimensions since the previous wave. Note: For values 1 through 5, the value shown is averaged across all states with the corresponding number of assimilated dimensions. The replication probabilities were estimated from six logistic regression models predicting assimilation along each of the six dimensions from levels of assimilation along each of the dimensions at the previous wave and from control variables. For each of the control variables a population average score was used to generate the replication probabilities shown. Consistent with Hypothesis 4 (‘‘stability of extremes’’), replication probabilities are highest at the extremes.

179

Assimilation Process of Immigrants

more stable than other states, as hypothesis 4 (‘‘stability of extremes’’) predicts. Table 5 shows the probabilities of forward assimilation and reverse assimilation along each dimension conditional upon either nonassimilation or assimilation in all other dimensions, estimated from logistic regression models. Consistent with hypothesis 5 (‘‘Movement toward extremes’’), for every dimension it is found that reverse assimilation is more likely and forward assimilation less likely when a migrant is unassimilated in all other dimensions than when s/he is assimilated in all other dimensions. For some dimensions these differences in likelihoods of assimilation seem much larger than for others.7 Temporal change in the migrant’s reported importance to carry on values and traditions of the homeland appears least affected by other dimensions while the directionality of linguistic assimilation of the migrant and her household members appear most strongly affected. Table 6 shows results of three Negative Binomial regression models,8 one for each wave, providing a test of the prediction that for the entire Table 5. Estimated Probabilities of Assimilation and Reverse Assimilation for Each Dimension, Conditional upon Either Nonassimilation or Assimilation in All Other Dimensions. Unassimilated Along All Other Dimensions

Dimension Language Identity Values Friends Work Household

Assimilated Along All Other Dimensions

Probability of forward assimilation

Probability of reverse assimilation

Probability of forward assimilation

Probability of reverse assimilation

.33 .14 .08 .15 .29 .31

.14 .73 .70 .54 .28 .41

.80 .42 .14 .46 .60 .79

.02 .37 .55 .19 .10 .08

Note: For each dimension, the probability of forward assimilation increases and the probability of reverse assimilation decreases with the number of dimensions along which the immigrant is currently assimilated. This provides support for Hypothesis 5 (‘‘movement toward extremes’’). The probabilities were estimated from six logistic regression models predicting probabilities of assimilation along each of the six dimensions from levels of assimilation along each of the other dimensions at the previous wave and from control variables. For each of the control variables a population average score was used to generate the probabilities of forward and reverse assimilation shown above.

180

ARNOUT VAN DE RIJT

Table 6. Results from Negative Binomial Models of Assimilation State Frequencies. Frequency coef (z) Wave 1 Two steps away One step away Maximal homophily Wave 2 Two steps away One step away Maximal homophily Wave 3 Two steps away One step away Maximal homophily df # migrants

.194 (1.32) .839 (5.24) 2.042 (14.80) .318 (1.65) .970 (4.72) 2.259 (14.29) .162 (.84) .931 (4.68) 2.265 (11.89) 10 7,031

po.05. Note: Effects of control variables and constants are omitted. Immigrants are disproportionately located in states of maximal homophily, consistent with Hypothesis 6 (‘polarity’). However, there is no significant increase in homophily with time, rejecting Hypothesis 7 (‘polarization’). Control variables: For each of the six assimilation dimensions a dummy variable is included indicating whether the focal cell involves above-threshold assimilation along the respective dimension. immigrant cohort, assimilation levels are distributed in a polarized fashion. The results provide strong support for hypothesis 6 (‘‘Polarity’’). At all waves, migrants tend to be found in or near a homophilous state. Least likely is a mixed state where migrants are assimilated along three dimensions while unassimilated along the other three. This mixed state serves as the baseline for the effects shown in Table 6. More likely is that they are assimilated in four or in two ways, thus two steps away from maximal homophily. Even more likely they are only one step away from one of the two homophilous states. And most likely they are assimilated in none

181

Assimilation Process of Immigrants

Table 7. Changes in Pairwise Correlations between Measures of Immigrant Assimilation from 12 year to 2 years to 4 years after entry (N=7,031). Change in Pearson Correlation Coefficient Identity Values Language Friendships Workplace

1 2

Values

Language

Friendship

Workplace

Household

# years

# years

# years

# years

# years

2

4

+.05 +.09

1 2

2

4

.04 +.06 +.02 +.05

1 2

2

4

+.02 +.09 +.03 +.04 +.06 +.08

1 2

2

4

+.00 +.00 +.00 +.03

+.03 +.02 +.04 +.08

1 2

2

4

.02 +.02 .02 +.05 .01

+.04 +.04 .02 +.05 .02

Note: Exceptions notwithstanding, change tends to occur in positive direction, providing weak overall support for Hypothesis 7 (‘‘polarization’’).

or all ways. For example, complete or nonassimilation at wave 3 is e2.2651=863% more likely than partial assimilation and this difference is strongly significant at all three waves (po.001). Comparing estimates across waves, the maximal homophily effects in Table 6 appear somewhat stronger for later waves, but w2 tests do not find this tendency to be significant. Thus, these results do not provide support for the predicted increase in polarity (hypothesis 7). As a second test of hypothesis 7 we examine changes in the pairwise correlations across measures of structural and cultural assimilation from wave 1 to wave 2 and from wave 2 to wave 3. Table 7 shows these correlations. For 24 out of 30 pairwise comparisons, correlation coefficients increase from wave to wave and for 13 out of 15 sequences the correlation is strongest at the last wave, 4 years after entry. The trends are thus in general agreement with hypothesis 7, some exceptions notwithstanding. The results as a whole mostly support the theory of homophilous assimilation. The data appear to exhibit the predicted bidirectional dynamic whereby with time immigrants come to occupy one of two homophilous positions.

DISCUSSION A theoretical account of the well-documented pluriformity in assimilation trajectories and outcomes need not draw on corresponding heterogeneity in

182

ARNOUT VAN DE RIJT

the resources and intentions of immigrants or any differential treatment by the host society. Rather, this variability can be explained as spontaneously emerging from adaptation in a culturally differentiated host society, where both the mainstream as well as the own ethnic group provide competing sociocultural alternatives. The immigrant enters the confines of the ethnic or cultural group that is structurally and culturally most proximate and whose boundaries are most permeable to her. Often this is not the mainstream but the immigrant’s own ethnic or cultural group. Our approach agrees with contemporary perspectives and critiques of classic ‘‘straight-line’’ assimilation theory that reject the notion of a universal transformation from newcomer to all-round member of a dominant mainstream, instead emphasizing the great diversity of assimilation trajectories and outcomes (Alba & Nee, 2003; Brown & Bean, 2006; Portes & Zhou, 1993; Rumbaut & Portes, 2001; Waldinger, Lim, & Cort, 2007). We differ, however, in where we look for the origins of this diversity. What we have argued and purported to show here is that the sources of heterogeneity in assimilation outcomes need not be found in exogenous factors differentiating immigrant groups but may for a significant degree lie in the internal mechanics of immigrant assimilation. Diversity may endogenously arise as an emergent property of a bidirectional dynamic. These mechanisms drive how newcomers naturally adjust to a changed environment. Moderately assimilated immigrants tend to further assimilate while unassimilated migrants strengthen their embeddedness in their own ethnic group. Responsible for this dual movement in opposing directions is a tendency for immigrants’ cultural behavior, beliefs, and values to come to match those of the people present in their immediate social environment. Such congruence is attainable not only by moving ‘‘forward’’ through immersion in the mainstream but also ‘‘in reverse,’’ through participation in the ethnic neighborhood, culture, and economy. This endogenous account does not by any means deny the existence of large a priori differences between migrants in social, financial, and human capital, settlement intentions, or in the differential opportunities that host societies bestow upon them. These factors can be seen as initial conditions that interact with the bidirectional forces we have identified, pushing most immigrants with greater resources, stronger intentions to adapt, and preferential treatment by the host population on a self-reinforcing integration trajectory while placing most others on the opposing path. Our point is rather that even if these important exogenous factors were not present, the adaptation process would still bifurcate and produce divergent outcomes.

Assimilation Process of Immigrants

183

While our investigation concerned the adaptation patterns of individual migrants, the results have implications for the assimilation of migrant groups. Namely, individual migrants do not assimilate independently. For example, if one migrant masters the official language and is the cohabitant of another migrant, then by implication the number of household members of that other migrant who speak the official language has gone up too. If the migrant subsequently develops friendships with those who speak the official language, then she is likely to introduce those friends to her partner. And so on. The self-reinforcing effects we identified within the lives of migrants might then also exist between migrants. Perhaps this can cause tipping effects at the population level, where the assimilation of one migrant triggers the assimilation of other migrants to that group (Bruch & Mare, 2006; Schelling, 1971; van de Rijt, Siegel, & Macy, 2009). It also suggests that the snowball effects that Massey et al. (1987) and Palloni, Massey, Ceballos, Espinosa, and Spittel (2001) find in the entry phase of migration, where one migrant increases the chance of a second migrant, may generalize to the adjustment phase: The more migrants approach a certain cultural group, the more likely subsequent migrants follow. As a result, we may see group-size effects at the population level. Large groups will naturally lead their members to develop more friendship, household, and work relations with others of the same group (Blau & Schwartz, 1984). The LSIC data show that the Chinese, forming the largest immigrant group in Canada, start off most behaviorally assimilated to their own group on all traits and in all relations and are more likely than others to remain so. Mexicans, a much smaller group in Canada than in the United States, acquire English speaking skills at a faster pace and more readily enter relations with members of other groups. The present study has an important implication for the social networks literature focusing on selection and influence processes. What it shows is that the scale of these processes is not limited to the empirical settings that are typically considered in this literature: teenagers influencing one another’s smoking habits (Mercken, Snijders, Steglich, Vartiainen, & De Vries, 2010) or socializing around common tastes and interests in online social networks (Lewis, Gonzales, & Kaufman, 2011). Immigrant assimilation concerns a metamorphosis that is initiated by much greater incongruity between self and other, that unfolds over a much longer stretch of time, and that concerns a much broader swath of life domains. Nonetheless, it is well captured by the same basic principles of network dynamics. We draw our conclusions in recognition of a number of important study limitations. First, the empirical analysis is restricted to three-and-a-half

184

ARNOUT VAN DE RIJT

years of residence. Extrapolation from the trend data suggests an equilibrating process with less and less change, but later-life data are needed to confirm this. Second, the data at hand could only measure assimilation toward or away from the mainstream. Cases of assimilation toward a third group have been observed for first and second generations of several nationalities and ethnicities such as the Sikh in California (Dadabhay, 1954; Gonzales, 1986; LaBrack, 1981; Leonard, 1982) and West Indians in the United States (Deaux, 2006; Portes & Stepick, 1993; Portes & Zhou, 1993; Waters, 2001). The theory can straightforwardly be expanded to accommodate such cases. Namely, migrants may assimilate to any of a multitude of cultural groups, most likely the one that is least socially distant. Third, past immigration studies have almost exclusively focused on change among immigrants, while ignoring changes among native populations. While this paper is limited in the same way, the model proposed is not tailored to migrants and the concepts of selection and influence may be analogously applied to autochthon adaptation to the foreign-born. Fourth, the measures used may suffer from measurement problems that cannot be assessed. We assumed but could not verify that reported cross-ethnic contact involved members of the mainstream and co-ethnic contact was with a minority culture. Last but not least, our results are limited to the early residence years of recent immigrants to Canada. A number of longitudinal datasets will allow future researchers to study individual assimilation in greater depth. Analysis of the New Immigrant Survey, the Longitudinal Survey of Immigrants to Australia, and the Longitudinal Survey of Immigrants to New Zealand permits an assessment of the generalizability of the bidirectional assimilation dynamics that we have here identified.

NOTES 1. Social capital also affects the probability of migration, prior assimilation (cf., Garip, 2008). 2. A third possibility for homophily is assimilation to another minority culture, a phenomenon that has in fact been observed (Dadabhay, 1954; Deaux, 2006; Gonzales, 1986; LaBrack, 1981; Leonard, 1982; Portes & Stepick, 1993; Portes & Zhou, 1993; Waters, 2001). The present two-group model can straightforwardly be generalized to scenarios with more than two groups; see also the discussion section. 3. This assumption of a preexisting state of cultural segmentation is consistent with theoretical models that show that selection and influence produce locally homogeneous communities within a cultural diverse population (Axelrod, 1997; Centola, Gonza´les-Avella, Eguı´ luz, & San Miguel, 2007).

Assimilation Process of Immigrants

185

4. Immigrants may simultaneously place value on contact with other ethnic groups and on maintenance of cultural identity and characteristics (e.g., Berry, 1997, 2006; Berry, Kim, Minde, & Mok, 1987; Sayegh & Lasry, 1993). The theory developed here does not exclude this possibility but posits a tension between crossethnic contact and ethnic identification based on the principle of homophily. 5. For simplicity we assume that relations are either cross-ethnic or with individuals who adhere to majority culture norms. In reality, there could be overlap and relations could also be with members of another minority group. The present theory is easily extended to more than two groups; see the discussion section. 6. This absolute value formulation constrains proximity to the one extreme to be equally likely to proximity to the other extreme. In the unconstrained alternative model with separate effects for all seven assimilation levels (0–6), the set of predictor variables would not be linearly independent. 7. An interpretation of the absolute probabilities would not be meaningful as they depend on the cutoff points chosen in the dichotomization of the variables. 8. We estimated three corresponding Poisson regression models, instead assuming Poisson distributed counts. These yielded lower standard errors, but on the grounds of overdispersion and goodness-of-fit, the negative binomial models are preferred.

ACKNOWLEDGMENTS I thank the editors of this volume, Shane R. Thye and Edward J. Lawler, for their care and commitment to putting together this excellent collection. I thank Damon Centola and Frank van Tubergen for useful discussion, and Jeong-han Kang, Gueorgi Kossinets, Yujun Wang, and participants of the Princeton Center for Migration and Development workshop and the Utrecht University Sociology departmental colloquium for helpful comments. I thank the Center for the Study of Economy and Society for financial support. The research and analysis are based on data from Statistics Canada and the opinions expressed do not represent the views of Statistics Canada.

REFERENCES Ajrouch, K. J., & Jamal, A. (2007). Assimilation to a white identity: The case of Arab Americans. International Migration Review, 41, 860–879. Akers, R. L. (1985). Deviant behavior: A social learning approach. Belmont, CA: Wadsworth. Akresh, I. R. (2007). Contexts of English language use among immigrants to the United States. International Migration Review, 41, 930–955. Alba, R. (1990). Ethnic identity: The transformation of white America. New Haven, CT: Yale University Press.

186

ARNOUT VAN DE RIJT

Alba, R. D., Logan, J. R., Stults, B., Marzan, G., & Zhang, W. (1999). Immigrant groups in the suburbs: A reexamination of suburbanization and spatial assimilation. American Sociological Review, 64, 446–460. Alba, R. D., & Nee, V. G. (2003). Remaking the American mainstream: Assimilation and contemporary immigration. Cambridge, MA: Harvard University Press. Allison, P. (2009). Fixed effects regression models. Thousand Oaks, CA: Sage. Andersen, R., & Milligan, S. (2010). Immigration, ethnicity and voluntary association membership in Canada: Individual and contextual effects. Research in Social Stratification and Mobility, 29, 139–154. Aral, S., Muchnik, L., & Sundararajan, A. (2009). Distinguishing influence-based contagion from homophily-driven diffusion in dynamic networks. Proceedings of the National Academy of Sciences, 106, 21544–21549. Arellano, M., & Bond, S. (1991). Some tests of specification for panel data: Monte carlo evidence and an application to employment equations. Review of Economic Studies, 58, 277–297. Axelrod, R. (1997). The dissemination of culture: A model of local convergence and global polarization. Journal of Conflict Resolution, 41, 203–226. Bandura, A. (1977). Social learning theory. New York, NY: Prentice Hall. Bashi, V. (2007). Survival of the knitted. Immigrant social networks in a stratified world. Stanford, CA: Stanford University Press. Bean, F. D., Brown, S. K., & Rumbaut, R. G. (2006). Mexican immigrant political and economic incorporation. Perspectives on Politics, 4, 309–313. Bean, F. D., & Stevens, G. (2003). America’s newcomers and the dynamics of diversity. New York, NY: Russell Sage. Berry, J. W. (1997). Immigration, acculturation and adaptation. Applied Psychology: An International Review, 46(1), 5–68. Berry, J. W. (2006). Mutual attitudes among immigrants and ethnocultural groups in Canada. International Journal of Intercultural Relations, 30, 719–734. Berry, J. W., Kim, U., Minde, T., & Mok, D. (1987). Comparative studies of acculturative stress. International Migration Review, 21, 491–511. Blau, P. M., & Schwartz, J. E. (1984). Crosscutting social circles: Testing a macrostructural theory of intergroup relations. Orlando, FL: Academic Press. Bloemraad, I. (2006). Becoming a citizen in the United States and Canada: Structured mobilization and immigrant political incorporation. Social Forces, 85, 667–695. Bosson, J. K., Johnson, A. B., Niederhoffer, K., & Swann, W. B. (2006). Interpersonal chemistry through nativity: Bonding by sharing negative attitudes about others. Personal Relationships, 13, 135–150. Brown, S. K. (2006). Structural assimilation revisited: Mexican-origin nativity and cross-ethnic primary ties. Social Forces, 85, 75–92. Brown, S. K., & Bean, F. D. (2006). Assimilation models, old and new: Explaining a longterm process. Migration Information Source. Washington, DC: Migration Policy Institute. Bruch, E., & Mare, R. D. (2006). Neighborhood choice and neighborhood change. American Journal of Sociology, 112, 667–709. Carliner, G. (2000). The language ability of U.S. immigrants: Assimilation and cohort effects. International Migration Review, 34, 158–182. Centola, D. (2011). An experimental study of homophily in the adoption of health behavior. Science, 334, 1269–1272.

Assimilation Process of Immigrants

187

Centola, D., Gonza´les-Avella, J. C., Eguı´ luz, V. M., & San Miguel, M. (2007). Homophily, cultural drift, and the co-evolution of cultural groups. Journal of Conflict Resolution, 51, 905–929. Chiswick, B. R., & Miller, P. W. (1995). The endogeneity between language and earnings: International analysis. Journal of Labor Economics, 13, 246–288. Christakis, N. A., & Fowler, J. H. (2007). The spread of obesity in a large social network over 32 years. New England Journal of Medicine, 357, 370–379. Crandall, D. J., Cosley, D., Huttenlocher, D., Kleinberg, J., & Suri, S. (2008). Feedback effects between similarity and social influence in online communities. Proceedings of the 14th ACM SIGKDD international conference on knowledge discovery and data mining (pp. 163–167). Dadabhay, Y. (1954). Circuitous assimilation among rural Hindustanis in California. Social Forces, 33, 138–141. Da´vila, A., & Mora, M. T. (2000). The English-skill acquisition of Hispanic Americans during the 1980s. Social Science Quarterly, 81, 261–275. Deaux, K. (2006). To be an immigrant. New York, NY: Russell Sage Foundation. De Vroome, T., Coenders, M., Van Tubergen, F., & Verkuyten, M. (2011). Economic participation and national self-identification of refugees in the Netherlands. International Migration Review, 45, 615–638. Diehl, C., & Schnell, R. (2006). Reactive ethnicity or assimilation? Statements, arguments, and first empirical evidence for labor migrants in Germany. International Migration Review, 40, 786–816. Ersanilli, E., & Saharso, S. (2011). The settlement country and ethnic identification of children of Turkish immigrants in Germany, France, and the Netherlands: What role do national integration policies play? International Migration Review, 45, 907–937. Espenshade, T. J., & Fu, H. (1997). An analysis of English-language proficiency among U.S. immigrants. American Sociological Review, 62, 288–305. Espinoza, K. E., & Massey, D. S. (1997). Determinants of English proficiency among Mexican immigrants to the United States. International Migration Review, 31, 28–50. Esser, H. (1987). The substitution of complete networks. An ‘‘Economic’’ explanation of development and change of friendship-networks. Working Paper. Mannheim, Germany. Festinger, L. (1957). A theory of cognitive dissonance. Evanston, IL: Row, Peterson. Finkel, S. E. (1995). Causal analysis with panel data. Thousand Oaks, CA: Sage. Fong, E., & Ooka, E. (2006). Assessing patterns of social integration among immigrants. International Migration Review, 40, 348–374. Fuller, S., & Martin, T. F. (2012). Predicting immigrant employment sequences in the first years of settlement. International Migration Review, 46, 138–190. Gans, H. (1973). Introduction. In N. Sandberg (Ed.), Ethnic identity and assimilation: The Polish-American community. New York, NY: Praeger. Gans, H. (1997). Toward a reconciliation of assimilation and pluralism: The interplay of acculturation and Ethnic retention. International Migration Review, 31, 875–892. Garip, F. (2008). Social capital and migration: How do similar resources lead to divergent outcomes? Demography, 45, 591–617. Glazer, N., & Moynihan, D. P. (1970). Beyond the melting pot: The Negroes, Puerto Ricans, Jews, Italians and Irish of New York city. Cambridge, MA: MIT Press. Golash-Boza, T. (2006). Dropping the hyphen? Becoming Latino(a)-American through racialized assimilation. Social Forces, 85, 27–55.

188

ARNOUT VAN DE RIJT

Gonzales, J. L. (1986). Asian Indian immigration patterns: The origins of the Sikh community in California. International Migration Review, 20, 40–54. Gordon, M. (1954). Social structure and goals in group relations. In M. Berger, T. Abel & C. H. Page (Eds.), Freedom and control in modern society. New York, NY: D. Van Nostrand Co. Gordon, M. (1964). Assimilation in American life. The role of race, religion, and national origins. New York, NY: Oxford University Press. Grenier, G. (1984). Shifts to English as usual language by Americans of Spanish mother tongue. Social Psychology Quarterly, 65, 537–550. Guarnizo, L. E., Portes, A., & Haller, W. (2003). Assimilationism and transnationalism: determinants of transnational political action among contemporary migrants. American Journal of Sociology, 108, 1211–1248. Hagan, J. (1998). Social networks, gender, and immigrant incorporation: Resources and constraints. American Sociological Review, 63, 55–67. Halaby, C. N. (2004). Panel models in sociological research. Annual Review of Sociology, 30, 507–544. Heider, F. (1946). Attitudes and cognitive organization. The Journal of Psychology, 21, 107–112. Houle, R., & Schellenberg, G. (2010). New immigrants’ assessments of their life in Canada. Statistics Canada Analytic Studies Branch Research Paper Series no. 322. Hurh, W. M., & Kim, K. C. (1990). Religious participation of Korean immigrants in the United States. Journal for the Scientific Study of Religion, 29, 19–34. Huston, T. L., & Levinger, G. (1978). Interpersonal attraction and relationships. Annual Review of Psychology, 29, 115–156. Itzigsohn, J., & Saucedo, S. G. (2002). Immigrant incorporation and sociocultural transnationalism. International Migration Review, 36, 766–798. Jasso, G., Massey, D. S., Rosenzweig, M. R., & Smith, J. P. (2000). The new immigrant survey pilot: Overview and new findings about U.S. legal immigrants at admission. Demography, 37, 127–138. Kalmijn, M., & Van Tubergen, F. (2010). A comparative perspective on intermarriage: Explaining differences among national-origin groups in the United States. Demography, 47, 459–479. Kandel, D. B. (1978). Homophily, selection, and socialization in adolescent friendships. American Journal of Sociology, 84, 427–436. LaBrack, B. (1981). Immigration law and the revitalization process: The case of the California Sikhs. In S. Chandrasekhar (Ed.), From India to America (pp. 59–66). La Jolla, CA: Population Review. Lazarsfeld, P. F., & Merton, R. K. (1954). Friendship as a social process: A substantive and methodological analysis. In M. Berger et al. (Eds.), Freedom and control in modern society (pp. 18–66). New York, NY: Van Nostrand. Lee, S. M., & Boyd, M. (2008). Marrying out: Comparing the marital and social integration of Asians in the U.S. and Canada. Social Science Research, 37, 311–329. Leonard, K. (1982). Marriage and family life among early Asian Indian immigrants. In S. Chandrasekhar (Ed.), From India to America (pp. 67–75). La Jolla, CA: Population Review. Lewis, K., Gonzales, M., & Kaufman, J. (2011). Social selection and peer influence in an online social network. Proceedings of the National Academy of Sciences, 109, 68–72.

Assimilation Process of Immigrants

189

Martinovic, B., Van Tubergen, F., & Maas, I. (2009a). Dynamics of inter-ethnic contact. A panel-study of immigrants in the Netherlands. European Sociological Review, 25, 303–318. Martinovic, B., Van Tubergen, F., & Maas, I. (2009b). Changes in immigrants’ social integration during the stay in the host country: The case of non-western immigrants in the Netherlands. Social Science Research, 38, 870–882. Martinovic, B., Van Tubergen, F., & Maas, I. (2011). Acquisition of cross-ethnic friends by recent immigrants to Canada: A longitudinal approach. International Migration Review, 45, 460–488. Massey, D. S., Alarco´n, R., Durand, J., & Gonza´lez, H. (1987). Return to Aztlan. the social process of international migration from western Mexico. Berkeley, CA: University of California Press. Massey, D. S., & Espinosa, K. E. (1997). What’s driving Mexico-US migration? A theoretical, empirical, and policy analysis. American Journal of Sociology, 102, 939–999. Matute-Bianchi, M. E. (1986). Ethnic identities and patterns of school success and failure among Mexican-Descent and Japanese-American students in a California high school: An ethnographic analysis. American Journal of Education, 95, 233–255. McConnell, E. D., & LeClere, F. B. (2002). Selection, context, or both? The English fluency of Mexican immigrants in the American Midwest and Southwest. Population Research and Policy Review, 21, 179–204. McPherson, M., Smith-Lovin, L., & Cook, J. M. (2001). Birds of a feather: Homophily in social networks. Annual Review of Sociology, 27, 415–444. Medvedeva, M. (2012). Negotiating languages in immigrant families. International Migration Review, 46(2), 517–545. Menjı´ var, C. (1999). Fragmented ties: Salvadoran immigrant networks in America. Berkeley, CA: University of California Press. Mercken, L., Snijders, T. A. B., Steglich, C., Vartiainen, E., & De Vries, H. (2010). Dynamics of adolescent friendship networks and smoking behavior. Social Networks, 32, 72–81. Nee, V., & Sanders, J. M. (2001). Understanding the diversity of immigrant incorporation: A forms-of-capital model. Ethnic and Racial Studies, 24, 386–411. Nee, V., Sanders, J. M., & Sernau, S. (1994). Job transitions in an immigrant metropolis: Ethnic boundaries and the mixed economy. American Sociological Review, 59, 849–872. Newman, M. E. J. (2002). Assortative mixing in networks. Physical Review Letters, 89, 208–701. Otten, S., & Epstude, K. (2006). Overlapping mental representations of self, ingroup, and outgroup: Unraveling self-stereotyping and self-anchoring. Personality and Social Psychology Bulletin, 1121, 117–127. Pagnini, D. L., & Morgan, P. S. (1990). Intermarriage and social distance among U.S. immigrants at the turn of the century. American Journal of Sociology, 96, 405–432. Palloni, A., Massey, D. S., Ceballos, M., Espinosa, K., & Spittel, M. (2001). Capital and international migration: A test using information on family networks. American Journal of Sociology, 106, 1262–1298. Park, R. E. (1950). Race and culture. Glencoe, IL: The Free Press. Park, R. E., & Burgess, E. W. (1921). Introduction to the science of sociology. Chicago, IL: University of Chicago Press. Paul, A. M. (2011). Stepwise international migration: A multistage migration pattern for the aspiring migrant. American Journal of Sociology, 116, 1842–1886.

190

ARNOUT VAN DE RIJT

Perlmann, J., & Waldinger, R. (1997). Second generation decline? Children of immigrants, past and present – a reconsideration. International Migration Review, 31, 893–922. Portes, A., Guarnizo, L. E., & Haller, W. (2002). Transnational entrepreneurs: An alternative form of immigrant economic adaptation. American Sociological Review, 67, 278–298. Portes, A., & Sensenbrenner, J. (1993). Embeddedness and immigration: Notes on the social determinants of economic action. American Journal of Sociology, 98, 1320–1350. Portes, A., & Stepick, A. (1993). City on the edge: The transformation of Miami. Berkeley, CA: University of California Press. Portes, A., & Rumbaut, R. G. (1996). Immigrant America: A portrait. Berkeley, CA: University of California Press. Portes, A., & Zhou, M. (1993). The new second generation: Segmented assimilation and its variants. Annals of the American Academy of Political and Social Science, 530, 74–96. Rose, A. M. (1942). A research note on the influence of immigration on the birth rate. American Journal of Sociology, 47, 614–621. Rosenthal, E. (1960). Acculturation without assimilation? The Jewish community of Chicago, IL. American Journal of Sociology, 66, 275–288. Roth, W. D., Seidel, M. L., Ma, D., & Lo, E. (2012). In and out of the ethnic economy: A longitudinal analysis of ethnic networks and pathways to economic success across immigrant categories. International Migration Review, 46, 310–361. Rumbaut, R. G., & Portes, A. (2001). Legacies: Stories of the immigrant second generation. Berkeley, CA: University of California Press. Sandberg, N. (1973). Ethnic identity and assimilation: The Polish-American community. New York, NY: Praeger. Sanders, J. M., & Nee, V. G. (1987). Limits of ethnic solidarity in the enclave economy. American Sociological Review, 52, 745–773. Sanders, J. M., Nee, V. G., & Sernau, S. (2002). Asian immigrants’ reliance on social ties in a multi-ethnic labor market. Social Forces, 81, 281–314. Sayegh, L., & Lasry, J. (1993). Immigrants’ adaptation in Canada: Assimilation, acculturation, and orthogonal cultural identification. Canadian Psychology, 34, 98–109. Schelling, T. (1971). Dynamic models of segregation. Journal of Mathematical Sociology, 1, 143–186. Sears, D. O., Du, M., Henry, P. J., & Bui, K. (2003). The origins and persistence of ethnic identity among the ‘‘new immigrant’’ groups. Social Psychology Quarterly, 66, 419–437. Shibutani, T., & Kwan, K. M. (1965). Ethnic stratification: A comparative approach. New York, NY: MacMillan. Staerkle´, C., Sidanius, J., Green, E. G. T., & Molina, L. (2005). Ethnic minority-majority asymmetry and attitudes towards immigrants across 11 nations. Psicologı´a Polı´tica, 30, 7–26. Statistics Canada. (2005a). Longitudinal survey of immigrants to Canada. Catalogue No. 80–614–XIE. Statistics Canada, Ottawa. Statistics Canada. (2005b). Longitudinal survey of immigrants to Canada, microdata user guide for wave 2. Ottawa: Statistics Canada. Steglich, C., Snijders, T. A. B., & Pearson, M. (2010). Dynamic networks and behavior: Separating selection from influence. Sociological Methodology, 40, 329–393. Stevens, G. (1992). The social and demographic context of language use in the United States. American Sociological Review, 57, 171–185. Stevens, G. (1999). Age at immigration and second language proficiency among foreign-born adults. Language in Society, 28, 555–578.

Assimilation Process of Immigrants

191

Stoll, M. A., & Wong, J. S. (2007). Immigration and civic participation in a multiracial and multiethnic context. International Migration Review, 41, 880–908. Strogatz, S. (1994). Nonlinear systems and chaos. Cambridge, MA: Perseus. Tajfel, H., & Turner, J. C. (1979). An integrative theory of intergroup conflict. In W. G. Austin & S. Worchel (Eds.), The social psychology of intergroup relations. Monterey, CA: Wadsworth. Tazelaar, F. (1980). Mental incongruities – social restrictions – behavior: A study of labor force participation of married women with an academic degree. Utrecht, Netherlands: Utrecht University. Trilla, C. C., Esteve, A., & Domingo, A. (2008). Marriage patterns of the foreign-born population in a new country of immigration: The case of Spain. International Migration Review, 42, 877–902. Van de Rijt, A., Siegel, D., & Macy, M. (2009). Neighborhood chance and neighborhood change: Comment on bruch and mare. American Journal of Sociology, 114, 1166–1180. Van Tubergen, F., & Kalmijn, M. (2005). Destination-language proficiency in cross-national perspective: A study of immigrant groups in nine western countries. American Journal of Sociology, 110, 1412–1457. Van Tubergen, F., & Kalmijn, M. (2009a). A dynamic approach to the determinants of immigrants’ language proficiency: The United States 1980–2000. International Migration Review, 43, 519–543. Van Tubergen, F., & Kalmijn, M. (2009b). Language proficiency and usage among immigrants in the Netherlands: Incentives or opportunities? European Sociological Review, 25, 169–182. Waldinger, R. (2007). The bounded community: Turning foreigners into Americans in twenty first century L.A. Ethnic and Racial Studies, 30, 341–374. Waldinger, R., Lim, N., & Cort, D. (2007). Bad jobs, good jobs, no jobs? The employment experience of the Mexican American second generation. Journal of Ethnic and Migration Studies, 33, 1–35. Warner, W. L., & Srole, L. (1945). The social systems of American Ethnic groups. New Haven, CT: Yale University Press. Wasserman, S., & Faust, K. (1994). Social network analysis: Methods and applications. Cambridge: Cambridge University Press. Waters, M. (2001). Black identities: West Indian immigrant dreams and American realities. Cambridge, MA: Harvard University Press. Watts, D. J., Dodds, P. S., & Newman, M. E. J. (2002). Identity and search in social networks. Science, 296, 1302–1305. White, M. J., Biddlecom, A. E., & Guo, S. (1993). Immigration, naturalization, and residential assimilation among Asian Americans in 1980. Social Forces, 72, 93–117. Wilson, K. L., & Portes, A. (1980). Immigrant enclaves: An analysis of the labor market experiences of Cubans in Miami. American Journal of Sociology, 86, 295–319. Zhou, M. (1997). Segmented assimilation: Issues, controversies, and recent research on the new second generation. International Migration Review, 31, 975–1008. Zhou, M. (2001). Straddling different worlds. In R. G. Rumbaut & A. Portes (Eds.), Ethnicities: Children of immigrants in America (pp. 187–227). Berkeley, CA: University of California Press. Zhou, M., & Logan, J. R. (1991). In and out of Chinatown: Residential mobility and segregation of New York City’s Chinese. Social Forces, 70, 387–407. Zimmerman, L., Zimmermann, K. F., & Constant, A. (2007). Ethnic self-identification of firstgeneration immigrants. International Migration Review, 41, 769–781.

