Theorizing Discrimination in an Era of Contested Prejudice : Discrimination in the United States [1 ed.] 9781592139149, 9781592139125

Citation preview

Theorizing Discrimination in an Era of Contested Prejudice

Theorizing Discrimination in an Era of Contested Prejudice Discrimination in the United States

Samuel Roundfield Lucas

T E M P LE UN I V E R S I T Y P R E S S

Ph ilade l p hi a

To Mom and Dad, whose joys, no matter how great, were never full because of race and sex discrimination

Temple University Press 1601 North Broad Street Philadelphia PA 19122 www.temple .edu/tempress Copyright © 2008 by Temple University All rights reserved Published 2008 Printed in the United States of America Text design by Kate Nichols The paper used in this publication meets the requirements of the American National Standard for Information Sciences—Permanence of Paper for Printed Library Materials, ANSI Z39.48-1992 Library of Congress Cata loging-in-Publication Data Lucas, Samuel Roundfield, 1963– Theorizing discrimination in an era of contested prejudice : discrimination in the United States / Samuel Roundfield Lucas. p. cm. Includes bibliographical references and index. ISBN-13: 978-1-59213-912-5 (cloth : alk. paper) ISBN-10: 1-59213-912- 4 (cloth : alk. paper) 1. Discrimination—United States. 2. Sexism—United States. 3. Racism—United States. 4. United States—Race relations. I. Title. JC599.U5L79 2008 305.0973—dc22 2008006411 2

4 6

8

9

7

5

3

1

Contents

Acknowledgments Introduction: Evidently

vii 1

1 Discrimination in the Era of Contested Prejudice:

Fundamental Bases 2 Experiential Realities and Public Contestation

8 23

3 From Condoned Exploitive Relations to the Era

of Contested Prejudice

53

4 Defining, Finding, and Remedying Discrimination:

Dominant Legal Perspectives

86

5 Defining, Finding, and Remedying Discrimination:

Critical Legal Perspectives and the Critique of the Dominant Legal View

103

6 Defining Discrimination Effects: An Asocial

Scientific Method 7 Discrimination as a (Damaged) Social Relation

143 175

vi / Contents

8 Epistemological Foundations for Studying Effects of

Discrimination as a Social Relation

205

9 Theorizing Discrimination in an Era of Contested Prejudice

235

Appendix A: Commentary on Methods of Data Analysis for Chapter 2

251

Appendix B: Commentary on Simulation for Chapter 5

257

References

259

Index

275

Acknowledgments

A

ll thought is social. My thinking on the definition and measurement of discrimination effects is the result of numerous conversations and correspondences. The opportunity afforded me to engage such issues is the result of a vast social network that delivered aid in a variety of forms, including technical assistance, library materials, supplementary data, computing time, data documentation, disk space, software assistance, and other essential resources. And, of course, I have been blessed with several willing informal conversation partners who helped me to explore the issues embedded in my research. Given this truth, though I wish to do so, it would be impossible to express my appreciation here to all those involved in my efforts to complete this project. This is all the more true because this work began as a dissertation. To aid my completion of the dissertation, I was fortunate to have received a Ford Foundation Minority Dissertation Fellowship. At the same time, the staff of Tuskegee University generously compiled a data series of lynchings by state and year, vastly speeding and simplifying my data collection effort. My dissertation advisor, Bob Hauser, provided supportive criticism, serious questioning, clarifying comments, and essential guidance, always with the greater aims of the project and my development as the supreme motivation. Adam Gamoran continued to provide me with a wealth of support and counsel, and helped me tremendously in paring away the countless meandering and therefore distracting digressions. Rob Mare

viii / Acknowledgments

called me to take seriously any given model or position I specified. Lauren Edelman pointed me toward legal literature I might not otherwise have found, encouraging me to make myself aware of the legal perspective on the issues I addressed and the positions I adopted. Charles Halaby encouraged me to refine my epistemological argument at many points and in many ways. And Glen Cain was always on my mind, the embodiment of a sister discipline from which I have often learned. Finally, together I can think of no more supportive dissertation committee; I have been blessed to have had the opportunity to study with such true scholars. The Wisconsin Model of Intellectual Work is such that many members of the community may be involved in the development of any given project. Thus, in addition to my committee, I profited from the attention of many faculty, staff, and students. Franklin Wilson and Gerry Marwell maintained a helpful interest in my progress, which provided further motivation for completion of the project. Discussions with Aimée Dechter and Ross Matsueda taught me much about principles of comparison and argumentation. I was able to obtain feedback at an early stage of the work from both the Race and Ethnic Colloquium and the Sociology of Gender Colloquium, as well as the Institute for Research on Poverty Seminar, all while in Madison. Clearly, it is impossible to express appreciation to all those involved. But a few other teachers were and remain supports. Stephen J. Ingels, Melvin Rogers, Dean Donna Mancini, and Professors Eric Chaisson, Judy Porter, Robert Washington, David Karen, William Hohenstein, Cornel West, and Paul Jefferson modeled through their integrity, rigor, and passionate commitment just how this project might be possible. Also, Dr. Piazza, Mr. Grant, Mr. Willis, Mr. Andreoli, Mr. Eagles, Mr. McCune, Mrs. Fort, Mrs. Cianelli, Mrs. Henderson, Mrs. Lipsitz, Mrs. Watson, Mrs. Hillaird, Mrs. Rogers, Ms. Zoerb, and others who taught me long before I entered higher education played vital roles in my developing understanding of how diverse peoples might live together and transcend the challenge that historic and contemporary discrimination imposes. I would be remiss in thanking only my most recent teachers if I also did not acknowledge the unpayable debt of gratitude I have for those who taught me such important lessons before I learned how to learn on my own. Not only was this project furthered in classrooms and at computer screens: I have received aid in coffee shops and over telephone lines. In many conversations, Eden Inoway helped me see and accept the fullest implications of the work, while Luci Thimm-Jurado helped me become a better scholar through her thoughtfulness and understanding. Wendy Carter, Min-Hsiung Huang, Daphne Kuo, Jane Mosley, Mario Sims, and Theresa Thompson-Colon reminded me time and again of the playful aspect of all that we do; some of

Acknowledgments / ix

the most draining moments of the project became a joy through their heartfelt care. I would not have finished the project had I not had someone to engage in passionate discussions over coffee, lunch, output, and after each presidential address at the American Sociological Association annual meeting—thanks, Lynn Magdol. And I would not have retained a sense of the deep connection among various fields of scholarship without the companionship of historian Mona Siegel and economist Donna Ginther. Donna Ulteig saw me through so many moments in Madison that no words can convey my gratitude. I owe her much. After leaving Madison, I continued to receive generous, essential aid. At crucial moments, I was supported by my hosts and conversation partners while on sabbatical at Harvard (2001–02) and at the Czech Academy of Sciences (spring 2007). Stanley Lieberson, Christopher Winship, Prudence Carter, Jason Kaufman, Cybelle Fox, Glenn Loury, Katherine Newman, Petr Mateˇju˚, Michael Smith, Blanka Javorova, Zuzana Uhde, Michaela Appeltova, Natalie Simonová, Petr Soukup, Marek Hrubec, Heather Fowler, and Judit Izsak, in various ways too numerous to mention, deepened my knowledge and furthered my efforts. Scholars elsewhere, most notably Allison L. Sneider of Rice University and Renee Monson of Hobart and William Smith Colleges, also helpfully answered my questions and pointed me to literature I would find of value. At Berkeley I received a junior faculty research grant to supplement the data. And, in the sociology department at Berkeley I have been fortunate to be exposed to an intellectually exciting environment built by scholars I long admired, an environment currently peopled by promising young thinkers such as Marion Fourcade, Dylan Riley, Sandra Smith, Robb Willer, Irene Bloemraad, Cıˇ han Tugˇal, Shannon Gleeson, Alex Janus, and Ellis Monk, whose future works will in part set the standards of the field and our own public understanding of society. At Berkeley, such visible examples provided models of utmost value. Joseph Chernick greatly aided my navigation of this project and the emotional toil it sometimes exacted. I owe him much. Supportive friendships with Susan Schacht and Sorayya Carr kept me going even when I doubted whether the work would ever be allowed to see the light of day. Sixteen years is a long time for a project, but some have been there in myriad ways throughout. Always keeping my philosophical senses from atrophying, Wai Kit Choi extended himself in numerous, supportive ways, without which the project would have foundered long ago. Kathleen Alice Crowley fired my mind with possibilities I would never have glimpsed otherwise. My sister, Allison, continually provided support, a sense of history, and understanding that only family can.

x / Acknowledgments

Support indeed has surrounded the project, even to the end of bringing the finished work forth. I write, of course, of having been wonderfully blessed with a supportive publishing team of Elena Coler, Heather Wilcox, and Gary Kramer. And, most notably, would that every author could have the experience of working with Mick Gusinde-Duffy, the editor who believed in the work, encouraged me along the way, and endeavored to make it better in every way. As I close, I want to recognize that my father and mother were the most important sources of this work. The project certainly bears the marks of the example and hopes of my mother and father, Hazel F. Lucas and Frederick N. Lucas, who modeled patience, acceptance, and perseverance. No thanks can ever be sufficient for what parents teach. Between the time I completed the scholarly work of the dissertation and moved to Berkeley, and the time I was enabled to finish the required typographical corrections before fi ling the final copy of the dissertation, my father passed away. I deeply regret that I shall never be able to hear, see, and feel his response to the journey I have taken. I remain mournful; having longed for the day when he would sit in my office and share with me the distance we have traveled, I feel cheated that I am here and he is not. In all our equations and theoretical terms there is, it often seems, no place for what really matters. The real cost of oppressive structures is not measured in dollars or occupational status. It is measured in the tears that flow behind closed doors, the blood that is spilled without understanding, the anguish that suffuses experience and comes to be who one is. These tears, blood, and anguish are translated into human relationships in complex and indeterminate ways. Damaged human relationships, not money or status— that is the cost of oppression, to oppressor and oppressed alike. Oppression reaches far. Throughout my odyssey I anxiously awaited the opportunity to share with my father a completed work that would convey my understanding of his history, his plight, his frustration. It pains me to note that will not happen. I have a habit of remarking that all thought is social. Never have I been more aware, throughout my being, of the complexity and simplicity of this truth. Sometimes the social nature of thought brings clouds of mourning, but I am hopeful that in the morning there is awakening. It is with that faith that I proceed. Madison, Wisconsin, USA Oakland, California, USA Cambridge, Massachusetts, USA Prague, Czech Republic

Evidently

T

he vast majority of evidence used to ascertain the effect of race discrimination on the success of blacks and the effect of sex discrimination on the success of women is, in a word, irrelevant. Not only is the evidence irrelevant to whether discrimination matters for blacks and women, but also the evidence is necessarily silent on the important question of whether whites and males gain or lose through the operation of antiblack and antiwoman discrimination. Because blacks and women are disadvantaged in many areas, any suggestion that whites and men might lose through antiblack and antifemale discrimination may appear ludicrous to many. And to suggest that silence characterizes any part of public attention to these issues may seem equally absurd given that scholars, politicians, and citizens across the political spectrum engage in raucous debates about the merits of policies purported to address prior and present discrimination. In these debates, the well-known facts of inequality often form the empirical basis for theoretical assertions of the essential or trivial nature of racial and sexual status, for substantive claims concerning the significance or nonexistence of discrimination, and for political demands to end or extend allegedly aggressive ameliorative strategies. Yet, despite their constant invocation, the oft-cited facts of race and gender inequality are irrelevant to whether discrimination matters; their invocation signals not clarity but, instead, the conflation of racial and

2 / Evidently

sexual inequality with the effect of antiblack and antifemale discrimination. The conflating of inequality and discrimination is just one of the consequences of the prevailing understanding of race and sex discrimination.1 In order to treat discrimination systematically, analysts must ask three important questions. First, what is discrimination? Second, what are the effects of discrimination? Third, what is to be done about discrimination? Three volumes address each of these questions in turn. In the process I detail why the existing definition of discrimination falls short, why the existing evidence of discrimination is irrelevant to the questions it is marshaled to answer, what the effect of discrimination on targets and nontargets of discrimination actually is, and how new approaches to addressing the social dislocations that follow in the wake of discrimination may be developed and enacted. However, in studying discrimination in the United States, we encounter particular socioemotional challenges, some of which hound the effort throughout, often threatening to derail the effort before it can even begin. Therefore, before proceeding, it is useful briefly to identify the primary source of these challenges and the ramifications for the enterprise pursued here. The hope is that, by identifying the source explicitly, we can, together, neutralize the threat it poses to our effort.

History, Biography, and Self-Consciousness: Facing the External Challenge In a well-known passage, C. Wright Mills calls on the field of sociology to nurture the sociological imagination, a capacity that “enables its possessor to understand the larger historical scene in terms of its meaning for the inner life and the external career of a variety of individuals. . . . By such means the personal uneasiness of individuals is focused upon explicit troubles and the indifference of publics is transformed into involvement with public issues” (1959: 5). Discrimination is one of the most difficult phenomena for people in the United States to discuss. Sadly, scholars of all stripes, sexes, statuses, ethnicities, and persuasions are no more immune to that difficulty than anyone else. With all due respect to C. Wright Mills, however, one of the reasons discrimination is most difficult to discuss is that too many persons appear to bring too much of a sociological imagination to the discussion—that is, as they consider the society or the institutions in which they live, they are too 1

In the 1970s, analysts regarded sex as an immutable biological construct and gender as a mutable social construct. However, more recent theorizing has revealed the weaknesses of this approach, such that now even the existence of a distinction between the two terms is questioned by some analysts. Because of this development in social theory, and because sometimes the term sex may be jarring or confusing, I use the terms sex and gender interchangeably.

Evidently / 3

aware of their own biographies, too focused on their own stations, such that they often cannot suspend awareness of their own stories long enough to allow the inquiry to proceed. Instead, they endanger the inquiry by their intense aim, in heart if not in mind, of confirming a fact of great interest to themselves. And what is this key fact that seeks confirmation in such discussions? To put the matter bluntly, if given voice, that fact would scream: “I am a good person!” This need for reassurance certainly is understandable, and one can be very sympathetic to the concern; that someone involved in a discussion of discrimination might be troubled by the prospect of being judged harshly certainly is sensible given this country’s painful history of racial oppression, sexual subjugation, and the smoldering hostility regarding these issues that characterizes the current period. However, this concern with how one may be judged constitutes what I call the external challenge for discrimination research and antidiscrimination activism, so labeled because it is not inherent in the actual enterprise but, instead, is imported into the process by some or many of those involved. The importers of this external challenge may be black, white, Asian, Latino/a, “multiracial,” “other,” male, female, gay, lesbian, bisexual, straight, transgender, and of any religious, cultural, or otherwise identifiable group. No sector has a monopoly on this import. Understandable though the concern may be, however, the research on stereotype threat (e.g., Steele 1997), the psychology of optimal performance and flow (e.g., Csikszentmihalyi 1990), and basic common sense should tell us that it is very difficult to conduct any kind of systematic inquiry when one is distracted by doubt as to one’s own character, others’ presumed assessment of one’s character, or the legitimacy of one’s placement in the status that allows one to conduct the inquiry. In the present context, those distracted by such doubts may become subconsciously invested in steering the inquiry to confirm that key fact with which they are concerned, or in undermining the inquiry if it appears that fact may not be confirmed (which is not to say that the fact actually would be disconfirmed). I have seen this dynamic unfold in dozens of scholarly and not-so-scholarly venues—in committee meetings at my home institution, on national panels charged (in every sense of the word) with systematically considering discrimination, in political groups working to transcend discrimination, in classrooms focused on issues of inequality, as well as in living rooms, dining rooms, and other rooms of the domicile. When inquiry is so infected with such self-conscious, defensive energy, the most that one can hope to do is muddle through, repair relationships if possible, leave intact existing understandings rather than further damaged ones, and hope that those involved will someday deal with these issues privately.

4 / Evidently

That this dynamic occurs is, of course, no shame; few of us want to be a “bad” person. Further, each of us has our blind spots and little inconsistencies, which make each of us vulnerable to some charge or another. For these and other reasons I certainly would prefer to ignore or sidestep this uncomfortable issue. Yet, I am forced to raise it here because ignoring it only strengthens its capacity for harm and because, unfortunately, this dynamic stands in our way. It stands in our way because, as we proceed, many of the largely takenfor-granted claims that have framed our understanding of discrimination will be called into question. And just as surely, therefore, this dynamic threatens to become unleashed. As we proceed, it may become difficult to resist the urge to evaluate the claims being made in this work in terms of their seeming implication for one’s own character. Indeed, to the extent the propositions that constitute the dominant perspective on discrimination have been uttered and popularized in part and through the same dynamics that are activated in public discussion of discrimination, it is quite possible that any question of those propositions is going to pose an existential challenge. But, that challenge must be resisted if we are to have any hope of communication, not to mention progress. Thus, I am making this uncomfortable plea: Please suspend the evaluation of what the project seems to mean for one’s own character. I make this plea for at least two reasons. First, the dynamic I have described can lead persons (of whatever race, sex, ethnicity, class, religion, culture, or sexual preference) to become so preoccupied with confirming their goodness that they jump to conclusions that do not follow from the claims that are made at various (perhaps early) points in the unfolding inquiry. And, when persons jump to conclusions they may easily misunderstand and mischaracterize their interlocutors. Attempting to defend their goodness, they may lash out and imply or even directly state that their interlocutors are “bad” not to have realized how “good” they are. The tragedy, of course, is that by jumping to conclusions, they can so damage the discussion that they prevent a resolution from emerging that might actually end up both confirming the fact with which they are deeply concerned while also building and reinforcing the structures and camaraderie needed to improve matters for everyone. The antidote to this dynamic is to accept that the inquiry, no matter how painful or disquieting in a given moment, has to fi nish before one can even begin to evaluate whether the inquiry has any implications for one’s character, to say nothing of discerning what those implications might be. My plea seeks to prescribe this antidote. Second, I make this plea because the task ahead is difficult enough. Any time an analyst proposes a rethinking of a phenomenon, it poses a serious challenge. Those evaluating a proposed reconceptualization must be able to

Evidently / 5

conduct the difficult task of suspending their current understanding enough to allow the reconfigured framework to take shape in their mind, while maintaining their critical faculties so they can evaluate the framework within its own terms. They must be comfortable with the unsettling prospect of what formerly were solid entities being broken into distinctly different phenomena, and the still more unsettling prospect of formerly distinct entities or concepts becoming fused, or perhaps being revealed to be merely different manifestations of phenomena formerly seen as unrelated. Thomas S. Kuhn (1962 [1970]) wrote eloquently of the anxiety, the visceral unease this can engender, and, highlighting the inherent difficulty of this task, he poignantly noted that often new approaches only win the day when devotees of the old approach, never able to be persuaded of the validity of the new conceptualization, eventually die off. Kuhn’s observations in this regard should give us pause—evaluating a reconceptualization of a phenomenon is demanding. Questions one must address in the effort include: Does the reconceptualization hang together coherently, or might it hang together coherently or even elegantly should additional research ensue on its basis? Does the reconceptualization allow us to address questions we have not been able to address? Does it hold out promise for new inquiries we have yet to pursue, or even imagine? These are extremely difficult questions to answer without benefit of hindsight, as any cursory check of the history of natural or social science can attest. To those exceedingly difficult fundamental questions we need not add the distraction of wondering what the claims seem to mean for one’s own character. Thus, I offer my uncomfortable plea. Of course, at this juncture I cannot say assuredly that the approach I propose is an advance. But, I believe it well may be an advance; it needs time, the work of others on its basis, and serious scrutiny to reveal its promise. The only way this ever will occur is if we set aside distractions and consider the matter of discrimination as a sociological phenomenon rather than as a matter of personal character.

Scope Conditions and Clarifications To aid that endeavor, I want to clarify three matters up front. First, this project is not about race—it is about discrimination. Discrimination on the basis of race occurs, but so does discrimination on the basis of sex, sexual preference, religion, and other factors. This project focuses on race and sex because of data limitations. But, it is imperative that I convey that this is not a study of African Americans in the United States. It is, instead, the study of a social phenomenon: discrimination. Therefore, though the discussion highlights sex and race to the virtual exclusion of other dimensions of discrimination, one has to start somewhere, and I have chosen,

6 / Evidently

for various reasons having to do with my own skill set and available resources, to start here. However, the foundation established herein may be applied to other dimensions of discrimination, and I would think such work will be essential for the framework to reveal its promise over time. Second, I do not use empirical research to validate the epistemological claims I make nor the measurement strategies that flow from those claims. Despite the existence of so-called “validity coefficients,” validity (as distinct from reliability) is a theoretical concern, not an empirical one. To establish that a measure is valid, one must ground the measure in a theoretical logic. This commitment is a general one, but it is extremely important in the case of discrimination. Of course, many scholars proceed in the opposite manner, using empirical research to confirm that their methods and measures are valid. This approach is most visibly, publicly championed in research purporting to measure intelligence. The obvious problem with this approach is that any method that is validated by confirming one’s expectations about the empirical world is a method that is unlikely to be regarded as accurate if instead it reveals a difference between what one expected to find and what one actually did find. Though the history of human inquiry is littered with the lab reports of those who failed to question their expectations in the face of anomalous information and thus failed to make the discovery we later came to regard as the foundation upon which subsequent knowledge would be built, this empiricist tendency remains strong. With such an empiricist approach, instead of presuming one’s expectations may be in error—a real possibility if one is conducting research—a measure is rejected as having been faulty. Because we use measures to tell us about the world, and need to be open to the possibility that the world is different than we believe it is, it is inappropriate in general to use empirical research to validate the method, and certainly so for areas imbued with confusion and contention. Thus, with a phenomenon as contentious as discrimination, using empirical research to validate the approach would be a colossal mistake. Hence, this volume lays out the theoretical, epistemological case, while later work will present an empirical analysis consistent with the epistemological claims. Those who do not find the theoretical case plausible need not be concerned with the findings from the empirical analysis. At the same time, if the empirical findings turn out to be surprising, well, sometimes the world is not as we believe it is. Third, my plea for suspending the question of individuals’ character is a necessary requirement for the framework to have any chance of being considered on its own terms. What is ultimately more important, however, is that this plea is essential for the effort, also pursued in later work, to reconfigure our solutions for discrimination, as citizens and human beings. This aim presupposes that discrimination is not primarily about individuals. Thus, in order to accomplish this aim, at its core the analysis cannot be primarily

Evidently / 7

about the actions or character of individuals. Indeed, a major motivation for the framework proposed herein is the belief that we have created major problems by focusing on individuals. That focus has animated the legal analyses of discrimination and it has infected the social scientific analysis of the effects of discrimination. If we are to ever break the current impasse and craft promising analytic approaches, inclusive political strategies, and transformative public policies, we must free ourselves from this individualistic perspective. What follows is a systematic effort to do so; what follows now is a brief guide to that enterprise.

Plan of the Work This volume describes the new era of race and gender relations that closed the twentieth century and has, so far, continued into the new millennium. I show how the old perspectives on discrimination were crafted for a reality that no longer exists, and how analysts’ efforts to address the well-known problems with those perspectives systematically fall short. I then craft a definition of discrimination devised for social scientific analyses, and show how this definition is both consistent with the reality of discrimination and appropriate for social scientific research. This book concerns epistemology, focusing on theories through which discrimination has been seen, and constructing a strategy to estimate discrimination effects. In later work I use the definition of discrimination developed in this volume to estimate the effects of race and sex discrimination in the United States. Titled Just Who Loses?, it presents empirical analyses of discrimination effects in education, labor force participation, job quality, earnings, and mortality. Thus, the analysis addresses every major phase of the socioeconomic lifecycle, as well as the important final outcome of death. Also in future work, I use the vantage point afforded by the theorized view of discrimination and the findings on its effects to critique both existing public policy concerning discrimination, racial inequality, and gender inequality, as well as the political strategies adopted by many who seek to redress the historic and contemporary wrongs of discrimination. I offer new policy options, and discuss a range of strategies that might prove successful. That work, unapologetically normative in its focus, closes with reflections on building the just society. Throughout the many pieces of the project the aforementioned plea remains relevant. Thus, I renew my request that we suspend the evaluation of what the project seems to mean for one’s character. In my mind and heart, no one’s character is at issue—the social world in which we live together is.

1 Discrimination in the Era of Contested Prejudice: Fundamental Bases

F

or the vast majority of United States’ history, interaction between men and women and blacks and whites was so governed by rigid and exploitive rules of conduct that the cost to many blacks and women of even being perceived as attempting to step beyond their station was social ostracism, physical battery, imprisonment, or a death calculated to maximize pain and suffering. Further, duly constituted authority not only ignored the violence against women and blacks, but frequently actively encouraged, abetted, or engaged in the savagery. This sordid history of condoned exploitation, however, has been eclipsed for long enough that its very existence has more than begun to fade from public memory. Surely, for all those who hope to someday realize the promise of America, the passing of the regime of Jim Crow race relations, brazen misogyny, and the taken-for-granted violence that characterized and maintained that state of affairs is cause for celebration. However, though revelry rightly should mark the progress that has been made, it would be a colossal mistake to conclude that the waning of the regime of explicit, rigid, brutal, restrictive, and exclusionary codes of interaction has ushered in an easy freedom and equality. What we have attained is not the reign of race and gender justice and equal opportunity. Instead, we occupy an interstitial moment in the history of race and gender relations in the United States, a moment that by its very nature constantly threatens to return us to the past from which we have come. It is this perilous time that I term an era of contested prejudice.

Discrimination in the Era of Contested Prejudice: Fundamental Bases / 9

Within this epoch progress commingles with forced retreat in effectively equal measure. Further, the contradictions of the era greatly facilitate misunderstanding of the time in which we live. Misunderstanding reigns because different analysts accurately describe some aspects of the complex reality, but systematically miss others. Yet the partial accuracy of many incomplete depictions often is sufficient to give the partisans to whom these analysts frequently speak not only increasing confidence in the overall truth of their limited visions, but also increasing doubt as to the integrity and goodwill of those with whom they disagree. Even more troubling, in academia the same kinds of misunderstandings not only occur but, owing to the strategic location of the professorate, ultimately have profound destructive ramifications. For our purposes here, the key manifestation of this understanding surrounds the definitions of discrimination as well as the strategies for studying discrimination effects. Presently many social researchers and commentators continue to use analytic tools no longer appropriate in the era of contested prejudice, as if nothing has changed. Equally problematic, others rightly question these tools, but fail to offer alternative analytic strategies that will address the problems they identify, maintain standard principles of scholarly work, and still document any effects of discrimination that remain. The use of outmoded research tools and the controversy surrounding them pose thorough-going challenges to both the academy and the wider society, as the ensuing confusion drives a wedge between communities and groups owing to their often equally accurate yet woefully incomplete and widely divergent readings of matters concerning discrimination. The result is that in the community of scholars—a community the entrée to which is ostensibly secured by a demonstrated willingness to explore and reflect—as well as in the wider society, persons all too often either talk past each other or, worse, speak not at all. The ultimate outcome of this impasse is to lower the likelihood that citizens will navigate the dangers and opportunities endemic to the time in a way that moves American society beyond contestation of prejudice to not only consensus against it, but further, to a realized commitment to address the damages inflicted by the terrible injustices that accompanied condoned exploitation and continue to accumulate in the era of contested prejudice. These brief comments serve to introduce the quandary that occupies this volume: namely, the arrival of a decidedly different epoch in race and gender relations, the challenges that era poses for the effort to understand race and gender inequality, and the implications of the age for the study of discrimination. In this chapter, I briefly highlight selected key aspects of the argument that will develop herein. The state of understanding and the controversy attending standard approaches to analyzing discrimination effects

10 / Chapter 1

make the issues pursued herein a necessary point of departure for our effort to systematically estimate the effects of discrimination in the United States. We begin by stating more directly the dominant definition of discrimination.

Defining Discrimination: Dominant Legal and Social Science Perspectives To study the social scientific definition of discrimination, one must attend to the legal definition of discrimination. The legal system elaborates an apparatus whose ostensible aim is to hold accountable perpetrators of race and sex discrimination. Yet, given the history of the United States, holding perpetrators accountable requires a major economic, temporal, and political capital investment in investigatory efforts if one seeks to distinguish perpetrators from others. Little evidence suggests such investments have been made. In the best case, this has left self-nominated targets of discrimination to bring cases to court in search of justice; in the worst case, even the private right of action is denied (Alexander v. Sandoval (99-1908) 532 U.S. 275 (2001)). In the best case, once inside the court house, the logics inherent in a legal system poised to punish transgressors continue to operate, requiring plaintiffs to show a wrong was committed and to identify the delimited set of actors who committed the wrong. In short, specific, identifiable acts, committed by specific, identifiable individuals (or corporate persons) are the focus. These acts, should they be proven, constitute discrimination. Of course, because the alleged discriminatory actions often were possible because the plaintiffs were locked out of the processes within which discrimination occurred, requiring them to identify the perpetrators is at least a little ironic, regardless of how much understanding one can have about the legal imperative to limit convictions to those who engaged in illegal action. The irony deepens, however, because an intriguing similarity between dominant legal reasoning and social science research on discrimination is evident. Courts in the United States are generally reluctant to broaden the scope of punishment beyond those individuals and corporate actors (i.e., legal persons) demonstrably involved in illegal action. Yet, social scientists—especially those in fields that usually focus on groups, such as sociology and economics— seem to have adopted views of discrimination as narrow and atomistic as that of the legal system, defi ning discrimination as the act, and only the act, of one or more individual persons. At the same time, drawing on the history of this country, some often have viewed the disadvantages of blacks and women as indicative of a role for acts of race and sex discrimination in reducing their success. Still, as analysts over time attempted to identify precisely the discriminatory acts and actors that were the source of the disadvantage,

Discrimination in the Era of Contested Prejudice: Fundamental Bases / 11

they became more and more committed to seeing discrimination as only the acts of individuals. However, the focus on narrow instances of discrimination and the par ticu lar associated actors, while perhaps defensible in the case of the courts, seems to contradict basic understandings of the complexity of social action long evident in the social sciences (Durkheim [1895] 1982; Sawyer 2002). Regardless of this contradiction, analysts have continued to use this approach in their efforts to estimate the effect of discrimination in a variety of domains. Ultimately, however, this effort has foundered in the analysts’ quagmire.

The Analysts’ Quagmire Estimating the effect of discrimination has proven extremely difficult—so difficult in fact that, at present, there is not a single instance in any domain of social life in the United States where some number or range of numbers not including zero has attained the status of a defensible, scholarly, broad consensus estimate of the effect of race or sex discrimination. Given the dominant definition of discrimination, three problems stand out as presenting basic challenges to the effort to estimate the effects of race and sex discrimination. First, unlike many social phenomena of potential interest, often (but not always) the action at issue is illegal. Therefore, key actors—discriminators— tend to try to conceal their unlawful behavior. Because the behavior is unseen, therefore, its effects may be difficult to trace. Second, the set of moments in which discrimination could occur is vast. Every interaction between student and teacher, applicant and employer, customer and proprietor, and more, is subject to the possibility of discrimination. Further, noninteractions that fail to transpire because the teacher has avoided the student, the employer has avoided the applicant, the customer has avoided the proprietor, the proprietor has avoided the customer, and so forth—all these noninteractions that may never rise to the level of conscious possibility may also reflect the operation of discrimination. The vast number of such possible instances makes it difficult to identify just where discrimination occurs, making it even more difficult to identify the effects of discrimination. As if these challenges were not enough, the third problem is that basic insights on studying causality imply that it may be impossible to estimate effects of discrimination. One elegant approach to the study of causality describes the conditions for appropriate causal inference (e.g., Holland 1986). This framework, known as Rubin’s Model, has gained increasing acceptance in the social sciences, and threatens to render impossible any effort to estimate the effect of discrimination.

12 / Chapter 1

Rubin’s Model maintains that analysts need to make two classes of observations simultaneously in order to infer a causal effect. Analysts need to observe a unit—a person, a family, a classroom, a school—exposed to a treatment, and also observe the same unit simultaneously “not being exposed” to a treatment. To identify the causal effect, then, we need to observe both cases— the case where unit i is exposed to the treatment and the case where unit i is not so exposed. The difference in outcomes between these two instances is the causal effect of being exposed to treatment. In popular culture, the fi lm Sliding Doors nicely illustrates the inferential challenge social scientists confront. In this fi lm, the main character rushes to but just misses the subway. We then see what happens to her. About halfway through the fi lm, the character, apparently in a parallel universe, encounters the exact same opportunity; this time she rushes to make the train and catches it. We then see what happens to her. Without giving anything away, the difference or lack of difference in her experiences identifies the causal effect of catching the train. The problem is that we observe only one of the conditions—either we make the train, or we do not. Assuming we catch the train, we can see what occurs after we take the train, but we cannot see what would have happened had we missed the train. Because we have no access to the alternative, and thus have no alternative for comparison, we cannot estimate the effect of catching the train. This missing data is fatal to our efforts because it is always possible that, had we missed the train, we might have produced the exact same outcomes as we observed when we caught the train. Thus, lacking one of the two required observations means that we cannot be sure that the alternative we observe had any effect per se. Therefore, researchers need to observe both conditions simultaneously. Of course, this is an impossible set of observations to make given our existing ability to move through the spacetime continuum. Hence, and unfortunately, we appear trapped in one universe, and we observe only one instance. To resolve this conundrum, analysts introduce more or less plausible assumptions to allow themselves to leverage the one world they observe in order to make causal inferences about that world. To do so, they try to make comparisons between what they observe (a factual observation) and what they would have observed had the same unit been exposed to the alternative condition (a counterfactual observation). They recognize that where appropriate comparisons cannot be made, causality cannot be established. In social science research, an experiment provides the closest approximation to this condition. Experimenters randomly assign persons to treatment and control. In an experiment, analysts make an assumption that random assignment leads to equivalent groups on average, and thus the researcher

Discrimination in the Era of Contested Prejudice: Fundamental Bases / 13

can give one group the treatment, the other group the control, and then compare the outcomes of the groups to identify the causal effect of treatment. A problem arises, therefore, in the study of discrimination, because we cannot in general randomly assign people to different racial groups or genders. Under the reasoning of the dominant social scientific perspective, if we want to identify the effects of discrimination on women, we need to randomly assign persons to different sexes. Of course, we encounter difficulty because we cannot change the sex of a person without fundamentally changing the person. Hence, under the dominant social scientific framework, we cannot conduct the random assignments necessary to resolve the inference challenge posed by Rubin’s Model. Comparing the job offers received by males and by females does not identify the causal effect, because one cannot compare such definitively different units to discern a causal effect. One way this quandary is summarized is by the phrase “no causation without manipulation” (Holland 1986: 959); because sex cannot generally be manipulated without fundamentally changing the unit, we cannot identify the effect of changing it by comparing males and females. For this reason, the logic of Rubin’s Model appears to imperil the entire study of discrimination effects.

Contrasting Eras of Race and Sex Oppression The individualistic definition of discrimination is used, and these three problems occur, in the context of changed social circumstances. Formerly, blacks and women were unable to vote, and often were prohibited from jury ser vice. It should be clear that exclusions from such basic citizenship rights and duties no longer hold directly. Thus, though analysts continue to debate the issue, the weight of the evidence indicates that matters of race and gender justice have changed, such that there are fewer restrictions and exclusionary practices, even though a non-negligible and potentially powerful set of restrictions and exclusions on blacks and women remain. Macrosociological studies of race relations (e.g., Wilson 1978) and survey research on white attitudes concerning racial justice (e.g., Schuman et al. [1985] 1997) corroborate each other, as both acknowledge continuities but point to important changes in race relations. Assessments of gender inequality (e.g., Goldin 1990) as well as analyses of the logic regulating gender relations (e.g., Rhode 1989) also point to important continuity amidst unmistakable change in the realities of gender disadvantage in the United States. Yet, some social commentators overstate the change that has occurred. For example, one social critic claims to believe “that the idea that white racism is the main obstacle to black success and achievement is now all but

14 / Chapter 1

obsolete . . . the black community today is the main obstacle to achieving the full integration our Civil Rights leaders sought” (McWhorter 2000: x.). While these assertions are hedged at different points in his polemic, the consistent claim is that race relations have changed so much that discrimination against blacks no longer stands in the way of blacks’ achievement. In contrast, Derrick Bell argues that, basically, matters of race will never change in the United States, contending: I want to set forth this proposition, which will be easier to reject than to refute: Black people will never gain full equality in this country. Even those herculean efforts we hail as successful will produce no more than temporary “peaks of progress,” short-lived victories that slide into irrelevance as racial patterns adapt in ways that maintain white dominance. This is a hard-to-accept fact that all history verifies. We must acknowledge it, not as a sign of submission, but as an act of ultimate defiance. (1992: 12; emphasis in original) Considering the state of women in the United States, Susan Faludi argues for a consistency of treatment as well, writing that: Some social observers may well ask whether the current pressures on women actually constitute a backlash—or just a continuation of American society’s long-standing resistance to women’s rights. Certainly hostility to female independence has always been with us. But if fear and loathing of feminism is a sort of perpetual viral condition in our culture, it is not always in an acute stage; its symptoms subside and resurface periodically. And it is these episodes of resurgence, such as the one we face now, that can accurately be termed backlashes. If we trace these occurrences in American history . . . we find such flare-ups are hardly random; they have always been triggered by the perception— accurate or not—that women are making great strides. These outbreaks are backlashes because they have always arisen in reaction to women’s “progress,” caused not simply by a bedrock misogyny but by the specific efforts of contemporary women to improve their status, efforts that have been interpreted time and again by men—especially men grappling with real threats to their economic and social wellbeing on other fronts—as spelling their own masculine doom. (1991: xviii–xix) Christina Hoff Sommers has a different take on the backlash possibility; pointing to progress on various fronts, Sommers essentially asks, “What backlash?” writing:

Discrimination in the Era of Contested Prejudice: Fundamental Bases / 15

For years, feminist activists have been wearing buttons claiming that women earn “59 cents to a man’s dollar.” Some journals have questioned this figure: Faludi calls them “spokesmen” for the backlash . . . The actual figure for 1988 is 68 cents, both for all women and for women with a college diploma. What of the remaining gap between male and female earnings? For the gender feminists, the answer is simple: the wage gap is the result of discrimination against women. But in fact, serious economics scholars who are trained to interpret these data (including many eminent female economists) point out that most of the differences in earnings reflect such prosaic matters as shorter work weeks and lesser workplace experience. (1994: 239–240) These writers represent the optimistic and the pessimistic poles of thought concerning race and gender relations. As such, they could scarcely be more different. Yet, at base, they share a foundational misrecognition of the complexity of social relations in the era of contested prejudice. Seizing on one area of the proverbial elephant while remaining blind to the remainder, they systematically misunderstand and, ultimately, mislead.

Implications of Misunderstood Change for the Research Enterprise It should be clear that misunderstanding greatly and negatively can affect the public debate concerning race and gender justice. What may not be so obvious is the expansiveness and intractability of the damage that occurs. When scholars fundamentally misunderstand a phenomenon, they are unlikely to devise appropriate means to study key aspects of the reality they seek to probe. Their failure to develop means to study key aspects of the social world will lead to the production of incomplete analyses and the reaching of possibly completely incorrect conclusions. Yet, because of the systematic misunderstanding of the phenomenon, analysts may fail to see fundamental problems with their means of investigation; seeing less fundamental problems, and mischaracterizing the import of the problems they do see, they will be unable to alter their methods to conduct more effective analyses. In the case of discrimination, this is exactly what has happened—applying the law’s perspective on discrimination to social science research has eventuated in a situation in which there is no general method of social scientific analysis that can convincingly demonstrate any effect of discrimination. It is this dearth of consensus approaches that has produced the sum total of zero consensus estimates of the effect of discrimination. And, while some analysts continue to seek solutions to the quagmire by making modifications from

16 / Chapter 1

within the dominant paradigm, there are many signs that such efforts will never bear fruit. Were scholars isolated individuals ensconced in irrelevant ivory towers, this state of affairs simply might provide the stuff of slapstick intellectual comedy. But, because the results of scholars’ analyses often provide the building blocks for the construction of legal structures, policy options, and political positions, scholars’ failure to understand the phenomena they study eventually will imperil societal understanding of the possibility of injustice and, ultimately, any effort to construct a successful policy response. Certainly, both race and sex oppression in the United States have continuities and discontinuities over time. Yet the ability to investigate discrimination and other social phenomena surrounding race and gender justice greatly depends first on recognizing the existence and character of change, and next on constructing approaches to study the phenomena surrounding race and gender justice and injustice that make theoretical and practical sense in the changed environment. That the study of such phenomena as discrimination must be sensitive to context should be an uncontroversial point, for successful disciplines routinely respond to changed circumstances first with changed strategies of data collection, analysis, interpretation, and eventually, if the problems persist, with paradigm change. For example, mid-twentieth-century economists and sociologists conducting empirical analyses of earnings, financial well-being, and wealth often relied on a short set of simple survey questions to measure persons’ economic resources. Indeed, the 1950 census asked only six questions concerning earnings and income. However, as the number and complexity of financial investments; health benefits; corporate executive compensation plans; and federal, state, and local government transfer programs has grown, the research community has responded with the Survey of Income and Program Participation (SIPP). SIPP collects data on seventy different sources of income, and sufficient detail to allow estimation of the tax burden for respondents. The complexity of SIPP is in direct response to the growing complexity of the process of financial resource accumulation. SIPP provides an excellent example of how analysts may respond to changes in the social reality under study by developing new tools (U.S. Census Bureau 2001). Sometimes new tools are insufficient; wholly new paradigms need to be developed. Of course, any effort to pursue a new paradigm may be fraught with controversy, as practitioners in the field will disagree about the need for a new conceptual scheme. Physicists recently encountered just such a challenge. Two major theories of physical existence are wildly successful, yet incompatible. Quantum physics (designed to describe the nature of the smallest particles of existence) and general relativity theory (designed to describe the

Discrimination in the Era of Contested Prejudice: Fundamental Bases / 17

largest scales of the universe) do not cohere. As Brian Greene writes, for some this is a problem, but not for others: the incompatibility between general relativity and quantum mechanics becomes apparent only in a rather esoteric realm of the universe. For this reason you might well ask whether it’s worth worrying about. In fact, the physics community does not speak with a unified voice in addressing this issue. There are those physicists who are willing to note the problem, but happily go about using quantum mechanics and general relativity for problems . . . as their research requires. There are other physicists, however, who are deeply unsettled by the fact that the two foundational pillars of physics as we know it are at their core fundamentally incompatible, regardless of the ultramicroscopic distances that must be probed to expose the problem. The incompatibility, they argue, points to an essential flaw in our understanding of the physical universe. This opinion rests on an unprovable but profoundly felt view that the universe, if understood at its deepest and most elementary level, can be described by a logically sound theory whose parts are harmoniously united. And surely, regardless of how central this incompatibility is to their own research agenda, most physicists find it hard to believe that, at rock bottom, our deepest theoretical understanding of the universe will be composed of a mathematically inconsistent patchwork of two powerful yet conflicting explanatory frameworks. (1999: 130) As Greene suggests, some may regard a new paradigm as necessary even as others remain committed to the dominant approach. And, as physicists are learning again, a staple fact of systematic research is that methodological and theoretical tools often apply in context and may become so inapplicable as contexts change or confront each other that new paradigms may be needed. That this possibility may also exist when it comes to measuring the effects of discrimination should, therefore, be a banal observation. However, though analysts have debated whether matters of race and sex oppression have changed over time, few of those contending that matters have changed seem to have considered the ramifications of the changes they emphasize for the tools analysts use to study the social world. But, it would be unlikely that massive change could occur in race and gender relations, but leave the investigatory technology forged in an earlier period equally useful for students of the later period. And if the investigatory technology and the theory undergirding it must change, then the needed change may be sufficiently broad and deep as to perhaps constitute a paradigm shift.