192

ARNOUT VAN DE RIJT

APPENDIX: A SIMPLE FORMALIZATION OF THE ASSIMILATION MODEL The assimilation model in the main text is left unformalized. Here we propose a simple formalization and derive the characteristic property of bidirectionality. Let t denote lifetime and a_k ¼ @ak =@t the rate of assimilation for an immigrant along dimension k. Selection (influence) can then be expressed as an immigrant’s rate of assimilation a_k along structural (cultural) dimension k being a positive linear function of the level of assimilation al along cultural (structural) dimension l6¼k. Furthermore, selection followed by influence (influence followed by selection) can then be expressed as the rate of assimilation along one cultural (structural) dimension depending on the level of assimilation along another cultural (structural) dimension. Thus, we may state that for all structural and cultural dimensions k, selection and influence produce positive linear cross-dimension effects: a_k ¼ ak þ

K X

bkl al

(A.1)

l¼1

Here, a is a K-sized vector of dimension-specific constants and b is a K  K matrix of coefficients. a_ is a linear system of differential equations with solution vector a(t) for a diagonalizable matrix b given by (e.g., Strogatz, 1994, pp. 123–44): aðtÞ ¼

K X

cm el m t vm

(A.2)

m¼1

where vm are eigenvectors and lm eigenvalues of b, and c is a vector of constants. For positive coefficients bkl the Perron–Frobenius theorem states that for some n, ln is real positive and strictly larger than all other eigenvalues, and that the corresponding eigenvector vn has only positive entries. As inspection of Eq. (A.2) establishes, this means that as t becomes large, the term cn eln t vn comes to dominate in magnitude, as it is asymptotically approached by a(t). Since cn eln t vn grows exponentially with t, the system has only two attractors, namely, maximal assimilation in all domains, ap, and minimal assimilation in all domains, a-p. After sufficient time has elapsed, the immigrant experiences forward assimilation along all

Assimilation Process of Immigrants

193

dimensions or reverse assimilation along all dimensions. Which of the two trajectories the immigrant will follow depends on whether cn is positive or negative, which in turn depends on initial conditions. Immigrants who start off rather assimilated further assimilate. Immigrants who start off not so assimilated reverse assimilate.

USING SIMULATED INTERACTIONS TO EXPLORE EMOTIONAL PROCESSES AND STATUS ORGANIZING PROCESSES: A JOINT APPLICATION OF EXPECTATION STATES THEORY AND AFFECT CONTROL THEORY Joseph Dippong ABSTRACT Purpose – The present chapter expands on recent research demonstrating an empirical link between theoretical constructs within affect control theory (ACT) and expectation states theory. I explore the utility of a joint application of the two theories, employing simulated interactions to examine status organizing processes. Design/methodology/approach – Although simulation results do not constitute data by which theoretical claims can be tested, they are useful for developing new research questions. I report results from a series of simulated dyadic interactions using ACT’s Interact program to Advances in Group Processes: Thirtieth Anniversary Edition Advances in Group Processes, Volume 30, 195–229 Copyright r 2013 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 0882-6145/doi:10.1108/S0882-6145(2013)0000030011

195

196

JOSEPH DIPPONG

investigate potential emotional and identity processes that underlie the enactment of status differences, and to explore affective responses to the legitimation and delegitimation of status orders. Findings – Simulation results call attention to a dynamic interplay between structural elements of the situation and the agentic behavior of interactants, suggesting that behavioral attempts to reduce deflections may lead to shifts in expectations over the course of interaction. Results raise the possibility that differences in affective impressions may produce expectations that are initially asymmetrical between interaction partners. Further, results suggest that the standardized tasks commonly employed in expectation states research may unintentionally generate affective responses that encourage status convergence. Originality/value – Drawing on insights and methods from ACT, expectation states researchers can improve the scientific understanding of small group interaction. Employing simulated interactions, researchers can promote theoretical advancement by uncovering new lines of inquiry at the intersection of two prominent social psychological traditions. Simulations also provide a further tool for methodological refinement within the standardized experimental setting. Keywords: Status; legitimation; affect; expectations; emotions; simulation

INTRODUCTION Expectation states theory and affect control theory embody two distinct and well-supported sociological approaches to explaining how individuals interpret and respond to the characteristics and behaviors of interaction partners. Given that affect control theory and expectation states theory focus on parallel – though certainly not identical – processes, it is only natural that social psychologists should consider how the two approaches might inform each other and be utilized together to shed new light on interaction processes. Indeed, Fararo and Skvoretz (1987, p. 1183) argue that ‘‘differentiated lines of research set up pressures for intellectual integration of the resulting theories and findings.’’ These integrative pressures between affect control theory and expectation states theory are clearly evident in a number of recent theoretical and empirical works aimed

Exploring Emotional Processes and Status Organizing Processes

197

at identifying and exploring connections between the two traditions. The continued examination of commonalities between expectation states theory and affect control theory can be seen as an intermediate stage, where the aim is to expand our existing knowledge and further unification.1 Building on previous theoretical and empirical work connecting affect control theory to expectation states theory, Dippong and Kalkhoff (2012) demonstrate a rather straightforward empirical linear relationship between the theoretical constructs central to these two research traditions. Employing both survey-based and experimental methods, Dippong and Kalkhoff (2012) empirically derive a set of regression weights by which measures of affective impressions at the heart of affect control theory research can be translated into a well-established quantitative measure of expectations. Further, Dippong and Kalkhoff (2012) begin to explore the utility of their model using simulated interactions to examine the processes by which differentiated performance expectations arise in initially statushomogenous dyads. The creation of a mathematical ‘‘bridge’’ linking concepts between affect control theory and expectation states theory represents an important step toward unification of the two (Markovsky, 2010; Markovsky et al., 2008). The purpose of the present chapter is to demonstrate the further value of a joint application of affect control theory and expectation states theory by examining how researchers can make use of simulated interactions to gain a better understanding of the interplay between expectations and emotions in status-differentiated groups. Specifically, I provide examples for how expectation states researchers can employ insights from affect control theory to predict specific emotional responses to the legitimation and delegitimation of status orders, as well as how emotional responses shape the expectations that actors develop for themselves and for their interaction partners. Aside from examining how emotions modify expectations for various higher and lower status actors, I also demonstrate how researchers can explore emotional processes within the status characteristics theoretical framework and refine experimental predictions by reconstructing and simulating patterns of interaction observed in the laboratory. In linking affect control theory and expectation states theory, the goal is to produce a better understanding of status organizing processes and affective identity processes by calling attention to social behaviors (and the situational factors that give rise to them) in a unique way and developing questions that neither theory would suggest on its own. By focusing on the intersection of emotions and status, researchers can more fully understand the relationship between structural facets of the situation and more dynamic

198

JOSEPH DIPPONG

elements of interaction by taking into account the affective processes by which actors construct and maintain a ‘‘definition of the situation’’ (Mead, 1934).

THEORETICAL BACKGROUND Affect control theory (hereafter ACT) originated as a social psychological explanation for how actors construct and maintain a ‘‘definition of the situation’’ in interaction (Heise, 1985). In explaining the processes through which individuals select and interpret social information, ACT provides a model for social action with an explicit emphasis on understanding behavior within the context of the sociocultural meaning systems in which interactions are embedded (Heise, 2007; Smith-Lovin, 1987a). To this end, ACT focuses on how the relationship between sentiments and identities influences individuals’ choice of behaviors, the labels they apply to themselves and interaction partners, and the emotions that actors experience within the situation. ACT, then, seeks to explain both how individuals recognize and filter the information that is available to them in interaction and how they interpret and respond to that information (Heise, 1987). As its name indicates, ACT centers on the generation of behaviors and emotions as a control process. According to the theory, as individuals interact, they experience affective responses that are specific to the situation, and compare those responses to internalized cultural standards that embody beliefs about the ways in which people are expected to respond (Heise, 1977; Heise, 1985). The affective control process involves an internal comparison between two different types of sentiments: (1) transient sentiments that represent situation-specific affective responses evoked through interaction; and (2) fundamental sentiments that encompass the internalized cultural standards by which social actors affectively define the elements of interaction (Heise, 1985; Smith-Lovin, 1987a). Transient and fundamental sentiments reflect the affective meanings that actors assign to identities, behaviors, and social settings, and form the basis for constructing and maintaining shared meanings through interaction. When an individual experiences a mismatch (or deflection) between situational meanings and internalized standards, ACT posits that she or he will enact the subsequent behavior that is most likely to reduce or eliminate the degree of deflection (Heise, 2007). For example, in an interaction between two actors – a professor and a student – fundamental sentiments define the cultural meanings assigned to the enactment of these two different identities.

Exploring Emotional Processes and Status Organizing Processes

199

If the transient sentiments that arise in the interaction do not match the respective fundamental sentiments, such as might occur if the student addresses the professor as a peer, the professor (operating under a different definition of the situation) would adjust her or his behavioral output to alleviate the deflection and create a consistent and shared understanding between the actors.

Quantitative Measures of Affective Impressions Both transient and fundamental sentiments consist of affective impressions that individuals experience across three separate dimensions. The first of these dimensions, evaluation, reflects people’s perceptions of the relative goodness (or badness) of an actor, behavior, etc. Second, the potency dimension captures impressions of powerfulness (or powerlessness). And third, the activity dimension embodies impressions of expressivity (or passivity). Research suggests that evaluation, potency, and activity (EPA) represent social dimensions that are culturally universal (Heise, 1985; Osgood, May, & Miron, 1975). Using semantic differential scales, researchers create EPA profiles that capture both the qualitative content of impressions and the magnitude of those impressions. To capture the qualitative component of impressions, researchers employ scales anchored by descriptive terms that reflect the affective meanings of actors, settings, and behaviors. Impressions along the evaluation dimension are commonly measured on a scale with semantic anchors that range from ‘‘good, nice’’ to ‘‘bad, awful’’; potency impressions are anchored by the terms ‘‘powerful’’ and ‘‘powerless’’; activity is anchored using terms such as ‘‘fast, noisy, young’’ to represent high levels of activity, and ‘‘slow, quiet old’’ to signify impressions of passivity. In terms of magnitude, affective impressions for all three dimensions range from –4.3 to +4.3, with larger absolute values reflecting stronger, more intense impressions. A given affective profile consists of ratings across the three EPA dimensions. ACT has proven extremely useful for producing quantitative predictions concerning the transient sentiments that are likely to arise in a given situation, the degree of deflection between transient and fundamental sentiments, the behaviors that individuals are likely to enact in response to experiencing deflection, and the emotional outcomes of these actions. Furthermore, the development of the computer-based Interact program has allowed researchers to simulate a wide variety of social situations, and to

200

JOSEPH DIPPONG

generate predictions regarding emotional and behavioral outcomes based on a number of affective meaning dictionaries.2 These dictionaries include ratings provided across multiple time periods and by samples from a variety of cultures, including the United States, Canada, Germany, Ireland, Japan, and China, making cross-cultural comparisons possible.

Expectation States Theory The expectation states theoretical research program encompasses a set of interrelated structural social psychological theories, including status characteristics theory (Berger, Cohen, & Zelditch, 1972; Berger, Fisek, Norman, & Zelditch, 1977), evaluations and expectations theory (Fisek, Berger, & Norman, 1991), reward expectations theory (Berger, Fisek, Norman, & Wagner, 1985), and double standards theory (Foschi, 2000), to name only a few. Although each of these unique branches of the expectation states research program addresses different facets of the structural underpinnings of social interaction, they are tied together by a shared focus on the expectations that individuals form for themselves and others as a basis for organizing behavior in small groups. The central argument of this group of theories is that expectations – whether they are performance expectations, reward expectations, or expectations that arise through behavioral interchange patterns – mediate the relationship between elements of the situation, such as the distribution of overt status characteristics within the group, and group members’ behavior. Evidence strongly suggests that in collective task groups, expectations form very quickly and once formed, act as self-fulfilling prophecies (Berger et al., 1977). The stable and observable patterns of influence, reward allocation, and task participation reflect a hierarchy of power and prestige that is rooted in people’s expectations. In this section I focus on the status characteristics and legitimation theoretical branches, which are relevant to the analyses below.

Status Characteristics Theory Status characteristics theory (hereafter SCT) explains the processes through which the cultural beliefs individuals hold about the relationship between a person’s status characteristics and her or his abilities are ‘‘imported’’ into micro-social interaction, producing inequality in collective task groups.

Exploring Emotional Processes and Status Organizing Processes

201

Beliefs about status characteristics serve as a basis for the formation of performance expectations, which are defined as ‘‘anticipations on the part of an individual of the abilities and task capacities of self and others’’ (Berger & Webster, 2006, p. 269). Performance expectations provide the foundation for observable hierarchies of power and prestige, in which relatively higher status actors are typically evaluated more positively, contribute more to group tasks, and are generally accorded a greater level of influence compared to relatively low-status actors. Over several decades, the central claims of SCT have amassed an impressive amount of empirical support (see Kalkhoff & Thye, 2006; Thye & Kalkhoff, 2009 for meta-analyses). Any trait can serve as a status characteristic if it meets two criteria: (1) individuals can be categorized according to at least two states, or ‘‘levels,’’ and (2) levels of the characteristic are differentially valued, such that possessing a relatively more valued level accords one greater social esteem and rewards (Berger et al., 1972). Furthermore, SCT distinguishes between two different forms of status characteristics; diffuse and specific, which are distinguished by the types of performance expectations to which they give rise. Whereas diffuse status characteristics, such as gender or social class, are connected to general performance expectations (which can be thought of as a global assessment of a person’s abilities, generally speaking), specific status characteristics are connected to expectations about a person’s abilities related to an explicit range of tasks (Berger et al., 1977). Interestingly, SCT postulates that even when status characteristics are not explicitly relevant to the group’s task, members nonetheless infer a connection and develop expectations accordingly (see the Appendix in Berger, Norman, Balkwell, & Smith, 1992 for a more in-depth description). Assumptions and Scope Conditions SCT employs a set of theoretical assumptions that describe the process through which beliefs about status characteristics become activated and produce performance expectations, as well as a number of scope conditions that delineate the specific situational context in which the theory’s predictions are expected to hold. Concerning assumptions, SCT states that status characteristics become salient (activated) when they differentiate between group members (such as in a mixed-sex group). The burden of proof assumption states that individuals treat all salient status characteristics as relevant to the group task unless they are provided with information or evidence that explicitly disassociates the characteristic from the task. Further assumptions specify how actors combine consistent and inconsistent status information, as well as how the informational structure of the

202

JOSEPH DIPPONG

interaction is affected when actors enter or leave the situation. In terms of theoretical scope, SCT offers predictions about the behavior of individuals interacting in small, collective task groups. That is, SCT specifies that status characteristics give rise to performance expectations when actors are working to complete a task which they are motivated to complete successfully, and when they believe that taking other group members’ opinions into account is necessary (Berger et al., 1972). See Appendices A and B for complete lists of SCT’s scope conditions and assumptions.

Legitimation Legitmation refers to the process by which beliefs and behaviors come to be seen as the way things ‘‘ought’’ to be or the way things should be done. Although status differences are sufficient for explaining inequality of influence within a group (a status-related outcome), research on legitimation examines how those initial differences lead to behavioral displays of groupsupported dominance from higher status actors (a power-related outcome) (Ridgeway & Berger, 1986). In essence, when a status order is legitimated, higher status actors exert more power because their position within the group is deemed appropriate, and they can, therefore, expect to be ‘‘backed’’ by the group in demanding compliance from particular lower status members (Berger, Ridgeway, Fisek, & Norman, 1998). The process of legitimation involves the activation of referential structures whereby individuals take into consideration the distribution of status and power positions within the group (Berger et al., 1998). When individuals assume that beliefs about valued status positions are normatively prescriptive, and when there is consensus within the group about the prescriptive nature of those beliefs, the power and prestige order is legitimated (Berger et al., 1998). This process often comes about when the group’s hierarchical structure is validated by an individual external to the group (Kalkhoff, 2005). Research suggests that power and prestige orders are more likely to be legitimated when they arise in status-heterogeneous groups (as opposed to status-homogenous groups), and when individuals are consistently at higher or lower status across multiple-status characteristics, creating larger status differences (Ridgeway, Johnson, & Diekema, 1994). Once a power and prestige order is legitimated, it can become delegitimated if events arise that create a mismatch between the previously

Exploring Emotional Processes and Status Organizing Processes

203

legitimate order and the actual distribution of expectations. For example, consider a task group composed of three coworkers – Stephen, Susan, and Jennifer – differentiated according to a single diffuse status characteristic (gender). If the group members believe that gender differences are prescriptive for differentiated expectations and behaviors, the status order is legitimated. If, however, the coworkers’ supervisor interjects support for Susan’s ideas and rejects Stephen’s suggestions, group members will develop new (lower) expectations for Stephen according to his demonstrated lack of task ability. And because expectations are symmetrical Susan and Jennifer would develop higher expectations for themselves. In this case, the two female members’ lower expectations for Stephen and higher expectations for themselves would produce a hierarchy of expectations inconsistent with the status order. Stephen’s authority would be delegitimated and a new status order would emerge.

Quantitative Measures of Expectations Many branches of expectation states theory employ graph–theoretic models of interactions to derive quantitative measures of the expectations that group members will form for each other. Computations based on the graph theoretic model produce a measure, termed the ‘‘expectation state value,’’ which ranges from –1 to +1, and captures an individual’s expectations for both self and other. Expectation state values are typically transformed to model expectation differences in dyadic relations (a measure termed expectation advantage), or in N-sized groups (termed expectation standing; Fisek et al., 1991; Kalkhoff, Friedkin, & Johnsen, 2010). These measures are especially useful for testing hypotheses derived from expectation states theory within the standardized experimental setting (SES). Subjective measures of expectations also exist, and are quite useful for verifying the successful manipulation of status differences within the SES, or for examining status organizing processes outside the SES. Zeller and Warnecke (1973) developed one such subjective measure of expectations. In constructing their index of general expectations, theyemployed questions assessing a variety of beliefs potentially linked to expectations, such as feelings of superiority or perceptions of morality for self and other. Zeller and Warnecke (1973) validated the scale through an experimental examination of the effects of education as a diffuse status characteristic. Their analyses indicated that the majority of variance in

204

JOSEPH DIPPONG

behavior across levels of educational attainment could be explained by the mediating effects of a single three-item index of expectations. Driskell and Mullen (1988) empirically demonstrate that subjective measures, including the Zeller and Warnecke scale, are reliable indicators of expectations.

Sentiments and Expectations Researchers within the expectation states tradition have given considerable attention to the role of sentiments and emotions in status organizing processes. For example, employing the theoretical framework of SCT, Lovaglia and Houser (1996) demonstrate that when individuals within a status hierarchy experience emotions that are consistent with their status position, the effects of structural inequality are reduced, while the experience of status-inconsistent emotions increases the effects of structural inequality. Stated differently, when higher status actors experience positive emotions they accord greater influence to lower status actors, and when lower status actors experience negative emotions, they accord less influence to higher status actors (Lovaglia & Houser, 1996). Noting that status and sentiments produce similar behavioral outcomes, Driskell and Webster (1997) seek to address the question of how sentiments, which possess no inherent task connotations, produce differentiated patterns of behavior. As a step toward answering this question, Driskell and Webster (1997) present evidence that an individual’s status predicts the type of sentiments that group members develop toward her or him, with higher status actors being generally more well-liked and evaluated more positively. Fisek and Berger (1998) shed further light on the issue by developing and testing models designed to examine whether sentiments function by mediating the relationship between expectations and behaviors (the mediation model) or by directly contributing to the expectations that individuals form for self and other (the constituent model). While Fisek and Berger (1998) find no theoretical or empirical basis to prefer one model over the other, Bianchi (2004; Bianchi & Lancianese, 2007) provides evidence that sentiments in fact moderate the effects of status characteristics on behavioral outcomes. Although more work is clearly needed to fully understand how sentiments and expectations work together in shaping behavior, existing evidence strongly suggests that emotional reactions play a key role in shaping behavioral responses to structurally based inequalities.

Exploring Emotional Processes and Status Organizing Processes

205

Connecting Expectations, Affective Impressions, and Emotions Social psychologists have recently begun to examine how EPA profiles map onto the ways in which individuals perceive themselves and others in terms of social status and power (cf. MacKinnon & Langford, 1994; Rogalin, Soboroff, & Lovaglia, 2007), as well as to explore theoretical connections between ACT and expectation states theory (Ridgeway & Smith-Lovin, 1994). At this stage, evidence indicates that the dimensions of evaluation and potency are positively associated with the perceived level of power and status accorded to a variety of occupational titles, though results suggest a more complex relationship between the activity dimension and levels of power and status.3 Research linking EPA profiles to power and status evaluations are an instructive first step, but in order to draw a clear connection between ACT and expectation states theory, it is necessary to go a step further and explore the potential links between EPA profiles and expectations. Addressing the relationship between affective impressions and expectations, Dippong and Kalkhoff (2012) employ both survey data and experimental data to demonstrate a positive relationship between affective profiles and general expectations. Dippong and Kalkhoff (2012) argue that all three dimensions of affective impressions are positively associated with expectations, and find empirical support for this prediction by linking participants’ affective impressions of other social actors with their subjective expectations of those actors. Across two studies, EPA significantly predicted expectations, as measured by Zeller and Warnecke’s (1973) three-item index (Dippong & Kalkhoff, 2012). A comparison of regression weights across survey-based and experimental results further revealed that the relationships present in the data were consistent in terms of both the direction of the effects and the magnitude of the effects. Aside from establishing a clear theoretical connection between the central constructs of ACT and expectation states theory, Dippong and Kalkhoff (2012) also point to the potential use of regression weights derived from their analyses to transform affective profiles into measures of predicted expectations. This possibility is intriguing for at least two reasons: (1) at a theoretical level, it provides a further avenue by which researchers can explore specific emotional reactions to status differentiation, as well as processes by which emotional reactions produce behaviors that reinforce or undermine status and power hierarchies; and (2) at a methodological level, it opens the door for expectation states researchers to draw on ACT’s Interact program to simulate status processes for the purpose of exploring and

206

JOSEPH DIPPONG

developing new hypotheses. As an example of how this might be done, Kalkhoff, Walker, and Webster (2013) discuss the utility of Interact simulations for helping social psychologists adjudicate between competing theoretical models of second-order expectations. Of course, no simulation can directly test a theoretical argument, but they are useful for illuminating new and previously unconsidered questions. As such, employing ACT and expectation states theory together provides yet another means to advance these two already prolific and well-supported traditions.

SIMULATIONS In this section I employ ACT’s Interact simulation program along with Dippong and Kalkhoff’s model to explore the connection between emotions and expectations in status organizing processes. Interact provides researchers with a number of options in constructing simulated interactions. First, as stated above, Interact allows simulations to be based on affective profiles contained in a variety of cultural dictionaries, which include data collected from samples in the United States, Canada, and a selection of European and Asian countries. Second, the program allows researchers to build interactions based on identities as modified by a variety of emotions and within a number of environmental contexts. And third, researchers can model interactions differently based on ratings from male and female actors, or they can employ a dictionary of averaged male and female ratings. All simulations presented here were conducted based on the Indiana University 2003 dictionary. Because Interact’s cultural dictionaries do not contain affective profiles for a number of the specific behaviors and situations of interest to expectation states researchers, in the simulations I present it was necessary to select from among a variety of behaviors and actor labels to serve as proxies. For modeling interactions in which there was no clear-cut proxy profile, or when multiple profiles captured the intended meaning equally well, I simulated interactions based on a range of descriptive terms that I deemed representative of the target behavior or identity. For all cases, I present the results obtained by averaging outcomes across multiple simulations. Admittedly, there are no objective criteria for choosing actor labels and behaviors to serve as proxies. Although the process of selecting proxies is necessarily imperfect, the simulation analyses I present are intended to illustrate a methodological approach to examining emotions and status

Exploring Emotional Processes and Status Organizing Processes

207

processes, as well as to suggest areas for further research. As such it should be noted that simulation results do not constitute data in and of themselves. Indeed, one goal of the present chapter is to encourage researchers to employ simulated interactions as a means to examine the connection between expectations and emotions in the status organizing processes I discuss, as well as in processes beyond those examined here.

Simulation One: Emotional Processes in Legitimation and Delegitimation As stated above, the power and prestige order of a group is legitimated when group members view the status order as normatively prescriptive. This can occur when an external authority authorizes or supports the contributions of higher status actors, as well as when group members themselves endorse the existing status structure (Walker, Thomas, & Zelditch, 1986). Likewise, a status order can become delegitimated when it is challenged by an external authority, such as when a supervisor approves of the contributions and ideas of lower status actors. In the case of both legitimation and delegitimation, it is logical to expect two distinct sets of emotional reactions from group members: (1) positive emotions from actors whose position has been supported; and (2) negative emotions on the part of actors whose position has not been supported. In other words, we would expect higher status actors to experience positive emotions when the status order is legitimated and negative emotions when the order is delegitmated, and the opposite for lower status actors. If and how these anticipated emotional reactions contribute to the power and prestige order of a group is an empirical question, and simulated interactions provide a means to generate more precise predictions concerning the actual emotions that actors will experience when a status order is legitimated (or delegitimated), while also allowing for an examination of resulting changes in expectations. Procedures To explore emotional responses to legitimation and delegitimation, I constructed a series of simulated interactions consisting of four distinct events. Simulations of legitimation involved two inter-related events: (1) a higher status actor receiving authorization/approval from an outside authority; and (2) a lower status actor receiving correction/disapproval from an outside authority. Similarly, simulations of delegitimation involved two related events: (1) approval directed toward lower status actors; and (2) disapproval directed toward higher status actors. I began by selecting

208

JOSEPH DIPPONG

actor labels from Interact’s dictionary of identities to represent higher and lower status actors.4 To ensure that the results I obtained were not directly attributable to some peculiarity of the identities I employed, I deemed it necessary to simulate each interaction multiple times using a variety of actor and partner labels, and using multiple different interpretations of participants’ behavior. In choosing actor labels, I employed pairs of identities that reflect relationships of clear status differentiation, such as professor–pupil and superordinate–subordinate. Once actor labels were selected, Interact provided an initial (unmodified) affective profile for each individual, from which I calculated predicted subjective expectation values. Expectations were derived from Dippong and Kalkhoff’s (2012; study one) regression weights, that generate predicted expectations (in terms of Zeller and Warnecke’s three-item index) based on EPA profiles. Predicted subjective expectations obtained from this model range from 0 (lowest expectations) to 9 (highest expectations).5 The left side of Table 1 presents the list of actor profiles on which simulations were based, as well as the resulting unmodified predicted expectations. Table 1. Unmodified Actor Label

Simulated Emotional Reactions to Legitimation.

Unmodified EPA and Expectations

Avg. Modified EPA and Expectations

Modified Actor Lables

Positive Modifiers: Tolerant, conscientious, contented, tactful, accommodating, moved, sympathetic, warm, self-righteous, introspective High Status E P A Ex E P A Ex Superordinate 0.53 1.24 0.95 7.10 0.94 1.11 0.45 7.07 Coworker, chum, lecturer Professor 1.72 1.73 0.44 7.38 1.62 1.31 0.19 7.22 Scholar, scientist, doctor Male 1.33 1.67 1.51 7.42 1.28 1.34 0.79 7.23 Foreman, superordinate Average: 1.19 1.55 0.97 7.30 1.28 1.25 0.48 7.17 Consultant, advisor, colleague Negative Modifiers: Defensive, anxious, angry, foolish, self-conscious, aggravated, flustered, disapproving, annoyed, careless Low Status Subordinate 0.28 1.27 0.49 6.10 1.07 1.12 0.37 6.02 Mental case, loser Pupil 1.29 0.35 0.08 6.70 0.63 0.59 0.04 6.28 Lackey, idiot Female 2.09 0.61 0.57 7.16 0.35 0.01 0.25 6.52 Mental case, loser Average: 1.19 0.34 0.53 6.65 0.68 0.57 0.05 6.27 Toady, parolee, suspect

Exploring Emotional Processes and Status Organizing Processes

209

Simulations involved two phases. First, I employ Interact to derive a list of the simulated group members’ predicted emotional reactions to legitimating or delegitimating feedback from an external authority. And second, I examine how the predicted emotional reactions influence group members’ affective impressions of themselves by generating emotion-modified EPA profiles. Following simulations, I calculated predicted subjective expectations for higher and lower status actors based on emotion-modified EPA profiles. I base my simulations on the assumption that the process of legitimation supports existing beliefs about the relative high value of task contributions from higher status actors as well as beliefs about the relative low value of task contributions from lower status actors, while delegitimation challenges such beliefs. Accordingly, in my simulation models, a show of support for higher status actors is tantamount to a rejection of the contributions of lower status actors, and vice versa. To generate predicted emotional reactions to legitimation and delegitimation, I constructed a set of interactions in which an external authority, as represented by the affective profile ‘‘role model’’ (E=2.47; P=2.31; A=1.57), interjects support for the existing status order. Results were obtained by simulating the four authority-group member interactions described above. To model legitimation I assigned positive interpretations (relative to self) of the external authority to higher status actors, using the terms ‘‘praise,’’ ‘‘back,’’ ‘‘defend,’’ and ‘‘glorify,’’ and assigned negative behavioral interpretations to lower status actors, using the terms ‘‘reprimand,’’ ‘‘defeat,’’ ‘‘correct,’’ and ‘‘dress-down.’’ To model delegitimation, I assigned positive interpretations to lower status actors and negative interpretations to higher status actors. Interact provided a list of several predicted emotional outcomes to the events modeled in each of the simulations described above. Across all simulated interactions, a number of emotional reactions emerged as common to both higher and lower status actors when faced with positive and negative feedback from an authority, and as such, I employ the same emotional modifiers to capture the effects of legitimation and delegitimation. Table 1 includes a full list of the positive and negative emotional modifiers employed in simulations. The final step in simulating the emotional effects of legitimation and delegitimation involved examining how the model-predicted emotional reactions modified the affective meanings of the identity labels actors ascribed to themselves and their partners. To obtain emotion-modified profiles, I applied each of the predicted emotional outcomes generated by Interact to each of the initial (unmodified) actor labels, which produced a

210

JOSEPH DIPPONG

new, altered EPA profile. Results reflect the average of the EPA profiles obtained by modifying each actor label by each of the predicted emotional outcomes. Also, I present predicted subjective expectations that are based on the averaged EPA profiles. Results Looking first at the top half of Table 1 (higher status actors), we see that Interact predicts a wide range of positive emotional responses to legitimation, including tolerance, conscientiousness, sympathy, and warmth. Table 1 presents the unmodified EPA profiles and predicted subjective expectations for three specific status-differentiated interaction pairs (superordinate–subordinate, professor–pupil, and male–female), as well as EPA profiles averaged across the three higher status identities and three lower status identities. The right side of Table 1 presents the affective profiles and expectations of the group members as modified by emotional reactions following delegitimation. Comparing the averaged unmodified identities to the averaged emotionmodified identities for higher status actors in the top half of Table 1, we see that modifying the initial affective profiles of higher status actors by the predicted positive emotional responses produces virtually no change in the overall evaluation, potency, or activity ratings of the simulated actors, along with a negligible decrease in predicted subjective expectations. While the slight decrease in expectations (–0.13) for higher status actors is unexpected, on the surface, such a change might be attributable to affective responses to power differentials between higher status actors and even more high-status authorities. Individuals necessarily lose a certain degree of potency by being objects of action rather than actors themselves (Smith-Lovin, 1987b). In the far right column, Table 1 also presents a list of actor labels derived from the emotion-modified EPA profiles. Looking at these emotionmodified identities, we see that higher status actors remain connected to relatively positive, powerful, higher status labels, including scholar, superordinate, and advisor. Thus, on the surface it appears that legitimation and its accompanying emotions produces little to no change in the affective impressions that higher status actors hold for themselves. Turning next to the results for lower status actors at the bottom of Table 1, simulations present a very different picture compared to that seen for higher status actors. As one might expect, simulations predict a number of negative emotional reactions to legitimation on the part of lower status actors, some of which include defensiveness, anxiety, anger, and selfconsciousness. Comparing the unmodified pre-legitimation affective profiles

Exploring Emotional Processes and Status Organizing Processes

211

to the emotion-modified ratings, the most noteworthy change in EPA profiles can be seen in the precipitous drop-off in evaluation ratings for lower status actors. That is, the average initial evaluation score for lower status actors is lowered by a magnitude of 1.87 following emotional reactions to legitimation. The substantial decrease in evaluation ratings for lower status actors corresponds to a drastic shift in the predicted identity labels that they ascribe to themselves. Specifically, the negative emotionmodified affective profile of lower status actors becomes associated with such devalued identities as mental case, lackey, idiot, and loser. Although results again reveal only a slight decrease in pre- and post-interaction predicted subjective expectations (–.38), this relatively minor shift belies a substantial change in the definitions that actors are applying to the situation. Because the average EPA profile for higher status actors remains essentially unchanged following authorization from an outside authority, while the profile for lower status actors changes markedly, simulation results suggest that the ability of higher status actors to assert dominance in legitimated status structures might be deemed ‘‘reasonable’’ within the group not by way of increased power in the affective profile of higher status actors, but by way of decreased evaluations of lower status actors. In essence, higher status actors may derive their legitimate power from a decreased perception of the relative goodness of lower status actors, and I consider possible explanations for this effect in the discussion section below. Furthermore, because the model-predicted subjective expectations also remain virtually unchanged following external authorization, the model raises the possibility that legitimation exerts its effects largely through emotional reactions that lead to a redefinition of the situation, and not entirely through changes in relative expectations. Results for simulations of delegitimation presented in Table 2 reveal a somewhat different pattern of predicted effects. Comparing the initial unmodified EPA profiles of higher status actors to the negative emotionmodified profiles on the top of Table 2, we see a noticeable decrease both in average evaluation ratings (a change of magnitude –1.82) and in average potency ratings (a change of magnitude –0.96). In turn, the negative emotion-modified affective profile of higher status actors following delegitimation associates them with negative identity labels that include opponent, foe, and novice. Thus, it appears that delegitimation not only causes higher status actors to identify themselves as less competent (i.e., novice), but it also diminishes commitment from the collective group, as evidenced in identity labels connoting competition, rather than partnership.

212

JOSEPH DIPPONG

Table 2. Unmodified Actor Label

Simulated Emotional Reactions to Delegitimation. Unmodified EPA and Expectations

Avg. Modified EPA and Expectations

Modified Actor Labels

Negative Modifiers: Defensive, anxious, angry, foolish, self-conscious, aggravated, flustered, disapproving, annoyed, careless High Status E P A Ex E P A Ex Superordinate 0.53 1.24 0.95 7.10 –0.85 0.42 0.55 6.58 Opponent, foe Professor 1.72 1.73 0.44 7.38 –0.42 0.67 0.20 6.68 Demagogue, quack Male 1.33 1.67 1.51 7.42 –0.62 0.69 0.89 6.73 Opponent, goof-off Average: 1.19 1.55 0.97 7.30 –0.63 0.59 0.55 6.66 Tease, opponent, novice Positive Modifiers: Tolerant, conscientious, contented, tactful, accommodating, moved, sympathetic, warm, self-righteous, introspective Low Status Subordinate 0.28 1.27 0.49 6.10 0.80 1.10 0.49 7.05 Coworker, righthand man Pupil 1.29 0.35 0.08 6.70 1.33 0.06 0.05 6.81 Assistant, neighbor, coworker Female 2.09 0.61 0.57 7.16 0.35 0.01 0.25 7.10 Assistant, advisor, colleague, partner Average: 1.19 0.34 0.53 6.65 0.59 0.38 0.26 6.99 Chum, coworker, scholar

Conversely, as shown at the bottom half of Table 2, when lower status actors experience the predicted positive emotions following the delegitimation of a disadvantaging status order, they experience a small but noticeable bump in their average potency ratings (an increase of 0.72). As a result, the average affective profiles of lower status actors flowing delegitimation come to be associated with identity labels that reflect generally equal-status relations, such as coworker, neighbor, and partner. These labels also reflect a certain degree of communality (partner, coworker), suggesting that delegitimation may increase lower status actors’ commitment to the group, while at the same time undermining higher status actors’ commitment. It appears, then, that when an existing status structure is delegitimated, emotional reactions from higher status actors may compromise the collective nature of the task, in that higher status actors take on the affective profile an outsider; they are no longer seen as task partners, but instead come to be viewed (and come to view themselves) as competitors. Furthermore, simulation results suggest that legitimation and delegitimation may encompass slightly different affective processes. Because the positive emotional outcomes for higher status actors in a legitimated group convey no apparent advantage, it seems that legitimation may be driven by the

Exploring Emotional Processes and Status Organizing Processes

213

negative emotional reactions of lower status actors. At the same time, in delegitimated structures, both the negative emotional outcomes for higher status actors and the positive emotional outcomes for lower status actors produce a substantial shift in situational definitions.