18 / Chapter 1

Hence, determining whether and how the facts of race and gender relations have changed or remained stable is not only of substantive interest. Of course, the question of change or stability is important simply in terms of establishing the facts of the matter. However important this question is on its own, it has even more important indirect ramifications owing to its implications for the continued adequacy of the methods and underlying theories analysts currently marshal to study discrimination in the United States. If it turns out that matters have changed, then it is quite possible that different approaches of investigation are needed. And, if analysts have been using methods designed for the earlier period, then it is easy to imagine that contemporary social scientists, working in good faith, could be using out-dated approaches to studying discrimination effects in their effort to document the impact of discrimination. If so, two observations follow. First, if the above proves true, one does not need to question the motives of analysts or posit hidden agendas to explain the failure of discrimination researchers to establish firmly an estimate of the effects of discrimination. Instead, basic sociology of knowledge underlies the failure of the endeavor, and opens the door to devising an inclusive resolution. Second, if the changed context calls for changed research approaches, it may necessitate a new paradigm, with a wholly different definition of discrimination, perhaps a reconfigured constellation of phenomena that count as discrimination, and a different understanding of what does not constitute discrimination. I contend that matters have changed, not as dramatically as McWhorter and Sommers indicate, but more so than Bell and Faludi recognize. A pattern of partial change, of course, is likely to be characterized by confusion, as different persons highlight different aspects of the complex reality. It is just such a pattern that might so complicate on-the-ground matters as to require new theoretical tools for discerning discrimination. Seen in this way, one such set of tools is particularly promising, based in redefi ning discrimination as more than the isolated acts of individuals, as, instead, a fundamentally social relation.

Discrimination as a (Damaged) Social Relation Gunnar Myrdal ([1944] 1962) highlighted a contradiction between the ideals of individual freedom and equality and the practice of oppressive discrimination, terming this predicament an American dilemma. For society to move beyond the epoch of condoned exploitive relations may have required an intense focus on the contradiction between rampant exploitation of blacks and women on the one hand, and the rhetoric of individualism and democracy on the other. Yet, moving forward from the era of contested prejudice may

Discrimination in the Era of Contested Prejudice: Fundamental Bases / 19

require analysts and the wider society to first move beyond the narrow individualistic conception of action to fully appreciate the social relations that make up discrimination. Regardless of this political possibility, conducting effective analyses of discrimination effects certainly requires bursting through the limitations inherent in the dominant, individualistic approach to discrimination research. Thus, in this case what may be true in the world of politics surely is true in the research enterprise; in order to estimate discrimination effects in the era of contested prejudice, we must appreciate the deeply social character of discrimination. Recall that the dominant social scientific approach essentially depends upon identifying the specific set of discriminatory actors. This is in contrast to the norm in social science; in general, social scientists are not concerned with identifying the specific actors in a situation. So, for example, social science as such is not concerned with whether the Realtor, the home-seller, or the bank stopped the mortgage transaction; nor is it concerned with whether the school board, the superintendent, the principal, the vice principal, or some consultant drew the lines that construct neighborhood catchment areas for schools. Instead, social scientists usually proceed to study the characteristics associated with continuing the transaction or constructing certain boundaries, not the specific persons or roles involved in stopping the transaction or drawing the boundaries. Surely, some questions may be pursued to the point of identifying actors at that level (e.g., who decided to fire on students at Kent State on May 4, 1969? did the president authorize torture?), but the vast majority of social scientific research efforts never approach this level of specificity. Yet, in the case of discrimination, this level of specificity is par for the course, as researchers routinely agonize over whether, for example, residential segregation is owing to credit market discrimination on the part of banks, housing discrimination on the part of Realtors, labor market discrimination on the part of employers (which reduce the fi nancial resources of targets of discrimination), and more. Indeed, with discrimination, even analyses that pursue this level of specificity are oft-regarded as insufficiently sophisticated in that they are criticized for not going far enough in the effort (e.g., contrast Turner’s [1993] detailed analysis and Butters’s [1993] dissatisfaction with the remaining distance between observation and act). As one might expect, attempting to draw causal connections to more and more specific sources magnifies the number of candidate sources, and thus dissipates the causal energy, greatly multiplying the difficulty of finding any part of the causal story large enough to sustain the claim that it is above zero in magnitude. Further, by magnifying the expectation of specificity, one also magnifies, possibly exponentially, the set of alternative explanations that could be entertained—any one of which may be sufficient to drive the estimated magnitude even closer to zero—while, at the same time, decreasing the precision

20 / Chapter 1

of the story, blurring matters even more, such that the probability of finding a defensible estimate of discrimination effects above zero tends toward zero itself. In contrast to the individualist approach that has led to the analysts’ quagmire, one might adopt a more social view of discrimination. A truly social view of discrimination will recognize that no one person ever discriminates against another alone. Instead, discrimination is necessarily a social relation that transcends the actors most obviously and directly implicated to involve the groups from which they come and the wider society as well. Further, a discriminatory social relation entails a limiting view of persons that serves to constrain the actions of some parties in a disadvantaging manner. As the social relation entails a truncated view of personhood for some, it is characterized aptly as a damaged social relation. Of course, one could dismiss the claim that social relations matter. Indeed, one hallmark of social science in the United States is its decidedly individualistic emphasis. But note that many eminent voices point to the transindividualistic nature of action, often concerning issues other than race and gender injustice. For example, in Bowling Alone political scientist Robert Putnam (2000) laments the passing of civic involvements that connected individuals to fundamentally social networks. Putnam maintains that this decline in the amount of social capital has created major problems in U.S. society, lowering the quality of life overall. R. H. Coase (1959) adds transaction costs to neoclassical economics, simultaneously making it possible that advantageous contracts that benefit individuals never will be entered because the costs of negotiating the contract may push the benefits below the threshold needed to enter the contract, and pointing analysts to consider the extraeconomic context in which contracts may be negotiated. This extraeconomic context is arguably the noneconomic dimensions of the social. And, it should go without saying, sociologists long have regarded analyses that abstract individuals from their social context as, at best, way stations on the road to understanding (e.g., Coleman 1988), despite the challenges one encounters in estimating the role of context (Hauser 1970; Manski 1993). Of course, all these currents occur amid countercurrents and debate. But, the point is that the par ticular claim—discrimination is a social relation and thus one cannot understand its operation and effects by using an analytic lens that reveals persons only as atomistic individuals—is not necessarily a radical departure. Quite the contrary, the perspective is consistent with aspects of all these currents in mainstream social science. The conclusion is clear: A great deal of mainstream social science is consistent with the proposition that, to understand discrimination, one must deeply appreciate that it

Discrimination in the Era of Contested Prejudice: Fundamental Bases / 21

entails a social relation, indeed, a fundamentally damaged one. That mainstream social science has not incorporated this definition into the study of discrimination is not because the theoretical resources for doing so are inconceivable. Viewing discrimination as a damaged social relation involving all parties available in the wider environment, rather than as simply the transgression of one atomistic actor against another, dramatically improves our ability to estimate the effects of discrimination: As I will show, it responds to all three of the problems that make up the analysts’ quagmire. Thus, taking seriously the social character of discrimination opens the door to estimating the effects of discrimination not only on a wide variety of important outcomes but, what is more, also estimating the effects of antiblack discrimination on whites, and the effects of antiwoman discrimination on men. The possibility of such effects rarely is acknowledged, but the reconceptualization of discrimination required to study effects of discrimination in the era of contested prejudice both necessitates and facilitates investigating the effects of discrimination on those who occupy both poles of the damaged social relation. With methods formed from and for the era of contested prejudice, such an investigation is exceedingly straightforward, and offers potentially surprising results.

Concluding Remarks The foregoing text traces a subset of the major points that systematically are developed in the pages of this volume. I contend that we live in an era of contested prejudice, yet we have been using methods and logics for understanding discrimination that were constructed to reveal the truth of discrimination during an epoch of condoned exploitive relations. The era of contested prejudice necessarily is confusing, because it stands between a past of brazen exploitation and a possible future characterized by the irrelevance of race and gender for educational, economic, and physical well-being. The confusion endemic to this time inhibits efforts to construct analytic techniques based on the phenomenon of discrimination as it has occurred during this epoch. But such techniques can be devised; in the following chapters, I articulate a theoretical foundation for just such an approach. In the remainder of this volume, I first support my contention that the best way to understand this time is to view it as an era of contested prejudice. I begin by reassessing the implications of publicly available data on the prejudicial attitudes of whites and men. This effort is crucially important for a phenomenon like discrimination that often provokes blame and anguish, for the effort will clarify how those of different sociodemographic groups, including social scientists, may experience systematically the same world in

22 / Chapter 1

vastly different ways, and thus draw dramatically different conclusions from the same reality. After establishing a fundamental asymmetry of experience that, I contend, underlies cross-group misunderstanding, in Chapter 3 I describe the sociohistorical factors that led to the emergence of the era of contested prejudice. The conflicted nature of the process by which this era came to be gives good reason to eschew both the optimistic and pessimistic visions of race relations and gender inequity in the United States. This chapter, as well as Chapter 2, serves to buttress my claim that sociology of knowledge, not maliciousness or ambivalence, best explains how the dominant perspective on discrimination came to hold sway despite its limitations for contemporary inquiry. Thus, Chapters 2 and 3 set the sociopolitical context for our epistemological turn. Chapter 4 begins to make that turn in earnest, describing the dominant legal perspective on discrimination and analyzing the legal system’s response to an allegation of discrimination. Chapter 5 presents a critical legal perspective on discrimination, and reanalyzes an allegation of discrimination to reveal the distinction between critical and dominant legal perspectives. Taking the conflict between different legal theories as a resource, Chapter 6 establishes that the social scientific approach to studying discrimination has more affinities with dominant legal reasoning concerning the defi nition of discrimination than with a potentially more useful critique of dominant legal perspectives. After showing the limitations the dominant social scientific perspective creates for the social scientific study of discrimination in the current era, Chapter 7 develops a deeply social perspective for discrimination research in the United States, and assesses the implications of this view for two coherent existing philosophical perspectives. Chapter 8 reconfigures discrimination research given the epistemological implications of the social definition of discrimination, while Chapter 9 closes the volume with reflections on the possibilities and perils raised by the goodwill of social scientists who have sought to estimate effects of discrimination in the United States. Before such reflections can occur, it is important to concern ourselves first with the thesis that sets the foundation for our theoretical analysis. Do we really live in an epoch that can be termed an era of contested prejudice? And, if so, what are the implications of this era for the systematic effort to understand the effects of race and sex discrimination?

2 Experiential Realities and Public Contestation

I

n order to make our way through the confusion engulfi ng the present day, we must come to appreciate that this is an era of contested prejudice. Chapters 2 and 3 will articulate fully this proposition. Chapter 3 outlines the history that has led us to this era, whereas this chapter focuses on documenting some of the important conditions that determine persons’ experience of the period. Documentation of current experience occurs first, for in this way we can establish important facts that should sensitize us to the difficulty that lies ahead, before we entertain the history, sociolegal critique, and implications of both for social scientific research. The accuracy or inaccuracy of the characterization of the present period as an era of contested prejudice is consequential. Certainly, many of the difficulties with estimating the effects of discrimination are well known. Yet, if the characterization I offer of this period, a characterization that highlights this era as fundamentally different from an earlier epoch, is incorrect, then it is at least plausible to contend that analysts who continue to employ the dominant definition of discrimination and simply adjust and fine-tune existing strategies of analysis eventually will resolve the problems that currently hound research on discrimination. However, if the contemporary period is different, and if the characterization of the current period as an era of contested prejudice is correct, it will mean that maintaining the dominant definition and continuing to

24 / Chapter 2

tweak existing strategies alone will never produce an effective means of estimating the effects of discrimination in the United States. Instead, what is needed is, for want of a better term, a paradigm shift. As I believe the latter correctly reflects our predicament, the first order of business is to establish that this characterization aptly captures the essential features of this period. Thus, whether we live in an era of contested prejudice or not is an important issue for our investigation. What we establish on this question will help us to understand how it is that different, well-meaning scholars and citizens may live in the same epoch characterized by the same reality concerning discrimination, yet have vastly different experiences of that reality and thus draw diametrically opposed conclusions about the time in which they live. And it will help us see how misunderstanding may be based in such different experiences and how people, attempting to make sense of matters from their par ticular social locations, may come to regard others as insincere, manipulative, or simply malicious when disagreements about these issues become visible. This treatment is, therefore, key to our epistemological project, for here we will trace the outlines of an important impediment we face owing to the character of our time. In any matter, few find it straightforward to adopt perspectives that fail to match their experience of reality; with discrimination, this difficulty likely is heightened. With respect to discrimination, therefore, the different experiences people have will mean that any single definition of discrimination we propose may be unlikely to easily resonate with a wide and sociodemographically diverse set of scholars, in part owing to the systematically discordant experiences they bring to the enterprise. However, by becoming sensitized to the discordant nature of these experiences, we may become better able to move forward despite the impediment our diverse experiences erect. Therefore, this chapter not only focuses on beginning to establish the character of the age in which we live and work—a task that, owing to the many problems that are widely seen to attend the analysis of discrimination, might not appear essential if one still believes the problems can be resolved within the existing paradigm—but also focuses on sensitizing social scientists to the differences in experiences they and their fellow citizens bring to the issue of discrimination, differences that, absent some consciousness and reflection, can easily make one person’s sense be mere nonsense to another. Thus, the work of this chapter is crucial for preparing the way for the kind of cross-group dialogue necessary for any consensus defi nition of and response to discrimination to emerge. Consequently, this effort accomplishes essential preparatory work for the ultimate success of our aim. I proceed, therefore, on the belief that to appreciate the character of this moment we must first reconsider both whites’ attributions about blacks and

Experiential Realities and Public Contestation / 25

adults’ attitudes on the exercise of women’s rights, and then expose the submerged implications of these views for the experience of blacks and women. These usually invisible implications underlie the gulf between those who see progress and those who see stagnation or retreat in race and gender inequality. Yet, once we excavate these implications, it will be clear that those who see stagnation are closer to the truth of race and gender injustice, but still noticeably far away.

Submerged Implications of White Attributions Concerning Blacks A reconsideration of survey research evidence on the attitudes of whites toward race prejudice and equity provides the key race-related evidence bearing on the character of our time. Survey research, like all research, has its limitations. First, and with par ticular reference to race relations, the time series of national data on racial attitudes begins during World War II, missing a great deal of social change. Second, differences in question wording can alter the meaning of questions in subtle ways; if questions differ, then some comparisons technically become inappropriate. Third, even when questions do not change, uncontrollable changes in the meaning of words can alter the meaning of questions. This latter problem bedev ils much research, but in the case of attitudinal data it may be especially problematic. The most definitive broad studies of racial attitudes to date are those conducted by Howard Schuman et al. ([1985] 1997) and the National Academy of Sciences report titled A Common Destiny (Jaynes and Williams 1989). These studies probed the complexities of attitudes toward blacks as well as toward policies designed to address racial inequality. On the basis of their analysis, Jaynes and Williams conclude that race still matters, and significant forms of resistance to policies redressing racial inequality remain. Yet, progress in attitudes and behavior is real, and often remaining resistance is motivated by factors other than race. In short, they support a more complex reading of the realities of race in the United States than either an unabashed optimist or pessimist can embrace. They draw their conclusions partly on the basis of an expansive analysis of a large set of survey questions. Indeed, a wide variety of survey questions have been treated in the trend analyses for racial attitudes; analysts have studied, among other issues, whether whites are willing to live near, work with, travel beside, or intermarry with blacks. The set of questions one could consider is thus large. However, a key motivation for even asking questions about race concerns is directly traceable to the stubbornness of racial inequality. And, analysts regard racial attitudes as potential factors in processes that maintain racial inequality. How respondents explain the disadvantage

26 / Chapter 2

of groups affects whether they believe members of the groups are deserving of assistance or not, and whether they believe assistance to members of the groups would make any difference. Further, evidence suggests that a policy can be tainted when it is identified with a group regarded as undeserving; Martin Gilens (1996) shows that opposition to welfare is connected to the degree to which the policy is racialized in the mind of citizens. Thus, racial attitudes appear to play an important role in molding the terrain upon which policy proposals are contested, affecting not only support for race-targeted policies but also evaluations of non-race-targeted means-tested programs. Given these relations, a promising lever for considering racial attitudes is provided by trend data on whites’ explanations for black disadvantage. Largely comparable trend data since the mid-1970s exist because researchers at the National Opinion Research Center (NORC) have asked respondents their explanation for blacks’ failure to reach socioeconomic parity with whites as part of the General Social Survey (GSS). The GSS has provided the nation’s social seismograph, tracking the landslides, tremors, and quakes in the social world on a continuing basis. Data from the GSS have been used in over five thousand studies spanning the disciplines of economics, sociology, political science, demography, and more (General Social Survey 2001). The GSS is the premier source for nationally representative data on attitudes, and thus provides the best data for our reconsideration. The GSS offered respondents four ways to explain blacks’ failure to reach socioeconomic parity with whites, two of which blame blacks: One response is that blacks “have less inborn ability to learn,” and another is that blacks “just don’t have the motivation or willpower to pull themselves up out of poverty.” The first explanation might imply that social policies to close the gap will be useless, whereas the second might imply that social policies to close the racial gap in outcomes would help the undeserving. Either way, support for race-targeted assistance and perhaps other policies may be undermined, at least partially, for those who hold such views. How many whites hold such views? Schuman et al. ([1985] 1997) find that, as of the early 1990s, whites’ most common explanation for persistent black socioeconomic disadvantage was that blacks lack willpower, and the least common explanation whites accepted was that blacks lack ability. Thus, it appears that whites are more likely to regard blacks as undeserving of help than as unable to benefit from assistance. Figure 2.1 illustrates changes in the proportion of whites who select either of these two explanations for black poverty between 1977 and 2000 (persons were allowed to select more than one response out of the four provided in the question). Obviously, the late start to the series misses a great deal of social change. Some analysts have pooled data collected from several different data collection efforts to extend the series backward in time, but as

Proportion Explaining Black Disadvantage 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1

Experiential Realities and Public Contestation / 27

1975

1980

1985 Year

Low Ability

1990

1995

2000

Lacks Willpower

Fig 2.1 Proportion of whites using “low ability” and “lacks willpower” explanations of black disadvantage.

Schuman et al. ([1985] 1997) note, such an effort is fraught with peril. However, by any reading the “low ability” explanation is preferred rarely, while the “lacks willpower” explanation generally is trending downward. Thus, the omission of the series prior to 1977 does not alter the general point—antiblack attitudes have declined. A brief word about the figure is required, however, for it contains features that will become more relevant as we proceed. Note that the solid and dashed lines do not connect the symbols. Instead, the lines move through the relevant series. The technical basis for the smoothed lines in the figure is provided in Appendix A, but suffice it to say here that smoothing is useful when data are sparse. In this case we have data only for some years, and sometimes not even consecutive years. Later figures will be even more sparse, for often we will have fewer than ten respondents for any given year upon which to base the analysis. The smoothing technique can aid our use of such sparse data, for it essentially averages across nearby years to produce an estimated curve. The resulting curve reflects an estimated proportion of whites holding limiting views of blacks. Figure 2.1 reflects the attributions of whites as a whole. Yet, some whites play potentially important roles in the maintenance or dissolution of black disadvantage by virtue of their occupations. Thus, the first step to revealing

Proportion Using Low Ability Explanation 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1

28 / Chapter 2

1975

1980

1985 Year Realtors Police

Fig 2.2

1990

1995

2000

Teachers

Proportion of whites using the “low ability” explanation, by occupation.

the submerged implications of explanations of racial inequality is to estimate trends in the racial attitudes of important authorities blacks may encounter. Three important authorities blacks may encounter are white Realtors, white teachers, and white police officers. Adults, including parents, encounter Realtors when they seek to move into neighborhoods, students encounter teachers in school, and age is no bar to potentially encountering the police. Figure 2.2 graphs the proportion of white Realtors, teachers, and police who believe that lack of ability explains black socioeconomic disadvantage. Note that this figure is smoothed, which de-emphasizes extreme and likely spuriously high or low values in the estimates owing to small sample size for any given estimate. Thus, these data are tailor made for smoothing, as, for example, only nine Realtors provided useable answers to the relevant questions in the 1977 sample, and only four did so in 1988. The advantage of smoothing is that it avoids extreme and unlikely swings in the estimates that would occur simply because some years provide imprecise estimates by themselves. But smoothing works best if extreme values are not observed for the first year in the series; in this case, we do observe an extreme value in the first year of the series, in that zero of nine Realtors in the 1977 sample explained black disadvantage as a result of blacks’ alleged low ability. It is unlikely that not a single white Realtor in the United States in 1977 explained black disadvantage on the basis of an alleged lack of ability—the smoothing

Experiential Realities and Public Contestation / 29

reflects this fact in that the line in the figure is imperceptibly above zero. Even with this limitation in the data, however, the smoothed results are more likely to accurately reflect the attitudes under study than the raw proportions, so I report smoothed results. That noted, attitudes of police dramatically changed during the 1977–2000 period. And even though teachers began the period with a low proportion claiming blacks’ low ability explained black socioeconomic disadvantage, a smaller proportion of teachers made that claim by 2000. Realtors, however, appear to be different, being more likely to claim blacks’ low ability as an explanation for inequality than the general population of whites by 2000. Figure 2.2 isolates key occupations for analysis, but doing so produces evidence as muddied as any surveyed by optimists and pessimists. Again, however, delving a little below the surface of these results will pay dividends in understanding persons’ experience of race in the United States. Figure 2.3 uses the statistical results reflected in Figure 2.2 to calculate the chance that a black person will encounter white authorities who explain black disadvantage as resulting from the low ability of blacks. Figure 2.3 assumes that police patrol in pairs, and a realty transaction such as a home purchase will need two Realtors. The pattern evident in Figure 2.3 is a raised version of Figure 2.2; the highs are higher, and the lows are higher as well. Given the proportion of police who claimed that black disadvantage flowed from the low ability of blacks, a black person who encountered two white police officers had, in 1977, about a 75 percent chance of meeting a duo containing at least one officer who believed black disadvantage was based in alleged black inabilities. Chances of encountering a duo with at least one member believing in the low ability of blacks were substantially lower by 2000. The probability stood at 20 percent in the year 2000. Note, however, that this was substantially larger than the proportion of white police on the force who harbored such views. Similarly, even though by 2000 only approximately one out of five Realtors claimed black disadvantage was based in blacks’ low ability, a black person dealing with two white Realtors had over a 40 percent chance of encountering a Realtor who harbored such views. In short, Figure 2.3 indicates that the common focus on raw proportions of whites holding unflattering views of blacks actually understates the implication of those attitudes for the experience of blacks who require support from the authority figure. To develop the estimates that provide the data for Figure 2.3 as well as many later figures, I used a basic statistical calculation and, in doing so, made some simplifying assumptions. The basic statistical calculation estimates the probability that a black encountering the police will meet one white police officer out of two who views blacks as lacking in ability, and the probability that a black encountering the police will meet two white police

Probability of Encountering Doubting Authority 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1

30 / Chapter 2

1975

1980 Two Realtors

1985 Year

1990

1995

2000

Two Police Officers

Fig 2.3 Probability that a black person encounters authority who doubts blacks’ ability, by occupation.

officers out of two who view blacks as lacking in ability. The statistic in the figure is the sum of those probabilities. One assumption of the calculation is that the authorities one encounters are independent of each other. Relaxing this assumption would change the results, but not clearly change the finding. For example, if we assume that police sort themselves so that partners share attitudes about blacks, this would lower the probability that a black person will encounter a police officer who doubts the ability of blacks, all else equal. However, if such sorting occurred, should a black person encounter one police officer who doubts the ability of blacks, the partner would also likely doubt the ability of blacks. Thus, relaxing the assumption of independence lowers the incidence of such encounters but raises the intensity of the exposure. Therefore, the assumption is important, but changing it does not change the qualitative result. A second assumption is that the baseline proportion of whites holding certain attitudes is unchanging for the duration of the encounter. For some figures this assumption is clearly incorrect (see Figure 2.4), but relaxing this assumption would not alter the basic finding. Thus, this simplifying assumption is maintained. Finally, the calculation reflects mean probabilities; the data are far too sparse to calculate varying probabilities by region, urbanicity, or other factors.

Probability of Encountering Doubting Teacher 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1

Experiential Realities and Public Contestation / 31

1975

1980

1985 Year

Twelve Teachers

Fig 2.4

1990

1995

2000

Thirty Teachers

Probability that black student encounters teacher who doubts blacks’ ability.

However, considerable variation in exposure probabilities may exist, and may indeed account for some of the variation analysts observe in blacks’ and women’s views on prospects for social justice. Compared to Figure 2.3, Figure 2.4 provides an even starker and arguably more troubling picture, not least of which because the figure concerns the experience of children. Recall that negligible proportions of teachers claimed that black disadvantage was based in the low ability of blacks; about 4 percent of teachers held such views in 2000. Yet, because students commonly encounter at least twelve focal teachers during their passage from grade one to twelve, and perhaps as many as thirty (or more) when curriculum differentiation at the high school level is considered, the likelihood that a black student who studied with white teachers will encounter at least one white teacher who doubts his or her ability simply because of race at some time during elementary and secondary school is shockingly high. Black students are very likely to encounter one or more white teachers in the classroom who doubt their inborn ability to learn during their transition from childhood to adulthood. These results support the view that black students may encounter potentially unsupportive and perhaps racist teachers in school. Existing research from the early 1980s indicated that teachers were less likely to be prejudiced

Proportion Using Lacks Willpower Explanation 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1

32 / Chapter 2

1975

1980

1985 Year Realtors Police

Fig 2.5

1990

1995

2000

Teachers

Proportion of whites using the “lacks willpower” explanation, by occupation.

than those in other professional occupations (Lacy and Middleton 1981). The results in Figure 2.2 show that teachers were exceedingly unlikely to explain black disadvantage as a result of blacks’ innate inability to learn. Yet, even so, many black students are likely to encounter such teachers. The probability that prejudicial teacher attitudes will become reflected in overt behavior is unknown. What is known, however, is that teacher expectations, of whatever origin, matter for students’ achievement (e.g., Kenealy, Frude, and Shaw 1991; Saracho 1991; Jussim and Eccles 1992; Kolb and Jussim 1994).The low incidence of teachers explaining black socioeconomic disadvantage by recourse to a belief in black inability eventuates in exposing a great deal of black children to teachers who doubt their ability; this state of affairs cannot have positive implications for the success of black students. Figure 2.5 graphs the proportion of white teachers, Realtors, and police officers who explain black disadvantage as owing to blacks’ lack of willpower. Recall that Schuman et al. ([1985] 1997) report that lack of willpower is the leading explanation whites offer for persistent black socioeconomic disadvantage. Still, throughout the period, the proportion of police and teachers who held such a position declined. The proportion of Realtors who held such a view declined, rebounded, then declined again, such that by 2000 approximately 50 percent of Realtors doubted blacks’ willpower.

Probability of Encountering Doubting Authority 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1

Experiential Realities and Public Contestation / 33

1975

1980 Two Realtors

1985 Year

1990

1995

2000

Two Police Officers

Fig 2.6 Probability that a black person encounters authority who doubts blacks’ willpower, by occupation.

Again, however, considering only the proportion of authorial whites who doubt the willpower of blacks understates the experience of blacks who encounter white authorities in important transactions. Figure 2.6 shows that, in the mid-1990s, blacks who encountered a pair of white police officers had about an 80 percent chance of encountering at least one functionary of order who traced persistent black disadvantage to presumed lack of motivation on the part of blacks. And, even as late as the year 2000, blacks who entered a transaction with two white Realtors had a better than three-in-four chance of dealing with a Realtor who doubted the strength of blacks’ willpower enough to explain persistent black disadvantage on that basis. These results again are undeniably pronounced, but pronounced as they are, they pale in comparison to the results for white teachers and black students evident in Figure 2.7. The figure appears nearly blank, save for lines barely visible at the top. The solid line indicates that black students are virtually assured of encountering at least one teacher in grades one to twelve who doubts the willpower of blacks, if the students have a total of twelve white teachers during those years. If they have thirty white teachers, it is so likely that they will encounter at least one teacher who explains persistent black disadvantage as a result of low motivation on the part of blacks that, given

Probability of Encountering Doubting Teacher 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1

34 / Chapter 2

1975

1980

1985 Year

Twelve Teachers

1990

1995

2000

Thirty Teachers

Fig 2.7 Probability that black student encounters teacher who doubts blacks’ willpower.

the resolution of the figures in this chapter, a line discernibly different from certainty could not be drawn! Probabilistic statistical results of social phenomena so rarely produce an appropriate estimate of virtual certainty that such results usually indicate something is wrong with the analysis. Thus, the above result certainly is worthy of protracted reanalysis. And, surely, there are complexities that the above analysis has set aside. Yet, many of these complexities would not alter the basic finding nor reduce what should be societal concern about the implications of the results. For example, clearly, the vast majority of black students do not have only white teachers. Indeed, evidence suggests black students are more likely to have black teachers than nonblack teachers. Ronald F. Ferguson (1998: 357) reports that for school districts in Alabama, the correlation between percentage of students who are black and percentage of teachers who are black is .91. Thus, black students are unlikely to be exposed to only white teachers. However, using this evidence to question the implications of the results above is a mistake, for doing so is to imply that there is no societal cost if black students are protected from the potentially subtly destructive experience of studying with white teachers only by the proportion of black teachers and the tendency of teacher assignments to match black children with black teachers.

Experiential Realities and Public Contestation / 35

This is not only a backdoor defense of segregation—problematic, for it implies black students will be okay as long as they and black teachers stay in their place and out of institutions dominated by whites—but the position is also incorrect, for it ignores that white teachers may stand in judgment of black students in many important pedagogical moments even when they do not directly teach them (e.g., on the playground, during field trips, in the lunchroom, and more). Also, white teachers teach and socialize white students, and can demonstrate disdain or support for black achievement, even if the classroom is lacking in black students, by how issues are discussed and even by the selection of topics for attention. Further, teachers make decisions about where they will teach as well, and evidence suggests teachers have lower job satisfaction in diverse schools (e.g., Freeman, Brookhart, and Loadman 1999), a result that may encourage white teachers to seek districts and schools that are overwhelmingly white. This set of relationships and actions could transform the observation that black students tend to have black teachers from a staid result of perhaps nonracialized assignment processes into possibly a manifestation of some white teachers’ discomfort with or disdain for black students. Both processes noted above can have a negative impact on black students. Thus, though one may point to complexities in the proportion of white teachers a black student is likely to have in the course of his or her educational career, the complexities belie other intricacies that restore the qualitative judgment—even with the low incidence of white teachers who doubt the innate abilities of blacks, the attitudes teachers do hold are enough to greatly and negatively affect the learning environment of black students. Hence, at this juncture and while we await detailed reanalyses of the result above, the result—black students are virtually certain to encounter white teachers who doubt blacks’ willpower—is worthy of a brief pause and some reflection on the meaning of the results for black students’ experience in school. What do these results mean for the social world? Principally, for our epistemological concerns, they indicate that the on-the-ground reality experienced by blacks is represented poorly by tallies of whites’ attitudes toward blacks and racial issues. When we consider an obvious, though oft-ignored, feature of social life—namely, the degree to which encounters with authorities are not singular but plural exposures—the failure of the effort to move directly from estimates of the incidence of white prejudice to the claim that matters have improved substantially for blacks is explained easily. Observe first that these attitudes are held by whites, some of whom occupy positions of mundane power—power deriving from the regular and rarely spectacular exercise of their occupational duties on a day-to-day basis. Then note that the organization of social, economic, and political life often necessitates repeated

36 / Chapter 2

interaction with a string of such authorities, as in the case of education, or requires one to do business with those one cannot select, as in the case of Realtors (i.e., the Realtor for the other party to the transaction) or police. Taken together, these features lead to the conclusion that blacks in school, in the economy, and in the criminal justice system navigate a minefield of white authorities who doubt their abilities and question their motivation. Like any minefield, the absolute density of threats may be low, but the likelihood of encountering the threat depends on both the threat density and the duration of risk, and the costs of encountering the threat are potentially high. Hence, unlike the anecdotes of a pundit, an analysis of high quality, systematically gathered data based on probability sampling reveals that the likelihood of encountering white authorities who harbor doubts about blacks unmistakably is high. Of course, evidence suggests that few whites harbor such negative views of blacks. Given this evidence, can anyone seriously maintain the view of pessimists: that nothing has changed, and blacks are as targeted today as they ever have been? Yet, given the evidence in Figures 2.1 through 2.7, can anyone seriously maintain the view of optimists: that race is irrelevant to the fortunes of blacks in the United States at the turn to the twenty-first century? Even granting a complex and still incompletely understood relation between attitudes and behavior, the results make it abundantly clear that the optimistic read of changes in whites’ racial attitudes grossly misunderstands the implication of those changes for the experience of blacks.

Submerged Implications of Judgments of Women’s Rights Trends in men’s attitudes concerning gender issues also have been a focus of research attention. As in the case of race, researchers have drawn several samples over many years, and asked a fairly stable series of questions concerning gender role attitudes. One question analysts have asked is whether persons approve or disapprove of a married woman working for pay outside the home if she has a husband capable of supporting her. Andrew Cherlin and Pamela Barnhouse Walters (1981) report that, as of 1972, approximately 38 percent of white men and 32 percent of black men did not approve of such labor force participation on the part of women. By 1978, the last year for which they had data, 27 and 36 percent of white and black men, respectively, disapproved. Jane Riblett Wilkie (1993) extended the assessment of this question forward, covering the 1972–1989 period, and studied responses to another question asked on the GSS. NORC also has asked respondents whether they agreed or disagreed that “Women should take care of running their homes and leave running the country up to men.” Wilkie reported that the

Proportion Saying Women Should Stay Home 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1

Experiential Realities and Public Contestation / 37

1970

1975

1980

1985 Year Men

1990

1995

2000

All

Fig 2.8 Proportion of men and women who asserted that women should stay at home and let men run the country.

proportion of males disapproving of women working outside the home continued to decline, as did the proportion of males saying that women should take care of running the home and let men run the country. Thus, existing research shows a major decline in the limiting attitudes of men about women. Figure 2.8 replicates this finding, as the solid line graphs the proportion of men who believe that women should stay home. Clearly, since the early 1970s, the proportion of men holding such positions has declined dramatically. Still, by 1998, approximately one out of six men held such limiting views of women’s rights. In the case of race it was defensible to confine attention to whites’ attitudes, not only because they are the dominant group in society, but also because blacks’ attitudes about blacks differ appreciably from the attitudes whites have about blacks. For example, in the 1977–2000 samples, not a single black teacher, black police officer, or black Realtor stated that blacks were inherently unable to achieve. Thus, combining the responses of whites and blacks would distort the on-the-ground reality. However, though women’s attitudes about women’s rights differ from those of men, they do not differ by nearly such a margin as blacks’ views about blacks’ status differ from whites. For example, Karen Oppenheim Mason and Yu-Hsia Lu (1988) report that, in 1985, approximately 51 percent of men claimed that “It is much better for

Proportion Saying Women Wrong to Work 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1

38 / Chapter 2

1970

1975

1980

1985 Year Men

1990

1995

2000

All

Fig 2.9 Proportion of people who stated that it is wrong for women to work in paid labor force if their husbands can support them.

everyone if the man is the achiever and the woman takes care of the home and family.” However, in the same year, nearly 47 percent of women agreed with the statement as well. Figure 2.8 reflects this reality, for the dashed line graphs the proportion of men and women who asserted that women should stay home and let men run the country. It is difficult to discern a difference between the two lines. This difficulty is traceable both to the nearly equal representation of men and women in the samples, and to the basic similarity in the proportion of people holding this view regardless of sex. Figure 2.9 graphs the men and everyone who believed that women were wrong to work if they had spouses who could support them. Though men and women diverged somewhat in their views early in the series, by the end even this small divergence had eroded. Thus, by 1998, fewer than one out of five men and women stated that a woman with a spouse who could support her was wrong to work. In the analysis of whites’ views of blacks and blacks’ experience, it was also fairly easy to identify key actors. However, this is a more difficult enterprise for women, as we shall see below. But, at the outset we can focus our attention on police officers and teachers. These are certainly among the set of potentially important authorities upon whom women may need to rely.

Proportion Believing Women Should Stay Home 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1

Experiential Realities and Public Contestation / 39

1970

1975

1980

1985 Year Teachers

1990

1995

2000

Police

Fig 2.10 Proportion of police officers and teachers who stated that women should stay at home and leave the running of the country to men, by occupation.

Figure 2.10 graphs the proportion of police officers and teachers who believed women should stay at home and leave running the country to men. Note the declining incidence of teachers who held a limiting view of women’s rights throughout most of the period, and the uptick around 1997. In comparison, the proportion of police officers who regarded women’s involvement outside the home as inappropriate changed irregularly throughout the period, but ended up at about the same level as when the study period began. Figure 2.11 illustrates the proportion of police officers and teachers who believed that it is wrong for a woman to work if her partner could support her. The patterns for both police officers and teachers are erratic. By the mid-1990s, nearly three of ten police officers and teachers agreed with the statement that a woman was wrong to work if her partner could support her, revealing a convergence in attitudes for these figures of authority. Again, however, the proportions in the figures do not indicate the risk to which the denigrated group is exposed in their dealings with these offices of power. Figures 2.12 and 2.13 illustrate the likelihood of a woman encountering authorities who have limiting views of women’s rights. Figure 2.12 indicates that, in 1998, a female who encountered two police officers had nearly a 40 percent chance of having to deal with one or more partners who doubted the legitimacy of her active engagement beyond the household. The implications

Proportion Believing Women Wrong to Work 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1

1970

1975

1980

1985 Year Teachers

1990

1995

2000

Police

Probability of Meeting Sexist Authority 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1

Fig 2.11 Proportion of police officers and teachers who stated that it is wrong for a woman to work if her husband could support her, by occupation.

1970

1975

1980

1985 Year

Twelve Teachers Two Police Officers

1990

1995

2000

Thirty Teachers

Fig 2.12 Probability that a female encounters authority who states that women should stay at home, by occupation.

Probability of Meeting Sexist Authority 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1

Experiential Realities and Public Contestation / 41

1970

1975

1980

1985 Year

Twelve Teachers Two Police Officers

1990

1995

2000

Thirty Teachers

Fig 2.13 Probability that a female encounters authority who states that it is wrong for women to work in the paid labor force if their husbands can support them, by occupation.

of this result for the fervor with which police officers will enforce women’s safety, especially in their own homes over and against potentially controlling and abusive behavior of male partners, cannot be known at this time. However, that a women seeking safety from an abusive partner had such a high probability of encountering a police officer whose views of her rights may match those of the partner from whom she may be seeking protection cannot bode well for the life chances of women seeking to survive the onset and continuation of threats of domestic violence and murder. Of course, one may regard any suggestion that women have something to fear from police officers as mere hyperbole. And, one may believe that few women need to worry about what might happen should they need to rely on the police. Unfortunately, research conducted by the National Institute of Justice (NIJ) and the National Centers for Disease Control (CDC) estimates that one quarter of women have been assaulted or raped by an intimate partner. In comparison, approximately 8 percent of men have been raped or assaulted by an intimate partner. Further, 5 percent of women and .6 percent of men report having been stalked by a current or former intimate. Yet, fully 80 percent of rapes of women, 75 percent of assaults of women, and 50 percent of all stalkers of women go unreported (Tjaden and Thoennes 2000). It is quite possible that one factor in the low incidence of reporting is concern

42 / Chapter 2

women may have about the treatment—perhaps dismissiveness—they may experience at the hands of some police officers. And, the analysis of this chapter suggests that if there is a connection between attitudes and behavior, which seems entirely plausible even if not fully understood, then such concern is not unwarranted. Despite the importance of the implications of the findings concerning police officers, the probability of encountering a teacher who believes women should not take leadership roles is far higher. Between 1974 and 1998, the likelihood a girl would encounter at least one teacher out of twelve who believed girls should follow varied from 35 to 62 percent. If we consider girls’ experience of a broader range of teachers, as we should to account for the way in which high schools are organized, the likelihood that a girl would graduate having been exposed to one or more teachers who doubted the appropriateness of her taking a leading role is dramatically higher; girls had somewhere between a 69 and 91 percent chance of encountering one or more teachers who believed women should not be in leadership roles. The findings for police in Figure 2.12 are largely mirrored in Figure 2.13, but the attitudes of the teachers to whom girls were exposed is such that girls appear even more likely to have encountered one or more teachers who believed it was wrong for a woman to work outside the home for pay. Considering the full set of teachers a student might have, between the early 1970s and the year 2000, girls never had less than a 93 percent chance of encountering teachers who believed a woman should stay at home and allow her man to support her financially. Recall that the proportion of teachers who held these views declined precipitously through the period and was, by 1998, approximately one to two out of ten. But these relatively low proportions translate into strikingly high likelihood of exposure for girls to teachers who doubt women’s legitimacy in the paid-labor force. All that we know about teacher expectations, and all that we know about the subtle process of socialization, suggests that this state of affairs is not positive for girls’ development. The similarity in men’s and women’s attitudes, coupled with occupational sex segregation that makes the position of elementary school teacher a mostly female occupation, made it necessary to include both men and women in considering the views of authorities and the exposure of girls and women to unsupportive members of key occupations. However, even more important is another feature of sex that creates a major difference between the implications of racial attitudes and gender attitudes. Parents and children typically are of the same race and thus both equally are targets of racial attitudes; this may create an important protective bond. In contrast, sex is different; parents may or may not match the sex of their children, and the nature of sex may bring complex gender dynamics relevant to children’s growth into the domicile itself. Thus, one profound difficulty with the above

Proportion Sexist 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1

Experiential Realities and Public Contestation / 43

1970

1975

1980

1985 Year

1990

1995

2000

Parents All

Fig 2.14 Proportion of persons saying that women should stay home and let men run the country, by parental status.

analysis of attitudes toward women’s rights is that those with authority by virtue of their occupational locations may not be the most pernicious authorities when it comes to the development of women’s capacities. The vast majority of women are first girls raised by parents, and parents (of whatever sex) who harbor limiting views of their female children may play a role in limiting the later success of women. Not only may they inadvertently socialize their girls in ways that undermine girls’ capacity-building efforts, but they also may socialize boys in ways that ultimately serve to maintain male disdain for women’s rights and personhood. Figure 2.14 attempts to focus attention on parents who, in their parental roles, may, however unintentionally, do more to threaten the success of women than anything anyone does in an occupational role. This figure provides only one step in this direction, for major obstacles stand in the way of fully analyzing the role of parents using these nationally representative data. For example, one obstacle is that the data do not allow one to identify which parents have sons and which ones have daughters, which would be useful for teasing out the mechanisms through which women’s success becomes threatened or undermined. Another obstacle is that one cannot identify which parents live with their own children, for the household composition data does not indicate whose children live in the household. Both of these limitations greatly reduce

Proportion Sexist 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1

44 / Chapter 2

1970

1975

1980

1985 Year

1990

1995

2000

Parents All

Fig 2.15 Proportion of persons saying that it is wrong for women to work in paid labor force if husbands can support them, by parental status.

the possibility of isolating the direct risk to daughters posed by parents who think women should leave running the country to men. Despite these limitations, the figure is useful to consider, for it provides our best evidence about the attitudes of key persons in girls’ potential support network. The figure reveals a large decline in the proportion of persons, parents, and nonparents who held limiting views of women’s rights. The large decline is a sign of a major social transformation. But the major transformation should not be overstated: Nearly one in five parents in 1998 believed their daughters should leave running the country to men. Whether such parents make exceptions in the cases of their own daughters cannot be known with the data available, but, if such parents act to avoid cognitive dissonance, daughters of such parents may be unlikely to receive unfettered support as they make the transition to adulthood and to acting themselves in the wider world. Indeed, even the older daughters of such parents may continue to be relatively unsupported throughout their lives. A similar finding is obtained in Figure 2.15, which tracks the proportion of parents and nonparents who believed a woman whose husband can support her financially is wrong to work. Again, nearly 20 percent of parents held such views. And, again, those holding these views may not have been

0

Probability Sibship Has Sexist Parent .1 .2 .3 .4 .5 .6 .7 .8 .9 1

Experiential Realities and Public Contestation / 45

1970

1975

1980

1985 Year Low

Fig 2.16 home.