Simulation Two: Reconstructing Laboratory Interactions Researchers across several branches of expectation states theory examine status organizing processes following a well-developed experimental protocol, termed the standardized experimental setting. The SES provides a framework by which experimenters can manipulate bases of expectations and assess the effects of differentiation. The most common approach to measuring the behavioral outcomes of differentiated expectations involves assessing rates of influence between group members. Measuring influence in the SES typically involves presenting research participants with a series of ambiguous binary-choice questions and asking them to choose between the two (under the belief that one of the two options is correct). Following an initial choice, participants are informed of a partner’s initial choice, and are then given the opportunity to either retain their initial choice or to change it based on feedback from the partner. Over a series of trials, then, individuals make a number of initial and final choices, creating unique patterns of interaction between partners. Openness to influence is operationalized as the proportion of trials on which the participant makes a stay response (rejects influence), which is denoted by p(s). One way in which ACT and SCT researchers can utilize Interact to explore affective processes and status organizing processes is to employ participants’ binary data to reconstruct patterns of interaction observed in the laboratory. By working with binary data to construct simulations, we can begin to model how affect and identity operate together with status and power differences to influence behavior in small groups. Simulating reconstructed experimental interactions can call researchers’ attention to facets of interaction that might go unnoticed when examining only p(s) values. Observations based on simulations, then, can offer fodder for new theoretical questions and further experimental investigation. Procedures To examine interplay between status organizing processes and affective identity processes, I constructed another set of simulations, this time drawing directly on behavioral data from a recent laboratory experiment

214

JOSEPH DIPPONG

(i.e., Dippong, 2013) conducted within SCT’s SES. Simulations involved recreating patterns of agreement, disagreement, ‘‘stay’’ responses, and ‘‘change’’ responses observed between male and female participants interacting in status-differentiated dyadic groups. In these simulations, I employ a range of status-differentiated actor labels (and associated affective profiles) along with influence-related behavioral terms available within Interact to shed light on possible dynamic affective processes that play out over the course of an extended interaction. Thus, rather than focusing directly on p(s) scores, I focus on the processes by which those scores are generated. I began the process of constructing simulations by randomly selecting data from ten participants from the recent study cited above, selecting five participants from a high-status condition, and five from a low-status condition. Data for the 10 participants consisted of the binary information reflecting the specific pattern of ‘‘stay’’ and ‘‘change’’ responses each participant demonstrated when faced with disagreement from their ostensive task partner. The original experiment consisted of 25 task trials, with 5 of the trials creating agreement between participants and their partners, and 20 creating disagreement.6 Individuals in the original study interacted in mixed-sex dyads, and were differentiated according to a single diffuse-status characteristic. Concerning the data from the 10 participants on which the simulations are based, the mean p(s) for higher status actors was .63 and for lower status actors the mean p(s) was .38.7 Thus, higher status actors rejected influence at a higher rate than their lower status counterparts. I created status-differentiated dyads employing the same actor–partner labels described above (male–female, professor–pupil, superordinate– subordinate). Once again, the left side of Table 3 presents pre- and postinteraction EPA profiles for the three sets of status-differentiated dyads I employed in simulations. Unlike the simulations described above, which employed the sex-averaged EPA ratings, for the second set of simulations I allowed for sex differences in ratings by varying the dictionaries on which simulations were based. Because I model all interactions from the perspective of focal actor p (i.e., all simulation results reflect p’s perceptions of the situation), p’s sex determined which affective ratings were employed in generating profiles.8 In addition to affective profiles for actors and partners, Table 3 also presents predicted general expectations based on preand post-interaction EPA scores. In addition to employing a variety of initial identities, I also varied actors’ interpretations of the events. All initial agreements were modeled using the

215

Exploring Emotional Processes and Status Organizing Processes

Table 3.

Pre- and Post-Interaction Affective Profiles and Model-Derived Expectations. Pre-Interaction

Post-Interaction

High-Status Actors

High-Status Actors

Eval.

Pot.

Act.

P: Male (Male) 1.63 1.99 1.43 O: Female 1.60 0.22 0.42 P: Professor (Female) 1.82 1.87 0.52 O: Pupil 1.38 0.90 0.30 P: Superordinate (Male) 0.23 0.62 0.81 O: Subordinate 0.44 1.61 0.41 Avg. P 1.23 1.49 0.92 Avg. O 1.14 0.76 0.10 Diff. 0.09 2.25 1.02

Expect.

Eval.

Pot.

Act.

Expect.

7.5465 6.9463 7.4440 6.5215 6.8553 5.9926 7.2813 6.4868 0.7945

0.660 0.729 0.329 0.557 0.669 0.455 .5528 .2087 0.3441

0.203 0.746 0.217 1.165 0.496 1.100 0.305 1.004 1.309

0.417 0.023 0.383 0.913 0.366 0.050 0.133 0.028 0.161

6.414 6.187 6.598 6.442 6.528 6.169 6.513 6.266 0.247

Low-Status Actors P: Female (Female) O: Male P: Pupil (Male) O: Professor P: Subordinate (Male) O: Superordinate Avg. P Avg. O Diff.

2.57 0.99 0.71 7.3607 1.03 1.34 1.58 7.2806 1.19 0.2 0.46 6.8725 1.61 1.58 0.35 7.3087 0.44 1.61 0.41 5.9926 0.23 0.62 0.81 6.8553 1.11 0.14 0.25 6.7419 0.96 1.18 0.91 7.1482 0.15 1.32 0.66 .4063

Low-Status Actors 0.345 0.181 0.330 1.043 0.277 0.167 0.317 0.343 0.026

0.709 0.391 0.463 1.213 0.523 0.357 0.256 0.654 0.899

0.818 0.0627 1.340 0.507 0.841 0.579 0.106 0.354 0.460

6.778 6.478 6.217 5.982 6.771 6.276 6.555 6.245 0.310

affective profile for ‘‘agree with,’’ which is associated with impressions that are slightly positive and powerful, and essentially neutral on the activity dimension (E=1.66, P=1.27, A=0.87 for males, and E=1.75, P=1.16, A=0.35 for females). Conversely, there is no specific term in Interact to clearly capture the behavioral meaning of attempted influence for disagreement trials. To address this issue, I employed three different terms, each of which reflects some facet of an influence attempt. The terms ‘‘challenge’’ and ‘‘advise’’ are associated with impressions that are moderately good, powerful, and lively. The term ‘‘contradict’’ is relatively affectively neutral (E=0.39, P=0.54, A=0.85 for males, and E=0.38, P=0.36, A=1.07 for females). To create behavioral responses reflecting acceptance or rejection of influence, again there were no directly corresponding terms in Interact, so I

216

JOSEPH DIPPONG

employed two different terms. First, because SCT argues that p(s) measures openness to influence as an indicator of deference, I employed the term ‘‘defer to’’ to signify acceptance of influence. For both sexes, the affective profile associated with deference is generally negative (E=0.69, P=0.12, A=0.90 for males, and E=0.69, P=0.65, A=0.50 for females). As an alternate to deference, I also employed the more positive term ‘‘compromise with’’ (E=1.95, P=1.42, A=0.06 for males, and E=2.20, P=1.71, A=0.35 for females). For all ‘‘stay’’ responses, I employed the term ‘‘disagree with,’’ which is generally affectively neutral for males (E=0.06, P=0.75, A=0.63), and slightly more powerful and lively for females (E=0.14, P=1.15, A=1.60). Starting with the affective profiles presented in Table 3 and the behavioral terms described above, I estimated a series of simulated interactions based on the actual laboratory data from 10 experimental subjects, the structure of which was determined by each participant’s pattern of ‘‘stay’’ and ‘‘change’’ responses in the original study. Each simulated interaction was allowed to play out based on 50 individual events – 25 actions from actor p and 25 contrived actions from partner o – corresponding to the 25 trials in the original experiment. As in the original study, five trials consisted of initial agreement between partners (and consequently, stay responses for all experimental participants), and as such, all simulations I report have ten events in common: five partner agreements and five stay responses, occurring in the exact same sequence across all interactions. Aside from the five initial agreements, the sequence of simulated events was customized based on the actual pattern of stay and change responses from each participant, so that each of the interactions (based on the unique data from ten participants) represents a unique pattern of responses. In all, for each research participant, I estimated nine separate simulations combining each of the three identity pairings with three different configurations of behavioral definitions for disagreements, stay responses, and change responses. In total, the results I present are based on 90 simulated interactions (45 low status, 45 high status). Results Once again, Table 3 presents EPA profiles and predicted subjective expectations for three separate status-differentiated dyads, as well as the averaged results and average differences across the three groups. Looking first at the average pre-interaction EPA profiles for higher status self (p) and lower status other (o) on the upper left side of Table 3, we see that on the whole, the pattern for higher status actors is close to what we might expect

Exploring Emotional Processes and Status Organizing Processes

217

based on previous research, and in accordance with the general predictions of SCT: the self-profiles for higher status actors reflect more positive evaluations in comparison to their lower status partners (an average difference of +0.09), along with impressions of greater potency (an average difference of +2.25) and more activity (an average difference of +1.02). Predicted expectations also line up as expected, with higher status actors relating to higher expectations (though by a relatively small margin – an average difference of +0.795). Thus, on the surface it appears that the identities selected as the basis for simulated interactions capture the intended affective meanings associated with occupying relatively high-status positions. Concerning the pre-interaction affective profiles for lower status actors and their partners on the lower left side of Table 3, once again the anticipated pattern of results holds, but with one noteworthy exception. The affective profiles for one set of identities (female and male interactants based only on female ratings) produce model-derived expectations contrary to the basic predictions of SCT. Although females perceive male interaction partners as more powerful and more active, they perceive themselves much more positively in terms of evaluation (a difference of +1.54), and this female advantage along the evaluation dimension translates into a small initial advantage in predicted expectations. I discuss this unexpected finding in more detail below. Turning to post-interaction ratings, the upper right side of Table 3 presents the affective profiles for higher status actors provided by Interact following the simulated experimental task. For higher status actors, several differences between average pre- and post interaction ratings emerge. First, whereas average pre-interaction ratings revealed that identities for both self and partner were associated with positive evaluations, post-interaction ratings produce diminished average evaluations for both self and partner. A comparison of pre- and post-simulation evaluations reveals that the interaction reduced evaluations of both self and partner, but that self evaluations took somewhat of a larger hit, with evaluations for self being reduced by a magnitude of 1.78 and evaluations for partners reduced by 1.35. A similar pattern holds for both the potency and activity dimensions. Average potency ratings were reduced between pre- and post-interaction, with ratings for self being diminished more (an average change in potency of –1.19 for self and –0.24 for partner). The result is that although higher status actors maintain an advantage in terms of potency, the gap between higher and lower status actors is reduced by over 40 percent with respect to this

218

JOSEPH DIPPONG

affective dimension. Similar to changes along the evaluation dimension, the initial advantage for higher status actors along the activity dimension is reduced to a disadvantage following simulated interaction. The overall reduction of EPA for both self and other ratings (with a greater reduction relating to self-ratings) produces a narrowing of differences in modelpredicted general expectations. Specifically, higher status actors hold onto a narrow margin of advantage in subjective expectations, but compared to pre-interaction ratings, that margin is reduced by close to 70 percent. In other words, the initially small but clear status differentiation has been substantially blurred through the course of interaction. Lastly, looking at post-interaction ratings for lower status actors at the lower right side of Table 3, we see a different pattern of results compared to that observed with higher status actors. Once again, average evaluation and activity ratings for self and partner are substantially reduced. Along the potency dimension, however, an unexpected result arises: for lower status actors, interaction raises perceptions of power for self and reduces perceptions of power for partners (an average change of +0.40 for self and –1.83 for partner). The overall effect of changes in affective profiles leads to a case in which an initially minor status disadvantage becomes a slight advantage in predicted subjective expectations. I discuss changes in pre- and post-interaction EPA profiles and expectations more in-depth in the discussion section below. Along with information on affective profiles and likely behaviors and emotions, Interact also provides information on the degree of deflection (i.e., the discrepancy between transient sentiments and fundamental sentiments) caused by behaviors from other social actors. Table 4 presents the average predicted deflections for higher and lower status actors in response to

Table 4. Deflections. Status

Disagreeing Choice from Partner

Stay Response

Change Response

High

9.4466 (3.2172) 7.6277 (2.5873) 8.4106 (3.0841)

8.1425 (3.8404) 10.5456 (3.1822) 9.3171 (3.7004)

9.1209 (4.4314) 10.2991 (3.3210) 9.6405 (3.9323)

Low Total

Column superscripts denote significant differences between status categories; po.001; po.01.

Exploring Emotional Processes and Status Organizing Processes

219

disagreements from a partner, stay responses, and change responses. Two particular features of the deflections stand out. First, compared to lower status actors, higher status actors experience a significantly greater degree of deflection when faced with a disagreeing response from a partner (average deflections of 9.45 for higher status actors and 7.63 for lower status actors), suggesting interplay between status, identity, and emotion in the generation of behavior. And second, the act of rejecting influence affects the magnitude of deflection differently for higher and lower status actors. Whereas for higher status actors, rejecting influence (i.e., a ‘‘stay’’ response) reduces the deflection created by an initial disagreement (a reduction of 1.33), for lower status actors, deflection increases regardless of their behavioral response. Thus, for higher status actors, the choice of behaviors is clear: rejecting influence verifies their high-status identity. In contrast, for lower status actors motivated to maintain an accurate self-identity (as ACT assumes all actors are), the task is essentially insoluble.

DISCUSSION The preceding simulations point to several new questions and directions that researchers within the areas of legitimation, status characteristics, and expectation states might pursue. Because simulations do not constitute a direct test of either theory, any results are merely suggestive of issues that may warrant further empirical and theoretical attention. As Lovaglia and Willer (2002, p. 252) note, ‘‘Representative simulationsyprovide a picture that shows how a theory works.’’ In the remainder of this section, I consider how the simulation results presented above can potentially contribute to the theoretical growth of expectation states theory by incorporating theoretical processes and arguments from ACT.

Legitimation and Delegitimation Borrowing an argument from Walker and colleagues (1986), existing research on legitimation demonstrates that the validation of a status order from an outside authority makes a difference in how power is exercised in status-differentiated groups, but it does not necessarily explain why it makes a difference. Simulation results presented here suggest the interesting possibility that legitimation via external validation solidifies status orders by creating negative emotional responses in lower status actors. Emotional

220

JOSEPH DIPPONG

responses to legitimation modified the affective impressions of lower status actors primarily by lowering their evaluation ratings, while leaving the affective profile of higher status actors largely unaltered. This suggests that legitimation processes reinforce (but do not substantially alter) existing positive beliefs about higher status actors, while at the same time amplifying the effects of negative beliefs about lower status actors. Although the effects of legitimation and delegitimation on affective profiles and expectations appear slight in a quantitative sense, the qualitative difference in affective labels following both is noteworthy. Simulations predict a wide array of negative emotions from higher status actors in response to the delegitimation of a status order, and when these negative emotions modify initially positive labels, the net result is the activation of negative affective identity profiles. Likewise, simulations predict positive emotions from lower status actors in response to delegitimation, and the resulting changes in affective profiles produces identities more akin to equalstatus relationships than status-differentiated groups. Simulations of legitimation predict negative emotional reactions by lower status actors, raising questions about to the extent to which status orders legitimated through external validation can remain stable over time. In order to maintain positive evaluations, powerful actors are expected to act in a gentle manner toward good objects and direct powerful acts toward negative objects (Smith-Lovin, 1987b). To the extent that group members are motivated to avoid status loss (cf. Troyer, Younts, & Kalkhoff, 2001), they must take into account the evaluation that group members hold for each other in choosing whether or not to direct powerful behaviors at certain individuals. It may be, then, that legitimation functions by creating negative emotions in lower status actors, which reduces group members’ evaluations of them, and renders them suitable objects of powerful acts. That is, in a non-legitimated-status order, higher status actors who direct powerful acts toward lower status actors would be expected to lose social esteem (Smith-Lovin, 1987b). But in a legitimated-status order, higher status actors can gain esteem by directing powerful actions at lower status actors because emotional responses have rendered lower status actors deserving objects of such actions. We know, however, that the experience of negative emotions by a lower status actor reduces her or his propensity to conform to the requests of a higher status counterpart (Lovaglia & Houser, 1996). This creates something of a paradox of authority (Weber, 1978), whereby higher status actors gain the legitimate right to demand compliance from lower status actors, but in a manner that renders lower status actors less likely to conform to such

Exploring Emotional Processes and Status Organizing Processes

221

demands. Collins (1992, p. 68) notes a similar dynamic when he suggests that the use of force or dominance ‘‘tends to limit itself.’’ Although Collins (1992) does not refer here to emotional reactions to power, his statement remains applicable to the present discussion, in that dominance asserted through the generation of negative emotions may be inherently limited in its usefulness. Negative emotion, in essence, breeds resistance. Finally, Lawler, Thye, and Yoon (2009) point out that positive emotional experiences promote commitment to certain types of social relations. To the extent that certain processes of legitimation create negative emotions, individuals should be less likely to experience commitment within such a context. This presents yet another potential paradox: whereas legitimation provides higher status individuals the right to demand compliance from others, the process may rely on forces that increase the propensity of lower status actors to exit the relationship. More research is needed to determine if the emotional reactions predicted by simulations – as well as the behavioral implications discussed here – bear out in actual group interactions. What is important, however, is that simulations provide a means to examine legitimation in a way that can potentially expand our knowledge by leading us to questions that may not ever arise during the course of a laboratory experiment. The uncovering of potential emotional processes underlying legitimation might shed further light on how status differences contribute to behavioral patterns of dominance and deference in task groups.

Status Dynamics In considering the simulations of laboratory interactions, interesting patterns emerged in the differences between predicted expectations for higher and lower status actors prior to interaction. Whereas pre-interaction EPA profiles produce an average difference in predicted subjective expectations of just over +.79 in favor of higher status actors, the model predicts a disadvantage of approximately –.41 for lower status actors. Although it has yet to be determined how subjective measures of expectations (e.g., the Zeller and Warnecke index) map onto theoretically derived expectations (e.g., expectation advantage), it is worth noting that in status-heterogeneous groups, there may be a small amount of disagreement between group members pertaining at least to perceptions about the degree to which certain status characteristics convey advantages to higher status actors.9

222

JOSEPH DIPPONG

The observed difference in expectations based on EPA profiles may be in part due to the imperfect relationship between affective impressions and performance expectations, but it also points to the possibility that in some situations, the meanings that higher and lower status actors ascribe to their relative positions may produce status structures that are initially asymmetrical. In other words, actors may start out with different (though complementary) expectations and through interaction negotiate a mutual understanding of the role of certain status characteristics. It is beyond the scope of the present chapter to explain the cognitive processes that might promote an asymmetrical status structure, though there is certainly substantial evidence within the field of psychology to demonstrate that in explaining their own behaviors, individuals often weight self-advantaging information more heavily than they weight self-disadvantaging information (cf. Duval & Silvia, 2002). Psychological research on the negativity bias also suggests that lower status actors evaluate higher status counterparts more negatively than higher status actors evaluate lower status actors (Amabile & Glazebrook, 1981). Both of these processes provide a plausible basis for asymmetrical status relations. The possibility that some status structures may be at least initially asymmetrical is highlighted in the unexpected differences in pre-interaction expectations I derive for males and females. Because expectations reflect the combined influence of all three EPA dimensions (Dippong & Kalkhoff, 2012), it may be possible for lower status actors to compensate for large disadvantages in perceived power by forming affective impressions that advantage themselves on other dimensions. As such, another potential advantage to using simulated interactions to examine status organizing processes is that Interact allows researchers to clearly delineate the effects of multiple social dimensions on the expectations that group members form for each other. Aside from pointing to possible sources of status asymmertry, simulations also suggest the possibility that initial differences in expectations narrow (and in some cases reverse) over the course of interaction within the SES. Because expectations are typically measured at a single time point, the possibility of temporal variation in expectations between partners has remained largely unexamined.10 ACT provides a potential explanation for this narrowing of expectations. Recall that ACT posits that when faced with deflection, individuals tend to enact the behavior that is most likely to reduce or eliminate the deflection. However, in the SES, only a select range of behaviors is possible: accept influence, reject influence, or exit the interaction. Given that partner disagreements create substantial deflection

Exploring Emotional Processes and Status Organizing Processes

223

for both higher and lower status actors, the limited behavioral responses available may in essence ‘‘bottleneck’’ actors and partners into converging identity profiles. In fact, following simulated interaction, higher and lower status actors ascribe to their partners affective profiles that share many identities in common. It may be, then, that in creating nearly constant disagreement between actors, while lacking alternative behavioral options, the SES naturally produces at least some degree of status convergence. In modeling how individuals will act when faced with multiple disagreements from an interaction partner, an intriguing number of behavioral options predicted by Interact require direct physical copresence between group members. This is evident in model predicted behaviors that include ‘‘show something to,’’ ‘‘sit next to,’’ ‘‘glance at,’’ ‘‘turn to,’’ and ‘‘look at.’’ In the absence of overt physical cues between themselves and their partners, individuals in the laboratory must construct a definition of the situation from minimal information and with limited behavioral tools. The limited range of behaviors available in the lab constrains the extent to which actors can fully enact status differences, which may create pressures toward equal status relations. Again, this is an issue that calls for empirical data.

Future Directions As with any new methodological application, the procedures employed in the present chapter warrant further refinement. Although the simulations I present required the use of proxy labels and behaviors, one of the strengths of Interact is that the program allows researchers to import their own affective profiles into the existing dictionary. A logical step toward refining the capacity of simulations to generate useful questions and hypotheses is to collect EPA ratings for the specific behaviors of interest to expectation states researchers. Nonetheless, using the existing dictionaries available in Interact, expectation states researchers can gain some confidence in their simulation results by drawing on multiple simulations that represent a variety of alternative affective interpretations of the situation, as I have here.

CONCLUSION In this chapter I have discussed the potential joint application of ACT and expectation states theory to examine status organizing processes using

224

JOSEPH DIPPONG

simulated interactions. As a first step, I have demonstrated how expectation states researchers can employ ACT’s Interact simulation engine to explore questions pertaining to emotional processes in legitimation, as well as to reconstruct laboratory interaction to examine how affect, identity, and status work together. In this chapter I have focused on two particular branches of expectation states theory, though the approach is applicable to a variety of other branches. Additionally, because Interact allows simulations of groups of up to four members, it may be possible to use simulations to develop new theoretical insights in group structures beyond the dyadic interactions typically in the SES. The questions and issues raised above call attention to the potential usefulness of examining status organizing processes by employing ACT and expectation states theory together to simulate social interactions. To draw any substantive conclusion, results obtained through simulation require empirical validation. Because the purpose of the present chapter is primarily to outline a new methodological approach, I have given direct attention to only a few of the many questions that could feasibly be derived from my simulations. Any of these questions could potentially produce further theoretical knowledge through empirical validation. More importantly, the present chapter highlights how methods and theoretical insights from ACT can be applied to generate substantive questions within the expectation states theoretical framework. Aside from potentially expanding the explanatory capacity of the individual theories, the joint application of ACT and expectation states theory can also point to new connections that will open the door for a more complete and formal unification of the two.

NOTES 1. Although Friedkin and Johnsen (2003) offer a formal tripartite integration of affect control theory, status characteristics theory, and social influence network theory, the aim here is to work toward the possibility of an integrated affect control theory and expectation states theory directly, without the use of a third theory. 2. Interact can be accessed from http://www.indiana.edu/Bsocpsy/ACT/ interact.htm. 3. See Dippong and Kalkhoff (2012) for a thorough discussion of the theoretical linkages between affective impressions and perceptions of status and power. 4. Simulations of legitimation and delegitimation are based on Interact’s dictionary of sex-averaged responses. 5. Predicted expectations are based on the following equation provided by Dippong and Kalkhoff (2012): e=6.554+(0.178Evaluation)+(0.271Potency)+ (0.114Activity).

Exploring Emotional Processes and Status Organizing Processes

225

6. The specific task I employed was termed ‘‘contrast sensitivity.’’ See Moore (1968) for a thorough description of the contrast sensitivity task. 7. The p(s) values reported here reflect the condition means only for the 10 subjects included in the present analyses. Condition means for the larger study from which these subjects were selected are .60 for higher status actors and .54 for lower status actors. 8. Following the approach common to SCT (and expectation states theory more generally), I employ the symbols p and o to differentiate between focal actors and their interaction partners. 9. Previous studies employing empirical data have found that compared to higher status actors, lower status actors report a narrower difference in subjective expectations, possibly attributable to a self-serving bias (cf. Thye, 2000; Dippong, 2013). 10. I know of only one recent study in which researchers measure subjective expectations both before and after interaction; although Kalkhoff and colleagues (2013) explore status processes in a setting distinct from the SES, their results also suggest a narrowing of status differences over the course of interaction.

REFERENCES Amabile, T. M., & Glazebrook, A. H. (1981). A negativity bias in interpersonal evaluation. Journal of Experimental Social Psychology, 18, 1–22. Berger, J., Cohen, B. P., & Zelditch, M. (1972). Status characteristics and social interaction. American Sociological Review, 37(3), 241–255. Berger, J., Fisek, M. H., Norman, R. Z., & Zelditch, M. (1977). Status Characteristics and Social Interaction: An Expectation-States Approach. New York, NY: Elsevier. Berger, J., Fisek, M. H., Norman, R. Z., & Wagner, D. G. (1985). Formation of reward expectations in status situations. In J. Berger & M. Zelditch, Jr. (Eds.), Status, rewards, and influence (pp. 215–261). San Francisco, CA: Jossey-Bass. Berger, J., Norman, R. Z., Balkwell, J. W., & Smith, R. F. (1992). Status inconsistency in task situations: A test of four status processing principles. American Sociological Review, 57(6), 843–855. Berger, J., Ridgeway, C. L., Fisek, M. H., & Norman, R. Z. (1998). The legitimation and delegitimation of power and prestige orders. American Sociological Review, 63(3), 379–405. Berger, J., & Webster, M. (2006). Expectations, status, and behavior. In P. J. Burke (Ed.), Contemporary social psychological theories (pp. 268–300). Stanford, CA: Stanford University Press. Bianchi, A. J. (2004). Rejecting others’ influence: Negative sentiment and status in task groups. Sociological Perspectives, 47(4), 339–355. Bianchi, A. J., & Lancianese, D. A. (2007). Accentuate the positive: Positive sentiments and status in task groups. Social Psychology Quarterly, 70(1), 7–26. Cohen, B. P. (1989). Developing sociological knowledge: Theory and method (2nd ed.). Chicago, IL: Nelson Hall. Collins, R. (1992). Sociological insight: An introduction to non-obvious sociology. New York, NY: Oxford University Press.

226

JOSEPH DIPPONG

Dippong, J. C. (2013). Priming effects and gender-based performance expectations: Expanding the informational structure of the interaction. Ph.D. dissertation, Department of Sociology, Kent State University, Kent, OH. Dippong, J. C., & Kalkhoff, W. (2012). Predicting status evaluations from affective impressions. Presented at the 24th Annual Group Processes Conference, Denver, CO. Driskell, J. E., & Mullen, B. (1988). Expectations and actions. In M. Webster. & M. Foschi (Eds.), Status generalization: New theory and research (pp. 399–412). Stanford, CA: Stanford University Press. Driskell, J. E., & Webster, M. (1997). Status and sentiment in task groups. In J. Szmatka, J. Skvoretz & J. Berger (Eds.), Status, network, and structure: Theory development in group processes (pp. 179–200). Stanford, CA: Stanford University Press. Duval, T. S., & Silvia, P. J. (2002). Self-awareness, probability of improvement, and the selfserving bias. Journal of Personality and Social Psychology, 82(1), 49–61. Fararo, T. J., & Skvoretz, J. (1987). Unification research programs: Integrating two structural theories. American Journal of Sociology, 92(5), 1183–1209. Fisek, M. H., & Berger, J. (1998). Sentiment and task performance expectations. Advances in Group Processes, 15, 23–39. Fisek, M. H., Berger, J., & Norman, R. Z. (1991). Participation in heterogeneous and homogeneous groups: A theoretical integration. American Journal of Sociology, 97(1), 114–142. Foschi, M. (2000). Double standards for competence: Theory and research. Annual Review of Sociology, 26, 21–42. Friedkin, N. E., & Johnsen, E. C. (2003). Attitude change, affect control, and expectation states in the formation of influence networks. Advances in Group Processes, 20, 1–29. Heise, D. R. (1977). Social action as the control of affect. Behavioral Science, 22(3), 163–177. Heise, D. R. (1985). Affect control theory: Respecification, estimation, and tests of the formal model. Journal of Mathematical Sociology, 11, 191–222. Heise, D. R. (1987). Affect control theory: Concepts and model. Journal of Mathematical Sociology, 13(1-2), 1–33. Heise, D. R. (2007). Expressive order: Confirming sentiments on social actions. New York, NY: Springer. Kalkhoff, W. (2005). Collective validation in multi-actor task groups: The effects of status differentiation. Social Psychology Quarterly, 68(1), 57–74. Kalkhoff, W., Friedkin, N. E., & Johnsen, E. C. (2010). Status, networks, and opinions: A modular integration of two theories. Advances in Group Processes, 27, 1–38. Kalkhoff, W., & Thye, S. R. (2006). Expectation states theory and research: New observations from meta-analysis. Sociological Methods and Research, 35(2), 219–249. Kalkhoff, W., Walker, L., & Webster, M. (2013). Comparing models of second-order expectations. Presented at the Annual Meetings of Group Processes, Eastern Sociological Society, Boston, MA. Lawler, E. J., Thye, S. R., & Yoon, J. (2009). Social Commitments in a Depersonalized World. New York, NY: Russell Sage Foundation. Lovaglia, M. J., & Houser, J. A. (1996). Emotional reactions and status in groups. American Sociological Review, 61, 867–883. Lovaglia, M. J., & Willer, R. B. (2002). Theory, simulation, and research: The new synthesis. In J. Szmatka, M. J. Lovaglia & K. Wysienska (Eds.), The growth of social knowledge: Theory, simulation, and empirical research (pp. 247–262). Westport, CT: Praeger.

Exploring Emotional Processes and Status Organizing Processes

227

MacKinnon, N. J., & Langford, T. (1994). The meaning of occupational prestige scores: A Social psychological analysis and interpretation. The Sociological Quarterly, 35(2), 215–245. Markovsky, B. (2010). Modularizing small group theories in sociology. Small Group Research, 41, 664–687. Markovsky, B., Dilks, L., Koch, P., McDonough, S., Triplett, J., & Velasquez, L. (2008). Modularizing and integrating theories of justice. Advances in Group Processes, 25, 211–237. Mead, G. H. (1934). Mind, self and society. Chicago, IL: University of Chicago Press. Moore, J. C. (1968). Status and influence in small group interactions. Sociometery, 31(1), 47–63. Osgood, C. E., May, W. S., & Miron, M. S. (1975). Cross-culture universals of affective meaning. Urbana, IL: University of Illinois Press. Ridgeway, C. L., & Berger, J. (1986). Expectations, legitimation, and dominance behavior in task groups. American Sociological Review, 51(5), 603–617. Ridgeway, C. L., Johnson, C., & Diekema, D. (1994). External status, legitimacy, and compliance in male and female groups. Social Forces, 72(4), 1051–1077. Ridgeway, C. L., & Smith-Lovin, L. (1994). Structure, culture, and interaction: Comparing two generative theories. Advances in Group Processes, 11, 213–239. Rogalin, C. L., Soboroff, S. D., & Lovaglia, M. J. (2007). Power, status, and affect control. Sociological Focus, 40(2), 202–220. Smith-Lovin, L. (1987a). Affect control theory: An assessment. Journal of Mathematical Sociology, 13(1-2), 171–192. Smith-Lovin, L. (1987b). Impressions from events. Journal of Mathematical Sociology, 13(1-2), 35–70. Thye, S. R. (2000). A status value theory of power in exchange relations. American Sociological Review, 65(3), 407–432. Thye, S., & Kalkhoff, W. (2009). Seeing the forest through the trees: an updated meta-analysis of expectation states research. Current Research in Social Psychology, 15(1), 1–14. Troyer, L. C., Younts, W., & Kalkhoff, W. (2001). Clarifying the theory of second-order expectations: The correspondence between motives for interaction and actors’ orientation toward group interaction. Social Psychology Quarterly, 64, 128–145. Walker, H. A., Thomas, G. M., & Zelditch, M. (1986). Legitimation, endorsement, and stability. Social Forces, 64(3), 620–643. Weber, M. (1978). Economy and society. Los Angeles, CA: University of California Press. Zeller, R. A., & Warnecke, R. B. (1973). The utility of intervening constructs in experiments. Sociological Methods and Research, 2(1), 85–110.

228

JOSEPH DIPPONG

APPENDIX A: SCOPE CONDITIONS OF SCT Five Scope Conditions (Cohen, 1989): 1. Actors are participating in a valued task with clearly defined criteria for success or failure. 2. Associated with the valued task is a specific skill necessary for the successful completion of the task. 3. Actors are collectively oriented, such that they believe it is both legitimate and necessary to consider the input of other group members in forming a decision. 4. Actors are task-oriented, such that they are motivated toward the successful completion of the collective task. 5. Actors are socially differentiated according to at least two levels of a salient status characteristic.

APPENDIX B: THEORETICAL ASSUMPTIONS OF SCT Five Assumptions (Berger et al., 1977): 1. Salience: A status characteristic becomes activated (salient) if members of the group are differentiated according to at least two levels of the characteristic. 2. Burden of Proof: Any salient status characteristic will lead actors to form expectations for task competence unless the characteristic is explicitly disassociated from the collective task. 3. Sequencing: The status structure develops as actors enter or exit the interactional setting. When a new actor enters the interaction, her or his status-relevant information is incorporated into the structure of the situation. When an actor exits the interaction, her or his status-relevant information remains a part of the informational structure. 4. Formation of Aggregated Expectation States: Actors combine all relevant status information into positive and negative ‘‘subsets,’’ and these subsets are combined to form a total aggregated expectation state value. According to the attenuation principle, the introduction of multiple

Exploring Emotional Processes and Status Organizing Processes

229

consistent paths attenuates the relative contribution of each consistent path. 5. Basic Expectation Assumption: Each actors’ position within the observable power and prestige order that emerges within a group is a direct function of her or his expectation advantage/disadvantage relative to other group members.