1990

1995

2000

High

Probability that a sibship has parent(s) who believe women should stay at

as unambiguously supportive of their daughters as they may have been of their sons. Ideally we would study a child’s chance of having parents who hold limiting views of women. However, GSS data does not indicate how many children each parent has. Thus, using only the data available in the GSS, were I to try to estimate the probability that a child will be exposed to antifemale parental attitudes, I likely would produce downwardly biased estimates. Owing to more traditional attitudes about family associated with gender role attitudes, partners who feel that women should stay home and let men run the country are likely to have more children. If so, the proportion of children from such families may be large. Hence, because the data do not contain enough information to make precise estimates of probability of exposure to antifemale parents for an individual child, and because each parent can be connected to one sibship—i.e., one set of one or more children—I refer to sibships rather than children. Figure 2.16 graphs the maximum and minimum likelihood of a set of siblings having one or more parents who said that women should let men run the country. The minimum estimate is based on assuming that birds of a feather flock together; men and women pair off so as to share positions on this

0

Proportion Sibship Has Sexist Parent .1 .2 .3 .4 .5 .6 .7 .8 .9 1

46 / Chapter 2

1970

1975

1980

1985 Year Low

1990

1995

2000

High

Fig 2.17 Probability that a sibship has parent(s) who believe that it is wrong to work in the paid labor force if their husbands can support them.

issue; thus couples are formed so that as many couples as possible contain two people who share views as to whether women should let men run the nation. Under such a condition, few sibships would be exposed to such parents, but both parents most likely would have the same view, thus potentially intensifying children’s exposure to the perspective. In contrast, the maximum estimate is based on assuming that opposites attract, that men and women pair off so as to disagree on this issue; thus as many couples as possible have both views on the question represented. Likely, the true answer falls somewhere in between. The results show that, in the mid-1970s, sibships had somewhere between a 40 and 75 percent chance of having at least one parent who believed men, not women, should be running the country. By the late 1990s, sibships’ chance of having one or more parents who held such views had fallen to a lower but still sizable 20 to 35 percent. Analogously, Figure 2.17 indicates that, in the early 1970s, sibships had a 50 to 70 percent chance of having at least one parent who believed a woman was wrong to work if the husband could support her. By 1998, sibships had somewhere between a one-in-five and a two-in-five chance of having one or more parents who believed women should stay home. Figures 2.16 and 2.17 indicate that significant proportions of sibships encountered a situation where those who loved them most held attitudes antithetical to girls’ intellectual

Experiential Realities and Public Contestation / 47

and emotional development. These analyses suggest that many girls, in the privacy of their families of origin, navigate a difficult, potentially subtly unsupportive environment. Taken together, Figures 2.8 through 2.17 suggest that a nonignorable proportion of key actors in the lives of women have views of women’s status that may make them hesitate or fail to lift their hand to aid the development and success of their daughters and the women they encounter in the wider society. Those who ignore the presence and perhaps powerful implications of the action or inaction of such persons not only fundamentally mischaracterize the truth of gender in U.S. society, but by their emphases and de-emphases also provide telling evidence in favor of the feminist who claims, rightly, that the real lived experience of women routinely is given short shrift in national discourse.

Concluding Remarks This chapter presents evidence based on survey questions that have been asked for decades. Many survey researchers claim that for studying trends there is no substitute for asking the same question survey after survey (e.g., Converse and Presser 1986). Using such data, analysts have revealed dramatic changes in attitudes toward blacks by whites, as well as men’s and women’s views of women’s rights. Yet, some analysts have criticized efforts to maintain a series of unchanging questions. For example, some have maintained that whites have learned how to speak about race in ways that allow them to easily answer standard survey questions in acceptable ways while nevertheless harboring prejudicial attitudes (e.g., Sears 1988). Further, Eduardo Bonilla-Silva (2001) argues that maintaining an unchanging series of questions ignores how change in the civil rights period has altered the mechanisms through which discrimination occurs. According to Bonilla-Silva, these mechanisms make direct prejudice less important for maintaining discrimination. Thus, analyses using old-style attitudinal data both understates the proportion of prejudiced whites and mistakes such change for real progress. Yet, social psychological research suggests that many of the trend questions are well suited to measuring direct, conscious prejudice (Fiske 2000). Further, social psychological research using a variety of instruments indicates that the proportion of U.S. residents who hold strong prejudicial views is well captured by standard surveys (Fiske 2000; Banaji and Hardin 1996). Even so, I accept Bonilla-Silva’s suggestion that racial discrimination may not be primarily about the behavior of explicitly prejudiced whites, and I also accept Sears’s criticism that standard questions may no longer tap the breadth of prejudicial attitudes. A similar case can be made for gender;

48 / Chapter 2

compare, for example, the data on actual housework and the claims of men to do more housework. Julie Press and Eleanor Townsley (1998) find that men with more stated commitment to gender equality tended to over-report their housework involvement more than others, suggesting that (some) men may have learned how to talk about gender and equality, while their behavior has still left them reclining in the La-Z-Boy. These critical observations, and research indicating the questions remain effective in measuring prejudice, both concern the questions asked. The findings above, however, concern not so much the questions asked but, instead, researchers’ use of the answers respondents provide. From this analysis, it appears useful to continue asking the longstanding trend questions from the past. Usually researchers halt the analysis after tallying the proportion of different groups holding par ticu lar views, or after using statistical models to reveal the correlates of those attitudes. However, any visceral understanding of the implications of the incidence of those holding prejudicial attitudes for the lived experience of the targets is lost when the analysis stops at that point. Such an analysis forgets that investigations of change in racial prejudice and gender role attitudes are not only about the people holding the attitudes; instead, they are also about the experiences of members of groups those attitudes target. The usual analysis, therefore, ignores the question of how the experiences of targeted groups may be changing or not changing owing to the pace and direction of change in race and gender attitudes. Of course, if there were better means of measuring prejudicial attitudes, the approach used here could be used with such data. Indeed, research on persons’ implicit prejudicial attitudes, notably exploiting increasingly powerful computer technology, has deepened our understanding of how social evaluation works (e.g., Dovidio, Evans, and Tyler 1986; Banaji and Hardin 1996; Cunningham, Preacher, and Banaji 2001). The main finding from this work is that far more people are prejudiced than is revealed by old-style attitudinal questions, even as the old-style questions correlate with scores on implicit prejudice. The implications of this line of research for discrimination are still unclear, but one obvious possibility is that the figures reported herein on blacks’ and women’s exposure to authorities prejudiced against them, high as they are, may be grossly understated. Is there value to the figures presented here? Well, one scholar I informed of these findings maintained that, though the level is changed by considering persons’ exposure, the trend is the same. This claim actually is incorrect for some of the key variables considered; for example, the change in the incidence of prejudicial teachers in Figure 2.5 is generally negative, while the change in students’ exposure risk in Figure 2.7 is virtually zero. Yet, the claim misses the point at a more fundamental level: Even if the trend were the same, the figures show that while whites and men and women will ob-

Experiential Realities and Public Contestation / 49

serve a sharply declining incidence of stated prejudice in the overall society, at the same time blacks and women will continue to have unignorably high chances of exposure to authorities prejudiced against them. The mathematical function and the additional parameters that link the incidence of prejudiced persons in the society with the level of risk blacks and women have of encountering prejudicial authorities are such that even massive decreases in the former may eventuate in slight or even nonex istent decreases in exposure risk for the latter. And it is this conclusion that sits at the center of the asymmetry of experience, this conclusion that makes sense of the perhaps increasing frustration many blacks and women may express at claims that prejudice has declined in society when, as the analyses above indicate, their personal experiences easily may suggest otherwise. It does not take much imagination to see how this condition can warp the cross-group communication necessary for progress to be made by reducing everyone’s trust in the sincerity of others. When the marked declines in stated prejudicial attitudes are considered, it should be clear that the dire pessimist position cannot be sustained. There was a time of condoned exploitive relations, during which harboring or displaying prejudicial views was often a matter of pride. Thus, it is clear that there has been change. But this story is well known. What is not as well known is the implication of the changing incidence of prejudice for the experience of blacks and women. When the likely experience of blacks and women is considered, as I endeavor to do above, it should be clear that matters have changed, but not nearly enough to make the encounter with unsupportive authority figures uncommon, let alone rare enough to be ignorable. Thus, the analysis indicates that the optimist position also is unsustainable. Considered together, both sides of the analysis should make clear that, for all the documented change in the proportions of persons espousing limiting views of the capability of blacks and the legitimate behavior of women, the oft-ignored experience blacks and women have with key authorities contradicts the common conclusion concerning change in race and gender relations obtained by considering only change in the proportion of persons possessing prejudicial opinions. And, because an important motive for studying race and gender attitudes is to understand the poor treatment of targeted groups, the experience of the targets should be of central interest. Seen in this way, the analysis above—which reveals that, despite apparent unparalleled declines in prejudicial attitudes, women and blacks are still virtually assured of encountering prejudiced authorities—can only indicate that the pessimist position, forged with a recognition of the experience of discrimination and prejudice, is closer to the truth than that proposed by optimists. Disagreements stem in part, therefore, from the implications of this asymmetry of experience. Some see precipitous declines in stated prejudicial

50 / Chapter 2

attitudes, and wonder when, if ever, “blacks” and “women” will be satisfied. At the same time, the low incidence of prejudicial attitudes still means that, in the course of their day-to-day business, blacks and women will encounter authority figures whose attitudes may prevent them from providing fair service to blacks and women. Yet, when blacks and women describe their experiences dealing with prejudiced authorities, they frequently are disbelieved, often because members of their audience “know” the truth of the declining proportions of persons holding prejudicial views (e.g., see McWhorter’s [2000: 4] dismissal of a student’s report of having experienced discrimination, even though McWhorter was not present during the incident—surely, the appropriate response to any such charge is cautious consideration, neither outright dismissal nor uncritical acceptance). Knowing they are not believed, many blacks and women come to doubt the integrity and commitment to justice of those who disbelieve their honest reports of their life experiences; knowing that prejudice has declined precipitously, many whites and adults come to doubt the integrity and even the sanity of the individuals who make such allegations or indeed any statement remotely interpretable as a charge of discrimination. Hence, the asymmetry of experience, hidden by analyses of only one side of the attitudinal data, erects a difficult-to-scale wall of misunderstanding between communities. These conclusions may be interesting, but, one may ask, why establish them here? What is the point of this analysis for our larger concern with developing a means to study the effects of discrimination? For one, the analysis provides grounds to suggest that the experience of discrimination may be sufficiently widespread as to be a concern. With some prominent commentators suggesting the opposite, any effort to reconceptualize discrimination must confront the issue of whether discrimination may be serious enough to warrant reconceptualizing. The evidence shows that, even if the major basis of discrimination is not prejudicial discriminators, and even if the incidence of prejudice has declined markedly, the exposure risk remains unignorably high, suggesting that in the contemporary period the phenomenon of discrimination remains worthy of attention. Second, the analysis also is important for our larger enterprise because social scientists of the United States live in the same world we have been studying here, and thus also are implicated in the asymmetry of experience documented above. These results suggest that social scientists must become sensitive to the possibility that they may misunderstand each other when they turn their attention to discrimination. They may harbor unexamined assumptions that may trace back to their different experiences. For these reasons, social scientists must make a special effort to transcend their own par ticular experiences when their work brings them face to face with needing to evaluate the phenomenon of discrimination, even more so as they enter-

Experiential Realities and Public Contestation / 51

tain the prospect of rejecting the paradigm that has guided social science research on discrimination for over fi fty years. Special difficulty may attend that process, both because paradigm change is always difficult, and because matters that concern a social factor implicated in society-wide, thorough-going asymmetry of experience are always challenging. For these reasons, the foregoing analysis, as a means of preparing the way for cross-group dialogue, is essential. Third, the asymmetry of experience is an important resource to which we shall return much later, after we address in Chapter 3 the emergence of the era of contested prejudice, and consider in Chapters 4 through 7 the various definitions of discrimination. Its establishment here will pay dividends when we take up the epistemological implications of different understandings of discrimination most directly, in Chapter 8. Of course, the finding of an asymmetry of experience states that blacks and whites and women and men have different experiences. This idea is hardly novel; the work of Joe Feagin (1991) and Candace West and Don H. Zimmerman (1983; Zimmerman and West 1975) effectively document important differences in experience. The difference here is that the case is made using the very same data that has documented greatly declining incidences of prejudice against women and blacks. By proceeding in this manner our understanding of how different researchers may obtain diametrically opposed findings when studying the same world is deepened. Excavating data long known to show the decline of prejudice, and by doing so revealing the dramatically large risk of exposure to prejudicial authorities that blacks and women face, makes it apparent that both pessimists and optimists may have grasped some portion of the truth. Given that both have some purchase on the truth, perhaps one need not accuse anyone of acting in bad faith in order to explain discrepancies in their conclusions. Indeed, as we fully grasp the gravity of the asymmetry of experience, it becomes easier to see how persons of good will, including social scientists, may systematically pursue analytic strategies and political projects based on systematically incomplete readings of reality. And, given this realization, perhaps it will become a bit easier to set aside the defensiveness and accusatory tone that may easily come to animate the effort to study the effect of discrimination. Perhaps all scholars involved simply are attempting to move forward on challenging terrain. Perhaps what is needed at this juncture is systematic evaluation of the dominant approach as well as an alternative, an evaluation shorn of defensiveness and fear that anyone is attempting to draw any conclusions about the character of those who pursued one approach or the other in good faith. To be sure, the interstitial moment I have termed the era of contested prejudice is nicely reflected in the asymmetry of experiences noted above.

52 / Chapter 2

Indeed, it is just these asymmetries that partially underlie the contested nature of the period. Different commentators draw on subsets of the experience, confident in the essential truth of the piece they have selected. But these partial analyses, however true in themselves, remain fundamentally incomplete and therefore misleading, ultimately delaying the development of cross-community understanding. It is this easy-to-recognize basis for misunderstanding that, I submit, provides the fundamental foundation for intense, sustained, conflict surrounding race and gender relations at the turn to the twenty-first century in the United States. Of course, the period of sustained conflict to which I refer did not arrive through some mysterious process. Instead, it is the result of the challenges created by the ways we navigated larger societal transformations during the latter half of the twentieth century. Analysts have deepened our understanding of the present by showing its placement in historical context; in the same way, we may deepen our understanding of the era of contested prejudice by considering it in the context of analyses of the transformation of race and gender relations in the United States. It is to that task we now turn, a task that will provide an answer to one important question while solidifying the foundation for considering another: How did we move from a regime of condoned exploitation to an era of contested prejudice, and no further? And, most important for our epistemological focus, how did this process eventuate in a par ticular social scientific definition of discrimination becoming the dominant definition, despite the existence of viable alternative definitions at the same time?

3 From Condoned Exploitive Relations to the Era of Contested Prejudice

T

he previous chapter establishes important oft-missed implications of widely known facts of race and gender attitudes. It is clear that the proportion of whites, and women and men holding unabashedly prejudicial views toward blacks and women declined through the latter parts of the twentieth century. Some regard this decline as indicating that race and sex discrimination is a weak factor in the lives of blacks and women. Yet, Chapter 2 reveals that when blacks and women encounter persons in key positions of authority, they are likely to encounter persons who doubt their motivation, ability, or legitimacy. Thus, men and women see rising support for women, but women still are likely to confront unsupportive figures of authority. And, whites see declining white prejudice, while blacks virtually are assured of encountering hostile white authorities in the course of their lives. Because individuals are unlikely to accept a perspective on race and gender that conflicts with their own lived experience, the asymmetric experiences, and the divergent understandings to which they give birth, are key underlying supports to the rampant confusion that characterizes the present time. But asymmetric personal experience is not the only factor underlying the era of contested prejudice and the misunderstandings that are common in the period. In addition, a thoroughgoing misrecognition of the sociohistorical developments that ultimately dissolved the regime of condoned exploitive relations holds sway as well. Because the history of the

54 / Chapter 3

period is misrecognized, the changes that have occurred may seem deeper than they actually are. And, most important, misrecognition of the process by which change occurred leads to faulty diagnoses of how to improve our efforts to estimate the effects of discrimination and how to move beyond the current era of contested prejudice. For this reason it is essential to reconsider the historical facts, in order to lay the groundwork for the epistemological task that occupies the bulk of this volume, and to identify the context within which any social policy and political strategy prescriptions must be developed. Had there been an earnest and sustained effort to eradicate discrimination, it is unlikely that, at the turn to the twenty-first century, black students would be nearly certain to encounter one or more white teachers who doubt the students’ motivation, and females would be unlikely to have as high as a two-in-five chance of having parents who doubt the appropriateness of women joining the workforce irrespective of a spouse’s livelihood. But, as I show below, no such sustained effort was ever made. An accurate characterization of the contemporary period based on a reconsideration of mid- to late-twentiethcentury history and a rereading of existing macrosociological analyses of relevant economic transformations will make clear that we stand in this interstitial moment because the confluence of events and logics that brought us here cannot, by themselves, move us beyond the era of contested prejudice. Existing analyses have chronicled the history of our period in a nuanced way by focusing on par ticular aspects of different communities’ experience in detail. Yet rarely have the set of analyses been brought together to examine prejudice and discrimination broadly. That is the aim of this chapter. To accomplish this aim, I first describe the regime of condoned exploitation in a bit more detail than I hitherto have. I then contrast this period with the era of contested prejudice. Next, I turn to the two definitive battles of the period—the black struggle for racial justice, and women’s campaign against misogyny—that, together, ultimately undermined the regime of condoned exploitation in favor of the current era. This restatement of sociohistorical analyses will establish that the current period was not constructed by virtue of a thoroughgoing effort to transcend the exploitation of the past but, instead, was the result of a confluence of social trends and national challenges that made some accommodation to movements for basic civil rights a palatable policy for key elites. This view, the implications of which are sketched in more detail below, moves conflict about the concept of discrimination to the center of the debate about race, sex, and social policy in the United States. Discrimination moves to the center because discrimination, formerly enacted through practices codified in law, is now formally regarded as against the law; this transforms “discrimination” from essentially a moral epithet lacking consistent legal force to a technical contested issue formally adjudicated in the courts. When discrimination was regarded as legal, its definition

From Condoned Exploitive Relations to the Era of Contested Prejudice / 55

was of little consequence. Once discrimination became regarded as formally illegal, defining it became of central importance. In the present period, therefore, to declare certain realities discriminatory or nondiscriminatory is to partially construct the balance of powers for contestation and resolution. Hence, understanding the history that created this pivotal role for the concept of discrimination, and how that history affects social scientific efforts to define discrimination, will be important for comprehending how we have advanced this far, but no further.