AGE AND THE EXPERIENCE AND MANAGEMENT OF EMOTION Kathryn J. Lively ABSTRACT Purpose – To determine how the correlational structure of emotion differs for individuals age 60 and above, compared to those under age 60, and to discuss the profound implications these differences may have for the experience and management of emotion. Design/methodology/approach – Structural equation modeling and shortest path analysis of emotion items from the General Social Survey (GSS)’s (1996) emotions module. Findings – Some positive and negative emotion pairs are more distant for individuals over age 60, while others are in fact closer. This variability leads to differences in available shortest paths between emotions, especially when emotional transitions require segueing through intermediary feelings. The segueing emotions most readily available to those over 60 are limited to the poles of affective meaning, whereas those used by ones under age 60 are more variable. The majority of negative emotions are more tightly correlated, whereas the majority of positive emotions are less so, among those over age 60. Research limitations/implications – Although the measures are limited to 18 of the 19 emotions recorded by the GSS, and are based on self-report Advances in Group Processes: Thirtieth Anniversary Edition Advances in Group Processes, Volume 30, 231–265 Copyright r 2013 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 0882-6145/doi:10.1108/S0882-6145(2013)0000030012

231

232

KATHRYN J. LIVELY

data regarding feelings felt over a period of seven days, these results suggest that attempts at intrapersonal and interpersonal emotion management may differ depending up the age of the actor/object. Originality/value – Addresses the need for more nuanced analyses of emotional experience that goes moves beyond simple frequencies. Also suggests potential bridges between sociological and psychological approaches to the study of emotion. Keywords: Age; emotion; emotional complexity; emotion management; emotional segueing

Sociologists and psychologists agree that older individuals often have different emotional experiences than their younger counterparts. Sociological arguments regarding the effects of age on emotion often stem from structural and cultural considerations. Structural arguments suggest that age acts as a buffer for negative emotions, largely due to changes in social networks, living conditions, economic resources, and the quality of one’s social networks, particularly one’s intimate relationships (Mirowsky & Ross, 2003). Cultural arguments, particularly those based on identity, suggest that individuals may select out of and into different roles and identities as they move through the life course and that these roles carry with them culturally shared emotional and behavioral expectations and opportunities (Lively & Heise, 2004). These identity-based arguments, for example, suggest that there are certain emotions deemed more normatively appropriate for grandmothers and grandfathers than for mothers and fathers. Psychological arguments regarding age and emotion tend to look to changes in cognitive processes and in the brain itself. The former suggests that older individuals (e.g., age 60 and above) develop short-term schemas that allow them to filter out negative stimuli and to more quickly, easily, and efficiently attune to, process, and recall positive stimuli (Carstensen, 2006). The latter suggests that structural declines in emotion-sensitive brain areas may selectively impair the processing of negative stimuli, which protects older individuals against threats to well-being. Similarly, this approach posits that structural degradation and functional slowing of the autonomic system may also diminish physiological arousal after exposure to a negative event. Regardless of the cause – structure, culture, cognition, or neurological degradation – it is clear that older individuals have different experiences of everyday emotion.

Age and the Experience and Management of Emotion

233

Although understanding how emotional frequency is distributed throughout society is important, recent studies suggest that the frequency of emotional experience may only be part of the story (Carstensen, 2006; Lively, 2008; Lively & Heise, 2004; also see Ross & Mirowsky, 2008). Sociologists have recently examined the underlying structure of emotional experience – that is, how multiple emotions covary simultaneously. Looking to correlations, as opposed to simply examining the frequency of distinct feelings (such as anger, or sadness), may provide a more nuanced understanding of emotional experience. Psychologists, too, have investigated correlations, though most often the relationship between positive and negative affect. Moving beyond the relationship between distinct, yet largely undifferentiated emotional states, Lively and Heise (2004) argue that the correlational structure of specific emotions indicates the process of emotion management (Hochschild, 1979; also see Gross, 1998). For example, the correlation between tranquility and distress represents the relative nearness or accessibility from one emotion to another. ‘‘Emotional segues’’ (or ‘‘shortest paths’’) may reveal how actors traverse multiple emotions to achieve specific emotional outcomes. For example, it may be that it is easier to move between tranquility and distress if one first moves through feelings of anger and fear (Lively & Heise, 2004). In a separate analysis, Lively (2008) showed that correlational structure of nine distinct emotional states, as well as the shortest paths between them, differed for women and men, suggesting that men and women may have access to different pathways for effective emotion management. Given that older individuals’ emotional experiences differ from those of their younger counterparts, I address the following questions: Does the correlational structure of specific emotional experiences differ for individuals under the age of 60 and for those 60 and older? And if so, do these differences suggest variation in the ways in which these groups manage their emotions? I answer these questions by examining (1) the correlational structure of emotional experience – that is the ways in which distinct emotional states covary – by age group and (2) the shortest paths between both similar and dissimilar emotions.

BACKGROUND Age Differences in the Experience of Emotion Emotional experience, to a certain extent, is a function of age. Sociological and psychological inquiries consistently find that older individuals, with few

234

KATHRYN J. LIVELY

exceptions, experience more positive and less negative feelings (Carstensen et al., 2010; Mirowsky & Ross, 2003). Whereas sociological studies tend to examine changes that occur outside of the individual, such as changing access to resources and participation in social roles and networks, psychological theories tend to focus on changes that occur within, including shifts in cognitive processing and physical and neurological degeneration. I review each of these literatures below before turning to recent developments in both the psychological and sociological literature: that is, the study of emotional complexity (Carstensen, 2006) and emotional segueing (Lively, 2008; Lively & Heise, 2004).

Psychological Approaches to Explaining Age Differences in Emotion Socioemotional Selectivity Theory Psychological theories of emotion and age tend to rely on intra-individual shifts due to cognitive, physiological, or neurological changes. Theories of emotional–motivational life span development, for example, propose normative shifts in individuals’ emotional goals and strategies across adulthood. Socioemotional selectivity theory (Carstensen, 2006) holds that anticipated endings (e.g., impending death) give primacy to enhancing emotionally gratifying experiences in the moment as opposed to maximizing future rewards. This shift activates mood-enhancement goals and reduces the willingness to accept purely negative experiences for the sake of long-term goals (Scheibe & Carstensen, 2010). Because aging is naturally associated with endings, they argue that the process itself predicts motivational changes – and, subsequently, affect – over time.1 Compensatory Strategies of Emotion Associated with Aging Whereas socioemotional selectivity theory emphasizes selective processes leading to the experience of more days associated with positive emotions, other psychological theories conceptualize age-related emotional– motivational changes as a compensatory means to adapt to declining resources. Dynamic integration theory (Labouvie-Vief, 2003) posits that diminishing cognitive capacities among the elderly make it more difficult to integrate and accept negative feelings, which leads to more positive affect overall.2 Similarly, the life span theory of control (Heckhausen & Schultz, 1995) holds that older individuals’ capacity to control their environment and achieve their developmental goals declines. Consequently, older adults may increasingly use secondary control strategies, such as emotional regulation,

Age and the Experience and Management of Emotion

235

rather than using primary control strategies aimed at changing the predicating situation.3 Learning and Practice Adding to selective and compensatory changes in emotional preferences and strategies, some psychologists point to the role that learning and practice have on older adults’ greater competence at emotional regulation (Blanchard-Fields, 2007; Scheibe & Blanchard-Fields, 2009). Some argue that long-term experience and practice in dealing with emotional situations lead older adults to acquire situational, strategic, and procedural knowledge about emotional processes, which increase their effectiveness in handling emotional situations (Scheibe & Carstensen, 2010). Thus, older individuals may be more knowledgeable about the emotional effects of future events (Scheibe, Mata, & Carstensen, 2009), more adept at tailoring their emotionregulatory strategy to contextual demands (Blanchard-Fields, 2007), and more efficient in terms of their emotion-regulatory processes (Scheibe & Blanchard-Fields, 2009). A recent study by Kafetsios (2004) reveals that older adults score higher than their younger counterparts on the ability to facilitate, understand, and manage emotions. Neurological Decline as Emotional Advantage Some researchers have suggested that improved affect in later adulthood is actually an unforeseen by-product of biological decline (Cacioppo, Berntson, Bechara, Tranel, & Hawkley, 2011). As noted, structural decline in emotion sensitive brain areas may selectively impair the processing of negative stimuli, just as structural degradation and functional slowing of the autonomic system may diminish physiological arousal after the exposure to emotional stimuli, thereby reducing the impact of negative events (Cacioppo, Berntson, Klein, & Poehlmann, 1998). However, once an autonomic reaction is elicited, the same mechanism can actually prolong physiological reactions, thereby increasing the duration of negative emotional states (Charles & Piazza, 2009). Although older individuals may, as a general rule, be protected from negative emotional experiences, those unfortunate enough to experience distress may actually suffer longer than their younger counterparts. Sociological Approaches to Explaining Age Differences in Emotion Age as Maturity Unlike psychological theories of emotion and age, which tend to rely on intraindividual shifts due to cognitive, physiological, or neurological changes,

236

KATHRYN J. LIVELY

sociological theories tend to focus more on social structural conditions, interactional encounters, or social roles. The age as maturity perspective holds that older individuals are more experienced and accomplished in dealing with life’s difficulties. Simply put, age increases practice with living. In keeping with this tradition, scholars have identified a number of social and individual changes that come with age, including lower crime rates (Hirshi & Gottfredson, 1983; Ross & Mirowsky, 1987), safer habits, a more orderly lifestyle (Umberson, 1987), and greater satisfaction (Loscocco and Kalleberg (1988). Studies also reveal that older individuals tend to hold a more positive self-image and greater sense of purpose that lasts, for the most part, until the mid seventies (Campbell et al., 1976; Gove, Ortega, & Style, 1989; also see Mirowsky & Ross, 2003). This perspective also suggests that as individuals age, they garner resources that may assist them with a shift from the more negative emotions associated with adolescence and young adulthood to more positive emotions. Thus, younger individuals, who are still developing emotionally and still in the process of seeking out valued social roles such as employment, marriage, and parental status, may suffer more emotional upset than their older counterparts. Often associated with greater access to resources (e.g., improved social and economic stability), increased age is likely to be accompanied by greater maturity, which is often accompanied by a more even, if not positive, temperament (Gove et al., 1989; Mroczek & Kolarz, 1998; Schieman, 1999; also see Carstensen, 2006, Carstensen, Isaacowitz, & Charles, 1999; Gross et al., 1997; Pinquart, 2001). Further, older people, by virtue of having lived longer lives, have more experience in dealing with problems and frustrations than their younger counterparts. Their lives also tend to be more orderly, characterized by smaller networks, more meaningful relationships, and greater emotional capital – all of which assists them in dealing with troublesome events. It is important to note that the buffering effects of age tend to taper off in the late 60s and early 70s. Indeed, numerous of epidemiological studies of emotion typically reveal a curvilinear age effect, particularly as it relates to distress (Ross & Mirowsky, 1987). Age, Emotion, and Identity According to affect control theory (Heise, 2007), emotions arise out of our ability to confirm our fundamental sentiments regarding events. Affect control theory is based on the notion that each element in a social interaction – actors, behaviors, objects (of that behavior), settings, attributes, and emotions – can be delineated into three dimensions of affective meaning

Age and the Experience and Management of Emotion

237

(Evaluation, Potency, and Activation (EPA)). Generally speaking, Evaluation refers to how good or bad something is, Potency refers to how potent or impotent something is, and Activation refers to how active or passive something is. In addition to underscoring the importance of affective meaning to social life, affect control theory assumes that individuals desire affective consistency (also see Stryker, 1980). Thus, individuals who occupy certain social roles, or identities, are also expected to engage in behaviors and experience emotions that are consistent with those roles. Individuals who exist in certain settings are expected to experience emotions that are consistent with those settings. When individuals move into roles characterized by less power and less activity (which is often the case of roles held by the elderly), their emotions are expected to change with them. If individuals move into smaller settings, retirement communities, or nursing homes, their emotions are expected to change in ways that are consistent with these types of environments as well. According to this perspective, age is often associated with more positive, less powerful, and less active affect. This has less to do with our cultural understanding of what it means to be over 60 than with our affective understanding of the social identities likely to be hold by individuals over the age of 60 (grandmother as opposed to mother, or retiree versus entrepreneur), as well as the behaviors they are likely to engage in (travel versus assault), the interaction partners with whom they are likely to be engaged (grandchild versus employee), the settings in which they are likely to be embedded (home or community center versus workplace or amusement park), and the attributes they are likely to hold (kind or sweet, versus arrogant or aggressive).

New, Yet Complementary, Directions in Measuring Emotions: Emotional Complexity and Emotional Segueing Although sociologists and psychologists studying emotion and aging rarely dialogue across disciplinary lines, both have turned their attention to the structure of emotional experience – that is, the correlations, or the co-occurrences, of emotion. Psychologists, focusing more on intrapersonal aspects of emotion, have tied the correlation of positive and negative affect – that is, emotional complexity (Carstensen, 2006; Ong & Bergeman, 2004) – directly to individual outcomes, including well being and physical health (also see Epel, McEwen, & Ickovies, 1998; Lane & Pollermann, 2002;

238

KATHRYN J. LIVELY

Pennebaker, Mayne, & Francis, 1997). In contrast, sociologists, focusing more on social processes, have examined correlations and shortest paths between discrete emotion states – that is, emotional segues (Lively, 2008; Lively & Heise, 2004) – for the purpose of elucidating strategies of intrapersonal, interpersonal, and organizational emotion management (also see Britt & Heise, 2000; Francis, 1997; Hochschild, 1979; Thoits, 1995). Emotional Complexity Relatively new on the topic of aging and emotion among psychologists is the discussion of emotional complexity (Ong & Bergeman, 2004; also see Carstensen, 2006). Simply put, emotional complexity is indicated by relatively greater independence in affective indicators (but see Ready, Akerstedt, & Mroczek, [2011]). In other words, emotional complexity is indicated by little or no correlation between positive and negative emotions. As a general rule, older individuals tend to report higher levels of emotional complexity than their younger counterparts (Carstensen, Pasupathi, Mayr & Nesslroade, 2000; Ong & Bergeman, 2004; Ready, Carvalho & Weinberger, 2008). Indeed, greater independence among positive and negative affect (i.e., greater emotional complexity) is found more for older persons than younger ones using several different methodologies (Carstensen et al., 2000; Ready et al., 2008; Ready, Robinson, & Weinberger, 2006).4 Notably, the capacity to preserve and maintain the boundaries between positive and negative affect is associated with less sustained negative affect (Pennebaker et al., 1997; also see Epel, McEwen, & Ickovies, 1998). Although these studies reveal considerable evidence that older populations enjoy greater independence among positive and negative affect, there has been considerably less attention paid to the relationship, or lack thereof, between specific emotions, such as, for example, distress and tranquility or joy and rage. Emotional Segueing In an attempt to determine how individuals transition through intervening emotions in order to achieve specific emotional outcomes, Lively and Heise (2004) used structural equation modeling to introduce the idea of emotional segueing (also see Lively, 2008). Using data from the 1996 General Social Survey’s (GSS) emotion module, they examined the correlational structure of nine latent emotion factors: tranquility, hope, joy, pride, self-reproach, anger, rage, fear, and distress, arguing that emotions that co-occur frequently within a seven-day period may be relatively accessible to one

Age and the Experience and Management of Emotion

239

another, while those that co-occur infrequently (or not at all) may be less accessible. To further explore this idea, they applied shortest path analyses (Rosen, 1988) to the correlations among these factors, and constructed a remoteness index of emotional states, yielding shortest paths between any two emotions. The remoteness scores, they argued, provide quantifiable representations of how much emotion management it takes to move from one emotional state to another. These shortest paths are consistent with emotional transformations documented previously in studies of social support groups (Francis, 1997; Thoits, 1995), social movement organizations (Britt & Heise, 2000), political speeches (Wasielewski, 1985), and total institutions (Goffman, 1961). In a separate analysis, Lively (2008) showed that the correlational structure of emotion, and thus, the corresponding emotional segues, differed for men and women. Her analysis demonstrated that women’s shortest paths tended to be longer than men’s, more complicated (meaning that they required transitioning through more emotions in order to achieve the same goal), and relied on emotions that have typically been cast as ‘‘feminine’’ (Lively, 2008; also see Lively & Heise, 2004). Lively (2008) suggested that these sex differences in emotional segueing may shed insight into women’s higher self-reported rates of depression; because the shortest paths for women to transition out of a negative, impotent, and inactive state was both more complicated and less efficacious than it was for men. Despite these rather stark sex differences in emotional segueing, sociologists have paid less attention to variation among other groups. However, given the ample evidence of age differences in emotional experience more generally and in emotional complexity (which, again, refers to the relationship, or lack thereof, of positive and negative affect), several questions remain: How does age affect the overall structure of emotion, including the relationship between positive and negative affect, and also between particular pairs of emotions, simultaneously? And to the degree that the co-occurrence of particular emotions may have implications for emotional accessibility, what implications might these differences have for emotion management, either the intrapersonal methods used by an individual or the interpersonal methods used by a loved one, friend, adult child, or caretaker?

Expectations Given the consistent findings regarding age and the frequency of emotional experiences from both sociologists and psychologists, I expect there are

240

KATHRYN J. LIVELY

significant age differences in the correlational structure of emotions and these differences should lead to different shortest paths between emotional states. If such differences do exist, based on the above, one might expect emotional pathways to vary in terms of (1) length, (2) complexity, and (3) the emotional segues used in the paths themselves. Expectation One: The shortest path between emotions will be longer for older individuals than their younger counterparts.5 Expectation Two: The shortest paths between emotions are more complex, or more complicated, for older individuals than for their younger counterparts.6 Expectation Three: The shortest paths for older adults that require segueing emotions will rely on emotions that are more positive, but less powerful and less active than those more common to younger adults.

DATA AND METHODS Data The 1996 GSS (Davis & Smith, 1996) employed a probability sample of 2,904 English-speaking adults residing in the United States. A random subset of 1,460 respondents was asked questions from the ‘‘emotions module’’ – 90 questions regarding emotion, including experiences in the last week with 19 different emotion states, details of anger episodes, and methods of emotion management. Respondents receiving the emotion module were given the following instruction for assessing recent emotional experiences: ‘‘Now I’m going to read a list of different feelings that people sometimes have. After each one, I would like you to tell me on how many days you have felt this way during the past 7 days.’’ Although ‘‘restless’’ was included as one of the 19 emotion states in the GSS questionnaire, it is not included in these analyses because it does not fit Ortony, Clore, and Foss’ (1987) definition of a pure emotional state. Although the GSS (1996) emotions module remains the only national representative dataset containing information on a wide array of emotions, there are particular limitations to the data that must be addressed. First, there are only 18 emotions in these analyses and these 18 fail to tap into any emotion that could be classified as unpleasant-dominant-quiescent such as bitterness or disgust (Morgan & Heise, 1988; MacKinnon & Keating, 1989). Second, the GSS (1996) data deal mainly with salient emotions and, thus,

Age and the Experience and Management of Emotion

241

may underrepresent those less intense flashes of emotion that make up the tapestry of one’s day-to-day emotional experiences. Third, the data are cross-sectional and, thus, it is impossible to determine causality or even directionality (see Heise, 1979). Therefore, I do not contend that emotions cause one another. Rather, I suggest that one emotion can facilitate the experience of some emotions and hinder the experience of others. I also do not posit that these paths are more likely to lead to positive or to negative emotions. Instead, I assess whether these paths exist between positive and negative emotions, which facilitates movement between the emotions in either direction.

Methods Because I expand Lively and Heise’s (2004) earlier analyses of emotional segueing within the general population, I use the nine emotion factors that they derived jointly from a classification of emotions proposed by psychologists Ortony, Clore, and Collins (1988) and insights from affect control theory (Heise, 1979; Smith-Lovin & Heise, 1988). Ortony et al. (1988) distinguish emotions that are reactions to events, agents, or objects, and propose that each kind of emotional reaction might be characterized by a number of different emotion tokens. For example, they propose that being pleased about a desirable event produces an affective condition of joy that might be recognized as feeling contented, happy, cheerful, delighted, ecstatic, etc. What distinguishes affect control theory’s perspective on emotion from the Ortony–Clore–Collins classification is the delineation of emotion into the three fundamental dimensions of affective meaning noted above: EPA. Affect control theory assumes that emotions, identities, attributes, and settings that have similar EPA ratings are more similar to one another than they are to those that have dissimilar EPA profiles. By looking at the EPA ratings of each of the individual emotion states within each of the latent factors and on the basis of their disparate activation values, Lively and Heise improved the overall fit of their confirmatory factor model by separating [over]joy from other emotions that they called tranquility and [out]rage from other emotions that they labeled anger. Upon improving the fit of the measurement model to acceptable levels, they concluded that the Ortony, Clore, and Collin’s classification seemed to obscure differences in emotionality, by treating activation and intensity synonymously.7

242

KATHRYN J. LIVELY

The Confirmatory Factor Model In the structural equation analysis presented here, each affective condition is a latent variable, the specific emotions are indicators, and I empirically estimate the coefficients connecting latent variables and indicators. Given the emotion indicators that are available in GSS 1996, this analysis is limited to Ortony et al. (1988) event-based emotions (specifically the well-being emotions of joy and distress and the prospect-based emotions of hope and fear) and agent-based emotions (specifically the attribution emotions of pride and self-reproach and the compound emotion of anger).8 Ortony et al. (1988) did not mention the tokens of ‘‘calm’’ and ‘‘at ease’’ that were included in the GSS 1996 and identified as emotions by Ortony et al. (1987). Though these two tokens are subjectively similar to ‘‘contented,’’ they do not really seem to fit in the category of joy (MacKinnon & Keating, 1989; Morgan & Heise, 1988). Consequently Lively and Heise (2004) established a supplementary category of tranquility to include them (also see Lively, 2008). I distinguish between two age groups, the younger group (which is comprised of individuals less than 60 years of age) and the older group (which includes individuals ages 60 and over). Although this division may seem arbitrary, the literature on aging typically views the period from 55 to 65 as an important point in individuals’ lives. At 55, for example, adults are beginning to shift from a more achievement-oriented outlook and to think about retirement and how they really want to spend the rest of their lives (Carstensen, 2006; Mirowsky & Ross, 2003). The late fifties to mid sixties also signal a time where biological changes occur (changes in estrogen and testosterone levels, for example) that may have important implications for emotional consequences (Dolomie-Faguar, Gatta, Nguyen, & Corcuff, 2008). While it is true that many neuroscientists consider age 65 the point at which important degenerative changes are likely to occur (or may even limit their investigations to differences among individuals 65 to 80 and 80 and above), it is important to note that age and its effects are continuous. In other words, things do not simply change on one’s 65th birthday. Instead, social, biological, and neurological changes occur through out the life course, although, admittedly, they may accelerate with advanced age (Mirowsky & Ross, 2003). Further, although there may be increased neurological deterioration or change once the individual passes age 65, there are also changes, such as role occupancy and living conditions, as well as other individual level

243

Age and the Experience and Management of Emotion

characteristics that may undergo shifts at earlier ages, say the middle to late fifties. Auxiliary analyses reveal some slight differences in the emotional experiences of individuals who are 55 and above and those who are 65 and above. If these changes continue in a linear fashion, the results from the current analyses may actually underestimate the effects of age on emotional experience.

RESULTS Table 1 presents the simple frequencies for the nine factors in the measurement model, by age group. As evidenced here, the data are highly skewed toward the positive in the sense that individuals, regardless of age, are much more likely to report experiencing positive than negative emotions. With the exception of tranquility, an emotion marked by high levels of Table 1.

Descriptive Statistics, Mean, and (Standard Deviation) by Age Group.

Emotion Factor

Tranquility Hope [excitement] Joy [overjoyed] Pride Self-Reproach Anger Rage [outrage] Fear Distress N

Younger (Less than 60)

Older (60 and older)

p-Value

4.59 (1.79) 3.72 (2.23) 1.85 (2.14) 3.03 (2.26) .55 (1.01) 1.75 (1.78) 1.63 (1.91) 2.20 (1.80) 1.46 (1.57) 1087

5.34 (1.68) 3.21 (2.51) 1.55 (2.09) 2.74 (2.58) 0.42 (0.98) .88 (1.36) 1.06 (1.66) 1.54 (1.59) 1.27 (1.64) 267

  

ns ns   

ns

Note: The data upon which these analyses are based were weighted by a factor of 2, essentially doubling the original sample size. Significance levels: po.05; po.01; po.001.

244

KATHRYN J. LIVELY

positive evaluation and lower leve1s of potency and activation, the older group reports fewer emotions (both positive and negative) than does the younger group. With few exceptions, the older group is significantly less likely to report days in which they experienced hope, joy, anger, rage, and fear than their younger counterparts. Although the older group is also likely to report fewer days of pride, self-reproach, and distress, these differences are not statistically significant.9 It is important to remember that these coefficients represent the main effects of age, so what is evidenced here is the overall conditions – including the social, biological, and neurological conditions – of these groups’ lives.10 Although these results are somewhat consistent with the sociological literature (which tends to focus on the negative emotions of anger, depression, etc.) and psychological literature (which tends to focus on positive and negative affect), they also stress the importance of examining an array of distinct emotional outcomes in order to truly understand the effect of age on emotional experience. Not only does age seem to facilitate the experience of tranquility, it also seems to subdue the experience of other pleasant emotions, such as hope and joy, in addition to feelings of anger, rage, and fear. Table 2 shows the specifications and results of Lively and Heise’s (2004) final measurement model, on which these analyses are based, for both the younger (under 60) and older (60 and over) age groups.11 For example, calm was used as the scale-defining item for the tranquility latent variable by setting its factor coefficient to 1.0, and then the data analysis estimated the factor coefficients for contented, at ease, and happy. Latent factors were allowed to co-vary freely. Error terms for the individual indicators were assumed to be uncorrelated. Results for the younger respondents appear in Model 1, whereas the results for older respondents are found in Model 2 (all coefficients are significant at the 0.00 level).12 Overall the model fits both age groups reasonably well, as shown by most of the goodness-of-fit indicators at the bottom of the table. The ratio of chisquare to degrees of freedom, which is often used in place of the chi-square value as a measure of fit when dealing with large samples, is 2.40 for the younger group and 2.17 for the older group. These ratios are well within the value of 3.00, which is frequently accepted as the maximum value indicating an acceptable fit (Kline, 1998, p. 128). Additionally, the incremental and the comparative fit indices for the younger group reaches 0.91 and .90, respectively, the standard levels of acceptance, although both of these markers fall slightly short for that of the older group (0.85 and 0.84). The goodness of fit index for each group is also well within acceptable limits.

245

Age and the Experience and Management of Emotion

Table 2.

Unstandardized Regression Weight for Model, by Age Group.

Emotion Item

Calm At Ease Contented Happy Overjoyed Excited Proud Ashamed Embarrassed Fearful Worried Anxious-Tense Sad Blues Lonely Angry Mad Outraged Fit Statistics x2/df IFI CFI GFI RMSEA N

Factor

Tranquility

Joy Hope Pride Self Reproach Fear

Distress

Anger Rage

Factor Coefficient (Younger Group)

Factor Coefficient (Older Group)

EPA, Combined Male–Female Data (Indiana 2002–2004)

1.00 1.18 1.16 1.05 1.00 1.00 1.00 1.00 1.21 1.00 1.56 1.43 1.00 1.07 0.92 1.00 1.09 1.00

1.00 1.69 1.78 1.47 1.00 1.00 1.00 1.00 1.33 1.00 2.38 1.03 1.00 1.11 1.13 1.00 0.94 1.00

2.33, 1.56, 0.13

2.40 .91 .90 0.96 0.04 1,087

2.17 .85 .84 0.94 0.07 267

2.29, 1.64, 1.63 2.56, 2.26, 2.54 2.27, 2.22, 1.70 1.73, 1.35, 1.26 1.32, 0.83, 0.14

2.09, 1.67, 2.11

1.50, 0.26, 0.99 1.65, 0.38, 1.81

Note: All factor coefficients have p-valueso.001; N=1,356 after list-wise deletion of missing data cases.

Finally, the root mean square error of approximation is near 0.05 for both age groups suggesting that the model’s ability to reproduce indicator covariance for both groups is acceptable.13 Also in Table 2 are the basic EPA profiles for each of the latent factors in the model. These profiles were derived by taking the mean of the EPA profiles for each of the emotion indicators in the model – for example, I averaged the EPA profiles for shame and embarrassment to create the mean EPA for the latent factor self-reproach.14 It is important to note that the EPA values presented here do not vary by age group and have been drawn from a sample of college students from a

246

KATHRYN J. LIVELY

large Midwestern University. Although Lively (2008) reported some slight sex differences in EPA values that corresponded to gendered pathways of emotion management, most studies suggest affective sentiments, including those for emotions, are more similar than dissimilar (Heise, 2007). Most studies of this genre assume that the affective sentiments regarding actors, objects, behaviors, settings, attributes, and emotion are culturally shared, unless the elements of events are specific to a particular subculture (Hunt, 2008; Smith-Lovin & Douglass, 1992). Although the measurement model does fit that data for both groups reasonably well, the overall fit is slightly better for the adults who are younger than 60 than for their older counterparts. The most obvious differentials lie in the factor loadings of at ease, contented, happy, and overjoyed on tranquility and the loadings for worried and anxious-tense on fear, for the older group, compared to the younger. These differences suggest that the factors themselves may be capturing different feeling states for the two groups. Specifically, the tranquility factor may be more positive, more powerful, and more active for the older group (given the relative EPA values of happy and contented compared to that of calm). Further fear may be capturing a feeling state that is, while still negative, also slightly less powerful and less active for the older group, compared to that experienced by the younger group. To the degree that segueing occurs along EPA dimensions (Lively, 2008; Lively & Heise, 2004), these relatively subtle differences may result in different shortest paths for the two groups. Table 3 shows the correlation matrix for the nine latent emotions: tranquility, hope, joy, pride, self-reproach, anger, rage, fear, and distress, with the older group represented in the upper triangle and the younger group in the lower.15 Again, emotions that are correlated positively are expected to be more readily accessible to one another than emotions that are negatively correlated. As might be expected, given their greater similarity, correlations within unpleasant emotions and correlations within pleasant emotions generally are higher than the correlations between pleasant and unpleasant emotions, for both groups. However, correlations between pleasant and unpleasant emotions are not simply negative for either group. Specifically, the older group has seven positive–negative emotion correlates that are positively related with correlations significantly different from zero at the 0.05 level. Pride, for example, correlates positively with selfreproach, rage, fear, and distress, whereas joy correlates positively with selfreproach, rage, and fear, and hope correlates positively with rage. In comparison, the younger group has six such positive–negative emotion correlates. In particular, joy is positively correlated with self-reproach, rage,

247

Age and the Experience and Management of Emotion

Table 3. Correlations among Emotion Factors for 60 and Older (Upper Triangle; N=267) and Younger than 60 (Lower Triangle; N=1,087). Tranquility Tranquility Hope Overjoyed Pride Self-reproach Anger Rage Fear Distress

– .43,b .28,b .29,b .16,b .32,b .33 .61 .67

Hope .18,a – .41 .41,b .11 .02b .03b .04 .14,b

Overjoyed a

.01 .46 – .39 .14,b .05 .08 .11 .03

Pride a

.11 .29,a .31 – .07b .11 .08 .10,b .03b

Self-Reproach Anger .53,a .09 .32,a .31,a – .30,b .18,b .48,b .36,b

Rage

Fear

Distress

.70,a .30 .61 .70 .17,a .12,a .06 .00a .01 .15 .17 .03 .09 .13 .27,a .22,a .48,a .55,a .63,a .78,a – .69,a .74,a .61,a .57,b – .62,a .40,a .65,b .53,b – .75,a .46,b .49,b .82,b –

 Correlations, generated by AMOS 16.0, are significantly different from zero at the .05 level. a

Relationships differ significantly from those in the middle-age group (36–55 years) at the .05 level in a two-tailed test based on Student’s t after Fisher’s r to z transformation (Winkler & Hays, 1975, p. 653). b Relationships differ significantly from those in the older group (56 and over) at the .05 level in a two-tailed test based on Student’s t after Fisher’s r to z transformation (Winkler & Hays, 1975, p. 653).

and fear (as it was in the older group), whereas hope is positively correlated with self-reproach (as opposed to rage as it was in the older group). Finally, pride is positively correlated with anger and rage (but not fear and distress, as was the case with the older group). Because the correlations may reveal the distance – or the relative accessibility – associated with two feeling states, these differences, regardless how minor, may have important implications for emotion management processes. For instance, this first set of results also suggest that individuals over the age of 60 may have an easier time moving among negative emotions and among positive emotions than their younger counterparts. And, to the extent that some emotional transformations may require moving through intervening feeling states (Britt & Heise, 2000; Francis, 1997; Lively, 2008; Lively & Heise, 2004; Thoits, 1995) the two age groups may also utilize different pathways – including having different shortest paths – when moving between positive and negative feeling states. Lively and Heise (2004) assume that emotions correlating positively during a weeklong period may be relatively accessible from each other and emotions correlating negatively may be relatively inaccessible and that these correlations have implications for emotion management (also see Lively, 2008). In keeping with this, I offer an index of remoteness based on correlation coefficients to quantify this notion more exactly (see Table 4

18 (12 via joy)

18 (13 via joy)

Distress

12 (11 via joy) 12 (10 via selfreproach) 14 (11 via selfreproach)

9

4 7

4

–

6

Hope

8

7

7

8

4 6

–

3

9

Overjoyed

12 (8 via fear)

7

7

7

– 7

4

5

13 (11 via Hope)

Pride

4

6 (5 via fear) 3

4

4 –

4

6

17 (12 via hope)

SelfReproach

3

2

3

–

7 3

9 (7 via self-reproach)

14 (9 via self-reproach)

18 (15 via self-reproach and hope)

Anger

3

3

–

2

6 3

6

7

16 (13 via hope

Rage

1

–

2

1

5 2

6

12 (8 via selfreproach)

18 (14 via self-reproach and hope)

Fear

–

1

4

2

18 (13 via selfreproach and hope) 10 (7 via selfreproach) 8 (5 via selfreproach) 5 1

Distress

Note: Each remoteness is computed from its corresponding correlation (r) by the formula 10  (1-{sign(r)  sqrt[abs(r)]}), where sign(r) is 1 if r is positive, 1 if r is negative, and 0 if r is zero. Shortest paths were determined with a Javascript program provided at Dijkstra’s Algorithm (n.d.). Retrieved January 3, 2012 from http://www.tutor.ms.unimelb.edu.au/dijkstra/dp_frame.html.

Rage

Anger

Fear

5

Overjoyed

5 14 (10 via hope) 16 (12 via pride) 16 (12 via joy)

3

Hope

Pride Self-reproach

–

Tranquility

Remoteness Scores and Shortest Paths for Sixty and Older (Upper Triangle; N=267) and Younger than 60 (Lower Triangle; N=1087).

Tranquility

Table 4.