The Regime of Condoned Exploitation I contend that we live in an era of contested prejudice, and I have contrasted the present with a previous period of condoned exploitive relations. The era of condoned exploitive relations was characterized by an enduring consensus that the interests, rights, and safety of blacks and women were a secondary concern at best. It is important to briefly distinguish exploitation from other forms of social and economic relations. Political sociologists and researchers in class analysis have considered the complexities of exploitation, and Erik Olin Wright’s definition provides a provisionally useful point of departure. Writing of economic exploitation, he states: Economic exploitation is a specific form of economic oppression defined by a par ticular kind of mechanism through which the welfare of exploiters is causally related to the deprivations of the exploited. In exploitation, the material well-being of the exploiters causally depends upon their ability to appropriate the fruits of labor of the exploited. The welfare of the exploiter therefore depends on the effort of the exploited, not merely on the deprivations of the exploited. (1994: 40; emphasis in original) Parsing this definition suggests that exploiters depend upon the presence of the exploited for benefits. Exploitation as defined above entails a transfer of material or other resources from the exploited to the unexploited group, on terms virtually unilaterally set by the exploiters, perhaps under conditions of extreme duress. Yet, even in such a state, the downtrodden targets of exploiters are still needed, somehow. Seen in this way, it becomes possible to identify a period of condoned exploitation. During the regime of condoned exploitive relations, positions in the economy, in the polity, and in society were allocated first to others, and blacks and women generally took whatever positions were left. In general, subservient positions, often involving menial labor, fell to women and blacks.

56 / Chapter 3

As of 1940, 60 percent of nonwhite working women were domestic laborers, and as late as 1960 more than one third worked as domestics. Further, one quarter of white women were employed in clerical positions in 1940. In that same year, fewer than 10 percent of nonwhite men were employed in professional; proprietor, manager, and official; clerical; sales; or craftsmen occupations, whereas at that time 46.2 percent of white men were employed in those occupations (Farley and Allen 1989). Thus, the positions allocated to blacks and women were on the lower rungs of the occupational ladder. These occupational positions were poorly paid, and the hold individual blacks and women had on even these positions often was tenuous, as suggested by the lower paid labor force participation rates of women and the higher unemployment rate of blacks in 1940 (Goldin 1990; Bonacich 1976). The exploitation was not only in the economic realm. For example, the Tuskegee Syphilis Experiment, begun in 1932 by the United States Public Health Ser vice, lasted through 1972. Study doctors withheld treatment for syphilis from poor African American sharecroppers, waited for the men to die, and autopsied their bodies to study the course of syphilis. The men were never informed that they had syphilis and thus their informed consent was never obtained (Jones 1993). This example reveals how the same logic of the disposability of disenfranchised persons that undergirded Nazi experimentation in Auschwitz (e.g., Spitz 2005) also existed in the United States. It is this disposability that is the essence of the regime of condoned exploitation, an essence that consistently permeates the allocation of resources, opportunities, and obligations. To be sure, few persons became the subjects of grotesque medical experiments. And, of course, a rare woman or black person might crack the monopoly on positions of power; yet the empowered black person or woman remained a rarity, and the general principle was undisturbed by their presence. Of course, changes in race relations and changes in the recognized rights of women did occur during the regime of condoned exploitive relations. For example, Reconstruction (1865–1877) was certainly a period of improvement in the status of blacks. And, the postsuffrage 1920s was a period during which many cultural limitations on women were relaxed. Still, Reconstruction is noted for the contestation around women’s rights and the subordination of women’s or universal human rights to a pursuit of improvements in the status of black males (e.g., Hoff 1991), and the 1920s were a time of rising nativist and antiblack violence spurred by growth in the Ku Klux Klan, a growth that counted hundreds of thousands of white women in its number (Blee 1991). Thus, though it is undeniable that changes occurred, never did advance for one group occur in concert with clear advance for the other during the period of condoned exploitive relations.

From Condoned Exploitive Relations to the Era of Contested Prejudice / 57

Indeed, often the very activists who sought to secure the rights of one group acted in ways, or spoke in terms, that disdained the rights of the other. For example, as Rosalyn Terborg-Penn documents, white feminist activists in the 1830–1920 period did not welcome the participation of black women or men in their organizations. The few black women, such as Sojourner Truth, who reached out to white women’s organizations often met resistance. Terborg-Penn relates an incident in 1851 in which some white women at the Women’s Rights Convention requested Sojourner Truth be prevented from speaking, lest she give the impression that the movement was “mixed with abolition and niggers” (1978: 20). Terborg-Penn notes that no less a figure than Susan B. Anthony asked Frederick Douglass not to attend the National American Women Suffrage Association convention, as Anthony feared his presence would offend the southern hosts of the conference. Such exclusionary acts were not uncommon (e.g., Sneider 1994; Newman 1999). Black leaders also were uneven in their support of women’s rights. Douglass, generally a supporter of women’s suffrage, wavered at par ticular strategic moments. Douglass clearly prioritized suffrage for blacks over the suffrage of women, by implication seeking to enfranchise black men first. Thus, though the rights and opportunities of women and blacks waxed and waned during the decades, because they did so in haphazard relation to each other, a thoroughgoing delegitimation of exploitive relations never materialized. Hence, despite the changes that occurred, they were not sufficient to usher in a qualitatively new period until the latter half of the twentieth century. Clearly, during the long and brutal regime of condoned exploitation, those attempting to further the rights of blacks often disparaged the rights of women, and those attempting to further the rights of women often disparaged the rights of blacks. However draining this dynamic was to the overall prospects for fundamental advance of human rights and civil liberties, this dynamic was not the central factor in the continuing exploitation of the period. Far more important was the elite and nonelite practice of disregarding the minimal rights of women and blacks in times of perceived crisis. A wellknown foundational example is provided by the three-fifths compromise that bound the Northern and Southern states into one nation. Prior to the compromise, Southern states demanded that enslaved blacks be counted by the census so as to apportion members of Congress to the states. The Northern states stood against counting slaves, for doing so would inflate the congressional power of the Southern states by counting nonvoting noncitizens. The impasse threatened to destroy the effort to construct a stronger national state on the shores of the western Atlantic. Eventually a compromise was reached: To wit, each slave should count as three-fifths of a person. With this compromise, the nation—and national unity—was forged on the continued

58 / Chapter 3

enslavement of blacks. Had that compromise not been reached, the United States might have died in 1787. It is easy to multiply examples to indicate the second-class status of women and blacks during the long era of condoned exploitive relations. The end of World War II provides one such example. As the war ended, soldiers returned to find that, during the wartime expansion, women had been hired for factory jobs and many others. The possible crisis of labor oversupply was solved by forcing women out of the workplace (e.g., Anderson 1981: 161–176). This had the further effect of fostering women’s economic vulnerability in a way that was consistent with the 1950s’ reprise of the cult of domesticity. The general point these two examples make is that deals that trade away the rights of women and blacks were common during the era of condoned exploitive relations. The foot soldiers in the regime of condoned exploitive relations were the lower-class white males and females threatened by the prospect of black advance and insecure about their own status in the land of opportunity. However, the regime was maintained primarily by elites who held local and national political, economic, social, and positional power, and who used their power to inspire others to trample the unrecognized rights of women and blacks when elites perceived it in elite interests to do so. Further, in order to sustain the ability to mobilize against blacks or women when needed, elites also nurtured rhetorics of difference and essential inequality that denigrated the status of blacks and women in general and, especially, any who dared move beyond the sphere to which their sex or race was constrained. Larry E. Tise (1987) and John Patrick Daly (2002) document widespread, though not necessarily universal, elite support of antiblack sentiment, while David W. Blight (2001) chronicles the processes of conscious forgetting that undermined black interests in the process of moving beyond the Civil War. At the same time, historians document that many powerful members of the elite judged appropriate behavior for men and women as distinctly different (e.g., Spruill 1938; Ryan 1975). These rhetorics both weakened the ability of advocates for the liberation of blacks and women to successfully appeal to the foundational documents of the nation in their efforts to secure basic civil rights, and invested lower-class white men and women with at least a psychological, and perhaps also a material, stake in both accepting and making real the alleged superiority of women in matters of feeling, the impotence of women outside the home, and the general inferiority of blacks.

The Era of Contested Prejudice In the era of contested prejudice, rhetorics of exploitation are no longer acceptable. Certainly exploitation itself has not been repudiated. However, elites

From Condoned Exploitive Relations to the Era of Contested Prejudice / 59

no longer support unabashed appeals to exploitation; i.e., exploitation rhetoric is no longer condoned. The vast majority of elites no longer defend exploitation as such; instead, other logics are used to support many of the same exact practices that were common in previous times. Surely, the retreat from unabashed support for exploitation is no small transformation, even though many (but not all) elite and nonelite sectors of society continue to support actual exploitation. The rejection of direct arguments in favor of exploitation is significant because when direct arguments for a policy cannot be used, indirect arguments must be employed. And, in this case as in others, indirect arguments may provide unexpected rhetorical resources the exploited and their advocates can use in something akin to an aikido move to transform the material conditions of the exploited. Further, indirect arguments may have other unintended effects that, appropriately nurtured, may further undermine the power of selected sectors of the elite. Thus, that the logic of exploitation is no longer one to which national or even most regional elites may appeal directly is an important alteration in the environment in which we live. However, it is a change that should not be overstated—exploitation rhetoric has been denounced, not exploitive practices. How did this repudiation of the language of exploitation come to be? Over time, the complexity of the developing socioeconomic order, as well as cultural transformation and geopolitical competition, finally undermined the regime of condoned exploitive relations. But, subsumed in the terms “developing socioeconomic order,” “cultural transformation,” and “geopolitical competition” is a decades-long struggle of a broad swath of society within a strategically pivotal context. A sufficient understanding of the implications of struggle in context cannot be obtained via a simple recitation of the labels analysts have developed to signify the processes constituting social change. No, the key aspects of the process by which the regime of condoned exploitive relations gave way to the era of contested prejudice need to be described in a bit more detail. It is necessary to discuss this process explicitly because one regrettable feature of public debate in the United States is a neglect or, what is worse, a romanticization of history. By relating a brief but slightly more detailed than usual history of race and gender relations, I hope to provide an antidote to both the ahistoricization of the present and the romanticization of our past. This is crucially important to do, for unless we realize the constellation of forces that led to demise of the regime of condoned exploitive relations, we will overstate the progress that has been made and become completely confused in our effort to understand and move beyond the present period.

60 / Chapter 3

A Brief Macrohistorical Account of Transforming Race Relations In The Declining Significance of Race, William Julius Wilson (1978) argues that black-white relations in the United States have gone through three periods. According to Wilson, each period is constituted by and within a confluence of racial, economic, and political arrangements, and as contestation around these arrangements occurs, the existing order eventually may be undermined in favor of an emergent arrangement. The three periods Wilson identifies are (1) the period of plantation economy and racial-caste oppression; (2) the period of industrial expansion, class conflict, and racial oppression; and (3) the period of progressive transition from racial inequalities to class inequalities. The first period ranges from the antebellum years to the early postbellum era, while the second period begins after the early postbellum era and continues until the New Deal era. Wilson dates the third period as beginning at the close of World War II. A key transformation that accompanied the inception of Wilson’s third stage was the migration of manufacturing jobs from central cities to suburban locations. Increasingly easy mobility fostered by the interstate highway system facilitated the decentralization of manufacturing and the accompanying jobs. Wilson reports that, in 1947, central cities accounted for two thirds of the manufacturing in the twelve largest metropolitan areas, whereas by 1970 they accounted for less than 40 percent (1978: 93). The shift was accomplished by altering patterns of industrial investment, such that central city manufacturing either declined or grew at a much slower rate than did manufacturing outside of central cities. The decentralization of manufacturing was important for race relations because manufacturing jobs of the period did not require advanced education and thus provided a toehold onto the short career ladder that led to stable blue-collar jobs and earnings sufficient to secure a lifestyle of relative affluence. Prior to World War II, racial conflict occurred around access to these jobs. The conflict was intensified by rising black migration to the northern cities, where the core of the nation’s manufacturing was located. However, as manufacturing decentralized, the opportunity for direct racial antagonism within the industrial order was greatly diminished. The unresolved racial conflict around access to jobs was, according to Wilson, displaced onto sociopolitical institutions within two decades of the end of World War II. At the same time that cities were losing their economic base, the racial composition of cities continued to change in ways that magnified the withincity power of blacks. The racial composition of cities underwent a dramatic transformation during the latter half of the twentieth century, owing to both out-migration of whites and in-migration of blacks. For example, the

From Condoned Exploitive Relations to the Era of Contested Prejudice / 61

proportion of blacks in Washington, D.C., in 1950 was 27.7 percent, while by 1970 it had risen to 71.1 percent. Washington, D.C., was an extreme but not an isolated case; in New York, the 1950 percentage was 9.5, while by 1970 blacks numbered 21.1 percent of the city’s population; in Chicago, 13.6 percent versus 32.7 percent; in Philadelphia, 14.8 percent versus 33.6 percent; in Cleveland, 16.2 percent versus 38.3 percent; in Baltimore, 23.7 percent versus 46.4 percent; and in Detroit, it was 16.2 percent versus 43.7 percent by 1970 (Census Bureau 1950; 1970). As blacks approached and achieved voting majorities in many cities between 1950 and 1980, they were able to obtain institution-based urban power. But black leaders often found it difficult to translate their newly attained positions into effective ser vices for their constituents. Ira Katznelson (1976) usefully contrasts the experience of white ethnics in the late 1800s and early 1900s and blacks in the urban Northeast and Midwest during and after the Great Migration, a period that established the strategic isolation against which blacks struggled in the aftermath of World War II. Whereas white ethnics were brought into existing political machines as a conduit to ethnic neighborhood-based constituencies, and were allowed some degree of autonomy and patronage power, with the limited exception of Chicago, blacks were not. Thus, when blacks took power in the cities, they not only had obtained control over municipalities whose economic base was eroding, but also they had little experience and, owing to their isolation from machine coalitions, few of the allies they would need to maintain and advance the interests of their communities. Thus, winning the cities was to win little, for it was at that historical moment a victory that awarded institutions whose power was being reduced to victors possessing limited resources. Further, with the decline in the urban industrial base, low-skill manufacturing jobs became scarce, decreasing job opportunities. Consequently, just as demographic shifts increased the chance that municipal political institutions would be responsive to the demands of black constituents, and just as the racial conflicts were moved to the political sphere, economic opportunities for urbanites precipitously declined, and the material resources necessary for the municipalities to respond in a transformative way were no more. These are main aspects of the view Wilson articulated in the late 1970s. In later work Wilson (1987; 1996) chronicles the implications of these macrostructural shifts for urban poverty and the social problems seen to attend urban environments. Certainly, Wilson’s macrohistorical account is correct to identify transformations in political economy as key factors in changes in race relations. However, evidence suggests that the first, second, and the opening stages of the third period Wilson identifies are, from one point of view, all part of

62 / Chapter 3

the same regime. As I argue below, the beginning of the third period Wilson identifies and terms the modern industrial stage of American race relations also constitutes the latter stages of the regime of condoned exploitive relations. This reclassification of time periods is based partly on more recent research that has made it necessary to elaborate the existing macrohistorical account in order to describe the destruction of the regime of condoned exploitation and the arrival of the era of contested prejudice. To relate this story with respect to race, it is useful to return to the process through which manufacturers vacated the urban core, and move forward from that point.

The Delegitimation of Racial Exploitation Rhetoric, 1948–1970 Wilson (1987; 1996) describes the ramifications of the retreat of manufacturing from central cities for the residents left behind. But the decision of manufacturers to abandon cities needs explanation as well. Why did manufacturers leave central cities between 1950 and 1980? One answer highlights the possibilities for decentralized manufacturing raised by the development of the interstate highway system and the growing suburbs. Clearly, infrastructural change did play a role in dissipating the advantage of urban manufacturing. However, it is clear that the citing of roads historically has been an arena of racialized policy. For example, Ronald H. Bayor (1988) shows how street planning, road-building, street naming, and policies of selective street paving in Atlanta were used to construct and maintain racial segregation. Bayor contends: Like zoning, the racial aspects of highway-road planning took many forms and had a long-term impact on the city. The most familiar use of this tactic, easily identified in most cities, was the utilization of roads to displace blacks. Atlanta used its highway development to clear slums and blacks out of downtown areas. For much the same reasons that civic centers, stadiums, hotels, and office buildings were built in or near downtown in cleared slum sections, highway construction was used to remove blacks from certain sections surrounding the central business district, set up racial buffers, and allow the city to redevelop the area commercially. (1988: 5) Not only were blacks displaced from neighborhoods in the cities through highway policy, but suburban growth was itself the direct result of a policy of federal subsidy and racialized housing policy. With respect to highway building, it is well known that, absent the federal subsidy of approximately 90 percent of the building cost, the highway system that has facilitated the rise of

From Condoned Exploitive Relations to the Era of Contested Prejudice / 63

suburban communities likely would not have been constructed, and it is that unequal subsidy that has underwritten suburban prosperity and siphoned off urban wealth (e.g., Rabin 1973). With respect to housing, Douglas S. Massey and Nancy A. Denton (1993) show how federal housing policy directly constructed white suburbs and an isolated, black, urban core. The context for this occurrence originates at the inception of explicit federal housing policy. The federal government first subsidized home purchases during the Great Depression by backing long-term mortgages and thus providing banks with an incentive to make long-term loans. Prior to that time, mortgages were usually no more than ten years in duration, and half the purchase price of the home was required in down payment (Fischer et al. 1996). As one might expect, home ownership rates were comparatively low. Albert Chevan (1989) reports that even by 1940, approximately 43.5 percent of male household heads, and 44.5 percent of female household heads, were home-owners. By 1980, 71.8 and 70.3 percent of male and female heads owned their homes, respectively. This dramatic increase in home-owning paralleled federal intervention in the housing market in the form of housing subsidies for home-ownership. But these subsidies were not race-neutral. Massey and Denton (1993) show that when the federal government moved to support home-ownership, it did so with procedures that could only serve to increasingly segregate blacks from the remainder of society. Moreover, this result was not produced by simple neglect; instead, the policies were constructed with the express aim of segregating blacks. Massey and Denton report that the first federal program, the Home Owners’ Loan Corporation (HOLC), was targeted to urbanites, and provided aid to borrowers about to default as well as those who already had defaulted on their mortgages. However, by refusing to make loans in racially mixed or black neighborhoods, HOLC institutionalized redlining. Ultimately, this procedure allocated resources to nonblack areas, and provided government support for disinvestment in black communities. The pattern of disinvestment continued as federal policy continued to reinforce racial segregation. Federal Housing Administration (FHA) policy followed the precedent established by HOLC, redlining black and racially diverse areas. In contrast, those moving into other communities could obtain FHA or Veterans Administration (VA) loans that allowed one to borrow as much as nine tenths of the cost of the home, putting down only 10 percent of the price up front. Further, it was the FHA and VA that spurred the transition to double- and triple-decade mortgage payment timetables. Massey and Denton write: [D]uring the 1940s and 1950s, the marriage of FHA and new construction techniques made it cheaper to buy new suburban homes than to

64 / Chapter 3

rent comparable older dwellings in the central city. As a result, the FHA and VA contributed significantly to the decline of the inner city by encouraging selective out-migration of middle-class whites to the suburbs. The bias in favor of suburbs was evident in FHA practices and regulations, which favored the construction of single-family homes but discouraged the building of multi-family units. In addition, FHA loans for the remodeling of existing structures were small and had a short amortization period, making it easier and cheaper for a family to purchase a new home than to renovate an older one. But the most important factor encouraging white suburbanization and reinforcing the segregation of blacks was the FHA requirement for an “unbiased,” professional appraisal of insured properties, which naturally included a rating of the neighborhood. In rating the home, the FHA established minimum standards for lot size, setbacks, and separation from existing structures that essentially eliminated from eligibility many inner-city dwellings, notably row houses and attached dwellings. In evaluating neighborhoods, the agency followed the HOLC’s earlier lead in racial matters; it too manifested an obsessive concern with the presence of what the 1939 FHA Underwriting Manual called “inharmonious racial or nationality groups.” According to the manual, “if a neighborhood is to retain stability, it is necessary that properties shall continue to be occupied by the same social and racial classes.” (1993: 53–54) Thus, as the federal government subsidized the establishment of white suburbs with such practices, it enabled the movement of manufacturing from cities to simultaneously entail a movement of manufacturing jobs from black communities. Is there evidence that racial composition of surrounding communities is associated with the siting or refurbishment of manufacturing plants? Yes, there is. Robert E. Cole and Donald R. Deskins Jr. (1988) document an important association between the absence of minorities and the siting and refurbishment of plants by Japanese and American automakers during the 1980s. They report that not only did Japanese automakers site plants in locations with relatively low percentages of minorities and employ even fewer minorities than the hiring pool would lead one to expect, but also U.S. companies were more likely to retool plants in areas with lower concentrations of minorities. Thus, not only did the influx of Japanese companies into American auto manufacturing transform work practices (Florida and Kenney 1991) and expand the rural manufacturing sector (Mair, Florida, and Kenney 1988), but

From Condoned Exploitive Relations to the Era of Contested Prejudice / 65

also the plant placement strategies of foreign and domestic companies further hastened abandonment of geographic areas characterized as largely black. Cole and Deskins are careful to emphasize their inability to link behavior to motivation. Yet their findings are consistent with discrimination as a key causal factor in the economic transformations that greatly weakened cities. However, even if the siting of plants was driven in part by a desire on the part of manufacturers to limit the access of blacks, why would manufacturers adopt this strategy in the first place? Indeed, U.S. history is replete with examples of blacks and other groups before their rise into whiteness being denied access to jobs. What changed after World War II to make it desirable to reorganize production outside of cities? I submit that what changed was elite opinion about the desirability of blatant discrimination against blacks. Employers sought to avoid the labor strife that characterized the industrial order prior to World War II. Changed elite opinion is manifest in growing judicial attention to discriminatory practices, rising success of legal challenges to standard discriminatory operating procedures, and increasing legislation outlawing conspicuous forms of discrimination. This change in the opinion of national elites made the perhaps unexpectedly emergent geographic distance a more palatable mechanism for excluding blacks than simply denying blacks the jobs that were within their own backyards. But even this observation begs a question: Why did elite opinion change? I contend that change in elite opinion was driven by a dawning awareness that blacks were essential to the effort to meet important national challenges. To make this final aspect of the case, we must return to World War II. Wars often unleash unintended social changes, and World War II was such a watershed event. Massive upheavals occurred in the rank of nations as a result of the war, such that the war’s conclusion marked the beginning of the American Century. Domestically, however, upheaval was at least as large. On the eve of war, the United States was a racially segregated nation. Though segregation reigned only in the South, blacks had no expectation of general enforceability of their rights anywhere in the country. The danger to blacks was so recognized and widespread that several guides, such as The Negro Motorist Green Book and The Traveler’s Guide: Hotels, Apartments, Rooms, Meals, Garage Accommodations, Etc., for Colored Travelers, were published by both private and federal government agencies to help blacks traveling in unfamiliar areas to avoid “problems” (Lewis 1997)—i.e., to make blacks traveling in unfamiliar territory aware of where they could and could not go. “Separate but equal” was the ostensible logic, though matters rarely were equal in practice, as whites continued to reserve for themselves a grossly disproportionate share of the nation’s resources.

66 / Chapter 3

The social forces unleashed during the war, and their ramifications for race relations, were enormous. As war approached, civil rights leaders advocated active enlistment and involvement in the war effort. Well aware of the second-class citizenship of black Americans, leaders argued that, by fighting foreign foes, domestic foes could be felled as well. As Bernard C. Nalty and Morris J. MacGregor report, “The name given this effort was the Double V campaign: victory against fascism abroad and discrimination at home” (1981: 103). However, the Selective Training and Ser vice Act of 1940 maintained the discriminatory thrust of American society vis-à-vis blacks. The act did provide that one of every ten persons drafted for ser vice would be black, yet it also continued segregation of the forces into black and white units. Thus, during the war, the overwhelming majority of black ser vice personnel were assigned to de jure segregated units. Consequently, by the end of the war, civil rights activists were concerned that fascism might have been beaten, but that domestic discrimination seemed as strong as ever. Fortunately for the civil rights agenda, the United States was not the sole victor, and an emerging and far more dangerous conflict appeared likely with one former ally, the Union of Soviet Socialist Republics (the USSR, or the Soviet Union). Accessorily, this emerging foe aided the civil rights struggle. The Soviet Union contested American leadership not only materially (in economics, science, and technology), but also symbolically (in a battle for the hearts and minds of the world). That latter battle was waged in part in debates concerning the apparent validity of the philosophical systems upon which the two nations purportedly were based. And, the treatment of blacks in the United States offered an important example by which to evaluate claims as to the justice, fairness, and liberty of the American system. The Soviets were well aware that U.S. oppression of blacks made it more difficult for the United States to successfully navigate the dawning postcolonial era. As colonies and former colonies struggled to secure independence, the two powers vied for access to the lands, resources, and allegiance of these states. The U.S. treatment of its own sons and daughters of Africa greatly complicated U.S. efforts in this regard. What gave power to this competition was the potential for military conflict between the United States and the Soviet Union. During World War II, the Soviets had distinguished themselves by winning some of the most brutal battles ever fought (Overy 1997). Few doubted the resilience and commitment of the Soviets, and if the United States was to find itself in battle against this former ally, it would need every able-bodied man it could muster for war. The Soviet threat thus afforded a classic opportunity for the advancement of civil rights. Black soldiers, segregated in past confl icts despite protestations, were now not only essential to a possible future war, but they were

From Condoned Exploitive Relations to the Era of Contested Prejudice / 67

widely recognized as such before the full joining of hostilities. Civil rights leader A. Phillip Randolph articulated the views of many blacks when in testimony before Congress on March 31, 1948, he indicated the likelihood of massive civil disobedience with the advent of a segregationist draft. Randolph stated essentially that blacks could not be counted on to soldier in an army that accorded them second-class citizenship ([1948] 1971). This brazen threat shocked many congressional leaders, and its value was in the implication of Randolph’s threat should full war mobilization be required. The threat also placed the incumbent president, Harry S. Truman, in an electoral bind as he sought re-election (Pfeffer 1990). Needing every black vote he could obtain, amidst a charged international environment, and in no small way because of the threat posed by civil rights activists such as Randolph, on July 26, 1948, President Truman signed Executive Order 9981, which renounced the principle of a racially segregated military in the United States. The battle, however, was not over, as the executive order had established a committee to implement the order. As the Fahy committee completed its work, integration of the military promised to be a long process (Nalty and MacGregor 1981: 243–294). Had matters been driven only by policy, integration of the military could have proceeded in a halting and uneven manner. Yet, once again war spurred social change, as the commencement of the Korean War sped racial integration of the U.S. armed forces, as perceived manpower needs in the theatre of war motivated military personnel assignments without consideration of their ability to maintain a segregated force (Nalty and MacGregor 1981: 295–296). The struggle and process of integrating the military offered at least two important lessons. First, it provided insights to civil rights groups that they could draw on in later struggles. And it played a role in adjusting non-Southern white elites’ views as to the costs of maintaining a commitment to de jure segregation in American society, complicating the story of how civil rights for blacks became more protected. Of course, World War II was not the first war in which blacks fought. Yet, other wars never interrupted the era of condoned prejudice. Indeed, as A. Phillip Randolph noted, Crispus Attucks was the first to die for the American Revolution, yet the patriots continued to enslave African Americans (Nalty 1986). Hence, the strategy of joining the military to bring about black liberation historically had failed. This raises an important question: What made the aftermath of World War II unique? An answer to this question can be developed by considering the aftermath of the previous war, World War I. At the inception of the Great War many black leaders, such as W.E.B. DuBois, called for blacks to enter the fight for democracy on the side of the United States (Nalty 1986). Blacks

68 / Chapter 3

served in the war, some quite honorably, but the postwar discussion did not emphasize the bravery of the black soldier (Butler 1980). Instead, a mixed message was articulated, greatly diminishing the value of their ser vice for pushing desegregation of the military, not to mention wider societal change. More fundamentally, at the end of World War I, the United States turned inward, away from playing a pivotal role on the world stage and thus away from the obligations and activities that would necessitate a large standing military. Thus, at the end of World War I, U.S. elites could dispense with black soldiers just as they had done at the end of all previous wars. Further, isolationism rendered international opinion concerning race relations largely irrelevant. World War II was unique in that its aftermath produced commitments that required continued military capability, in an environment in which newly liberated former colonial nations were important observers of U.S. domestic policies. The confluence of these conditions presented an opportunity for change, as key elites were motivated to accommodate the advocates for change under those circumstances. These observations suggest that federal authorities became concerned with protecting the rights of blacks only by virtue of concerns for the interests of nonblacks. Derrick A. Bell Jr. (1980) describes this perspective, pointing first to the long-running efforts of blacks to overturn the courts’ support for segregation, citing Roberts v. City of Boston (59 Mass (5 Cush.) 198 (1850)) to suggest that Brown v. Board of Education (347 U.S. 483 (1954)) was hardly a new effort on the part of blacks and their civil rights allies. What changed to make the Brown effort successful, Bell argues, was the calculus of political advantage. U.S. elites came to recognize how changed geopolitical conditions made it useful to move forward on civil rights for blacks, in order to more effectively contest Communist accusations of the racist character of U.S. society. These accusations had force in newly independent nations, as they feared the integrity of U.S. claims to support the interests of postcolonial regimes. In addition, Bell notes the economic interests of those who might want to further industrialize the South, and who might find the separation of the races economical ly inefficient, as well as Randolph’s stated possibility that blacks might not join the fight next time should hostilities beckon the nation into war. Upon comparing the Supreme Court’s decision in Brown v. Board of Education and a series of later cases in which the Court seemed to retreat from what some regarded as the promise of Brown, Bell concludes: “The interests of blacks in achieving racial equality will be accommodated only when it converges with the interests of whites. However, the fourteenth amendment, standing alone, will not authorize a judicial remedy providing effective racial equality for blacks where the remedy sought threatens the superior societal status of middle and upper class whites” (1980: 523).

From Condoned Exploitive Relations to the Era of Contested Prejudice / 69

Desegregation of the military, desegregation of the schools, desegregation of the lunch counters—all turned on nonblack elites’ perceptions of their own needs. Recent historical research documents the likely validity of this alltoo-commonly ignored perspective. Documentary evidence indicates that federal-level concern with the geopolitical implications of widely known incidents of flagrant discrimination against blacks motivated their engagement on the side of extending civil rights to blacks. Thomas Borstelmann (2001) situates the dawning federal concern with the international implications of domestic civil rights developments in the historical context of decades of national elites’ outright racism. Mary L. Dudziak (2000) documents how interested the executive branch became, sending State Department officials rather than Justice Department officials to testify before Congress so as to explicate the president’s position and the basis for civil rights legislation not in constitutional law but, instead, in perceived national security implications of failure to pass civil rights legislation. This burgeoning scholarship, and additional continuing examples (Lucas and Paret 2005), strongly support Bell’s interest-convergence thesis. Many facts of the civil rights struggle of the 1950s and 1960s are well known, and need not be chronicled here in any detail. Suffice it to say, blacks and their allies agitated in the courts to overturn the doctrine of separate but equal that had reigned explicitly since Plessy v. Ferguson (163 U.S. 537 (1896)). Then they attempted to integrate places of business, transportation facilities, public accommodations, and schools, and sought to secure the ballot as well. First in the South and then beyond, civil rights activists encountered hostile local authorities and endeavored to transform attitudes and material conditions. As this struggle waged, the regime of condoned exploitive relations began to waver. The removal of unabashed exploitation, the appeal to equity-based reasoning, and even the rejustification of exploitive relations on the basis of allegedly egalitarian logics, would eventually signify the end of the regime of condoned exploitive relations. But that end did not come solely by virtue of the struggle for the civil rights of blacks, and even that struggle’s gains were not won through moral suasion but, instead, owing to an explicitly recognized threat posed by new geopolitical realities that gave the status of blacks in the United States both symbolic and material import on the world stage.

The Delegitimation of Sex Exploitation Rhetoric, 1948–1973 It is clear that both the prospect and the reality of war played a major role in delegitimating overt racial discrimination shortly after 1945. The Korean War marked the end of a segregated military, and affirmed the principle of

70 / Chapter 3

equality in a wide-ranging way. But the treatment of blacks alone did not overthrow the regime. As the treatment of women in the United States at the close of World War II and into the 1950s suggests, the regime of condoned exploitive relations remained in place. Women had been recruited into the ranks of factory labor at the inception of the war to stave off labor shortages occasioned by the allocation of young men to the armed ser vices after 1941. Before describing the nature of this change in more detail, assessing its ramifications, and chronicling the demise of women’s factory labor force participation, it is imperative that we recognize just what did not change during this period. The paid labor force participation rate of single women was high prior to the war; nearly 80 percent of single, never-married white women ages twenty-five to thirty-four were in the paid labor force in 1940 (Goldin 1990: 18). Further, the paid labor force participation rate of married black women exceeded the paid labor force participation rate of married white women throughout the 1920–1980 period. But prior to the war, and regardless of race, the paid labor force participation rate of married women was not high; approximately one in six white married women ages twenty-five to thirty-four, and about one in three nonwhite married women of the same age, were in the paid labor force as of 1940 (Goldin 1990: 18–20). World War II occasioned a ramp up in production coupled with an increase in military labor needs. This classic conundrum was exacerbated in that long-running demographic shifts were reducing the proportion of single females available (Anderson 1981; Goldin 1990: 175). Should married women’s labor force participation remain at prewar levels, labor shortages were sure to follow. During the war, therefore, many actors, including the federal government, private agencies, manufacturers, and others, encouraged married women to enter the factory in support of the war effort (Anderson 1981: 4). Once there, they found themselves highly paid compared to women working in typically “female” jobs, though less well-paid than men doing the same work (Campbell 1984: 222). Further, women largely were shut out of union leadership, closing off a platform from which sustained organized protest on working conditions and pay could be launched. Women who entered the military encountered similar opportunities, similar limitations, and similar frustrations (Campbell 1984: 17–61). In the military, and during a war, women of course could not agitate for change easily, given the rigidly hierarchical nature of the military apparatus and the crisis situation. But even for civilian white women, the complexity of their position and the many options available undercut the likelihood of sustained protest. D’Ann Campbell writes: A feminist consciousness was scarcely visible in the factories. Women participated in official strikes, and occasionally led walkouts against

From Condoned Exploitive Relations to the Era of Contested Prejudice / 71

the hiring of black women co-workers, but few agitated for equal pay or equal rights. Numerous factors were involved. The turnover of women was so high that a leadership cadre seldom formed; men who totally controlled the unions rarely listened to women; and the lack of specialized skills meant women had weak bargaining advantages. Women, furthermore, were typically more dissatisfied with unequal burdens caused by home duties. They responded to bad conditions not by voice but by exit—switching jobs or returning to the home. Black women, who had vastly fewer job options and who needed to work, were more likely than whites to protest, even to the point of organizing strikes in Atlanta laundries. (1984: 222). Thus, during the war, women served the nation in many ways, and faced a mix of new opportunities and limiting expectations throughout. As the war came to a victorious close, and soldiers, sailors, and Air Force personnel began returning stateside, the prospect of labor oversupply became evident. And, as the postwar period began, business, union, state, and federal officials colluded in displacing—firing—women from factory jobs to make room for returning male warriors. Some women welcomed the prospect of relinquishing their factory positions, but others, both single and married, did not. In a 1944 survey, the Women’s Bureau found that about three quarters of the women employed in war time manufacturing wanted to remain in the paid labor force, and of those, the vast majority hoped to keep the same jobs they had (Weissbrodt 1946). Yet, women of course were not a monolithic block, and many hoped to leave the paid labor force once the war ended. Still, regardless of their wishes, women were tossed out of the factory jobs they had occupied during the war. One example, described by Karen Anderson, proves illuminating: Having been told by the UAW [United Auto Workers] during the war that they could rely on the union to safeguard the interests of all workers, women members soon found out how meaningful union protection was when the victims of management discrimination were female. When the automakers ignored the seniority provisions in their contracts with the UAW in order to limit women’s access to industry jobs, the union ignored the situation, filing no grievance actions. After their union proved of little help in fighting postwar discrimination, the dispossessed women autoworkers took action on their own, setting up picket lines outside of the Ford Highland Park plant in November 1945. Bearing placards reading “Stop Discrimination Because of Sex” and warning male workers that their seniority could also be threatened, women members of UAW Locals 400, 50, and 600 claimed that

72 / Chapter 3

2,200 men without seniority had been hired at the plant while 5,000 experienced women remained idle. Officials at Ford justified their refusal to hire women for the plant, where they had cut the female work force from a wartime peak of 5,849 to 300, on the grounds that the work had become too heavy for women when the factory shifted to the production of tractors. The company denied a charge by the women that the job classification system had been manipulated so that entrylevel jobs involved heavy work and served to exclude women. In response to the women’s action, John C. Carney, president of Local 400, promised to put the matter before the union’s executive board. In fact, however, UAW officials, who were engaged in contract negotiations with General Motors, preferred to concentrate only on the rights of male workers and privately reassured Ford management that it need not worry about the women’s protest. Abandoned by their union, the women failed to challenge successfully the discriminatory practices that excluded them from the vast majority of jobs in the industry that dominated their local economy. (1981: 165–166) Such examples were common. Moreover, World War II was certainly not the first time in which women had been called into the industrial factory, only to be thrown out of their jobs at the end of the war; for example, it appears that the same occurred at the close of the Civil War (Foner 1979: 122– 123). But the interesting contrast here is not with the past but with the treatment of blacks of the period. The inegalitarian implications of firing women and replacing them with men were just as visible as the aforementioned treatment of black males in the military. And, in comparison to the behavior of Soviet officials, U.S. officials clearly were moving in an antiegalitarian manner against women in the United States. Natalia Pushkareva (1997) notes that Soviet women were on active duty during the war, and engaged in combat in air, on land, and at sea. After the death of Stalin, the response of Soviet officials to gender inequality was far more supportive in many respects than that offered by U.S. officials. Thus, if contrasting unfair treatment were sufficient to place an issue onto the table in the U.S.-Soviet competition, then the treatment of women could have been an issue of as much debate at the international level as the treatment of blacks became. Why it did not become so is illuminating. Notably, the treatment of women in the United States was not substantially worse than the treatment of women in emerging third-world nations. Recall that one reason the treatment of blacks became a strategic issue was that Asian and African colonies might consider the treatment of U.S. blacks as indicative of how the United States eventually would treat them. Thus, conspicuous forms of exploitation, and brazen support of a policy because of

From Condoned Exploitive Relations to the Era of Contested Prejudice / 73

its racially exploitive aspect, lost favor among elites in the United States. Because women were equally or more oppressed in the rising independent states in comparison to the United States, and because women did not have decisive power in either locale, the question of how United States policy treated women did not become a sustained issue of international attention. In other words, the treatment of women was not taken into strategic account because there was no pivotal base, constituency, or resource whose allegiance or allocation seemed to turn on the treatment of women. Thus, key actors in the United States made a judgment that the economy did not need women in industrial positions, and thus women were fired from them. Labor and business leaders, with government acquiescence, agreed to sacrifice one sector of society for their own material or symbolic interests. The mistreatment of women and the terms in which the treatment was discussed, as evidenced in the above examples, strongly suggest the regime of condoned exploitive relations was not fully ended by immediate postwar changes. Clearly the war, the geopolitical contest it spawned, and the domestic social activism that accompanied the postwar period, greatly endangered the regime—we see that in the response to black protest. Yet, had this been all that occurred, the turn to the twenty-first century would likely still have found Americans living in the regime of condoned exploitive relations. But other factors, such as more fundamental economic transformations of the developing economy, also shifted the terrain in ways conducive to the death of the regime, by bringing the treatment of women to the fore at the same time as blatant antiblack prejudice as an explicit reason for policy was being delegitimated. To convey the nature of the economic transformations that played the final decisive role in undoing the acceptance of rhetoric of exploitation, we must consider that in actuality women have always been key actors in the U.S. economy, without whom it likely would have had serious difficulties avoiding collapse. This should be obvious, because it appears to be the human condition—most sectors of most societies labor. As Claudia Goldin (1990) observes, women have been active in the U.S. economy since colonial times. But economic transformations first hid then revealed again to mid- to late-twentieth-century social scientists women’s contribution to economic activity. In the mid–twentieth century, the divergence between home and work was made visible by a sharp distinction in the physical location of economically productive activity. Since the 1990s, this division has appeared to be breaking down, but, if so, it will not usher in an entirely new phenomenon. From the inception of the union to the end of the nineteenth century, the household was a site of significant economic production. Much of that production moved to other sites through the twentieth century, and accounting systems of the

74 / Chapter 3

period tallied work done outside the home as a contribution to the economy but ignored the same work when it was done in the home. This approach to recording economic activity certainly led to some underestimation of the economic activity of women (see Goldin 1990: 42–55; Goldin 1986). As work moved outside the home, other social forces were unleashed owing to the organization of work in the United States during the period under study. Concomitant with the movement of productive economic activity outside the home was the increase in the size and scale of factories and firms. The increasing concentration of capital, coupled with the rapacious behavior of business interests, also engendered conflict over the degree to which economic activity could be regulated, and the legality of worker organization. Through most of this period, the productive activity of married women was not recognized. After World War II, and after the factory doors were closed to women, paid labor force participation rates of women continued to rise. Cynthia B. Lloyd and Beth T. Niemi (1979) report that the paid labor force participation rate of women ages twenty-five to fifty-four was approximately 37 percent in 1950, and rose steadily to approximately 60 percent by 1979. Throughout the period, the paid labor force participation rate of nonwhite women exceeded that of white women. Further, Lloyd and Niemi (1979: 39) show consistently increasing rates of paid labor force participation at all ages. Noting that single women generally had high rates of paid labor force participation, and were concentrated in lower-level white-collar ser vice occupations, Goldin argues that the steady increase in women’s paid labor force participation occurred principally because of the rising paid labor force participation of married women. She also points to a key impediment to married women’s paid labor force participation that was dismantled during the 1950s—marriage bars. Marriage bars forbade married women from obtaining some positions, most notably clerical or teaching positions, but also some professional jobs such as lawyer and medical doctor. Sometimes firms required single women to give up their jobs if they married; at other times firms retained such women, but did not hire married women. Goldin (1990) shows that the decline of marriage bars seems to have occurred because of labor undersupply. Thus, it appears that transformations in the economic structure may have figured in the rise of female paid labor force participation. As the ser vice sector began to grow, businesses faced potential labor shortages should they refuse to hire or retain relatively educated women who happened to marry or be married. Thus, the most vibrant, expanding part of the economy was that part that had employed single women, rather than men. Hence, the jobs available for women threatened to outstrip

From Condoned Exploitive Relations to the Era of Contested Prejudice / 75

the supply of single women. Perhaps to entice married women into the paid labor market, businesses responded by dismantling marriage bars. As women entered the paid labor market, they attained a foothold that created the conditions for successfully resisting some aspects of exploitation. As we saw in the case of blacks, the eventual end to the segregated military was based on the strategic importance of blacks to a potential future war, and the leverage they gained thereby allowed for successful agitation for social change. In the same way, the ability of women to successfully agitate for social change came not from a change of heart on the part of others as to the inappropriateness of the exploitive behavior, but by virtue of both the growing awareness of the strategic value of women for activities of central importance and a general retreat from conspicuous defense of exploitation. As for the latter factor, the general retreat from overt support of exploitation per se, women certainly piggybacked on the advances of and the logic espoused by black activists in their struggle for civil rights. Proponents of black and female rights were not alone in sensing that the logic of one struggle was applicable to others. Indeed, it has passed into lore that during congressional debate on the Civil Rights Act of 1964, opponents offered an amendment to add women to the list of protected classes to, as Paul Burstein (1985 [1998]) reports, the laughter of the floor. The late hour in which the amendment was offered occasioned a furious debate among advocates of the bill, some of whom believed that women should be included, and others who felt their inclusion threatened passage. The amendment threatened to split the coalition supporting the measure. In the end, the amendment was offered, it passed as supporters rallied behind the amendment, and passage of the Civil Rights Act was not derailed (e.g., Burstein 1985 [1998]: 22–23; Skrentny 2002). The strategy of civil rights opponents had both political and intellectual implications. Politically, they would win the day if there were no majority willing to support the advance of both women and blacks. Intellectually, the alleged absurdity of either women or blacks advancing could be revealed by suggesting that whatever legislation would apply to one should also apply to the other. Such episodes suggest that actors recognized, or at least sought to manipulate, a similarity between the two groups, to forestall the advance of both. Hence, it is fair to surmise that, ultimately, pushing over the regime of condoned exploitation required advance for both. Thus, had women not succeeded in tapping into the logic used by blacks, or found some other way to at least somewhat erode support for women’s oppression at the same time that blacks were making gains, the regime of condoned exploitive relations would not have fallen. For the era of contested prejudice to dawn, multiple rhetorics of exploitation had to lose legitimacy.