248 KATHRYN J. LIVELY

Age and the Experience and Management of Emotion

249

for the formula). If two emotions correlate 1.0 then they have a remoteness of 0, if they correlate 0.0 they have a remoteness of 10, if they correlate –1.0 then they have a remoteness of 20. In-between correlations are curvilinearly related to remoteness; for example, a correlation of 0.5 gives a remoteness of 3.16 Table 4 shows the results of shortest path analyses of the correlations presented in Table 3 (with the older group, again, represented in the upper triangle and the younger group represented in the lower). In graph theory, the simplest shortest path problem is that of finding a path between two vertices such that the sum of the weights of its constituent edges is minimized. The shortest paths presented here were generated by Dijkstra’s algorithm – a model that solves single source problems if all edge weights are greater than or equal to zero (Cormen, Leiserson, Rivest, & Stein, 2001). As noted previously, the remoteness scores signify the relative distance between two emotions and may serve as a marker of the emotion management effort required to move from one emotion to another. Based on the notion that small shifts in similar emotions are more manageable than large leaps between dissimilar ones, the shortest paths (reported in parentheses) use shortcuts via intervening emotions that reduce the remoteness between less accessible emotions. Indeed, prior analysis of the general population reveals that the basic correlation between tranquility and distress is actually smaller (and therefore, more experientially distant) than the combined correlations between tranquility and anger, anger and fear, and fear and distress (Lively & Heise, 2004). As can be seen, the majority of remoteness scores, particularly those emotion pairs in the middle, are single digits: these are emotion pairs that do not have any shortest paths. The path is simply what it is, meaning that the accessibility of these remoteness scores cannot be improved upon with the introduction of another emotion. However, several of the scores around the edges, particularly those in the corners representing relatively disparate emotion pairs, do have shortest paths that can be improved by segueing through an additional feeling state. It is here, among the seemingly distant emotions, that potential emotional segues exist for both groups. Notably, a number of shortest paths for the older group are more complicated (or require an additional segue) than the same transitions for the younger group. In particular, the shortest paths between distress and tranquility and fear and tranquility involve segueing through two emotions, self-reproach and hope, for the older group; whereas these same shortest

250

KATHRYN J. LIVELY

paths, among the younger group, involve transitioning through only one segueing emotion: pride. Further, the shortest paths between hope and anger and joy and anger involve self-reproach for the older group, whereas no segues are identified for that transition among the younger group, suggesting that the remoteness between these emotions, at least for individuals under the age of 60, cannot be shortened by transitioning through additional emotions. Similarly, the shortest path between distress and joy involves segueing though selfreproach for the older group, but there is no corresponding segueing emotion for the younger group. Additionally, the shortest path between pride and tranquility (two positive emotions) for the older group is via hope, whereas as these two emotions are relatively near and do not require an additional step among the younger group. There are three exceptions, however, to the pattern of greater complexity among the older group. First, the shortest path between rage and selfreproach, for the younger group, can be shortened by the introduction of fear. Second, the shortest path between rage and hope can be shortened via the introduction of joy. Third, the shortest path between distress and pride, among the younger group, may be reduced through the intervening emotion of fear. None of these paths require a segueing emotion in the older group: that is, these three pairs are more experientially close among the older group than they are the younger group. In terms of the actual emotions used as segueing emotions, those identified among the older group include hope (3 out of the 11 shortest paths for the older group involve hope, as a single segueing emotion), selfreproach (5 pathways), or a combination of the two: hope and self-reproach (3 pathways). In contrast, the younger group reveals hope and self-reproach as single points of transition in only two of their shortest paths (the pathway between self-reproach and tranquility and the pathway between hope and distress, respectively). The remaining 8 (out of 10) pathways occur either through pride (for example, the shortest path between anger and tranquility), fear (i.e., the shortest paths between self-reproach and rage and distress and pride), joy (the shortest path between hope and fear), and joy (the shortest paths between distress and tranquility, fear and tranquility, rage and tranquility, and rage and hope). It is worth nothing that all of the shortest paths identified for the older group involve hope, self-reproach, or some combination thereof. Selfreproach is considered to be a negative, impotent, and quiescent emotion (1.73, 1.35, 1.26), whereas hope is viewed as positive, powerful, and active (2.56, 2.26, 2.23).

Age and the Experience and Management of Emotion

251

Although the data from the younger group also reveal hope and selfreproach as potential segueing emotions, the shortest path analyses also reveals a number of other segueing emotions, including joy (2.29, 1.64, 1.63) and fear (1.32, 0.83, .014). Interestingly joy is slightly less positive, powerful, and active than hope and fear is slightly less negative, weak and quiescent than self-reproach, which suggests that younger individuals have access to a segueing emotions that exist along the fundamental dimensions of affective meaning. As noted earlier in the chapter, there is also considerable variation in the remoteness among negative emotions, particularly those transitions that do not require segueing emotions. For example, all of the remoteness values between negative emotion pairs (with the exception of distress and rage and distress and fear) are significantly smaller among the older group than among the younger group. In other words, each of these pairs is experientially closer among individuals 60 and above than they are among individuals less than 60. The remoteness value between distress and selfreproach is four times larger for the younger group than the older group, meaning that these two negative emotions are much less likely to co-occur among the younger group compared to their older group. The opposite pattern is found when examining the remoteness values among the positive emotion pairs. Essentially, all of the remoteness scores between positive emotion that are significantly different across age groups (po.05) indicate greater distance for the older group than for the younger group. The remoteness value between joy and pride, for example, among the older group is 9, compared to younger group, which has a remoteness of 5 for the same path. Again, what this suggests is that older individuals may have a more difficult time accessing pride from a state of joy than their younger peers. There is a similar discrepancy for the remoteness of tranquility and hope, and, as discussed above, tranquility and pride.

CONCLUSION It has long been understood that the individuals’ self-reported emotional experience varies with age. Although early studies focused almost exclusively on the experience of emotion in terms of frequencies, scholars recently have shifted their focus to the underlying structure of emotions. Psychological studies examining the correlation between positive and negative affect reveal that the two may become increasingly independent with age (Carstenson, 2006), which, in and of itself, has implications for

252

KATHRYN J. LIVELY

both well-being and physical health (Epel et al., 1998; Lane & Pollermann, 2002; Pennebaker et al., 1997). While this approach informs scholarly understanding of the effect of age on generalized measures of affect, a closer look at the simultaneous correlations between discrete emotions states provides us with a more nuanced understanding of emotional experience and the process of emotion management – particularly emotion management that may require segueing through one or more intervening emotions (Britt & Heise, 2000; Francis, 1997; Lively & Heise, 2004; Thoits, 1995). Psychological studies suggest that relatively undifferentiated positive and negative affect is more dispersed among older individuals than younger ones (Carstensen, 2006). This analysis, with its more granular focus on the correlations between specific emotions and their corresponding shortest paths, suggests that may be just part of the story. While six of the remoteness values between specific positive and negative emotion correlates are larger for the older group, four of them are shorter, and ten are not significantly different (po.05). Thus, focusing on composite measures as a way to tap into emotional complexity may obscure important nuances when it comes to understanding the effects of age on emotional experience (Expectation: Length). Moreover, within positive and negative emotions – something that cannot be determined by simply examining the relativeness nearness of negative and positive affect – older individuals have shorter and longer paths, respectively. Despite the rather mixed support for the first expectation (that older individuals will have longer shortest paths between positive and negative emotions), the results do support the notion that older might have more complex or convoluted shortest paths than their younger peers (Expectation: Complexity). Results show that older individuals have more shortest paths that require transitioning through a segueing emotion (11, compared to 10 identified for the younger group), and also that three of those paths require transitioning through more than one emotion. Specifically, moves between tranquility and anger, fear, and distress, can all be expedited by segueing through self-reproach and hope. Notably, none of the shortest paths for the younger group involve more than one segueing emotion.17 These last results bring us to the third expectation: Emotional Segues. Consistent with theoretical predictions from affect control theory (Heise, 2007) and empirical results from multidimensional scaling (Lively & Heise, 2004; Ross & Mirowsky, 2008), older individuals’ shortest paths tend to use emotions characterized by lower levels of potency and activation, with the exception of hope, which is characterized in terms of high potency and activation.18 Moreover, in three of the times that hope was identified as a

Age and the Experience and Management of Emotion

253

segueing emotion, it was coupled with self-reproach (an emotion that is viewed culturally as negative, weak, and inactive). Finally, only 3 out of the 11 pathways for the older group relied solely on positive, powerful, and active emotions (i.e., hope), but 6 out of 10 did for the younger group (i.e., either hope or joy). And although the younger group also relied on selfreproach in two of the shortest paths (i.e., the paths between fear and hope and distress and hope), the remaining paths relied on fear, (an emotion that is considerably less negative, less weak, and less quiescent than self-reproach). The results also reveal stronger positive correlations between negative emotions for the older group than for the younger group, and weaker correlations between positive emotions. The relative nearness of negative emotions among the older group may explain why 8 out of 11 of the shortest paths identified for older individuals involve negative emotions, compared to the 4 out of 10 shortest path involving negative emotions identified for younger individuals. These findings suggest that older individuals, in the course of a seven-day period, may be more likely to find themselves mired down in negative emotions. In contrast, younger individuals may find themselves moving more easily between positive emotions and, to the degree that the complexity of shortest paths matters, between positive and negative feelings. Psychological studies suggest that this emotional flexibility may be detrimental, to the degree that individuals may become unable to separate positive and negative feelings. However, it may also be the case that such flexibility may also benefit individuals who are required to manage their emotions – for example, in occupations or family situations that require substantial and sustained amounts of emotional labor (Hochschild, 1983, 1989; Lively, 2000; Lois, 2003, 2013).

DISCUSSION As recent studies have shown, age is an important predictor of emotional experience, at least as it pertains to the likelihood of experiencing or expressing particular emotions. However, it also has real implications for more subtle aspects of emotion – for example, the correlation of positive and negative affect as well as how specific emotions co-occur, both of which potentially seem to influence pathways of efficient and effective emotion management. These findings, based on the correlational structure of specific emotions, may provide us with insight into how the process of emotion management operates by age and offer additional clues as to why it is that older individuals seem more often protected from negative emotional

254

KATHRYN J. LIVELY

experiences, with the exception of distress which tends to increase in the mid 60s and early 70s, than their younger counterparts. Although psychologists are beginning to explore the relationship between largely undifferentiated positive and negative affect, this analysis hopes to reorient scholarly work on age and emotion, and emotion more generally, to the importance of how every day emotions co-occur. The idea that co-occurrences may be as important, if not more important, than absolute frequencies is not new. But it is consistent with recent lines of research in such fields as mental and physical health and juvenile delinquency, which have shifted their analyses to include questions of comorbidity (Rosenfield, Phillips, & White, 2006). Having said this, four caveats should be addressed. First, as mentioned earlier, self-report data may be tapping into individuals’ most salient or most memorable emotional experiences, as opposed to the entirety of their experiences. This suggests that the pathways reported here may be memory structures rather than emotional structures; in fact, affect control theory (Heise, 1979) builds in memory through its reliance on fundamental and transient sentiments. Even if these findings reflect memory rather than emotional transitions, the pathways nonetheless suggest that emotion memories are storied in ways that are consistent with cultural stereotypes and social roles associated with age. Although it is beyond the scope of this study to determine whether the items are tapping into real emotion or emotional memory, the concern raises the need to collect data (1) over a shorter period of time or (2) in real time. Such an undertaking could use diary methodology, where subjects are routinely asked to write down emotional experiences or reactions randomly throughout the day. Another strategy would entail creating multiple emotional experiences in experimental settings that could be measured in real time (Robinson, Rogalin, & Smith-Lovin, 2004). A second caveat to this work is that it only relates to age differences broadly defined, without taking into account variation within each group. To the degree to which age, like gender, is an institutionalized arrangement based in large part on power and status, the effects of age considered alone may be overstated (Ridgeway & Smith-Lovin, 1999). However, several sociological studies have consistently found that age does have important effects on emotional experience, even when controlling for other personal and social characteristics (Mirowsky & Ross, 2003; Ross & Mirowsky, 2008; Simon & Nath, 2004). Future work in this area would need to determine whether the emotional paths of older individuals remain different than those of younger individuals once other social characteristics and structural

Age and the Experience and Management of Emotion

255

considerations, such as parental status (Evenson & Simon, 2005), have been taken into consideration. Additionally, because of the changes we see in the emotional quality of elders’ lives dip dramatically at age 75 (Mirowsky & Ross, 2003), it is also important to take a more nuanced look at the aging process. And indeed, some psychological studies of the effect of age on emotion make age-graded distinctions among the elderly – that is, individuals over the age of 65 (Ong & Bergeman, 2004). A third caveat is the possibility that these analyses are picking up cohort effects, instead of the effects of aging. Individuals who were 60 and above who answered questions in the GSS Emotions module grew up in an era marked by two World Wars. Thus, many of individuals in ‘‘the older group’’ are members of what some have termed ‘‘The Greatest Generation’’ (Brokaw, 1998; Elder, 1999). This group, more than any other in U.S. history, has been celebrated for the sacrifices it made during the war years and for its efforts at rebuilding the nation. In contrast, ‘‘the younger group’’ contains the Baby Boomers. According to Jones (1980), the baby boomers, as a group, were the wealthiest, most active, and most physically fit generation to date and among the first to grow up expecting the world to improve with time. Because of increased postwar incomes, Baby Boomers were also the first generation that experienced an abundance of food, apparel, and access to retirement programs (Owram, 1997). Given the vast discrepancies between these two generations, as well as subsequent generations, it is possible that cohort effects may also be at play. A fourth caveat has to do with the nature of the EPA data, which guided the analysis of the qualitative differences between the shortest paths. As mentioned earlier, the EPA data on which this part of the analysis is based were collected from college-age students at a large Midwestern university. Although affect control theorists, as well as others using semantic differentials, assume that cultural sentiments are shared, there has been some recent evidence that the EPA values for emotion words, as well as selfreports of emotion, may differ by sex (Lively, 2008; Lively & Heise, 2004). Many scholars have theorized at length about the gendered nature of the emotional culture in the United States. (Cancian, 1987; Gordon, 1989; Simon & Nath, 2004). Despite consistent differences reported in studies of age and emotion, no one, to date, has theorized about whether there might be systematic differences in the affective meaning of emotion, by age. However, this study suggests that there may, in fact, be some differences worth exploring. Previous segueing studies have illustrated that emotional segues tend to correspond to shifts along affective dimensions of Potency and Activation

256

KATHRYN J. LIVELY

(Francis, 1997; Lively, 2008; Lively & Heise, 2004; also see Thoits, 1995). Moreover, Lively’s (2008) study of sex differences in emotional segueing revealed that women’s shortest paths between emotional states were significantly more likely to use emotions that are less potent and less active; notably, in that analysis, she was able to use the EPA values for women and men, separately.19 In this study, it is true that older individuals have greater variety in segueing emotions (their paths use twice as many different emotions as the younger group) and, for the most part, use emotions that are less powerful and less active, as would be expected as age results in shifts away from powerful and active affect (Lively & Heise, 2004; Ross & Mirowsky, 2008). They also are more likely than the younger group to use negative emotions, which is consistent with the finding that older individuals’ negative emotions are more tightly correlated than those of their younger counterparts. However, two of their shortest paths, the ones between self-reproach and tranquility, and between anger and tranquility, both use hope, an emotion that is characterized – at least among college-age students – as very good, very powerful, and very active. That said, it is worth noting that hope, like fear and self-reproach, has an inherent temporal component that several of the other emotions, such as anger, rage, distress, tranquility, and joy, do not (Charmaz, 1991). Specifically, hope is a forward-looking emotion and it cuts both ways. In other words, people hope for things to happen, such as getting a new job, getting a promotion, or getting engaged, and they hope for things not to happen, such as having to have surgery, being placed in an institution, or the death of a friend or spouse. It seems that the things we hope for – or the likelihood that they will actually happen – may have implications for the quality of the emotion itself. The degree to which younger people have more expansive desires, and longer time horizons than their older counterparts (Carstensen, 2006; Charmaz, 1991), may amplify the EPA scores associated with the emotion. Similarly, the types of desires that older individuals have, coupled with their shorter time horizons may lower associated EPA scores. In addition, the pride that younger individuals feel may be more self- or futuredirected, whereas the pride that older individuals feel may be more otherdirected and focused on the past. In order to know for sure whether these sentiments differ by age, we must collect EPA data for emotion words from a broader, more representative population that is not limited to 18–24-years-old college students. Given the important role that affective meaning seems to play in emotional segueing, another area ripe for this type of analyses would be cross-cultural or subcultural variations in emotion management. Building

Age and the Experience and Management of Emotion

257

on the work of Morgan and Heise, 1988 and MacKinnon and Keating (1989), Smith and colleagues have headed an ambitious project that has documented numerous subtle variations between the affective sentiments of Americans and Japanese (Smith & Francis, 2005; Smith, Matsuno, & Ike, 2001; Smith, Matsuno, & Umino, 1994), between Americans and Chinese (Smith, 2002), and between Japanese women and men (Smith, Matsuno, & Umino, 1994; also see Hunt, 2008; Smith-Lovin & Douglass, 1992). Given the degree to which sentiment structures are culturally influenced – as well as their apparent roles in emotional segueing – it is likely that the shortest paths between emotions may also be cultural-specific. Although the findings reported here might appear relatively complex, the differences associated with the ways in which discrete emotions co-occur seem to have real implications for the experience and the duration of emotional experiences, as well as the nature and effectiveness of emotion management strategies. Sociologists to date have paid little attention to how the structure of emotional experience may vary as a result of group membership or how these differences may impact one’s emotional wellbeing. And although psychologists have posited that having distinct positive and negative affect is positive for well-being and health, there has been very little investigation as to how correlations between negative emotions may affect the individual or what these types of tight correlations may tell us about the nature of individuals’ lives. According to several contemporary theories of emotion, many negative emotions are the result of negative identity-disconfirming events (Heise, 2007; also see Burke, 1991; Thoits, 1992) or interactions in which individuals are denied what they perceive as their due, in terms of power or status (Kemper, 1978). Given that older individuals’ negative emotions tend to co-occur more frequently than those of younger individuals, this may suggest that older individuals may be more likely to face events where their identities, possibly multiple identities simultaneously, are disconfirmed – such as an event that might disrupt one’s sense of self as a homeowner, a father, an employee, a driver, a husband, or even as a man. Additionally, these results suggest that older individuals may be more sensitive to the potential interpersonal damage caused by anger, which may explain the greater co-occurrence of anger and distress and rage and distress. Notably, the variations between the older and younger groups are, in part, consistent with psychological studies of aging and emotion as well as neurological studies of the aging brain. In particular, these results are also consistent with the finding that while there is a form of neurological degeneration that may prevent negative physiological stimuli from landing

258

KATHRYN J. LIVELY

(Cacioppo, Berntson, Klein, & Poehlmann, 1998), the same degeneration can also extend the negative experience once the physiological change has occurred (Charles & Piazza, 2009). That said, correlational differences in the older and younger groups’ emotions may also be attributable to complex social arrangements, subcultural variation in affective sentiments, or even preferred methods of emotion management. As sociologists continue to cultivate an understating of how social characteristics affect individuals’ emotional lives, future research may need to consider the ways in which the correlational structure of emotions (and the shortest paths between them) vary as well as to formulate more systematic understandings of why and to what ends such differences may or may not occur. And finally, to the degree that correlations seem to coincide with efficacious means of coping or emotion management, further analyses of them and their corresponding shortest paths may allow us to uncover systematic differences in emotion management styles, both personally and interpersonally. From a practical standpoint, such an occurrence might allow us to determine which styles are more efficient and effective. Moreover, from an equity standpoint, does older individuals’ utilization of emotion management pathways marked by weaker emotions lower their status, particularly in interactions in the work place, where laboratory and ethnographic studies alike reveal that individuals who exhibit powerful emotions are more likely to be attributed more status or judged as being of higher status (Tiedens, 2000)? Or is it possible that older individuals (who, up to a point, tend to hold higher status in both informal and formal organizations) may occupy a different emotional reality than their younger subordinates, constituents, and followers? These observations lend themselves to further empirical questions, such as whether younger individuals, who tend to have lower levels of emotional well-being, would be better off using the shortest paths more common to their elders, even though it might mean claiming (and thus transitioning through) less powerful and active emotions? If these pathways are social (and therefore learned) can individuals be resocialized to use more effective emotion management strategies – strategies that might, in turn, reduce frequent experiences of distress and to make more frequent feelings of wellbeing (marking the poles of emotional experience) even if it means claiming emotions that are typically – or culturally and, perhaps, structurally – off limits? And to what degree may some groups, such as the elderly, be affected by their relative inability to access pride, and their overutilization of negative emotions during complex emotional transitions? These questions are beyond the scope of the data presented here. They may, however, set an

Age and the Experience and Management of Emotion

259

agenda encouraging researchers to bridge the disparities within research on age differences in emotion. They may also push our understanding of yet another way one’s social location influences one of the most intimate and seemingly individual aspects of one’s life.

NOTES 1. Fung, Carstensen & Lultz (1999) suggest that the same motivational changes can also occur in contexts other than aging that are also associated with a limited time perspective. 2. In a recent review of emotion and aging, Scheibe & Carstensen (2010) point to several studies that call this assumption into question (Isaacowitz, Toner & Neupert, 2009; Knight et al., 2007; Mather & Knight, 2005; Petrician, Moscovitch, & Schimmack, 2008). Instead, they argue that negative stimuli automatically demand attention and that cognitive control is necessary to disattend to negative stimuli and attend to positive stimuli. 3. See Gross (1998) for an early, integrated review of the emotional regulation literature within psychology. 4. Neuroticism tends to lower emotional complexity, across all ages, explaining away individual differences among older adults (Ong & Bergeman, 2004; Ready et al., 2011). 5. Length, here, refers to the simple correlation (or the higher order correlation) that exists between distinct emotional states. It is used, here, synonymously with how the psychologists use the term complexity (Carstensen, 2006; Ready et al., 2011). 6. Complexity, here, does not refer to the simple correlations (or the relationships) between the emotional states (Carstensen, 2006; Ready, Akerstedt, & Mroczek, 2011). Instead, complexity refers to the number of segueing emotions required to move from one emotion state to another (Lively, 2008; Lively & Heise, 2004). 7. See Lively and Heise (2004) for more details regarding the steps they took in creating the final measurement model, which was derived from insights taken from Ortony et al. (1988) emotional classification and affect control theory (Heise, 1979, Smith-Lovin & Heise, 1988). 8. For additional details concerning the Ortony, Clore, and Collins classificatory scheme, plus assignments from the GSS, see Lively & Heise, 2004). 9. Given that the data here are based on self-reports, older individuals’ less frequent reporting of emotions may be a function of memory, a point to which I return in the conclusion; however, that said, ancillary analyses reveal fewer significant differences between the 65 and above group than for the 55 and above group, which may be capturing the increase in distress that tends to occur in the late 60s to mid 70s (Ross & Mirowsky, 2008). 10. In light of the data requirements of the measurement model, it is beyond the scope of this analyses to determine whether the shortest paths differ by age, while controlling for a variety of sociodemographic and social characteristics, such as sex, gender, employment, marital status, and number of children living in the home. However, auxiliary analyses reveal that the relationships revealed in Table 1 do

260

KATHRYN J. LIVELY

indeed remain statistically significant (po.05), even after the inclusion of controls, with the exception of hope (po.10) and joy. Further analysis suggests that differences in joy that appear when examining main effects only may be caused in differences in parental status, a point to which I return in the discussion. These results are available from the author. (Also see Ross & Mirowsky, 2008 for a review of studies on emotion and aging that also control for these and other variables in their analyses). 11. Because all of the emotion items were highly skewed (Lively & Heise, 2004), I estimated the model using asymptotically distribution free (ADF) estimation in AMOS 4.0. Unlike maximum-likelihood (ML) estimation, which assumes a normal distribution, ADF does not impose an assumption of normality and typically generates reliable estimates in large samples (Browne & Cudeck, 1992). The estimates using ADF estimation and ML estimation were largely similar. 12. Given that the data is self-report, it is possible that some of the results, especially those related to remoteness, may be a reflection of the similarity of the linguistic terms used rather than the actual emotions themselves. However, as the results of the measurement models reveal, the nine emotion factors are empirically distinct. Indeed, the purpose of using latent variables in the models, instead of the 18 indicators themselves, is to increase the likelihood of tapping into an underlying emotional state that may have one or more emotion words with which it is associated (see Ortony, Clore, & Collins, 1988, for more detail). 13. The Incremental Fit Index (IFI) estimates how well the model fits in comparison with a baseline model of independence (Bollen, 1989); the IFI statistic is preferable to other similar measures when using estimators that do not assume normality in the data (Hoyle & Panter, 1995). The Comparative Fit Index (CFI) also assesses the reduction in the amount of misfit of the target model relative to the baseline model (Hu & Bentler, 1995). The Root Mean Square Error of Approximation (RMSEA) was developed by Steiger and Lind (1980) as a means for compensating for the effect of model complexity. According to Browne and Cudeck (1992) a value of the RMSEA of about .05 or less would indicate a close fit of the model in relation to the degrees of freedom, however a value of .08 represents a reasonable fit. Although this figure is often based on subjective judgment, they argue that it is more reasonable than the requirement of exact fit with the RMSEA=0.0. 14. The EPA profiles are available at The Affect Control Theory Website (http:// www.indiana.edu/%7Esocpsy/ACT/interact/subcultures/sex_averages.htm). These averages are based on the Indiana 2002-4 Attributes Data. 15. AMOS yields correlations that are sometimes higher than zero-order correlations among indices, a standard result when creating latent trait models. In keeping with Lively and Heise’s (2004) original analyses, I used AMOS estimates for the correlations rather then computing actual scores and correlating them. 16. The remoteness scale creates arbitrary values. The values presented here were derived from Lively and Heise’s (2004) formula for ease of comparability (also see Lively, 2008). 17. Although the majority of shortest paths that require transitioning through segueing emotions link positive and negative emotions, there are two exceptions. For the older group, the shortest path between pride and tranquility requires hope, and, for the younger group, the shortest path between self-reproach and rage involves fear. These patterns may be a result of the greater dispersion of positive emotions

Age and the Experience and Management of Emotion

261

among the older group and the greater dispersion of negative emotions among the younger group. 18. As noted earlier, the hope variable is actually measured by the question: ‘‘How many in the days in the last seven days did you experience excitement?’’ According to Ortony et al. (1988) excitement and hope are affectively similar and tend to cluster together. Lively and Heise (2004) used the term hope as a marker to be consistent with Ortony, Clore, and Collins’ well-known classification. Assuming that excitement and hope are largely interchangeable, and hope’s relationship to faith, this result may reflect older individuals’ greater tendency to pray when attempting to deal with angry feelings (Simon & Nath, 2004). Future work is required to determine nuances regarding the interchangeability of hope and excitement. 19. The data used in the computer simulation program, INTERACT, divide the EPA values of emotions, identities, behaviors, and settings by sex. Unfortunately, this is the only social characteristic of record.

REFERENCES Blanchard-Fields, F. (2007). Everyday problem solving and emotion – An adult developmental perspective. Current Directions in Psychological Science, 16, 26–31. Britt, L., & Heise, D. R. (2000). From shame to pride in identity politics. In S. Stryker, T. J. Owens & R. W. White (Eds.), Self, identity, and social movements (pp. 222–268). Minneapolis, MN: University of Minnesota Press. Brokaw, T. (1998). The greatest generation. New York, NY: Random House Publishing. Browne, M. W., & Cudeck, R. (1992). Alternative ways of assessing model fit. Sociological Methods and Research, 21, 230–258. Burke, P. J. (1991). Identity processes and social stress. American Sociological Review, 56, 836–849. Cacioppo, J. T., Berntson, G. G., Bechara, A., Tranel, D., & Hawkley, L. C. (2011). Could an aging brain contribute to subjective well being? The value added by a social neuroscience perspective. In A. Todorov, S. T. Fiske & D. Prentice (Eds.), Social neuroscience: Toward understanding the underpinnings of the social mind (pp. 249–262). New York, NY: Oxford University Press. Cacioppo, J. T., Berntson, G. C., Klein, D. J., & Poehlmann, K. M. (1998). Psychophysiology of emotion across the life span. In K. W. Schaie & M. Powell (Eds.), Annual review of gerontology and geriatrics: Focus on emotion and adult development (Vol 17, pp. 27–74). New York, NY: Springer. Cancian, F. M. (1987). Love in America: Gender and self-development. Cambridge: Cambridge University Press. Carstensen, L. L. (2006). The influence on sense of time on human development. Science, 312, 1913–1915. Carstensen, L. L., Isaacowitz, D. M., & Charles, S. T. (1999). Taking time seriously: A theory of socioemotional selectivity. American Psychologist, 54, 165–1681. Carstensen, L. L., Pasupathi, M., Mayr, U., & Nesslroade, J. R. (2000). Emotional experience in everyday life across the adult life span. Journal of Personality and Social Psychology, 79, 644–655.

262

KATHRYN J. LIVELY

Carstensen, L. L., Turan, B., Scheibe, S., Ram, N., Ersner-Hershfield, H., Samanez-Larkin, G., y Nesselroade, J. R. (2010). Emotional Experience Improves with Age: Evidence Based on Over 10 years of Experience Sampling. Psychology and Aging, 11, 21–33. Charles, S. T., & Piazza, J. R. (2009). Age difference in affective well-being: Context matters. Social and Personality Psychology Compass, 3, 1–14. Charmaz, K. (1991). Good days, bad days: The self and chronic illness in time. New Brunswick, NJ: Rutgers University Press. Cormen, T. H., Leiserson, C. E., Rivest, R. L., & Stein, C. (2001). Introduction to algorithms (2nd ed.). Cambridge, MA: MIT Press. Davis, J., & Smith, T. (1996). General social survey, 1972–1996, cumulative codebook and datafile. Chicago, IL: NORC and University of Chicago. Dolomie-Faguar, L., Gatta, B., Nguyen, T. D., & Corcuff, J. B. (2008). Bioavailable Estradiol in Man: Relationship with age and testosterone. Clinica Chimica Acta, 398, 145–147. Elder, G. H., Jr. (1999). Children of the great depression: Social change in life experience (25th Anniversary ed.). Boulder, CO: Westview Press. Epel, E. S., McEwen, B. S., & Ickovies, J. R. (1998). Embodying psychological thriving: Physical thriving in response to stress. Journal of Social Issues, 54, 301–322. Evenson, R., & Simon, R. W. (2005). Clarifying the relationship between parenthood and depression. Journal of Health and Social Behavior, 46, 341–358. Francis, L. (1997). Ideology and interpersonal emotion management: Redefining identity in two support groups. Social Psychology Quarterly, 60, 153–171. Goffman, E. (1961). Asylums: Essays on the social situation of mental patients and other inmates. Garden City, NY: Anchor Press. Gordon, S. (1989). The socialization of children’s emotions. In C. Saarni & P. Harris (Eds.), Children’s understanding of emotions (pp. 319–349). New York, NY: Cambridge University Press. Gove, W. R., Ortega, S. T., & Style, C. B. (1989). The maturational and role perspectives on aging and self through the adult years: An empirical evaluation. American Journal of Sociology, 82, 1327–1335. Gross, J. J. (1998). The emerging field of emotional regulation: An integrative review. Review of General Psychology, 2, 271–299. Gross, J. J., Carstensen, L. L., Pasupathi, M., Tsai, J., Go¨testam Skorpen, C., & Hsu, A. Y. C. (1997). Emotion and aging: Experience, expression, and control. Psychology and Aging, 12, 590–599. Heckhausen, J., & Schultz, R. (1995). A life-span theory of control. Psychological Review, 102, 284–304. Heise, D. R. (1979). Understanding events: Affect and the construction of social action. New York, NY: Cambridge University Press. Heise, D. R. (2007). Expressive order: Confirming sentiments in social actions. New York, NY: Springer Press. Hirshi, T., & Gottfredson, M. (1983). Age and the explanation of crime. American Journal of Sociology, 89, 552–584. Hochschild, A. R. (1979). Emotion work, feeling rules, and social structure. American Journal of Sociology, 85, 551–575. Hochschild, A. R. (1983). The managed heart: The commercialization of human feeling. Berkeley, CA: University of California Press.

Age and the Experience and Management of Emotion

263

Hochschild, A. R. (1989). The second shift: Working parents and the revolution at home. New York, NY: Penguin. Hoyle, R. H., & Panter, A. T. (1995). Writing about structural equation models. In R. H. Hoyle (Ed.), Structural equation modeling: Comments, issues, and applications (pp. 56–77). Thousand Oaks, CA: Sage. Hu, L. T., & Bentler, P. (1995). Evaluating model fit. In R. H. Hoyle (Ed.), Structural equation modeling: Comments, issues, and applications (pp. 76–99). Thousand Oaks, CA: Sage. Hunt, P. M. (2008). From festies to tourrats: Examining the relationship between Jamband subculture involvement and role meanings. Social Psychology Quarterly, 714, 356–378. Isaacowitz, D. M., Toner, K., & Neupert, S. D. (2009). Use of Gaze for real-time mood regulation: Effects of age and attentional focus. Psychology and Aging, 24, 989–994. Jones, L. (1980). Great expectations: America and the baby boom generation. New York, NY: Coward, McCann & Georghegan. Kafetsios, K. (2004). Attachment and emotional intelligence capabilities across the life course. Personality and Individual Differences, 37, 129–145. Kemper, T. D. (1978). A social interactional theory of emotions. New York, NY: Wiley. Kline, R. B. (1998). Principles and practice of structural equation modeling. New York, NY: Guilford Press. Knight, M., Seymour, T. L., Gaunt, J. T., Baker, C., Nesmith, K., & Mather, M. (2007). Aging and goal-directed emotional attention: Distraction reverses emotional biases. Emotion, 7, 705–714. Labouvie-Vief, G. (2003). Dynamic integration: Affect, cognition, and the self in adulthood. Current Directions in Psychological Science, 12, 201–206. Lane, R. D., & Pollermann, B. Z. (2002). Complexity of Emotion Representations. In L. F. Barrett & P. Salovey (Eds.), The wisdom in feeling: Psychological processes in emotional intelligence (pp. 271–294). New York, NY: Guilford Press. Lively, K. J. (2008). Emotional segueing and the management of emotion, by women and men. Social Forces, 87, 911–936. Lively, K. J., & Heise., D. R. (2004). Sociological realms of emotional experience. American Journal of Sociology, 109, 1109–1136. Lois, J. (2003). Heroic efforts: The emotional lives of search and rescue volunteers. New York, NY: New York University Press. Loscocco, K. A., & Kalleberg, A. L. (1988). Age and the meaning of work in the United States and Japan. Social Forces, 672, 337–356. MacKinnon, N. J., & Keating, L. (1989). The structure of emotion: A review of the problem and a cross-cultural analysis. Social Psychology Quarterly, 52, 70–83. Mather, M., & Knight, M. (2005). Goal-directed memory: The role of cognitive control in older adults’ emotional memory. Psychology and Aging, 20, 554–570. Mirowsky, J., & Ross, C. E. (2003). Social causes of psychological distress (2d ed.). New York, NY: Aldine de Gruyter. Morgan, R., & Heise., D. R. (1988). Structure of emotions. Social Psychology Quarterly, 51, 19–31. Mroczek, D. K., & Kolarz, C. M. (1998). The effect of age on positive and negative affect: A Developmental perspective on happiness. Journal of Personality and Social Psychology, 75, 1333–1349.

264

KATHRYN J. LIVELY

Ong, A. D., & Bergeman, C. S. (2004). The complexity of emotions in later life. Journal of Gerontology: Psychological Sciences, 59B, 117–122. Ortony, A., Clore, G. L., & Collins, A. (1988). The cognitive structure of emotion. New York, NY: Cambridge University Press. Ortony, A., Clore, G. L., & Foss, M. A. (1987). The referential structure of the affective lexicon. Cognitive Science, 11, 341–364. Owram, D. (1997). Born at the right time. Toronto: University of Toronto Press. Pennebaker, J. W., Mayne, T. J., & Francis, M. E. (1997). Linguistic predictors of adaptive bereavement. Journal of Personality and Social Psychology, 72, 863–871. Petrician, R., Moscovitch, M., & Schimmack, U. (2008). Cognitive resources, valence, and memory retrieval of emotional events in older adults. Psychology and Aging, 23, 585–594. Pinquart, M. (2001). Correlates of subjective health in older adults: A meta-analysis. Psychology and Aging, 16, 414–426. Ready, R. E., Akerstedt, A. M., & Mroczek, D. K. (2012). Emotional complexity and emotional well-being in older adults: Risks of high neuroticism. Aging Mental & Health, 16(1), 17–26. Ready, R. E., Carvalho, J. O., & Weinberger, M. I. (2008). Emotional complexity in younger, midlife, and older adults. Psychology and Aging, 23, 928–933. Ready, R. E., Robinson, M. D., & Weinberger, M. I. (2006). Age differences in the organization of emotion knowledge: Effects involving valence and time frame. Psychology and Aging, 21, 726–736. Ridgeway, C. L., & Smith-Lovin, L. (1999). The gender system and interaction. Annual Review of Sociology, 25, 191–216. Robinson, D. T., Rogalin, C. L., & Smith-Lovin, L. (2004). Physiological measures of theoretical concepts: Some ideas for linking deflection and emotion to physical responses during interaction. Advances in Group Processes, 21, 77–115. Rosen, K. H. (1988). Discrete mathematics and its applications. New York, NY: Random House. Rosenfield, S., Phillips, J, & White, H. (2006). Gender, race, and the self in mental health and crime. Social Problems, 53, 161–185. Ross, C. E., & Mirowsky, J. (1987). Normlessness, powerlessness, and trouble with the law. Criminology, 25, 257–278. Ross, C. E., & Mirowsky, J. (2008). Age and the balance of emotions. Social Science and Medicine, 66, 2391–23400. Scheibe, S., & Blanchard-Fields, F. (2009). Effects of regulating emotions on cognitive performance: What is costly for younger adults is not so costly for older adults. Psychology and Aging, 24, 519–529. Scheibe, S., & Carstensen, L. L. (2010). Emotional aging: Recent findings and future trends. Journal of Gerontology: Psychological Sciences, 65B, 135–144. Scheibe, S., Mata, R., & Carstensen, L. L. (2009). Yes, they can! Older adults accurately forecast emotional responses to wins (though not to losses) in the 2008 U.S. presidential election. Unpublished Manuscript. Schieman, S. (1999). Age and anger. Journal of Health and Social Behavior, 40, 273–289. Simon, R. W., & Nath, L. K. (2004). Gender and emotion in the U.S.: Do men and women differ in self-reports of feelings and expressive behavior? American Journal of Sociology, 109, 1137–1176.

Age and the Experience and Management of Emotion

265

Smith, H. W. (2002). Are emotions cross-culturally intersubjective? A chinese test. Paper presented at the International Sociological Association. Brisbane, Australia. Smith, H. W., & Francis, L. E. (2005). Social vs. self-directed events among Japanese and Americans. Social Forces, 84, 821–830. Smith, H. W., Matsuno, T., & Ike, S. (2001). The affective basis of attributional processes among Japanese and Americans. Social Psychology Quarterly, 64, 180–194. Smith, H. W., Matsuno, T., & Umino, M. (1994). How similar are impression-formation processes among Japanese and Americans? Social Psychology Quarterly, 57, 124–139. Smith-Lovin, L., & Douglass, W. (1992). An affect control analysis of two religious subcultures. In V. Gecas & D. Franks (Eds.), Social perspectives in emotions (Vol. 1, pp. 217–248). New York, NY: JAI Press. Smith-Lovin, L., & Heise, D. R. (1988). Analyzing social interaction: Advances in affect control theory. New York, NY: Gordon & Breach. Steiger, J. H., & Lind, J. (1980). Statistically based tests for the number of common factors. Paper presented at the annual meeting of the Psychometric Society, Iowa City, IA. Stryker, S. (1980). Symbolic interactionism: A social structural version. Menlo Park, CA: Benjamin/Cummings. Thoits, P. A. (1992). Identity structures and psychological well-being: Gender and marital status comparisons. Social Psychology Quarterly, 55, 236–256. Thoits, P. A. (1995). Managing the emotions of others. Symbolic Interaction, 19, 85–109. Tiedens, L. Z. (2000). Powerful emotions: The vicious cycle of social status positions and emotions. In N. M. Ashkanasy, E. J. C. Hartel & W. J. Zerbe (Eds.), Emotions in the workplace: Research, theory, and practice (pp. 71–81). Westport, CT: Quorum. Umberson, D. (1987). Family status and health behaviors: Social control as a dimension of social integration. Journal of Health and Social Behavior, 28, 306–319. Wasielewski, P. L. (1985). The emotional basis of charisma. Symbolic Interaction, 8, 207–222. Winkler, R. L., & Hays, W. L. (1975). Statistics: Probability, inference, and decision. New York, NY: Holt, Rinehart, and Winston.

A GENERALIZED THEORY OF CONFLICTING LEADER IDENTITY ON GROUP PERFORMANCE Christabel L. Rogalin ABSTRACT Purpose – This chapter seeks to theoretically answer the question: under which circumstances do groups succeed under female leadership? Further, is it possible to conceptualize the engineering of groups such that group success under female leadership is a likely outcome? Design/methodology/approach – In this chapter, I draw on identity control theory (Burke & Stets, 2009; Stets & Burke, 2005) and role congruity theory (Eagly, 2003) to discuss the implications for female leaders of the discrepancy between the female gender identity and the leader identity. Next, I draw upon status characteristics theory (Berger et al., 1972) to further illustrate the negative consequences of being a female leader. Then, drawing on group processes research, I make the explicit link between the negative expectations for female leaders on group performance through the endorsement of group members. Finally, I utilize innovative research using institutionalization of female leadership to propose a possible solution for improving group performance. Research implications – I present nine testable hypotheses ready for empirical test. Advances in Group Processes: Thirtieth Anniversary Edition Advances in Group Processes, Volume 30, 267–291 Copyright r 2013 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 0882-6145/doi:10.1108/S0882-6145(2013)0000030013

267

268

CHRISTABEL L. ROGALIN

Social implications – I propose that training materials underscoring the skills that females have as leaders can subvert the development of conflictual expectations facing female leaders, thus removing the deleterious effects on group performance. That is, if group members receive training that emphasizes the competencies and skills women bring to the group’s task and to the leadership role, then group performance will not be threatened. Keywords: Leadership; female leadership; institutionalization; identity; status

INTRODUCTION This chapter develops a theory to answer the questions: Under what circumstances do groups succeed under female leadership? How may groups be engineered to increase the likelihood of success under female leadership? In this chapter, I distill an array of disparate theories and empirical findings in the form of a set of concise propositions representing the current state of our knowledge to answer these two questions and set forth nine testable hypotheses. Though women and men are employed at nearly equal rates, women are still visibly underrepresented in high-powered positions of leadership in management and politics. Based on the current rate of change, Catalyst (2007) projects that it will take an estimated 47 years for women to hold corporate office positions at rates equal as men. To improve women’s status in organizations and society, some policymakers advocate recruiting talented women for leadership positions and placing them in those positions. While the identification and placement of talented women in leadership positions is an important step toward equality, alone it will be insufficient for gaining equality in leadership outcomes. When women do achieve leadership positions, all too often they fail to endure. Effective female leadership of successful groups depends on the expectations of group members and, specifically, whether they expect their leader to possess the traits that facilitate successful leadership. The general expectations people hold for women (i.e., nurturing, deferential, and concerned with the emotional well-being of others) are inconsistent with the general expectations that people hold for group leaders (i.e., assertiveness, dominance, and

Conflicting Leader Identity

269

concerned about the task-performance of others) (Eagly, 1987; Eagly & Karau, 2002). Male leaders, however, do not suffer from such inconsistent expectations, as the traits expected from leaders are consistent with traits expected of men. While many researchers have studied perceptions of female leadership and the effects of leadership on women (Eagly & Karau, 2002; Ritter & Yoder, 2004; Schein, 2001), this project is unique, in that it examines the conditions under which female leadership catalyzes stronger group performance. When social expectations for a person are inconsistent, as in the case of female leaders, people’s beliefs about the person’s competence decline. In the case of female leaders, conflicting cultural expectations for being a woman and being a leader undermine group members’ assessments of women’s general competency (Eagly, 2003; Eagly & Karau, 2002). Because they do not confront such conflicting expectations, male competency is not similarly questioned in this manner. Such competing expectations have consequences for both female leaders and for group performance under their leadership. If group members do not believe that their leader is competent, then the leader will fail in her attempts to lead the group (Bass, 1990). Further, group members have less confidence in a leader, and are less likely to endorse that leader, if they believe the leader is incompetent (Dornbusch & Scott, 1975; Hollander & Julian, 1970). In the absence of confidence in the leader, leadership effectiveness is undermined. Thus, the ramifications for women are that compared to maleled groups, female-led groups underperform because women are viewed as less competent (or effective) than men (Carli, 2001; Pugh & Wahrman, 1983; Wagner, Ford, & Ford, 1986). What does it mean to say that a leader is effective? Effective leaders are those who keep group members focused on the task and motivate group members to perform well. It has long been recognized that group performance depends on members’ attention to the task and willingness to contribute to the group’s work on the task (Steiner, 1972). Thus, groups whose members are focused on the task and are willing to contribute to the task will outperform groups whose members are less task-focused and less willing to contribute. In this work, I theorize about the ways conflicting cultural expectations facing leaders affect performance. Therefore, I propose a general theory of how identity conflict impacts performance. This theory allows me to address unanswered questions regarding conflicting expectations for women leaders and their negative effects on perceptions of female competency (Eagly & Karau, 2002) and group performance. More importantly, I propose a general

270

CHRISTABEL L. ROGALIN

theory regarding the effects of identity conflict on group outcomes. Specifically, I argue: Conflicting Leader Identity-Competence ExpectationsEndorsement-Group Performance A focus on the process through which expectations connected to a leader’s gender affects group performance also leads to possible solutions for this important social problem of the lack of female leaders. I propose that training materials underscoring the skills that females have as leaders can subvert the development of conflictual expectations facing female leaders, thus removing the deleterious effects on group performance. That is, if group members receive training that emphasizes the competencies and skills women bring to the group’s task and to the leadership role, then group performance will not be threatened (Cohen, 1972, 1982; Pugh & Wahrman, 1983). In previous studies, researchers have found that emphasizing women’s skills and abilities increases the likelihood that female leaders will be perceived as being highly competent (Lucas, 2003; Yoder, Schleicher, & McDonald, 1998). In this chapter I extend this research to present a theory outlining the impact of a conflicting leader identity on group performance. To accomplish this goal, I draw on identity control theory (Burke & Stets, 2009; Stets & Burke, 2005) and role congruity theory (Eagly, 2003; Eagly & Karau, 2002) to discuss the implications for female leaders of the discrepancy between the female gender identity and the leader identity. Then, using status characteristics theory (Berger, Cohen, & Zelditch, 1972; Berger, Fisek, Norman, & Zelditch, 1977; Webster & Foschi, 1988) I further illustrate the negative consequences of being a female leader. I use group processes research to make the explicit link between the negative expectations for female leaders on group performance through the endorsement of group members. Finally, I utilize innovative research using institutionalization of female leadership to propose a possible solution for improving group performance in groups with female leaders.

FEMALE LEADERSHIP AS ROLE-IDENTITY CONFLICT When women are in leadership positions, they are in a situation in which they must either act in ways inconsistent with being a woman or in ways

271

Conflicting Leader Identity

inconsistent with being a leader (Eagly & Karau, 2002). The general expectations that people in our society hold for women – that they be nurturing, deferential, and concerned with the emotional well-being of others – are not consistent with the general expectations that people in our society hold for group leaders. Leaders are typically expected to be assertive, dominant, and concerned for the task performance of others. Thus, female leadership is an example of an identity conflict. The leader identity tends to be associated with masculine characteristics (Eagly & Karau, 2002). While not specifically on leadership, Schein’s (1973, 1975) research finding ‘‘think manager, think male’’ has been consistently found among male respondents but not among female respondents (Brenner, Tomkiewicz, & Schein, 1989; Schein, Mueller, & Jacobson, 1989).1 This phenomenon has also been found in cross-cultural research by Schein and colleagues (Schein & Davidson, 1993; Schein & Mueller, 1992; Schein, Mueller, Lituchy, & Liu, 1996). Taken together there is a clear pattern of evidence for a discrepancy between how we view women and how we view positions of leadership, such as management positions. In the next section, I draw upon a variant of identity theory, identity control theory, to illustrate how female leadership is a conflicting identity (Burke & Stets, 2009).

Identity Control Theory Identity theory (Stryker & Burke, 2000) conceptualizes the self as comprising multiple identities. Drawing from McCall and Simmons’ (1978) concept of role-identity, identity theorists view identities as the self-definitions people use to describe themselves. Identities are the self-references a person applies to herself as a consequence of the different positions, or statuses, she occupies in society. Identities provide meaning by denoting specific behavioral expectations of a given position. For example, how a female leader views herself in terms of the social expectations for her being a woman and for being a leader comprises her identity as ‘‘female leader.’’ As I will discuss below, sometimes social expectations can involve contradictory behaviors. Identity theory posits a relationship between the social structure and the person, while identity control theory (a variant of identity theory) posits a relationship between the self-meanings a person holds and feedback received from others (Burke & Stets, 2009). If a person receives feedback that is inconsistent with an identity, then the person will adjust her behavior

272

CHRISTABEL L. ROGALIN

accordingly. Since social expectations for women are inconsistent with expectations for leaders, female leaders are likely to find themselves in a predicament. As I will describe later in the chapter, if the expectations associated with different identities conflict, then individuals have difficulty functioning and influencing the behavior of others. In the context of group leadership, I propose that this conflict will have adverse effects on the group’s performance. So far in this chapter I have outlined how female leadership is an example of a conflicting identity, drawing on identity control theory. In the next section I will briefly detail social role theory which describes some of the implications for female leaders for their conflicting identity and then turn to role congruity theory, which specifically addresses the issues that female leaders face because of the inconsistency of societal expectations for being a woman and being a leader.

Role Congruity Theory Drawing on social role theory (Eagly, 1987), role congruity theory (Eagly, 2003; Eagly & Karau, 2002) helps to explain why women are underrepresented in leadership positions. According to social role theory, different social positions (e.g., women and men) and corresponding roles perpetuate the expectations associated with social positions (for women and men). A role consists of the behavioral expectations that are associated with a particular social category. For example, some of the roles that correspond to holding the position of leader are being influential and competent at tasks. At the level of the individual, a role exists as a schema, or an abstract knowledge structure about groups of people. At the level of society a role exists as a shared ideology that is expressed among society members (Eagly, Wood, & Diekman, 2000; Eagly, Wood, & JohannesenSchmidt, 2004). A gender role consists of the behavioral expectations that are associated with a person of a particular gender and represent the characteristics that are believed to differentiate women from men. Making up gender roles are descriptive norms which describe how people are believed to actually be and injunctive norms which describe beliefs about how a certain kind of person ought to be (Cialdini & Trost, 1998). For example, women are not only believed to be concerned with the emotional well-being of others (descriptive norm) but it is also believed that they must be concerned with the emotional well-being of others (injunctive norm). Since injunctive norms contain

Conflicting Leader Identity

273

behaviors that are desirable or considered appropriate, these behaviors are likely to elicit approval from others when enacted and elicit disapproval when behaviors deviate from them. Thus, people engage in gender appropriate behaviors because others expect them to do so and will deliver penalties (or sanctions) to those people engaging in gender inappropriate ways and rewards to those people engaging in gender appropriate ways. Social role theory argues that gender roles impact people’s self-concepts. Research tends to find that, on average, self-concepts are gender-stereotypical (see Spence & Buckner, 2000). Self-concepts help to guide behavior when gender roles become internalized by a person. By internalizing a society’s gender roles, they become the standard by which a person measures her behavior. Research demonstrates that both women and men tend to judge themselves based on the extent to which they conform to gender expectations of their society, findings consistent with identity control theory (Burke, 1991; Burke & Stets, 2009; Stets & Burke, 2005). In their study, Wood, Christensen, Hebl, & Rothgerber (1997) found that women and men view themselves favorably when they behave in ways that conform to gender roles and unfavorably when they behave in ways that deviate from gender roles. Not only do gender roles influence behavior through expectancy confirmation and self-regulation, but they also influence behavior through their coexistence with other, more specific roles. For example, how a female leader behaves is dependent on her role as a woman and on her role as a leader. Social role theory argues that gender roles combine with specific roles in a given situation to help influence behaviors. While gender roles impact behavior in every situation, they have less impact on behavior than specific roles. In the example of a female leader, social role theory argues that the leader role has a greater impact on behavior than the gender role, an argument echoed in status characteristics theory (Berger et al., 1972, 1977; Webster & Foschi, 1988), on which I will elaborate below. As a result of gender roles and the internalization of gender roles, people in the United States have differing beliefs about the typical characteristics of women and men (Broverman, Vogel, Broverman, Clarkson, & Rosenkrantz, 1972; Diekman & Eagly, 2000). Specifically, women are expected to adhere to communal characteristics, while men are expected to adhere to agentic characteristics (Eagly, 1987; Eagly & Karau, 2002). Communal characteristics are associated with a concern for the welfare of others (e.g., being helpful, affectionate, kind, sympathetic, nurturing, and gentle) while agentic characteristics are associated with a tendency to be controlling, assertive, and confident (e.g., aggressiveness, self-sufficiency, ambition, independence,

274

CHRISTABEL L. ROGALIN

forcefulness, and a tendency to behave as a leader) (Eagly & Karau, 2002). That is, leaders tend to embody agentic characteristics.2 Based on these research findings, I offer the following theoretical propositions: Proposition 1. If others perceive a person to hold a female gender identity, then the person’s identity will be associated more with communal traits than a person who others perceive holding a male gender identity. Proposition 2. If others perceive a person to hold a male gender identity then the person’s identity will be associated more with agentic traits than a person who others perceive holding a female gender identity. Proposition 3. If others perceive a person to hold a leader identity, then the person’s identity will be associated more with agentic traits than communal traits. As noted previously, deviations from injunctive norms elicit disapproval (Cialdini & Trost, 1998). Behavior typical of the leader role violates the injunctive norms of the female gender role. For example, assertive and dominant women are thought of negatively yet these behaviors are expected of leaders (Costrich, Feinstein, Kidder, Marecek, & Pascale, 1975; Rudman & Glick, 2001). Thus, female leaders are caught in a catch-22 (Rudman & Glick, 2001). If females behave agentically in the effort to be perceived as qualified for leadership positions, the result is that they are rated as having equal competence ratings to those of agentic men, but they suffer a backlash effect not similarly experienced by men (Glick, Zion, & Nelson, 1988). The backlash effect consists of the negative evaluations of agentic women because they are violating the injunctive norm that women are supposed to be communal – in other words, those women are supposed to be nice (Rudman, 1998; Rudman & Glick, 1999, 2001). In sum, if women behave agentically, as the role of a leader dictates, then they are viewed as competent as men. Yet, they are neither liked nor respected.3 The disjuncture between leadership expectations and injunctive norms of the female gender role results in success being costly for women in terms of social approval. When competent women are compared to competent men, the former tend to be viewed as undesirable as group members (Hagen & Kahn, 1975) and even elicit cues of negative affect (Butler & Geis, 1990; Koch, 2005) from interaction partners. According to role congruity theory, female leaders experience two distinct forms of prejudice due to the role incongruity principle. The role incongruity

Conflicting Leader Identity

275

principle is the discrepancy (or conflict) between female gender role and the leader role (Eagly, 2003; Eagly & Karau, 2002; Ritter & Yoder, 2004). Due to this incongruity, women are perceived as less qualified for leadership than men; thus, there is less access to leadership positions for women. When women do achieve leadership positions, they are more likely to be presented with obstacles to overcome than do men in leadership positions. Several recent studies have demonstrated further empirical support for role congruity theory (cf. Boyce & Herd, 2003; Garcia-Retamero & Lo´pezZafra, 2006; Ritter & Yoder, 2004). On the basis of the research discussed in this section, I offer the following propositions: Proposition 4. If a person is a female leader, then others will perceive a greater discrepancy between the person’s gender identity and leader identity than a male leader. Proposition 5. If group members perceive a greater discrepancy between a person’s gender identity and leader identity, then the person will experience more negative evaluations from members of the group. Based on propositions 4 and 5, I derive the following: Derivation 1. If a person is a female leader, then she will experience more negative evaluations from members of the group than a male in the role of a leader. Previous research supports Derivation 1. Compared to males, females tend to be viewed as less competent (Pugh & Wahrman, 1983; Thomas-Hunt & Phillips, 2004; Wagner et al., 1986; Wood & Karten, 1986) and to have less expertise prior to group discussions (Thomas-Hunt & Phillips, 2004). Thomas-Hunt and Phillips (2004) also found that females were less influential on the group task than males. Females in turn reported less confidence in their ability to influence the group compared to males. Furthermore, female experts were perceived as holding less expertise than nonexpert females, a pattern that was not found among the males. Female experts were also found to be less influential than nonexpert females, while male experts were more influential than nonexpert males. Furthermore, groups had higher levels of performance when they had male experts, compared to groups with female experts – even when comparing female and male experts with the exact same level of expertise. As I described earlier, research also tends to find that, compared to men, women are more negatively evaluated, including negative affective reactions (Butler & Geis,

276

CHRISTABEL L. ROGALIN

1990; Koch, 2005). As I will describe in more detail in the next section describing status characteristics theory (Berger et al., 1972, 1977; Webster & Foschi, 1988) researchers have documented across a variety of contexts and settings that people presume that women are less competent than men and are less worthy to hold positions of leadership (Ridgeway, 2001). I will now describe how these discrepancies between the female gender role (or identity) and the leader role (or identity) affect group outcomes.

FEMALE LEADERSHIP AS A STATUS INCONSISTENCY One explanation for why groups may underperform when led by women can be provided by status characteristics theory. Status characteristics theory is part of the expectation states theoretical research program and documents the status generalization process in task-oriented groups (Webster & Foschi, 1988). This is the process through which observable characteristics, which are referred to as status characteristics, generate a status hierarchy. A status characteristic is any property (e.g., gender) of a person with two or more states that are differentially evaluated in the larger culture within which a task group is embedded. Associated with each state of the characteristic are performance expectations. Performance expectation states are the generalized conceptions of the likelihood of each group member’s abilities as they relate to the group’s task (Berger, Webster, Ridgeway, & Rosenholtz, 1986). These expectation states produce behavioral inequalities among group members due to the different competency expectations associated with the status characteristic – one state leads to higher performance expectations, the other state leads to lower performance expectations. People with higher performance expectations are given more opportunities to perform, they are more positively evaluated when they contribute, and they are more influential in group decision-making than people with lower performance expectations. Since higher status members are given the opportunity to perform more within the group, higher status members tend to contribute more to the task outcome. Thus, a person’s behavior reinforces the differential expectations for each member of the group (Berger et al., 1986). As I will later describe, the ability of a leader to influence group members is critical to group performance and an important mechanism through which a conflicting identity can undermine group performance.

Conflicting Leader Identity

277

Extensive theoretical development and empirical testing of status characteristics theory has demonstrated the impact of status characteristics on group interaction. Status characteristics theory focuses only on task and collectively oriented groups. A group is task-oriented when it has an established goal and where success or failure in achieving the goal is recognized (Berger, Rosenholtz, & Zelditch, 1980). Collective orientation refers to the notion that each group member believes it necessary and legitimate to consider the other group members’ contributions because each member is believed to affect the performance of the overall group. The assertion that gender acts as a status characteristic has been well supported by empirical findings (Meeker & Weitzel-O’Neill, 1977; Pugh & Wahrman, 1983; Ridgeway, 1982; see Wagner & Berger, 1997 for a summary of research, but see Foschi & Lapointe, 2002 for a study in which empirical support was not found4). In the absence of any specific information about ability or competence, men are expected to be more competent and are more influential than women in task groups (Carli, 2001; Lockheed, 1985; Wagner & Berger, 1997; Wood & Karten, 1986). Women are viewed as more competent and are more influential than men only when the task is tagged as feminine (Pugh & Wahrman, 1983). An example of a masculine-tagged task is changing the oil in a car; an example of a feminine-tagged task is sewing (Balkwell & Berger, 1996; Dovidio, Brown, Heltman, Ellyson, & Keating, 1988). These are examples of masculine-and feminine-tagged tasks because they are culturally understood as tasks that men perform more commonly than women (automotive maintenance), or that women perform more commonly than men (sewing). In task groups, both women and men have lower performance expectations for women (Lockheed & Hall, 1976; Meeker & Weitzel-O’Neill, 1977). Furthermore, even when women and men display equivalent levels of performance, women are perceived to have lower task abilities than men (Biernat & Kobrynowicz, 1997; Foschi, 2000). The only instances in which women are viewed as being more competent than men are those in which there is explicit evidence that the woman is significantly more competent than the man (Pugh & Wahrman, 1983; Shackelford, Wood, & Worchel, 1996; Wagner et al., 1986; Wood & Karten, 1986). Since both women and men hold higher performance expectations for men in mixed-sex interactions, men will be more likely than women to (1) contribute to the task, (2) be given opportunities to contribute, (3) receive positive evaluations for their contributions, and (4) influence others in group decisions (Berger et al., 1972, 1977). If the group task is masculine tagged, then gender differences are expected to be even greater.

278

CHRISTABEL L. ROGALIN

On the basis of these insights, I propose the following additions to my theory being outlined in this chapter: Proposition 6. If a person is a female leader, then she will be viewed as less competent by members of the task group than a male leader. Much, if not all, of the research focused on female leadership has investigated the negative consequences for female leaders (Eagly & Karau, 2002). Role congruity theory focuses on the negative consequences for female leaders due to the inconsistency between the female gender role and the leader role. I will now turn my attention to a question that has received far less attention: the impact of a leader’s conflicting identity on group performance. I argue that this is a general theory, in that conflict between the leader identity and any other salient conflicting identity will generate decrements in group performance through the mechanisms described below.

IMPACT OF A LEADER’S CONFLICTING IDENTITY ON GROUP PERFORMANCE Both role congruity theory (Eagly & Karau, 2002) and status characteristics theory (Berger et al., 1980, 1986) hold that female leaders will experience negative consequences. As described previously, role congruity theory posits that female leaders will receive more negative evaluations than male leaders, while status characteristics theory posits that female leaders will be viewed as less competent than male leaders and their contributions will be viewed less positively than the contributions of male leaders. Bass (1990) argues that if group members do not believe that the emerging task leader has task competency, then the leader will fail in her or his attempts to lead the group. Julian and Hollander (1966 as cited by Bass, 1990) found that people within a group are more likely to follow directions of a taskcompetent leader than a leader they view as not having task competence. Furthermore, Hollander and Julian (1970) found that group members were less willing to endorse a person as a leader (or a spokesperson) if they thought the leader was task incompetent than if they thought the leader was task competent. Endorsement is a key mechanism through which group members come to collectively support and adhere to the leader’s efforts to direct the group and focus their efforts on their task. Within the context of authority (i.e., leadership), endorsement refers to individually held beliefs that a person has

Conflicting Leader Identity

279

the right to exercise authority over others (Dornbusch & Scott, 1975; Lucas, 2003). Thus, Proposition 7. If members of a task group view the leader as less competent, then they are less likely to endorse the person as the leader of the group than members of a task group who view their leader as more competent. Based on propositions 6 and 7, I derive the following: Derivation 2. If a person is a female leader, then she will be less likely to be endorsed as the leader of a group than a male leader. Effective leaders keep group members focused on the task, and motivated to perform well on the task through members’ perceptions of their ability to reward contributions of group members (Bass, 1990). This effectiveness is undermined in the absence of endorsement. Thus, Proposition 8. If members of the group are less likely to endorse the person assuming the position of the group leader, then the group will be less task-focused than members of groups who are more likely to endorse the person assuming the position of the group leader. Based on derivation 2 and proposition 8, I derive the following: Derivation 3. If a group has a female leader, then its members will be less task-focused, compared to members of groups with a male leader. It has long been recognized that group performance depends on members’ attention to the task and willingness to contribute to the group’s work on the task (e.g., Steiner, 1972). Proposition 9. If a group’s members are less task-focused, then the group’s performance will be lower, compared to groups whose members are more task-focused. Building on Goldman and Fraas (1965), I conceptualize group performance as a combination of accuracy and the amount of time spent by the group on the task. Thus, I argue that groups having members who are taskfocused and willing to contribute to the task will outperform groups whose members are lacking in those areas. I propose the following hypotheses from derivation 3 and proposition 9: Hypothesis 1. If a group has a female leader, then its work will be less accurate, compared to groups with a male leader.

280

CHRISTABEL L. ROGALIN

Hypothesis 2. If a group’s members are less task-focused, then the group will take more time to accomplish the task, compared to groups whose members are more task-focused. I propose the following hypotheses from derivation 3 and hypothesis 2: Hypothesis 3. If a group has a female leader, then it will take more time to accomplish the task, compared to groups with a male leader. Proposition 9 and hypotheses 1–3 explicate one path through which two conflicting identities (the identity of a leader and identity of a female) affect group performance, as captured by accuracy and time-to-completion of the task. The first path runs through the group members’ task-focus. The second path runs through members’ contributions to the group task: Hypothesis 4. If a group’s members contribute less to the group’s task, then the group’s work will be less accurate, compared to groups whose members contribute more to the group’s task. Hypothesis 5. If a group’s members contribute less to the group’s task, then the group will take more time to accomplish the task, compared to groups whose members contribute more to the group’s task. Hypothesis 1 and 3 are also realized through derivation 4 and hypothesis 4, and derivation 4 and hypothesis 5, respectively. Yet, it is important to note that this represents an alternative path to group performance. That is, an important contribution of my theory is that it clearly explicates the different processes through which a conflicting leader identity affects group performance. To date, such fine-grained theorizing is absent in research on leadership and group dynamics.

Intervening in the Status Generalization Process Is it possible to intervene in the status generalization process? Status characteristics theory documents the process by which differences in status characteristics give rise to differences in performance expectations (Webster & Foschi, 1988). Numerous theoretical demonstrations have shown how it is possible to intervene in the status generalization process (see Berger et al., 1972; Freese & Cohen, 1973; Webster, 1977) and many studies empirically demonstrate this possibility (Cohen, 1972, 1982; Cohen & Lotan, 1995; Cohen & Roper, 1992; Markovsky, Smith, & Berger, 1984; Pugh & Wahrman, 1983).

Conflicting Leader Identity

281

A key finding of the work by Cohen and Roper (1992) is that effective status interventions require change in performance expectations of lower status people in the eyes of lower status people and higher status people. It is insufficient simply to change the lower status person’s self-expectations. As noted above, as evidenced in work on second-order expectations, the expectations other people hold for us influence both their behaviors toward us and our behaviors toward them (Troyer & Younts, 1997; Troyer, Younts, & Kalkhoff, 2001; Webster & Whitmeyer, 1999, 2002). Our behavior is further influenced by the expectations we believe others hold for us as well. This insight – that the expectations others hold for our behavior – will be brought up again in the next section on legitimation and institutionalization. In the next section, I turn to recent theory and research that reveals a mechanism, institutionalization of female leadership, which I believe can intervene in the processes linking a leader’s conflicting identity to adverse group performance.

PROPOSED STRATEGY FOR INCREASING THE EFFECTIVENESS OF FEMALE LEADERSHIP Yoder (2001) points out that many of the strategies effectively employed by male leaders, such as sitting at the head of the table (Porter, Geis, & Jennings, 1983) and behaving assertively (Butler & Geis, 1990; Carli, 1990, 2001) are not as effective for women. She further argues that since power and status are conflated with gender, female leaders do not have an equal footing when compared to male leaders. For women to achieve equal footing women’s status must be enhanced, which would reduce the status difference between women and men. Yoder (2001) makes recommendations on two levels, individual and organizational. At the individual level, she draws on work by Ridgeway (1982, 1988) to argue that women must appear group-oriented (i.e., communal) instead of self-oriented (i.e., agentic). As I will elaborate later in this work, Ridgeway’s insight helps to resolve the problem of legitimacy that female leaders face. At the organizational level, Yoder (2001) suggests that women need to appear to be exceptionally competent – simply being as competent as men is not enough (e.g., Pugh & Wahrman, 1983). Drawing on research by Brown and Geis (1984), DeMatteo, Dobbins, Myers, and Facteau (1996), and Offerman and Beil (1992), Yoder (2001) suggests that at the organizational level, female leaders need to be legitimated.

282

CHRISTABEL L. ROGALIN

Legitimation Legitimacy is the process by which a power and prestige order comes to be viewed as the way things should be. When a status hierarchy in a group is legitimated, performance expectations associated with each person present in the situation take on a moral quality (Ridgeway & Berger, 1986, 1988; Wagner & Berger, 2002). When people behave in ways inconsistent with their associated performance expectations, their behaviors are perceived to be both unexpected and inappropriate. The legitimacy of the leader may mediate the relationship between a leader’s conflicting identity and group performance. Status characteristics theory argues that it is illegitimate for lower status people to behave too assertively (Meeker & Weitzel-O’Neill, 1977; Ridgeway, 1982; Ridgeway & Berger, 1986, 1988). If a lower status person behaves too assertively, there are negative sanctions (Youngreen & Moore, 2008). Displaying competitive or dominant behavior as a lower status person can be viewed as an attempt to gain status and influence; therefore, this type of behavior is perceived as inappropriate for lower status people (e.g., women displaying competitive or dominant behavior in a mixed-sex group). Ridgeway and Berger (1986, 1988) argue that the negative consequences experienced by people of lower status are due to the nonlegitimated nature of their positions. Even if women obtain leadership positions within a task group, they are still perceived to have lower status than men due to the referential beliefs held by members of the group. Referential beliefs are widely shared cultural beliefs about the kinds of people who hold higher versus lower status positions within the larger society and the awards associated with these corresponding positions. For example, in our culture, we are prone to associate men rather than women with higher status positions. When referential beliefs are activated in a task group, they are used by group members to form status expectations. The legitimation assumption of status characteristics theory states that the stronger the status expectations (which are dependent on the number of activated referential beliefs), the greater the likelihood that group members will treat the existing status structure as legitimate (Ridgeway, Johnson, & Diekema, 1994). The stronger the status expectations, the more people will behave as if the existing power and prestige order is the way that things should be. Legitimacy arises within the group, while it is dependent on the larger cultural beliefs outside the group. Meeker and Weitzel-O’Neill (1977) suggested that lower status people may overcome their status disadvantage if they appear to be motivated by a

Conflicting Leader Identity

283

desire to help out the group rather than motivated to help out themselves. Drawing on this idea, Ridgeway (1978, 1982) argued that a person’s motivation affects status by affecting a group’s willingness to accept that person’s task contributions. Only contributions that are accepted by fellow group members will contribute to a person’s status standing within the group. The value of a contribution is dependent upon the perceived motivation of the contribution. Perceived motivation by fellow group members is what is important in this process, not the true motivation of the person. People appearing group-oriented are more likely to have their contributions accepted by the group, compared to people appearing selforiented. However, the extent to which motivation truly impacts status attainment is dependent upon the salient status characteristics of that person. Contributions made by higher status people are viewed as legitimate; thus, the motivations behind their contributions are not questioned and their contributions are accepted by the group. Contributions made by lower status people are viewed as illegitimate – as questioning the existing power and prestige order within the group – which causes other members of the group to question the motivations behind the contributions. Consequently, lower status people are more likely to be accepted by the group by appearing group-oriented rather than self-oriented. To counteract this problem of legitimacy, Ridgeway (1978, 1982) suggested that female leaders need to appear group-oriented rather than self-oriented to be influential within the group. If women are able to frame their attempts at influence as being group-oriented rather than self-oriented, they can minimize the problem of legitimation.5 Yoder et al. (1998) conducted a study on female leadership where they varied the basis of leadership: (1) appointed, (2) trained, and (3) trained credibly. In both the trained and the trained credibly conditions, the female was trained with task-relevant information, but it was only in the trained credibly conditions that the followers were made aware of her knowledge. Yoder and colleagues argued that making followers aware of the leader’s training would validate and legitimize the leader’s expertise. Groups led by appointed female leaders took significantly longer than trained or trained credibly female leaders to complete the task. Furthermore, groups lead by credibly trained leaders had better group outcomes than groups lead by appointed or trained female leaders. What these studies demonstrate is that competency alone is not sufficient to make female leaders effective; something must be done at an institutional level to make female leaders effective.

284

CHRISTABEL L. ROGALIN

Institutionalization Turning to the institutional level, Lucas (2003) found that institutionalization of female leadership minimized the gender gap in perceptions of the competence of male and female leaders. Institutionalization is the process by which ‘‘social processes, obligations, or actualities come to take on a rulelike status in social thought and action’’ (Meyer & Rowan, 1977, p. 341). In his experiment, Lucas institutionalized female leadership by having the research participants watch a video which demonstrated for the specific task in which they were about to engage that it was legitimate and proper for women to hold leadership positions. The findings by Lucas (2003) suggest that it may be necessary to institutionalize female leadership in order to obviate the gender difference in perceptions of competence. Based on Lucas’ work, I propose the following: Proposition 10. If a group’s members perceive female leadership as institutionalized, then they will view a female leader as more competent than groups whose members do not perceive female leadership as institutionalized. Proposition 11. If a group’s members perceive female leadership as institutionalized, then they will be more likely to view a female leader to be as competent as a male leader, compared to groups whose members do not perceive female leadership as institutionalized. While Lucas (2003) focused on how institutionalization affected the group leader, the focus of the theory presented here is on group performance. To summarize, effective leaders are those leaders who are viewed as competent. Competent leaders are endorsed by group members, allowing them to keep members focused on the task and secure their contributions to the task. Based on propositions 10, 7, and 8 and hypothesis 2, I propose the following hypothesis: Hypothesis 6. If a group’s members perceive female leadership as institutionalized, then a female-led group will take less time to accomplish the task of the group, compared to female-led groups whose members do not perceive female leadership as institutionalized. Based on propositions 10, 7, 8, and 9, I propose the following hypothesis: Hypothesis 7. If a group’s members perceive female leadership as institutionalized, then a female-led group’s task work will be more accurate, compared to female-led groups whose members do not perceive female leadership as institutionalized.

285

Conflicting Leader Identity

Based on propositions 11, 7, and 8 and hypothesis 2, I propose the following hypothesis: Hypothesis 8. If a group’s members perceive female leadership as institutionalized, then a female-led group will take the same amount of time to accomplish the task of the group as a male-led group. Based on propositions 11, 7, and 9, I propose the following hypothesis: Hypothesis 9. If a group’s members perceive female leadership as institutionalized, then a female-led group’s task accuracy will be comparable to the task accuracy of a male-led group.

CONCLUSIONS The purpose of this chapter has been to investigate the circumstances under which groups with female leadership succeed. This project draws from the sociological theories of identity control theory (Burke & Stets, 2009), status characteristics theory (Berger et al., 1980), and work on organizational authority (Dornbusch & Scott, 1975) and from the psychological theory of role congruity theory (Eagly & Karau, 2002). As discussed above, previous work has theoretically and empirically addressed the effect of statuses and characteristics of a group leader on the perceptions of the leader by group members. This work leads to questions about the process by which a leader’s conflicting leader identity impacts group performance and how this process might be circumvented by institutionalizing female leadership. My work provides several answers to these questions. First, I have made explicit the link between status, identity processes, and group outcomes. Other research within these two areas tends to focus only on individuals. For example, status research tends to examine the relative influence of group members and also how they are evaluated vis-a`-vis one another. I have taken these ideas another step to examine how these processes affect group outcomes. A second contribution is a new general theory of how conflicting identities held by a leader can undermine group performance. While this work is focused on the empirical example of female leadership, this theory can also be used to examine racial and ethnic identities, cross-cultural identities and disabilities as they relate to leadership outcomes. My theory outlines the mechanisms by which a leader’s identity conflict affects group performance, with particular application to cases of female

286

CHRISTABEL L. ROGALIN

leaders. This theory could be used to increase women’s success at being leaders. Furthermore, it provides a number of specific derivations that are primed for hypothesis formulation and empirical testing. Given that in 2012, only 14.3% of Fortune 500 CEOs were women (Catalyst 2012), the time is now.

NOTES 1. For other research providing support of gender-typing of management positions, see Duehr and Bono (2006); Martell, Parker, Emrich, and Crawford (1998); and Powell, Butterfield, and Parent (2002). 2. Research by Wood et al. (1997) supports the assertion that people tend to agree that men and women both should and actually do behave in ways corresponding to these characteristics. 3. This is not to say that men do not experience a backlash effect. If men violate the injunctive norm that they are to be agentic (and not communal), then they experience a backlash effect, in that communal men are viewed as nice but less competent (and employable) than agentic men (Rudman & Glick, 1999, 2001). 4. It is important to note that the findings of Foschi and Lapointe (2002) do not disprove status characteristics theory. The arguments of status characteristics theory only hold if a characteristic held by group members is a status characteristic. The research by Foschi and Lapointe is only suggestive of a change in our cultural expectations for men and women, which has also been documented by other researchers (Diekman & Eagly, 2000). 5. A reviewer pointed out a potential tension between Ridgeway’s (1978, 1982) suggestion that women who appear group oriented can minimize the problem of legitimation and Derivation 2 which suggests that group members are less likely to endorse a female leader. The idea that there is a tension between what the leader desires and what the group feels is a theoretical (and empirical) idea that warrants further inquiry, which is outside the scope of this paper.

ACKNOWLEDGMENTS I would like to thank Jane Brooks, Cheryl DeLeon, Laura Hirshfield, Shirley A. Keeton, Jeffrey L. Lucas, Michael J. Lovaglia, Barry Markovsky, Carla Pfeffer, Lisa Troyer, Reef Youngreen and the anonymous reviewer at Advances in Group Processes for comments on earlier versions of this paper. This research was supported by grants from the National Science Foundation to Christabel L. Rogalin and Lisa Troyer.

287

Conflicting Leader Identity

REFERENCES Balkwell, J. W., & Berger, J. (1996). Gender, status, and behavior in task situations. Social Psychology Quarterly, 59, 273–283. Bass, B. M. (1990). Bass & Stogdill’s handbook of leadership: Theory, research, and managerial applications. New York, NY: The Free Press. Berger, J., Cohen, B. P., & Zelditch, M., Jr. (1972). Status characteristics and social interaction. American Sociological Review, 37, 241–255. Berger, J., Rosenholtz, S. J., & Zelditch, M., Jr. (1980). Status organizing processes. Annual Review of Sociology, 6, 479–508. Berger, J., Webster, M., Jr., Ridgeway, C. L., & Rosenholtz, S. J. (1986). Status cues, expectations, and behavior. Advances in Group Processes, 3, 1–22. Berger, J. M., Fisek, M. H., Norman, R. Z., & Zelditch, M., Jr. (1977). Status characteristics and social interaction: An expectation-states approach. New York, NY: Elsevier. Biernat, M., & Kobrynowicz, D. (1997). Gender- and race-based standards of competence: Lower minimum standards but higher ability standards for devalued groups. Journal of Personality and Social Psychology, 72, 544–557. Boyce, L. A., & Herd, A. M. (2003). The relationship between gender role stereotypes and requisite military leadership characteristics. Sex Roles, 49, 365–378. doi:10.1023/ A:1025164221364 Brenner, O. C., Tomkiewicz, J., & Schein, V. E. (1989). The relationship between sex role stereotypes and requisite management characteristics revisited. Academy of Management Journal, 32, 662–669. Broverman, I. K., Vogel, S. R., Broverman, D. M., Clarkson, F. E., & Rosenkrantz, P. S. (1972). Sex-role stereotypes: A current appraisal. Journal of Social Issues, 28, 59–78. Brown, V., & Geis, F. L. (1984). Turning lead into gold: Evaluations of men and women leaders and the alchemy of social consensus. Journal of Personality and Social Psychology, 46, 811–814. Burke, P. J. (1991). Identity processes and social stress. American Sociological Review, 56, 836–849. Burke, P. J., & Stets, J. E. (2009). Identity theory. New York, NY: Oxford University Press. Butler, D., & Geis, F. L. (1990). Nonverbal affect responses to male and female leaders: Implications for leadership evaluations. Journal of Personality and Social Psychology, 16, 358–368. Carli, L. L. (1990). Gender, language, and influence. Journal of Personality and Social Psychology, 59, 941–951. Carli, L. L. (2001). Gender and social influence. Journal of Social Issues, 57, 725–741. doi:10.1111/0022-4537.00238 Catalyst. (2007). Census 2006: Women corporate officers. New York, NY: Catalyst. Catalyst. (2012). 2012 Catalyst census: Fortune 500 women executive officers and top earners. New York, NY: Catalyst. Retrieved from http://www.catalyst.org/knowledge/2012catalyst-census-fortune-500-women-executive-officers-and-top-earners Cialdini, R. B., & Trost, M. R. (1998). Social influence: Social norms, conformity, and compliance. In D. T. Gilbert, S. T. Fiske & G. Lindzey (Eds.), The handbook of social psychology (Vol. 2, pp. 151–192). Boston, MA: McGraw Hill. Cohen, E. G. (1972). Interracial interaction disability. Human Relations, 37, 648–655.

288

CHRISTABEL L. ROGALIN

Cohen, E. G. (1982). Expectation states and interracial interaction in school settings. Annual Review of Sociology, 8, 209–235. Cohen, E. G., & Lotan, R. A. (1995). Producing equal-status interaction in the heterogeneous classroom. American Educational Research Journal, 32, 99–120. Cohen, E. G., & Roper, S. S. (1992). Modification of interracial interaction disability: An application of status characteristics theory. American Sociological Review, 37, 643–657. Costrich, N., Feinstein, J., Kidder, L., Marecek, J., & Pascale, L. (1975). When stereotypes hurt: Three studies of penalties for sex-role reversals. Journal of Experimental Social Psychology, 11, 520–530. DeMatteo, J. S., Dobbins, G. H., Myers, S. D., & Facteau, C. F. (1996). Evaluations of leadership in preferential and merit-based leadership selection situations. Leadership Quarterly, 7, 41–62. Diekman, A. B., & Eagly, A. H. (2000). Stereotypes as dynamic constructs: Women and men of the past, present, and future. Personality and Social Psychology Bulletin, 26, 1171–1188. doi:10.1177/0146167200262001 Dornbusch, S., & Scott, W. R. (1975). Evaluation and the exercise of authority. San Francisco, CA: Jossey-Bass. Dovidio, J. F., Brown, C. E., Heltman, K., Ellyson, S. L., & Keating, C. F. (1988). Power displays between women and men in discussions of gender-linked tasks: A multichannel study. Journal of Personality and Social Psychology, 55, 580–587. Duehr, E. E., & Bono, J. E. (2006). Men, women, and managers: Are stereotypes finally changing? Personnel Psychology, 59, 815–846. doi:10.1111/j.1744-6570.2006.00055.x Eagly, A. H. (1987). Sex differences in social behavior: A social-role interpretation. Hillsdale, NJ: Erlbaum. Eagly, A. H. (2003). Few women at the top: How role incongruity produces prejudice and the glass ceiling. In D. van Knippenberg & M. A. Hogg (Eds.), Leadership and power: Identity processes in groups and organizations (pp. 79–93). London: Sage Publications. Eagly, A. H., & Karau, S. (2002). Role congruity theory of prejudice toward female leaders. Psychological Review, 109, 573–598. doi:10.1037//0033-295X.109.3.573 Eagly, A. H., Wood, W., & Diekman, A. B. (2000). Social role theory of sex differences and similarities: A current appraisal. In T. Eckes & H. M. Trautner (Eds.), The developmental social psychology of gender (pp. 123–173). Mahwah, NJ: Erlbaum. Eagly, A. H., Wood, W., & Johannesen-Schmidt, M. C. (2004). Social role theory of sex differences and similarities: Implications for the partner preferences of women and men. In A. H. Eagly, A. E. Beall & R. J. Sternberg (Eds.), The psychology of gender (pp. 269–295). New York, NY: Guilford Press. Foschi, M. (2000). Double standards for competence: Theory and research. Annual Review of Sociology, 26, 21–42. Foschi, M., & Lapointe, V. (2002). On conditional hypotheses and gender as a status characteristic. Social Psychology Quarterly, 65, 146–162. Freese, L., & Cohen, B. P. (1973). Eliminating status generalization. Sociometry, 2, 177–193. Garcia-Retamero, R., & Lo´pez-Zafra, E. (2006). Prejudice against women in male-congenial environments: Perceptions of gender role congruity in leadership. Sex Roles, 55, 51–61. doi:10.1007/s11199-006-9068-1 Glick, P., Zion, C., & Nelson, C. (1988). What mediates sex discrimination in hiring decisions? Journal of Personality and Social Psychology, 55, 178–186.

Conflicting Leader Identity

289

Goldman, M., & Fraas, L. A. (1965). The effects of leader selection on group performance. Sociometry, 28, 82–88. Hagen, R. L., & Kahn, A. (1975). Discrimination against competent women. Journal of Applied Social Psychology, 5, 362–376. Hollander, E. P., & Julian, J. W. (1970). Studies in leader legitimacy, influence, and innovation. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 5, pp. 33–69). New York, NY: Academic Press. Koch, S. (2005). Evaluative affect display toward male and female leaders of task-oriented groups. Small Group Research, 36, 678–703. doi:10.1177/1046496405281768 Lockheed, M. E. (1985). Sex and social influence: A meta-analysis guided by theory. In J. Berger & M. Zelditch, Jr. (Eds.), Status, rewards, and influence: How expectations organize behavior (pp. 406–429). San Francisco, CA: Jossey-Bass. Lockheed, M. E., & Hall, K. P. (1976). Conceptualizing sex as a status characteristic: Applications to leadership training strategies. Journal of Social Issues, 32, 111–124. Lucas, J. W. (2003). Status processes and the institutionalization of women as leaders. American Sociological Review, 68, 464–480. Markovsky, B., Smith, L. F., & Berger, J. (1984). Do status interventions persist? American Sociological Review, 49, 373–382. Martell, R. F., Parker, C., Emrich, C. G., & Crawford, M. S. (1998). Sex stereotyping in the executive suite: ‘‘Much ado about something’’. Journal of Social Behavior and Personality, 13, 127–138. McCall, G. J., & Simmons, J. L. (1978). Identities and interactions. New York, NY: Free Press. Meeker, B. F., & Weitzel-O’Neill, P. A. (1977). Sex roles and interpersonal behavior in taskoriented groups. American Sociological Review, 42, 91–105. Meyer, J. W., & Rowan, B. (1977). Institutionalized organizations: Formal structure as myth and ceremony. American Journal of Sociology, 83, 340–363. Offerman, L. R., & Beil, C. (1992). Achievement styles of women leaders and their peers: Toward an understanding of women and leadership. Psychology of Women Quarterly, 16, 37–56. Porter, N., Geis, F. L., & Jennings, J. (1983). Are women invisible as leaders? Sex Roles, 9, 1035–1049. Powell, G. N., Butterfield, D. A., & Parent, J. D. (2002). Gender and managerial stereotypes: Have the times changed? Journal of Management, 28, 177–193. doi:10.1177/014920 630202800203 Pugh, M. D., & Wahrman, R. (1983). Neutralizing sexism in mixed-sex groups: Do women have to be better than men. American Journal of Sociology, 88, 746–762. Ridgeway, C. L. (1978). Conformity, group-oriented motivation, and status attainment in small groups. Social Psychology, 41, 175–188. Ridgeway, C. L. (1982). Status in groups: The importance of motivation. American Sociological Review, 47, 76–88. Ridgeway, C. L. (1988). Gender differences in task groups: A status and legitimacy account. In M. Webster, Jr. & M. Foschi (Eds.), Status generalization: New theory and research (pp. 188–206). Stanford, CA: Stanford University Press. Ridgeway, C. L. (2001). Gender, status, and leadership. Journal of Social Issues, 57, 637–655. doi:10.1111/0022-4537.00233 Ridgeway, C. L., & Berger, J. (1986). Expectations, legitimation, and dominance behavior in task groups. American Sociological Review, 51, 603–617.

290

CHRISTABEL L. ROGALIN

Ridgeway, C. L., & Berger, J. (1988). The legitimation of power and prestige orders in task groups. In M. Webster, Jr. & M. Foschi (Eds.), Status generalization: New theory and research (pp. 207–231). Stanford, CA: Stanford University Press. Ridgeway, C. L., Johnson, C., & Diekema, D. (1994). External status, legitimacy, and compliance in male and female groups. Social Forces, 72, 1051–1077. Ritter, B. A., & Yoder, J. D. (2004). Gender differences in leader emergence persist even for dominant women: An updated confirmation of role congruity theory. Psychology of Women Quarterly, 28, 187–193. doi:10.1111/j.1471-6402.2004.00135.x Rudman, L. A. (1998). Self-promotion as a risk factor for women: The costs and benefits of counterstereotypical impression management. Journal of Personality and Social Psychology, 74, 629–645. Rudman, L. A., & Glick, P. (1999). Feminized management and backlash toward agentic women: The hidden costs to women of a kindler, gentler image of middle managers. Journal of Personality and Social Psychology, 77, 1004–1010. Rudman, L. A., & Glick, P. (2001). Prescriptive gender stereotypes and backlash toward agentic women. Journal of Social Issues, 57, 743–762. doi:10.1111/0022-4537.00239 Schein, V. E. (1973). The relationship between sex role stereotypes and requisite management characteristics. Journal of Applied Psychology, 57, 95–100. Schein, V. E. (1975). The relationship between sex role stereotypes and requisite management characteristics among female managers. Journal of Applied Psychology, 60, 340–344. Schein, V. E. (2001). A global look at psychological barriers to women’s progress in management. Journal of Social Issues, 57, 675–688. doi:10.1111/0022-4537.00235 Schein, V. E., & Davidson, M. J. (1993). Think manager, think male. Management Development Review, 6, 24–28. Schein, V. E., & Mueller, R. (1992). Sex role stereotyping and requisite management characteristics: A cross cultural look. Journal of Organizational Behavior, 13, 439–447. Schein, V. E., Mueller, R., & Jacobson, C. (1989). The relationship between sex role stereotypes and requisite management characteristics among college students. Sex Roles, 20, 103–110. Schein, V. E., Mueller, R. R., Lituchy, T., & Liu, J. (1996). Think manager-think male: A global phenomenon? Journal of Organizational Behavior, 17, 33–41. Shackelford, S., Wood, W., & Worchel, S. (1996). Behavioral styles and the influence of women in mixed-sex groups. Social Psychology Quarterly, 59, 284–293. Spence, J. T., & Buckner, C. E. (2000). Instrumental and expressive traits, trait stereotypes and sexist attitudes: What do they signify? Psychology of Women Quarterly, 24, 44–62. doi:10.1111/j.1471-6402.2000.tb01021.x Steiner, I. D. (1972). Group process and productivity. New York, NY: Academic Press. Stets, J. E., & Burke, P. J. (2005). New Directions in Identity Control Theory. Advances in Group Processes, 22, 43–64. Stryker, S., & Burke, P. J. (2000). The past, present, and future of an identity theory. Social Psychology Quarterly, 63, 284–297. Thomas-Hunt, M. C., & Phillips, K. W. (2004). When what you know is not enough: Expertise and gender dynamics in task groups. Personality and Social Psychology Bulletin, 30, 1585–1598. doi:10.1177/0146167204271186 Troyer, L., & Younts, C. W. (1997). Whose expectations matter? The relative power of firstorder and second-order expectations in determining social influence. American Journal of Sociology, 103, 692–732.

Conflicting Leader Identity

291

Troyer, L., Younts, C. W., & Kalkhoff, W. (2001). Clarifying the theory of second-order expectations: The correspondence between motives for interaction and actors’ orientation toward group interaction. Social Psychology Quarterly, 64, 128–145. Wagner, D. G., & Berger, J. (1997). Gender and interpersonal task behaviors: Status expectation accounts. Sociological Perspectives, 40, 1–32. Wagner, D. G., & Berger, J. (2002). Expectation states theory: An evolving research program. In J. Berger & M. Zelditch, Jr. (Eds.), New directions in contemporary sociological theory (pp. 41–76). Lanham, MD: Rowman & Littlefield Publishers, Inc. Wagner, D. G., Ford, R. S., & Ford, T. W. (1986). Can gender inequalities be reduced? American Sociological Review, 51, 47–61. Webster, M., Jr. (1977). Equating characteristics and social interaction: Two experiments. Sociometry, 40, 41–50. Webster, M., Jr., & Foschi, M. (Eds.). (1988). Overview of status generalization. In Status generalization: New theory and research (pp. 1–20). Stanford, CA: Stanford University Press. Webster, M., Jr., & Whitmeyer, J. M. (1999). A theory of second order expectations and behavior. Social Psychology Quarterly, 62, 17–31. Webster, M., Jr., & Whitmeyer, J. M. (2002). Modeling second-order expectations. Sociological Theory, 20, 306–327. Wood, W., Christensen, P. N., Hebl, M. R., & Rothgerber, M. (1997). Conformity to sex-typed norms, affect, and the self-concept. Journal of Personality and Social Psychology, 73, 523–535. Wood, W., & Karten, S. J. (1986). Sex differences in interaction style as a product of perceived sex differences in competence. Journal of Personality and Social Psychology, 50, 341–347. Yoder, J. D. (2001). Making leadership work more effectively for women. Journal of Social Issues, 57, 815–828. doi:10.1111/0022-4537.00243 Yoder, J. D., Schleicher, T. L., & McDonald, T. W. (1998). Empowering token women leaders: The importance of organizationally legitimated credibility. Psychology of Women Quarterly, 23, 209–222. Youngreen, R., & Moore, C. D. (2008). The effects of status violations on hierarchy and influence in groups. Small Group Research, 39, 569–587. doi:10.1177/1046496408320120

THE IVY AND THE TRELLIS: AGENCY, BIOLOGY, AND SOCIALIZATION David Peterson ABSTRACT Purpose – This chapter presents a theory of socialization that explains cultural transmission while balancing both biological aspects of development and the child’s agency and creativity. Design/methodology/approach – This chapter presents a synthesis of research in sociological theory, developmental psychology, and neuroscience. It is roughly divided into two complementary sections based around the metaphor of ivy growing upon a trellis. The discussion centered around the ‘‘ivy’’ utilizes psychological and neuroscience research to explain how early learning is guided by significant others. The ‘‘trellis’’ section synthesizes literature in developmental psychology and social theory to explain how the child’s experience is enframed both cognitively and emotionally in ways that guide the child into appropriate forms of action and feeling. Finally, I discuss how this model can explain other forms of socialization. Findings – I propose that the child’s innate capacities and motivations are enframed through significant relationships in order to direct the child’s emergent behavior into sequences of competent action. Isolated Advances in Group Processes: Thirtieth Anniversary Edition Advances in Group Processes, Volume 30, 293–315 Copyright r 2013 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 0882-6145/doi:10.1108/S0882-6145(2013)0000030014

293

294

DAVID PETERSON

competencies are guided into simple and delimited domains of social activity like games and, later, more complex and interpretive structures like paradigms and ideologies. Originality/value – This chapter synthesizes research in several literatures in order to develop a new theory that addresses some old questions regarding cultural transmission. Additionally, it represents another step in showing how sociology can integrate research from biological fields without deferring to them. Keywords: Socialization; childhood development; biological sociology; competency; scaffolding

The Parsonian (Parsons & Bales, 1955) theory of socialization in which the child internalizes norms and values has long been impossible to justify in the face of psychological evidence that suggests culture is actually much more fragmented and variably reproduced than previously thought (DiMaggio, 1997). Moreover, sociological studies of children have shown that children are actively and critically engaged in their own socialization (Corsaro, 2005; Thorne, 1993) and are not just simply the passive recipient of adult culture. In place of cultural holism and passive social reproduction, models like Swidler’s (1986) ‘‘toolkit’’ theory suggest that we learn more culture than we use and are selective about which cultural tools we employ. However, it is equally clear that our toolkits do not contain completely atomistic cultural schemata that can be combined and utilized with total freedom. Instead, many groupings of strategies and skills have a relatively stable structure. The formation of these meso-level cognitive structures is not yet understood. In an article for The Annual Review of Sociology, DiMaggio (1997) has argued that Perhaps the highest priority for students of culture and cognition is to develop models of thematization, by which I mean the ways in which diverse schemata aggregate to more general and sociologically interesting constructs like thought styles, stories, logics, paradigms, and ideologies. (p. 278)

Such constructs are neither the totalizing, oppressive norm structure of the Parsonian straw man nor the freely chosen, constructed, and abandoned tools of the unattached agent. Instead, they are emotionally charged vehicles of interpretation and expression that provide a common structure for social interaction. They are the outcome of a process of socialization which draws the learner into preexisting frames of interaction.

295

Agency, Biology, and Socialization

The socialization of children represents an ideal medium to uncover how and why schemata aggregate for three reasons. First, there has been a tremendous amount of work in both developmental psychology and neurology which is helpful in making more scientifically robust statements about the learning process. Bergesen has used such findings to launch incisive critiques against both Durkheim’s theory of mind (2004a) and Mead’s theory of language development (2004b), yet there has been very little work unpacking the positive implications of these findings on learning and socialization. Second, the development of cultural constructs in children results in knowledge that is widely shared but that is not universal. These constructs develop through individual experience, yet they are collective. Thus, neither biological nor psychological processes can account fully for their development. Third, as Freud (1966) argued, the interpretation of psychological phenomena is easier in children since they have less of the cultural ‘‘distortion’’ that complicates adult psyches. The predictability of children’s development allows the formal elements of socialization to be abstracted. The child’s relationship to culture can be loosely compared to ivy crawling up a trellis. The ivy, in a sense, consumes the structure, uses it to support its movements, and becomes defined in the way it winds across its bars. Each piece of the lattice lends support to the ivy’s next move upward. While its exact trajectory is unpredictable, the general direction of growth is assured as long as the structure provides continued support. While this is a simplistic metaphor, it illustrates the dualism of development. The child is not ‘‘instilled’’ with her culture. She takes it up as she finds it useful. But as the child ‘‘grows up,’’ social institutions ranging from parents to schools responsively adjust to her level of competency. They frame her experience, promoting new forms of growth. This essay is divided into two sections. The first deals with the ‘‘ivy,’’ the child’s learning mechanisms and agency. The second is concerned with the ‘‘trellis,’’ the way the child’s experience is structured in order to promote typical competencies.

THE IVY Piaget was perhaps the first developmental theorist to recognize the centrality of children’s agency to their cognitive development (Miller, 2002). The title of his book, To understand is to invent (1974), underscores the essentially creative nature of learning. This activity on the part of the child

296

DAVID PETERSON

helps explain why even rigid socialization only imperfectly transmits cultural schemata. Any theory of socialization must account for the fact that the child is an active agent in the process and not merely a passive vessel for society’s rules, norms, and values. Emirbayer and Mische (1998) reviewed the diverse uses of ‘‘agency’’ in fields ranging from rational choice to feminist theory. Although they found that agency had been used in a wide variety of ways, the authors developed a synthesis of these heterogenrous accounts. Following their lead, I conceive of agency as having three dimensions with respective time orientations. The iterational (past) deals with the schematization of past experience, the practical–evaluative (present) deals with moment-by-moment judgments, and the projective (future) deals with plans of action. Within their tripartite division, however, they give special standing to the past-oriented, iterational dimension: The concept of iteration is crucial for our conception of agency since we maintain that both the projective and practical–evaluative dimensions are deeply grounded in habitual, unreflected, and mostly unproblematic patterns of action by means of which we orient our efforts in the greater part of our daily lives. (1998, p. 975)

But what happens to agency when the actor has no past? No experience means no schemata and, thus, no system of value or way to project action. They have nothing at their feet but an empty ‘‘toolkit.’’ Of course, this cannot be the end of the story. Instead, it points back to the beginning, to the genesis of practical–evaluative and projective agency in the child that takes the form of universal capacities, mechanisms, and motivations. In fact, one might say without exaggeration, that what the infant begins with are the skills to absorb cultural skills. This is, of course, a bit vague. Given the profusion of work in developmental psychology and neurology we can now be much more specific in regards to the child’s innate capacities and primal motivations. The infant’s native ability to distinguish objects (Bloom, 2000; Spelke, 1990) and simple events (Gleitman, 1990; Nelson, 1986, 1996) provides the basis for what Goldman has called our ‘‘basic, intuitive ontology’’ (2006, p. 178), the universal aspects of perception that are the result of our common evolutionary heritage. However, from birth, and perhaps even earlier (Elman et al., 1996), we begin the lifelong process of organizing and reorganizing these fundamental schemata. Once infants gain enough strength and coordination to control their movements, they begin actively exploring their environment. An object interesting to the infant absorbs her and becomes absorbed by her. It is investigated, probed, turned around until its

Agency, Biology, and Socialization

297

fascination fades. The child has taken in the logic, at least what little of it is important to her, and its luminescence diminishes. It becomes another part of the world that radiates with the soft inner light of familiarity. The child comes to ‘‘know’’ the block, knows how it will act when pushed over the crib, and knows what it tastes like. And, in knowing it, it can fade into the background of newly transfigured streams of action. Of course, the child does not yet know that the block is made of ‘‘wood’’ and has a ‘‘letter’’ on each side. Nor does she know that the block is ‘‘cheaply made.’’ As she develops and the sphere of her activity expands, the block may be ‘‘discovered’’ again and again as an emerging social context makes it newly relevant. This phenomenological account of learning is supported by psychological research that has demonstrated an important division between two types of cognition. According to dual-process models of cognition, the movement from fascination to familiarity, from ‘‘deliberative cognition’’ to ‘‘automatic cognition’’ (DiMaggio, 1997, p. 269) manifests itself neurologically by a shift in brain activity. When confronted with a cognitive task that is novel, difficult, or variable, the cerebellum and prefrontal cortex are highly active (Diamond, 2000). The cerebellum has more neurons than all the other sections of the nervous system combined, and cells within the cerebellum have the most synaptic connectivity (up to 200,000 connections per cell) which allows for extremely fast and efficient communication. As the task becomes more routinized, activity in the cerebellum decreases. Cognition becomes ‘‘automatic’’ and no longer requires concerted effort. The task has been absorbed into the cognitive/motor schemata of the individual, freeing up precious attention for future tasks. Dual-process models of cognition have long been utilized by social theorists to account for the unconscious nature of most cultural knowledge. As Vaisey (2009) points out, a similar distinction is found in Giddens’ (1984) ‘‘practical’’ and ‘‘discursive’’ forms of consciousness. Similar strains of thought can be seen in the pragmatist notion of ‘‘habit’’ (Dewey, [1922] 2002) and Bourdieu’s (1977) ‘‘habitus.’’ According to these theories, cultural logics arise from schemata which have been absorbed in similar patterns across a social group due to the practical organization of the social environment. While this marriage between cognitive science and social theory helps us understand how social logics operate, they do not shed light on how they develop. In order to address how locally learned schemata aggregate into collective cognitive structures, I address two issues which dual-process models have thus far neglected – the importance of creativity and significant others.

298

DAVID PETERSON

Creativity and Significant Others Dual-process theories have been largely based upon reactive models in which deliberative cognition is activated by some problem in cognition. The failure of automatic cognition is caused by a conflict in processing (Barrett, Tugade, & Engle, 2004). The effects of failure, ranging from frustration to a perceived threat to one’s ‘‘ontological security’’ (Giddens, 1991), compel individuals to regain balance. This is not completely wrong. As Giddens (1991) has argued, the psychological need for order is what drives our tendency toward habituation. However, Giddens warns that an orderly life is not a monotonically increasing good. An overreliance on habit is a sign of neurosis. Habit must be balanced by innovation, automatic cognition by deliberation. Giddens, thus, attempts to fortify his concept of ‘‘ontological security’’ by adding an element which overcomes itself: ‘‘the experience of creativity as a routine phenomenon is a basic prop to a sense of personal worth and therefore to psychological health’’ (1991, p. 41). Too little routine is problematic but so is too much. True ontological security reveals itself to be a tension between routine and a force that overcomes routine, instead of a univocal concept. This destabilizing force is acutely visible in children who abandon scores of old habits with each new developmental stage. Yet the predictable pathways of children’s development illustrates that overcoming habit is not a random process. Creative action is not simply a ‘‘non-habitual behavior.’’ It has a comprehensible logic and discernable direction. This direction is guided primarily by significant others. Because most of the research on dual-process models has been concerned with problem solving, the complexities of interpersonal relationships have been downplayed. But, of course, our experience is not characterized by a flat wasteland of problematic objects that we overcome and render invisible. It is increasingly clear that cognizing other people is essentially different from cognizing objects. This difference has been the topic of the growing ‘‘mirror neuron’’ literature.1 Mirror neurons are neurons that fire both when performing an action and when witnessing another person perform that action (Rizzolatti, Fadiga, Gallese, & Fogassi, 1996). Early mirror neuron experiments performed on macaque monkeys provided the first evidence that social cognition is biologically different from the cognition of objects. In addition to mirror neurons that resonate with another’s physical action, studies have shown the presence of mirror neurons that respond to another’s emotional display. Recent studies have shown

Agency, Biology, and Socialization

299

that witnessing the expression of emotions like happiness, disgust, or pain activates the parallel affective centers of the brain in the viewer (Jabbi, Swart, & Keysers, 2007; Singer et al., 2004; Wicker et al., 2003). While in the first months this emotional projection is localized entirely within the infant/other interaction, by around nine-months children begin reliably sharing joint attention with their caregivers (Carpenter, Nagell, & Tomasello, 1998). By around the first year, infants gain the ability to follow other people’s gaze (Butterworth & Jarrett, 1991). Coupled with the infant’s understanding of the facial expressions (Haviland & Lelwica, 1987) of others, this ability allows the infant to evaluate actions based upon mutual gaze and the other’s emotion. The canonical example of this ‘‘social referencing’’ is Sorce, Emde, Campos, and Klinert’s article (1985) in which 75% of 10-month-olds whose mothers were smiling at them crossed a visual cliff to reach them whereas none of those whose mothers appeared anxious did so. The biological ability to mirror happiness or fear in conjunction with the cognitive capacity needed to define the event as a unity, in this case, ‘‘going-to-mommy,’’ allowed the act to be infused with emotion.2 Hobson (1993) has argued that the child, caregiver, and object form a ‘‘relatedness triangle’’ in which knowledge about and evaluation of the environment is mediated through significant others. By not acknowledging the foundational importance of significant others in cognition theorists have obscured the directionality of development and, thus, the development of social logics. Pleasure and pain are merely the crude surface of experience. Children learn far more than the sensual value of objects. Yet, as Merleau-Ponty argues, ‘‘The real lends itself to unending exploration; it is inexhaustible’’ ([1945] 2005, p. 378). No item is ever absorbed in its entirety. There are always new perspectives to be had, new properties to uncover, new contexts to bring to bear. Every object contains an infinite horizon of possible experiences, an unlimited amount of ways it can be woven into the rest of child’s experience. If the child’s environment were truly a wasteland of objects, her voracious learning would be as unpredictable as wildfire. Why do children follow predictable cultural trajectories? Why do they tend to absorb the same aspects of the same objects on roughly the same developmental schedule instead of getting sidetracked into the unending exploration of some trivial thing? In fact, some children do get sidetracked. Showing inordinate amounts of interest in trivial things is a symptom of autism (Grandin, 1996). Importantly, another symptom is a lack of social understanding like grasping another’s intentions (Baron-Cohen, Leslie, & Frith, 1986). The significant other is not an object amongst objects, but something which essentially disturbs the child’s cognitive system by both

300

DAVID PETERSON

projecting emotional and motoric information and by being the elusive object of the child’s desire. Several theoreticians have suggested that there is a primary cognitive distinction between people and objects. Bakhtin’s (1986) ‘‘monological’’ and ‘‘dialogical’’ forms of communication, Buber’s (1970) ‘‘I/It’’ and ‘‘I/Thou’’ relationships, and Levinas’ (1996) ‘‘Being’’ and ‘‘Other’’ each provide a fundamental distinction between relations with objects and relations with significant others. In each case, the cognition of ‘‘things’’ with its subject/ object structure is contrasted with cognition that is relational, that dissolves clean distinctions between self and object, ego and alter. However, these two realms are not merely coincidental and unrelated but form a recursive structure in which the drive to attain intimacy with the other person drives the individual toward greater control of their environment. This, I suggest, is a first step toward understanding how the tension between control and intimacy draws the infant into widening spheres of interaction. The direction of development is guided by the reciprocity of intimate relationships, not just the mastery of objects. The Tension between Control and Intimacy The control-intimacy tension is what calls the child’s attention toward those areas which give her greater access to and influence over significant others. While the child has an independent interest in objects, she usually reaches a balance with them. She will be absorbed by a new toy until she has absorbed it and then it will cease to enthrall her. She will learn that an object hurts and will be wary of it. Yet, her interest in significant others does not find the same type of balance. In the child’s experience, other people do not lose their radiance because they cannot be fully absorbed. This is because the mother of the child’s experience is not the mother in her full complexity. She simplifies her actions within the relatedness triangle. She hides behind her hands and emerges again to the delight of the child. She marches a stuffed animal across the floor. She engages in behavior far below her level of competence and interest in order to provide basic structures of interaction the child can grasp. The child, driven by the control-intimacy tension, wanting to be with the mother and regulate the interaction, seeks to gain the logic of the game. However, with each developmental step, the child finds only new complexities, new logics, and new skills to be attained. This development is not random. The child, in pursuit of intimacy, absorbs skills that improve her control and understanding of the interaction.

Agency, Biology, and Socialization

301

They grow toward the parent-as-an-acculturated-adult. As the mother pulls away from the simplistic interactions of infancy into the more complex interactions of toddlerhood, she draws back toward herself, toward the practices of an acculturated adult. She begins using more words to address the child and engaging her in more complex interactions. The child, seeking more relational power, acquires increasing skills. Early skills like dexterity and motility allow the child to physically move toward others. These skills give way to more culturally specific skills like language. With each step the child gains more control over interactions, more ability to have desires met and control the environment as she weaves her actions more completely into preexisting, socially defined fields of action. As the child enters social institutions like school, a similar pattern emerges. The child, in relation with both peer groups and teachers, gains skills and knowledge about the nature of the school day, social groups, student–teacher relationships, task orientation, etc. As these lessons are absorbed, the child incorporates them into herself only to have the structures recede into more sophisticated cultural formulations. Throughout development, the skills necessary to achieve reciporicity in significant relationships are always just beyond the child’s grasp.3

THE TRELLIS Because the child must ‘‘invent’’ each new behavior and ‘‘discover’’ each new understanding for herself, the development of new skills and knowledge is always somewhat unpredictable. Each new behavior or idea is an emergent phenomenon. It strikes consciousness in a flash of insight or is stumbled upon by accident. It then provides a ‘‘cognitive term’’ (Pylyshyn, 1984) with which to perform mental operations. It is the Husserlian (1982) ‘‘idea’’ or Schutzian ([1932] 1972) ‘‘act’’ which unifies a segment of empirical reality within a single ideational core. If the child gains more control over her environment or receives positive relational reinforcement, the emergent behavior is repeated and becomes absorbed. If it fails or receives negative reinforcement it does not find a foothold. These ‘‘abilities’’ become cultural competencies when they develop in relation to regularized cultural environments. Such competencies are fostered by both the intentional and unintentional structures of social life that confront the child. By organizing the child’s experience, agents of socialization exploit the control-intimacy tension to motivate the child to invent and discover for themselves standard ways of behaving within preestablished contexts of social action.

302

DAVID PETERSON

They enframe the child’s environment in order to foster typical forms of emerging competence, the spontaneous invention or discovery of cultural behavior or knowledge. Using their superior physical, intellectual, and relational power, caregivers erect frames (Goffman, 1974) around the child’s experience to direct her attention and behavior. These include interactional routines, physical restraints and guides, and projected emotional evaluations. Much of the interaction that occurs between caregivers and pre-linguistic infants is designed to create a frame for interaction, a simple conversation of gestures (Mead, [1934] 1952) that can draw the child into the relatedness triangle. For instance, ‘‘peek-a-boo’’ is a straightforward hide-and-reveal game drawing the infant into its structure because it uses the caregiver’s face as bait. The child, inherently interested in the significant other, is absorbed in the game and absorbs its simple logic, eventually ‘‘discovering’’ the skills which the frame elicits through design and/or accident. When the infant is younger than 12-months-old, the game is mostly controlled by the mother. After 12 months, however, the child has absorbed the game and takes the lead role (Bruner & Sherwood, 1976). This enframing is made possible by the mother’s ability to radically simplify her actions. This is the simple beginning of the interaction rituals (Collins, 2004; Goffman, 1967) that will increasingly structure her life. It is the genesis of what Bruner and Sherwood (1976) call ‘‘social synchronization,’’ in which the infant and caregiver enter into a back-and-forth of reciprocal re-contextualization-based, not upon a set of explicit rules, but on both party’s skillful maintenance of the interaction. Yet, the infant’s lack of expressive ability and physical strength makes this a different type of interaction than the negotiation of reality that occurs between adults who share a culture. For the very young child, this power imbalance manifests itself in physical constraints. Giddens (1984, p. 172) has written that Durkheim’s argument for the externality of social facts always struggled because they are not truly ‘‘external’’ in the same sense that nature is external. But, for the child who has not yet learned the difference between the society and nature, the cultural world has no unique reality to it. When she violates some unknown rule the social world suddenly descends upon her in the form of her mother’s arms (i.e., physical constraints) which impose themselves with the impersonal force of nature. Her world is populated by social agents imposing structure upon her. Other imposed physical constraints like cribs enframe the child’s environment in more stable ways. For the child, many of the constraints imposed by others demand the same unquestionable respect as natural events.

Agency, Biology, and Socialization

303

Such constraints follow a predictable pattern. As the child’s explosive biological potential emerges into a new ability, caregivers increase physical restrictions until she can act appropriately (in a way that is safe, not embarrassing, polite, etc). They impose limits to the new power. They enframe it to guide its growth. Thus, as the infant begins to toddle, she gets confined to locations where she can do little harm to herself or property. As her walking becomes more competent, her parents can allow her more freedom to explore and the frame recedes. As the toddler learns what she is allowed to play with, her parents can allow her more freedom to interact with the environment. When it becomes clear that the child can pour milk herself, they can further ease off their control. Yet, even after physical constraints loosen their hold, the child is constrained by the preexisting ideational and emotional structures that enframe her experience. Ideationally, the child’s experience is enframed within the symbol systems of her culture. While the caregiver and prelinguistic infant can become synchronized in simple games and even engage in private forms of communication, expression in such ad hoc systems is quite limited. In order to engage more meaningfully with the caregiver, the child must adopt the system which the caregiver understands since she does not have one of her own. The linguistic inequality limits the child’s expressive capabilities and, thus, her ability to influence others. Consequently, while children’s rapid acquisition of language can be explained neurologically by the overproduction of synapses in the areas of the brain related to language (Siegler, 1989), the reason they emerge into the linguistic structures of their culture is that it vastly increases their relational power. Emotionally, the child’s experience is enframed by the approval and disapproval of others. Caregivers use both positive and negative sanctions to elicit the absorption of the correct schemata. Emergent behavior that is incomprehensible through some preexisting frame goes unacknowledged while that which is harmonious often elicits praise. A smile can be enough of a stimulus to perpetuate a behavior even when the child does not understand its cultural meaning. Negative sanctions are imposed when the emergent behavior threatens a frame. Some counteraction (e.g., discipline, ridicule, unwanted attention, etc.) perpetually forces the contested behavior back into consciousness. It is not allowed to become automatic cognition because it is constantly being challenged. Schemata that get absorbed begin the process of aggregation due to the mind’s unconscious propensity to draw a ‘‘grammar’’ of interaction from frequent empirical association (Reber, 1967) and then to reorganize these cognitive structures into increasingly more explicit, symbolic, and flexible

304

DAVID PETERSON

representations (Karmiloff-Smith, 1992). For instance, by about ninemonths-old children begin to organize their knowledge around the routine events that regularize their lives (Meltzoff, 1988; Nelson, 1996). By threeyears-old, children can provide a temporal sequence of familiar events almost perfectly (Nelson, 1986), including events which have little significance for them. The collective nature of these aggregated schemata emerge from similarly structured environments. For example, Sharp, Cole, and Lave (1979) found that Mayan adolescents who had attended school classified the word ‘‘duck’’ along with other words in the same taxonomic class – for example, ‘‘goose,’’ ‘‘chicken,’’ ‘‘turkey.’’ Adolescents who had not attended school, however, tended to classify ‘‘duck’’ based upon their role in various action schemas (‘‘fly,’’ ‘‘swim,’’ ‘‘eat’’). This inductive process is utilized and guided by active strategies on the part of the caregiver.

Scaffolding and Contextualization The term used by psychologists to describe processes designed to guide the child into culturally harmonious ideation and action is scaffolding (e.g., Clark, 1997; Gauvain, 2001; Rogoff, 1990). Scaffolding is any process whereby a caregiver manipulates the child’s environment in order to aid learning. For instance, a mother may physically help the child perform a task that she does not have the strength to perform herself point out important features of the environment, or make the beginning sound of a word the child is searching for. It is the ideational aspect of enframing. Rogoff (1990), drawing from Wood, Bruner, and Ross (1976), finds six functions of instructional scaffolding: 1. Recruiting the child’s interest in the task as it is defined by the tutor. 2. Reducing the number of steps required to solve a problem by simplifying the task, to that the learner can manage components of the process and recognize when a fit with task requirements is achieved. 3. Maintaining the pursuit of the goal, through motivation of the child and direction of the activity. 4. Marking critical features of discrepancies between what a child has produced and the ideal solution. 5. Controlling frustration and risk in problem solving. 6. Demonstrating an idealized version of the act to be performed. (1990, p. 94) The caregiver defines the task and, thus, enframes the interaction. Within this frame they recruit and maintain the child’s attention, overcoming

Agency, Biology, and Socialization

305

boredom and frustration, while eliciting from the child an idealized action for an idealized purpose. What is sought is clarity of the ideational core, the reason for the act. It is agreement of schemata and the synchronization of acts that is the goal of scaffolding. But, as Rogoff points out, the parent’s goal is not to break the action into ‘‘minutely ordered steps to be mastered in a lockstep fashion. Rather, effective structuring [y] maintains children’s involvement with the purpose of the activity, integrating varying aspects of the talk in a manageable chunk’’ (1990, p. 94). Rogoff’s use of the term ‘‘chunk’’ is significant. In a classic study, George Miller (1956) demonstrated that subjects were only able to maintain seven (plus or minus two) pieces of perceptual information in their short-term memory at one time. Beyond saturation level, information is either forgotten or ‘‘chunked’’ together, a space-saving procedure whereby several bits of information become treated as one cognitive term. Caregivers break the project into digestible chunks in order to free up cognitive resources while maintaining focus on the meaning of the project. Blatant falsehoods are allowed when the correct interpretation of the situation is more important than the acquisition of specific facts. For instance, a mother may call a whale a ‘‘fish’’ or say that an electric outlet is ‘‘hot’’ (Rogoff, 1990) because the actual facts would muddle the frame the caregiver has established. Simple ‘‘monothetic’’ acts are chunked into ‘‘polythetic’’ structures (Schutz, 1932, 1972), enframing schemata within typical, cultural behavior. For instance, typical events like ‘‘going to the market’’ or ‘‘getting ready for bed’’ provide umbrella structures that enframe a diverse set of actions, giving them unified meaning (Nelson, 1986). A polythetic act like ‘‘getting ready for bed’’ is constituted by many ‘‘sub-acts’’ like putting away toys, brushing one’s teeth, changing into pajamas, turning on the nightlight, getting into bed, and awaiting a story. ‘‘Getting ready for bed’’ is too complex a process for the young child to grasp completely and it involves a number of schemata she does not yet have. The parent must continually subsume the whole project under one title, reminding the child of its purpose throughout its duration, and lending extra help during any sub-act that the child has not yet mastered (e.g., helping the child brush their teeth). As they develop, event representations undergo a number of changes including increasing schematization, the identification of optional and required acts, conditional pathways, and the evolution of hierarchized structures, that is, structures with subgoals and their associated action sequences embedded within larger goal structures. (Nelson, 1986, p. 232)

306

DAVID PETERSON

As the child moves through the event again and again, she absorbs its logic. Repetition ensures that the child experiences (roughly) the same sequence of events enough times that its formal elements begin to emerge. Some nights she may not have to put away her toys. At grandma’s house, there may be no story. She may sleep in her parent’s room on occasion. Aspects of the routine may change but the functional purposes of the subacts are always subjected to the overarching purpose of the polythetic act. This allows the child to draw out the necessary pieces of the act from the contingent. Through repetition the child learns the meaning of the polythetic act and her event representation becomes more ‘‘skeletal and general’’ (Nelson, 1986, p. 232) and, thus, more ideal and social. Yet, scaffolding, which emphasizes the emergence into and absorption of specific ideational structures and skills, is always also a contextualization, which provides emotional information about the meaning of the frame. What is the nature of this child/caregiver/object circuit? Why should this be learned? How important is this lesson when other things come up? What constitutes success and failure? What are the stakes? How strict does the performance need to be? Teaching always includes both semiotic content (i.e., the ideational core) and emotional/interpersonal content (i.e., the social context). While scaffolding aims at the harmonization of ideational cores, contextualization attempts to harmonize value, meaning, and significance through emotional socialization. Thus, the process of abstraction is steered into harmonious meaning by hundreds and thousands of small corrections that occur mostly through the emotional process of contextualization, which gives the frame its emotional structure. Collins (1981) and Scheff (1990) have made similar points, claiming that formal systems are too clumsy and inefficient to account for day-to-day social control. Emotional interaction provides a more parsimonious solution to the problem of social order than the inculcation of specific rules. Emotions which arise during social approval or disgrace, according to Scheff, function as ‘‘a highly efficient system of informal sanctions, a system which is almost invisible’’ (1990, p. 74). Again, however, the power imbalance makes the social structure of emotions appear to be objective features of the natural environment and, hence, not open to influence or debate. As Mead has said, ‘‘The only way in which we can react against the disapproval of the entire community is by setting up a higher sort of community which in a certain sense out-votes and one we find’’ ([1934] 1952, pp. 167–168). The child cannot offer a meaningful challenge to such disapproval because she has no higher community and, therefore, no alternative interpretive system with which to compare it. She cannot reframe it. She has

Agency, Biology, and Socialization

307

no method of challenging the judgment while maintaining relations. She can reject the shame only by refusing the interaction. While both are aspects of a unified enframing process, contextualization is more difficult to analyze than scaffolding since, by definition, its importance lies in what is not evident through symbolic media. This is ultimately due to neural properties. Ideation like language and reasoning are processed in the neocortex but it is the evolutionarily early limbic system which determines features of contextual importance like emotion, motivation, and judgment (Siegel, 1999). Through contextualization, caregivers seek to guide the child into the correct emotional relationship with the frame. This imposition of emotion is not wholly unpredictable but tends to conform to what Gordon (1981) has called ‘‘emotional culture.’’ It is the somewhat stable emotional structure of a group’s somewhat stable ideational frames. Neuroscientist Antonio Damasio’s (1994) ‘‘somatic marker hypothesis’’ suggests that our feelings are complex predictions about the future based upon the past association between a class of stimuli and an emotion. Frequent association and correction through approval and disapproval eventually allows the child to come to the correct interpretation on their own. According to Hochschild (1979), social forces contextualize the child’s emotions and bring them into social harmony in two ways. First, there are ‘‘feeling rules,’’ the direct association of an emotion with a specific frame. Weddings are joyous. Funerals are sad. When the child engages in appropriate behavior, she is rewarded with legitimacy within the group. When her behavior is inappropriate she is shamed, ignored, dismissed, or removed. However, although socialization into a culture’s ideational system is vital for interaction, few social situations can be reduced to this ‘‘clearly defined frame=emotion’’ type of association. Instead, most interactions involve a measure of creativity that presupposes shared dispositions, a somewhat stable system of collective evaluation. This is what Hochschild has labeled ‘‘framing rules.’’ This is the ability to emotionally understand novel situations based upon the schematic similarity of its components with another, known situation. This is the child’s creative interpretation of the situation which is treated as more or less legitimate.

From Playing Games to Negotiating Situations The world that children are drawn into is not a chaotic amalgam of skills and knowledge but is composed of somewhat stable and coherent

308

DAVID PETERSON

‘‘situations’’ (Collins, 2004). Games provide a clear example of how the control-intimacy tension is exploited through enframed interactions, not just to teach isolated skills, but to draw the child into a complex system of interrelated, aggregated schemata. Games are bounded by rules, principles, and goals that enframe collective behavior. If the child wishes to participate, she must absorb them. A player who does not know the rules is a destructive force on the game and must be disciplined, marginalized, or ignored if it is to continue. Of course, most games are too complex for the child to absorb in their full, aggregated form. Yet, the fact that children can and do learn games brings us back to the original question DiMaggio posed. Namely, how do schemata aggregate into somewhat coherent social logics? This problem is unraveled, however, when we see that aggregated schemata can be enframed and projected ahead of the individual’s skill level under a polythetic umbrella which is then grown into through increasing skill in different, monothetic domains. For instance, the ability to play baseball involves the absorption of many different schemata. The schema of each player must be learned in isolation from every other player. The rules of pitching, batting, fielding, etc. must also be learned before the entire set of schemata can be aggregated into a legal game of baseball. However, ‘‘American baseball,’’ as a preexisting, polythetic domain of action, stands at the beginning of the learning process, guiding the diverse schemata into a culturally delimited aggregation. Thus, although ‘‘British baseball’’ and ‘‘Finish baseball’’ are made up of roughly the same schemata, American children do not spontaneously begin playing either. Through enframing, the diverse schemata are drawn together into a coherent and socially harmonious structure. Of course, as Bourdieu frequently pointed out, while a useful metaphor for social life, the stable rules and boundaries of games are unrealistic in a world in which rules and boundaries are, themselves, always a part of social struggle. However, I think there is reason to believe that socialization, especially in the early stages, would be similar in both cases and that differentiation would only become evident much later in the socialization process. While the struggle over boundaries can occur in any situation, isolated individuals facing well-established institutions can have little effect on their rules and boundaries. The right and ability to contest the rules and boundaries of a field is usually reserved for the well-placed (and, thus, highly socialized) leaders and experts. Lower status members usually mount a serious challenge only by joining a collective movement. Isolated individuals, on the other hand, usually must accept the basic structure of the group in order to participate.

Agency, Biology, and Socialization

309

Thus, I believe this model can illuminate socialization processes beyond children’s learning relatively enclosed and simple social logics like games. Even socialization into more complex and dynamic frames follows the same pattern. The learner, with little or no understanding of the social logic of the group, is drawn into its web of meanings. Frames are imposed, by design or accident, to guide development. Socialization in early stages tends to be an uneven amalgam of isolated skills and simplistic idealizations of the group’s logic. However, the more the learner absorbs, the more interconnection appears between the isolated skills and the more the social logic is given flesh. Importantly, skill tends to be entwined with relational reciprocity. The more the child wants to be taken as a subject, one able to make legitimate claims about the nature of reality, the more she must become an object of the social groups which define her involvement. In order to gain reciprocity within her family, school, or peer group, she must become a competent daughter, student, and friend. In each of these domains, the same recursive structure appears in which the child’s desire for intimacy is exploited in order to guide the absorption of specific skills. Agents of socialization organize the satisfaction of desire through the imposition of frames. This fosters emergent competence which, in turn, opens up new cultural milieus with new horizons of control and intimacy. As the recursive structure expands with increasing competence, the child ceases to be just an object of the structure and becomes a legitimate agent of it. This process is neither deterministic nor teleological. According to Thorne (1993), typical socialization theories rely on a double standard in which ‘‘social scientists grant adults the status of full social actors, but they define children as incomplete, as adults-in-the-making’’ (3). This fetishizes adult culture and dismisses nonconformity as nonsensical or simply defiant. The trend in the sociology of childhood has been to treat children as full social actors who have complex relationships with the structures imposed upon them. Instead of simply absorbing them, they ‘‘play’’ with (Thorne, 1993) and ‘‘interpretively reproduce’’ them (Corsaro, 2005). But while social scientists should avoid the trap of normatively labeling children’s behavior based upon a thoughtless acceptance of the dominant culture, it is important to realize that social groups routinely do. This labeling, which can be as formal as a right of passage or as informal as an approving look, changes the child’s status within the group. The more the child accepts the group’s ideational and emotional structure, the greater the legitimacy granted to her. Socialization, in this sense, is less about instilling the child with culture and more about the changing relationship between child and group.

310

DAVID PETERSON

This relationship undergoes a qualitative change once the child becomes an adult. While the meaning of adulthood differs widely from culture to culture (and even within cultures), formal elements can be abstracted. The ‘‘child’’ as a construct, is defined by what it lacks. The concept of ‘‘competence’’ has been utilized by both developmental psychologists (Gauvain, 2001) and sociologists (Clausen, 1991; Giddens, 1984) to describe the balance between expectation and ability that characterizes adulthood. Parents and other authority figures impose definitions of reality on children because they lack basic competencies. The power imbalance between children and adults allows these practices to be imposed through a variety of both positive and negative sanctions. As skills increase toward competence, relational power increases and provides greater reciprocity. They are able to negotiate increasing portions of reality with others. Becoming an adult means being able to enter into reciprocal relationships with others by means of gaining cultural competencies.

CONCLUSION The development of children provides sociologists with an underutilized resource for exploring how schemata aggregate many domains. The development of scientific paradigms, political ideologies, institutional logics, and other forms of collectively aggregated schemata are all processes in which an individual’s capacities and motivations meet a preexisting social group. In each case, the ‘‘ivy’’ and the ‘‘trellis’’ have different features yet they share a common logic: Driven by the tension between control and intimacy, individuals emerge into new forms of behavior which are guided into culturally competent form through the physical, ideational, and emotional enframing of experience by authorities.

A few words of caution are in order, however. In order to provide a picture of the complex process by which schemata aggregate I have painted with broad strokes. The concepts introduced here are meant to be useful heuristics which should be applied cautiously. The control-intimacy tension, for instance, is a concept necessary for understanding why habits are overcome and how creativity gets channeled in specific directions. However, distinguishing ‘‘intimacy’’ from ‘‘control’’ in any empirical situation is an extremely difficult hermeneutic process (See Katz, [1999, pp. 229–273] for an excellent example of what an interpretation might look like). Similarly, most interpretive frames imposed upon children

311

Agency, Biology, and Socialization

are neither as explicit nor as coherent as games. ‘‘Going to the market’’ has a logic which is gleamed from repeated trips. But this is not the clear logic of baseball. Bourdieu calls this sort of frame ‘‘a fluid, ‘fuzzy’ abstraction’’ (1977, p. 112). Talk of ‘‘frames’’ a helpful tool for understanding how preexisting scenes of action function to draw schemata into a cultural form but they are more porous and complex than I have treated them here. Of course, this means that socialization into these ‘‘fuzzy’’ frames is also a good deal more variable than I have implied. The more coherent and rigid the frame, the easier it is for the researcher to map out socialization processes and vice versa. Finally, I have tried to avoid the extremes of ‘‘cognitive individualism’’ and ‘‘cognitive universalism’’ (Zerubavel, 1997) in order to draw out the importance of social structure on shaping meso-level cognitive structures. However, by looking at broad and uncontroversial issues like the acquisition of basic competencies I have glossed over important, within culture, differences in development. For instance, different racial or ethnic groups may share a general schema (e.g., that police are meant to ‘‘protect and serve’’) while differing widely in their evaluation of how accurate that schema is. Even within families, children face very different environments depending upon their birth order or sex. While collective ideational structures are necessary for social cohesion, they are not sufficient. Different contexts produce different interpretations. Systematic differences in context will produce systematically different interpretations and, thus, subcultural and countercultural forms of emergent behavior. Over time, deep fault lines develop that can erupt in a violent clash of ideas. More targeted work in the area of contextualization, especially dealing with emotion, is needed to elucidate these sociologically significant processes.

NOTES 1. See Rizzolatti and Craighero (2004) for a review. Lizardo (2007) and Turner (2007) have recently debated the relevance of mirror neurons for Bourdieu’s theory of cultural transmission. 2. It is not entirely clear what occurs during social referencing. Some argue that one-year-old infants lack the cognitive skills to recognize that their caregivers are actually ‘‘referencing’’ an object they are interested in (Baldwin & Moses, 1996). They suggest that security seeking behavior leads nervous infants to ‘‘check back’’ with their mothers. If their mother’s display fear, it resonates with them leading to a more apprehensive attitude toward the ambiguous object. While this view appears more parsimonious, for the sociologist’s purpose both versions lead to the same end. The mother’s emotional projection adds emotional content to the infant’s environment.

312

DAVID PETERSON

3. Healthy development in relation to the control-intimacy tension is hardly assured. Work in developmental ‘‘Attachment theory’’ (Ainsworth, Blehar, Waters, & Wall, 1978; Main & Solomon, 1990) has detailed different forms of insecure relationships that development when control and intimacy fall into a pathological balance.

REFERENCES Ainsworth, M., Blehar, M., Waters, E., & Wall, S. (1978). Patterns of attachment. Hillsdale, NJ: Lawrence Erlbaum Associates, Publishers. Bakhtin, M. M. (1986). Speech genres and other late essays. Austin, TX: University of Texas Press. Baldwin, D. A., & Moses, L. J. (1996). The ontogeny of social information gathering. Child Development, 67, 1915–1939. Baron-Cohen, S., Leslie, A., & Frith, U. (1986). Mechanistic, behavioral, and intentional understanding of picture stories in Autistic children. British Journal of Development Psychology, 4(2), 113–125. Barrett, L. F., Tugade, M. M., & Engle, R. W. (2004). Individual differences in working memory capacity and dual-process theories of the mind. Psychological Bulletin, 130(4), 553–573. Bergesen, A. J. (2004a). Durkheim’s theory of mental categories: A review of the evidence. Annual Review of Sociology, 30, 395–408. Bergesen, A. J. (2004b). Chomsky versus Mead. Sociological Theory, 22(3), 357–370. Bloom, P. (2000). How children learn the meaning of words. Cambridge, MA: MIT Press. Bourdieu, P. (1977). Outline of a theory of practice. Cambridge, New York: Cambridge University Press. Bruner, J. S., & Sherwood, V. (1976). Early rule structure: The case of ‘‘Peekaboo’’. In R. Harre (Ed.), Life sentences (pp. 55–62). London: Wiley. Buber, M. (1970). I and Thou. New York, NY: Charles Scribner’s Sons. Butterworth, G. E., & Jarrett, N. (1991). What minds have in common in space: Spatial mechanisms serving joint visual attention in infancy. British Journal of Developmental Psychology, 9(1), 55–72. Carpenter, M., Nagell, K., & Tomasello, M. (1998). Social cognition, joint attention, and communicative competence from 9 to 15 months of age. Monographs of the Society for Research in Child Development, 63, 4. Clark, A. (1997). Being there: Putting brain, body, and world together again. Cambridge, MA: MIT Press. Clausen, J. S. (1991). Adolescent competence and the shaping of the life course. American Journal of Sociology, 96(4), 805–842. Collins, R. (1981). On the microfoundations of macrosociology. American Journal of Sociology, 86(5), 984–1014. Collins, R. (2004). Interaction ritual chains. Princeton, NJ: Princeton University Press. Corsaro, W. A. (2005). The sociology of childhood (24nd ed.). Thousand Oaks, CA: Pine Forge Press. Damasio, A. R. (1994). Descartes’ error: Emotion, reason, and the human brain. New York, NY: Putnam.

Agency, Biology, and Socialization

313

Dewey, J. ([1922] 2002). Human nature and conduct. Mineola, NY: Dover Publications, Inc. Diamond, A. (2000). Close interrelation of motor development and cognitive development and of the cerebellum and the prefrontal cortex. Child Development, 71(1), 44–56. DiMaggio, P. (1997). Culture and cognition. Annual Review of Sociology, 23, 263–287. Elman, J. L., Bates, E. A., Johnson, M. H., Karmiloff-Smith, A., Parisi, D., & Plunkett, K. (1996). Rethinking innateness: A connectionist perspective on development. Cambridge, MA: The MIT Press. Emirbayer, M., & Mische, A. (1998). What is agency? American Journal of Sociology, 103(4), 962–1023. Freud, S. (1966). Introductory lectures on Psycho-analysis. New York, NY: W. W. Norton & Company. Gauvain, M. (2001). The social context of cognitive development. New York, NY: Guilford Press. Giddens, A. (1984). The constitution of society: Outline of the theory of structuration. Berkeley, CA: University of California Press. Giddens, A. (1991). Modernity and self-identity: Self and society in the late modern age. Stanford, CA: Stanford University Press. Gleitman, L. (1990). The structural sources of verb meanings. Language Acquisition, 1(1), 3–56. Goffman, E. (1967). Interaction ritual: Essays on face-to-face behavior. New York, NY: Pantheon Books. Goffman, E. (1974). Frame analysis: An essay on the organization of experience. Boston, MA: Northeastern University Press. Goldman, A. I. (2006). Simulating minds: The philosophy, psychology, and neuroscience of mindreading. Oxford: Oxford University Press. Gordon, S. L. (1981). The socialization of children’s emotions: Emotional culture, competence, and exposure. In C. Saarni & P. L. Harris (Eds.), Children’s understanding of emotion (pp. 319–349). Cambridge: Cambridge University Press. Grandin, T. (1996). Thinking in pictures: My life with Autism. New York, NY: Vintage Books. Haviland, J., & Lelwica, M. (1987). The induced affect response: 10-week-old infants’ responses to three emotion expressions. Developmental Psychology, 23(1), 97–104. Hobson, P. R. (1993). Autism and the development of mind. New York, NY: Psychology Press. Hochschild, A. R. (1979). Emotion work, feeling rules, and social structure. American Journal of Sociology, 85(3), 551–575. Husserl, E. (1982). Ideas pertaining to a pure phenomenology and to a phenomenological philosophy: First book. London: Kluwer Academic Publishers. Jabbi, M., Swart, M., & Keysers, C. (2007). Empathy for positive and negative emotions in the Gustatory Cortex. NeuroImage, 34(4), 1744–1753. Karmiloff-Smith, A. (1992). Beyond modularity: A developmental perspective on cognitive science. Cambridge, MA: The MIT Press. Katz, J. (1999). How emotions work. Chicago, IL: University of Chicago Press. Levinas, E. (1996). Basic philosophical writings. Bloomington, ID: Indiana University Press. Lizardo, O. (2007). ‘‘Mirror neurons,’’ collective objects and the problem of transmission: Reconsidering Stephen Turner’s critique of practice theory. Journal for the Theory of Social Behavior, 37(3), 319–350. Main, M., & Solomon, J. (1990). Procedures for identifying infants as disorganized/disoriented during the ainsworth strange situation. In M. Main (Ed.), Assessing attachment though

314

DAVID PETERSON

discourse, drawings, and reunion situations (pp. 121–160). New York, NY: Cambridge University Press. Mead, G. H. ([1934] 1952). Mind, self, and society: From the standpoint of a social behaviorist. Chicago, IL: University of Chicago Press. Meltzoff, A. N. (1988). Infant imitation and memory: Nine-month-olds in immediate and deferred tests. Child Development, 59(1), 217–225. Merleau-Ponty, M. ([1945] 2005). The phenomenology of perception. New York, NY: Routledge. Miller, G. A. (1956). The magical number seven, plus or minus two: Some limits on our capacity for processing information. The Psychological Review, 63(2), 81–97. Miller, P. H. (2002). Theories of developmental psychology. New York, NY: Worth Publishers. Nelson, K. (Ed.). (1986). Hillsdale, NJ: Lawrence Erlbaum Associated, Publishers. Nelson, K. (1996). Language in cognitive development: Emergence of the mediated mind. New York, NY: Cambridge University Press. Parsons, T., & Bales, R. F. (1955). Family, socialization and interaction process. Glencoe, IL: Free Press. Piaget, J. (1974). To understand is to invent. New York, NY: Penguin. Pylyshyn, Z. (1984). Computation and cognition: Toward a foundation for cognitive science. Cambridge: MIT Press. Reber, A. S. (1967). Implicit learning of artificial grammars. Journal of Verbal Learning and Verbal Behavior, 6(6), 855–863. Rizzolatti, G., & Craighero, L. (2004). The mirror-neuron system. Annual Review of Neuroscience, 27, 169–192. Rizzolatti, G., Fadiga, L., Gallese, V., & Fogassi, L. (1996). Premotor cortex and the recognition of motor actions. Cognitive Brain Research, 3(2), 131–141. Rogoff, B. (1990). Apprenticeship in thinking: Cognitive development in social context. New York, NY: Oxford University Press. Scheff, T. J. (1990). Microsociology: Discourse, emotion, and social structure. Chicago, IL: Chicago University Press. Schutz, A. ([1932] 1972). The phenomenology of the social world. Evanston, IL: Northwest University Press. Sharp, D. W., Cole, M., & Lave, C. A. (1979). Education and cognitive development: The evidence from experimental research. Monographs of the Society for Research in Child Development, 44, 1–112. Siegel, D. J. (1999). The developing mind: Toward a neurobiology of interpersonal experience. New York, NY: The Guilford Press. Siegler, R. S. (1989). Mechanisms of cognitive development. Annual Review of Psychology, 40, 353–379. Singer, T., Seymour, B., O’Doherty, J., Kaube, H., Dolan, R. J., & Frith, C. D. (2004). Empathy for pain involves the affective but not the sensory components of pain. Science, 303(5661), 1157–1162. Sorce, J. F., Emde, R. N., Campos, J. J., & Klinert, N. D. (1985). Maternal emotional signaling: Its effect on visual cliff behavior of 1-year-olds. Developmental Psychology, 21(1), 195–200. Spelke, E. S. (1990). Principles of object perception. Cognitive Science, 14(1), 29–56. Swidler, A. (1986). Culture in action: Symbols and strategies. American Sociological Review, 51(2), 273–286.

Agency, Biology, and Socialization

315

Thorne, B. (1993). Gender play: Boys and girls in school. New Brunswick, NJ: Rutgers University Press. Turner, S. P. (2007). Mirror neurons and practices: A response to Lizardo. Journal for the Theory of Social Behavior, 37(3), 351–371. Vaisey, S. (2009). Justification and motivation: A dual-process model of culture in action. American Journal of Sociology, 114(6), 1675–1715. Wicker, B., Keysers, C., Plailly, J., Royet, J-P., Gallese, V., & Rizzolatti, G. (2003). Both of us disgusted in my insula: The common neural basis of seeing and feeling disgust. Neuron, 40(3), 655–664. Wood, D., Bruner, J. S., & Ross, G. (1976). The role of tutoring in problem-solving. Journal of Child Psychology and Psychiatry, 17(2), 89–100. Zerubavel, E. (1997). Social mindscapes: An invitation to cognitive sociology. Cambridge, MA: Harvard University Press.