76 / Chapter 3

The Reconstruction period and the 1920s, and the reactionary resurgences that reversed them, reveal what happens when one group gains yet the general rhetoric of exploitation remains accepted. Thus, the success of women, limited as much as the success of blacks, actually was essential to the end of the regime.

Limitations of the Delegitimation of Condoned Exploitive Relations It is important to be clear about what did not happen in the process by which overt exploitive relations, and direct appeal to the exploitative nature of the relation, lost legitimacy. The reason I term the period that began sometime in the 1970s as the era of contested prejudice rather than the era of contested exploitation is that exploitation is not seriously contested. Instead, one explanation for exploitive relations is contested, and the contested explanation is prejudice. However, the same exploitive actions can be taken and exploitive structures can be maintained if another basis for them can be developed and articulated. Thus, exploitive relations still are treated as legitimate; appealing to the prejudicial or discriminatory aspect of the relation is not. The above distinction goes a long way to explaining how it is that social policy was not altered more fundamentally during the 1945–1975 period. Consider, for example, social security: Jill Quadagno (1988) reveals how southern congressmen intervened to prevent social security from reaching blacks when the program was established in the mid-1930s. Quadagno finds that southern congressmen sought to exclude agricultural and domestic workers from social security, in part owing to a fear that the federal program would weaken owners’ control over black labor. Southern congressional delegations were needed to pass the Social Security Act; thus, to gain sufficient support for the legislation, Roosevelt agreed to the exclusions, over the objections of black advocates (Quadagno 1988: 116). By 1954, however, congressional amendments to the program placed most farm workers under social security (Quadagno 1988: 149). Yet, these amendments do not appear to have been motivated by a repudiation of the exploitation entailed in the former system. Instead, southern members of congress saw federal resources flowing to other regions, and sought to obtain more federal dollars for their region. Further, the decline in the proportion of workers engaged in agricultural labor, reflecting in part more efficient productive techniques and the substitution of capital in the form of tractors, fertilizer, pesticides, and herbicides for labor, reduced the incidence of persons covered by the new extension. In short, large groups of agricultural laborers were no longer essential to farm production (e.g., Lemann 1991), making the policy change more palatable.

From Condoned Exploitive Relations to the Era of Contested Prejudice / 77

Thus, superficially it would appear that extending the Social Security Act to additional workers was a clear sign of progress against discrimination. Certainly, material well-being was improved by this extension of coverage; however, it is also true that the very process by which this occurred as well as the reasons articulated in congressional debate are consistent with the contention that what became repudiated between 1945 and 1973 were explicit appeals to exploitation, not actual exploitation itself. Quadagno (1994) shows how housing policy nicely captures the stability beneath the changed façade of U.S. social policy. During the mid-1960s, major changes were made in housing policy. The Johnson administration attempted to expand the funding available for low-income housing and to subsidize home purchases of low- and middle-income urbanites. The Housing and Urban Development Act of 1968 leveraged private banking funds to increase access for the poor. Yet, the act also removed protections by downgrading the quality level homes needed to reach to qualify for federal financing assistance, attaching fi nancing to the home rather than the potential buyer, and essentially allowing the purchase of homes at interest rates of 1 percent (Quadagno 1994: 105–106). The low interest rate reduced incentives to focus on housing cost, and, as the poor almost by defi nition contained inexperienced home-buyers, the new would-be home-owners often were vulnerable to unscrupulous Realtors. The resulting situation was ripe for abuse. Richard Nixon promised to resolve these problems and help the poor attain home-ownership. After an ambitious start, however, it became clear that the only way to keep this promise would be to move low-income housing into the suburbs. Confronted with complaints from suburban constituents, Nixon effectively reneged on the commitment. However, the way in which he did so is telling of the changed circumstances. Quadagno reports: Nixon could not afford to alienate the suburbs, the middle-American heartland, the core of his political support. On June 11, 1971, he declared that while his administration would support the law of the land, “open housing, open cities, open suburbs, open neighborhoods,” it would not go beyond the law by using federal money to force economic integration . . . Surely, it was wrong to deny housing to a person because of race, but neither should “a municipality that does not want federally assisted housing have it imposed from Washington by bureaucratic fiat.” (1994: 110; emphasis in original) The issue of housing is only one of several in which rhetorical change was not followed by substantive change. On issue after issue, the language justifying par ticular policies changed, but the overarching policy of deprioritizing

78 / Chapter 3

the interests of blacks and women and other selected out-groups was maintained through institutions’ operation of business as usual.

Defining Discrimination in the Context of (Which?) History Though the historical narrative may be persuasive, what does it have to do with the social science definition of discrimination? A great deal, because at pivotal moments social scientists construct defi nitions of social phenomena from within par ticu lar sociohistorical contexts in ways that are consistent with their understanding of those contexts. When social context changes, or social scientists’ understanding of context changes, simple inertia, tradition, analysts’ generational experiences, or the speed or complexity of the changes may create a lag between their methods and the social reality those methods are meant to probe, even when analysts are aware of a change in the context (Mannheim 1928 [1952]; Schuman and Rieger 1992; Lucas 1999). Hence, methods and theories are not everywhere and eternally useful; if the social context changes, or if social scientists’ understanding of the sociohistorical context changes, then it also may become necessary to change their conceptual tools. And, the nature of the changes in the tools will depend on the nature of the changes society is understood to have experienced. Thus, consider two versions of history. In one version, some malicious people committed acts of discrimination against some other people. Eventually, the nation rallied and stopped the malicious people from committing the oppressive acts. Once the nation stopped those individuals from committing those acts, the newly freed people no longer lived under the strain of discrimination. This version may be doubly familiar. It generally matches the mainstream version of history taught in textbooks and popularized in movies and television, wherein which villains such as Eugene “Bull” Connor personify prejudice and discrimination. In this version some mean, identifiable characters treat some other people badly, the poorly treated and their allies protest, law enforcement eventually makes the mean people stop their bad behavior, and the newly freed are now, well, free. This version also generally matches the fairy tales we tell our children, tales populated by mean fairy godmothers, mean stepmothers, and mean giants. In these fairy tales, all one needs to do to escape the ogres’ mendacity is put an end to the par ticular mean acts they commit, either by fleeing, cursing, or killing them. In this Disney-esque version of history, it is sensible to see discrimination as only the acts that individuals commit, because the social realm has been purged of virtually every entity except atomistic individuals, some of whom act on the basis of prejudice. Individual action becomes the focus of the defi-

From Condoned Exploitive Relations to the Era of Contested Prejudice / 79

nition of discrimination because essentially no other type of action or arrangement exists. Now consider the more complex story related above, a story that eventuates not in liberation but, instead, in the era of contested prejudice. In this story certain acts that are blatantly discriminatory are delegitimated. For example, companies will no longer explicitly bar women from certain occupations, and the military will no longer explicitly bar blacks from serving alongside others. At the same time, operations are rearranged in ways that end up allowing exclusion to be justified in other ways; for example, industrial manufacturing is reorganized spatially, far away from concentrations of blacks. Afterward, nonurban factory-owners employing few blacks could point to the long commute to explain the absence of blacks. The sum total of the rearrangements maintains a real possibility of effective, perhaps even more exclusionary, discrimination, even though blatant acts of discrimination may have diminished greatly. Though some obviously discriminatory edicts and practices are removed, the key fact to note is that a par ticular practice is removed only when doing so solves other geopolitical, reputational, or economic problems. This means, of course, that one might suspect that equally discriminatory practices whose removal would not further geopolitical, reputational, or economic aims might not be removed. This suspicion is apt; at the same time that many offices of the federal government went on the offensive against blatant acts of discrimination and blatantly discriminatory structures, many structures that make up discrimination were buttressed, while new discriminatory structures were fashioned. Indeed, consistent with Dudziak’s persuasive argument that federal civil rights policy emphasized impression management, the maintained structures were less visible than the dismantled ones. Further, new structures were arranged in ways that ultimately made them less visible than the blatant acts of the previous order. Of course, how can this more complex history jibe with the popularized, simplified version? The more complex version does not deny the explicit claims of the simplified history. Surely, individuals did agitate for change, and, surely, many blatant acts of discrimination were disavowed. The simplified story is generally true, as far as it goes; it just does not go far enough. The complex narrative supplements the simplified story with an appreciation of the supraindividual aspects of the social realm, aspects suppressed or ignored in the simplified version. Thus, the complex narrative can affirm that acts were disavowed, but also observe that many structures produced by those acts were left in place. The complex narrative can note also that other structures were produced and used in ways that eventuated in discrimination as well. The more complex story, considered in the context of an awareness of an asymmetry of experience, and highlighting the partial nature of

80 / Chapter 3

the revolution in patterns of race and gender interaction, renders our failure to move beyond the era of contested prejudice much more understandable than does the simplified story. How could political, business, and cultural leaders explain to the wider public the dismantling of some discriminatory structures and the simultaneous creation or maintenance of other discriminatory structures? The first observation to make is that the question presupposes a level of reasoned discourse and debate that just does not operate in the political sphere—politics is not a debating club. Further, it presupposes that actors on the ground, in the moment, will realize and react to possibly obscure connections between a multiplicity of distinct policy areas. Yet, one does not need to maintain that contractors for the interstate highway system explicitly colluded with manufacturers, developers, and real estate interests to build suburbs and move manufacturing from the central cities. A more complex possibility is more consistent with Ann Swidler (1986), who notes that people use culture as a tool kit. Drawing on Swidler, I would suggest that the resources developed by highway building, for example, were put in ser vice to aims that made sense to the actors involved, and often the possibilities for doing so only became visible after the resource had been produced. Hence, one does not need to posit some vast malicious conspiracy that planned to give with one hand and take away with another. All one need posit is persons’ tendency to use resources at their disposal to create and maintain conditions they understand to be consistent with their material sustenance and symbolic understanding of the world. Of course, all these observations culminate in the realization that there may have been no moment in which any actor was expected to explain the apparent contradiction alluded to by the question above. Yet, any possible contradiction might not have been apparent even if the different actions had been observed. When structures were removed, the justification for their removal was articulated through an antiprejudice rhetoric. However, because many structures may be relegitimated using other logics—see Nixon’s housing policy announcement—many of the structures erected in the era of condoned exploitation could easily live on, as their justifications were refashioned and the understanding of the new justifications diffused through sectors of society. The shifting of the justifications, therefore, allowed prejudice—a par ticular explanation for societal arrangements and practices—to be delegitimated, while many of the discriminatory arrangements remained intact. What the maintenance of old discriminatory structures and the creation of new arrangements of discrimination means, of course, is that in the history described in this chapter, discrimination is only reworked—it is never really escaped. Thus, the two stories differ appreciably. Of course, Disney-esque is not a pejorative term in general, as may be suggested by the observation that the

From Condoned Exploitive Relations to the Era of Contested Prejudice / 81

Disney-esque explanation is accurate, as far as it goes. Labeling an explanation “Disney-esque,” therefore, should not raise concern about that explanation if the explanation is meant for children. For children, by gently relating some aspects of the truth Disney-esque explanations can ease their maturation; for adults, by masking the truth and complexity of matters Disney-esque explanations can only ease their disempowerment. The two different histories have different ontological implications as well. In the Disney-esque version of history it is eminently reasonable to define discrimination as only the individual acts that one person does to another, because the social world is composed of only such acts. If that history is true, then the dominant definition of discrimination is appropriate, and any other definition may appear nonsensical. However, if the history is more complex, as reflected in the narrative provided here, then a more complex, nuanced definition of discrimination that takes into account discriminatory structures may be needed.

Concluding Remarks The common understanding of the postwar period through the late 1960s is that thousands of courageous Americans marched, chanted, challenged, and eventually converted the American people to egalitarianism in matters of race and gender. The conscience of the nation was awakened to the evil in its midst, and bold action to eradicate it ended segregation and ushered in an era of true opportunity. This picture is not entirely false. Surely, thousands of courageous citizens, blacks and whites, men and women, Jews and Muslims, Protestants and Catholics, young and old—indeed, people of all ages, ethnicities, and walks of life joined together to challenge the nation to make real the promise made at the inception of independence. But there is another dimension to the story, and that forgotten dimension virtually never is the object of movies or television documentaries. That dimension concerns the fall of the regime of condoned exploitation, a fall that occurred through contest that never produced unequivocal victory for blacks, women, or even the principle of justice. What was overturned was the legitimacy of a rhetoric of explanation, not an extensive class of deeds and arrangements. That is not to suggest that nothing has changed. Language has changed, and other changes have occurred as well, changes that in a deep sense determine the resources available for different communities. Wilson describes much of the relevant change in U.S. history in that regard, effectively illustrating how the nature and brutality of exploitation may vary as social, economic, and political realities change. And, illuminating studies of gender in the United States chronicle the complexity of gender relations and public

82 / Chapter 3

policy as well (e.g., O’Conner, Orloff, and Shaver 1999). The changes these researchers highlight matter greatly for the material, rhetorical, positional, and intellectual resources available both for day-to-day life and for any efforts to make social change. However, note that both the first and second period of black-white relations Wilson identifies fall squarely and completely under the regime of condoned exploitive relations I have identified, and much of Wilson’s modern period was actually a moment of transition in my analysis. On this basis one might argue that my regime of condoned exploitive relations is indefensibly long, for it subsumes within it the spread and eradication of African American enslavement as well as the disenfranchisement and then the enfranchisement of women, either one of which is clearly a turning point in race or gender relations. At the same time, my divisions mean that incidents as disparate in kind and time as the three-fifths compromise, which institutionalized slavery in the eighteenth century, and the hiring and later firing of women in industry at the start and end of wars (an arguably private act occurring as late as the 1940s) become evidence. Thus, one might argue, my carving of time into such broad categories ignores important changes that occurred. Any division of time is a judgment call. But, the length of the period is not relevant to the case for the existence of an identifiable era and for an identifiable set of moments of qualitative change. What is relevant is the motivation for the division. My aim is not to identify a change in the situation of blacks or women individually; my aim is to identify a change in the legitimate ideological placement of both blacks and women. I am searching for nothing short of a fundamental shift in the social relations of all groups, such that exploitation rhetoric is delegitimated. As long as one group is juxtaposed to another, or one rises while another falls deeper into vulnerability to exploitation rhetoric, then nothing fundamental has changed. Seen in this way, one should not be surprised that the divisions I identify differ from those produced by studying the experience of blacks or women alone. Further, looking over the span of human history in general, or western history specifically, when it comes to such fundamental matters as the social relations between groups and the legitimating ideologies for those relations, it is the rare epoch that lasts less than a century. Our sensitivity to our own time and well-documented events may easily magnify the import of tremors through which we ourselves have lived or for which we have more extensive information, making a small fluctuation appear to be a shift of continental proportions. Yet, major top-to-bottom shifts in social relations may occur no more often than do fundamental changes in technology (e.g., the age of carbon-based energy, circa 1750 to the present; see Roberts 2004), financial arrangements (e.g., mercantilism, circa 1500–1700), and other foundational arenas (e.g., political entities and systems, such as the nation-state system,

From Condoned Exploitive Relations to the Era of Contested Prejudice / 83

circa 1860 to the present [Palmer and Colton 1978]). For technology, we know that inertia bred from uncertainty inherent in new technologies, plus the stock of prior infrastructural investment coupled with the economic interests those investments represent, collectively slow the transition from one technology to another; for finance, we know that a web of debt, obligations, and local, regional, national, and international rules greatly reduces the incidence of fundamental, power-shifting change. At base these inertial tendencies flow from powerful actors’ unwillingness to risk losing in the transition from one epoch to another. If this is true in matters as contract-laden as finance and as material as technology, it is likely at least as true in an interactional arena wherein which a main activity is the articulation and enactment of interests. The social relations between men and women and blacks and whites clearly fall within the interactional arena. Hence, to expect short periods of fundamentally different social relations is to be disappointed. Certainly, as one tightens the focus on any given period of time, it is possible to discern shifts and changes of import for those living during the period. But the deeper ramifications for change, or lack of a ramification of what appear to be big changes, can often only be discerned by broadening the focus, not narrowing. Once we broaden the focus enough to consider simultaneously the plight of both blacks and women, it should be clear that the basis for the demarcations herein is that what does not change throughout the prewar period is the fundamental vulnerability of both groups as well as others, who basically were rightless during the entire era of condoned exploitive relations—rightless in that perceived crisis could motivate the restriction or elimination of whatever freedoms blacks and women temporarily enjoyed. It is not clear that such is the case in the present moment, and the dissolution of the legitimacy of exploitive rhetoric makes some actions much more difficult to pursue. Indeed, one tenet of the claim that the era of condoned exploitation has ended and an era of contested prejudice has begun is that the benefits of delegitimating the language of exploitation flow not only to blacks and women, the focal agitators for social change, but to all groups. A comparative case study germane to this hypothesis is the contrasting treatment of those of Japanese ancestry as World War II began, and those of Arab ancestry as the War on Terror commenced. Both wars were regarded as having begun with an attack on American soil that shocked the nation, leading to large loss of life. Both attacks were perceived by many U.S. residents to have been launched by mad members of a warmongering culture. However, whereas in the first case persons of Japanese ancestry were rounded up and interred for the duration of the war, losing their property in the process, all with the support of the Supreme Court (Korematsu v. United

84 / Chapter 3

States (323 U.S. 214 (1944))), in the second case such actions have (to date) been successfully resisted. Sporadic calls advocating racial/ethnic profiling have so far gone officially unheeded or even repudiated. Indeed, at the outset of the War on Terror, elites tended to denounce mob behavior that targeted those of Arab ancestry. Of course, it is unlikely that the arrival of an era of contested prejudice completely explains the differential treatment of these two groups in somewhat similar circumstances. Still, it is difficult to maintain that the general delegitimation of rhetorics of prejudice played no role in the different outcomes. For these reasons, therefore, I have argued that the changes Wilson identifies are important details for understanding the status of African Americans specifically, but for our purposes here we need keep sight of an important set of continuities—for most of U.S. history the interests of selected outgroups were routinely and unashamedly sacrificed for the interests or even the comfort of others, regardless of the illegality of the acts that would impel the sacrifice, or the threat that the vulnerability of the targets implied for the interests of those only slightly better off. And, in my reading of history, the national sacrificial lambs of choice in the United States from the beginning were blacks and women. That is not to imply that blacks and women were the only ones. At different times, and in different regions or states or cities, certainly different groups joined blacks and women in this status of vulnerability. But note, these other groups joined blacks and women; if not in fact then in the collective imaginary, they did not supplant blacks and women. For this reason, no matter how much it may be useful to know additional stories, one cannot understand the era in which we live without understanding change and stability in the historic status of blacks and women. Surely, many scholars, especially those who may have been born during or after the 1960s, are well aware of the mixed blessings of the 1948–1970 period. But the commentators who stoke the fires of public opinion often ignore the complex reality described herein, and amply documented by social scientists and social historians. This history has important ramifications for the current period. After the delegitimation of condoned exploitation, acts that had been directly defensible as exploitation are now potentially contestable as discrimination. Hence, determining what will and will not count as discrimination becomes of central importance. In this way, defining discrimination moves to the center of the effort to socially construct inter-racial and intergender interaction. As discrimination, variously defined, becomes key to the conflict, the courts as the final arbiters become key as well, even though the courts can only respond to some parts of the complex narrative provided above. The

From Condoned Exploitive Relations to the Era of Contested Prejudice / 85

courts, like any institution, could only be partial in their response because the legal sphere, like any institution, is not neutral; it has embedded within it interests, aims, and values of its own. Thus, as the courts become more prominent, the interests, aims, and values of the legal sphere also come to play a big role in determining what counts as discrimination and what does not, not only solidifying the context social scientists seek to study, but also greatly affecting which of the particular theoretical tools and concepts social scientists would propose to employ in their research would be perceived as intersubjectively rational, and which would seem clearly irrational. Thus, to critically consider the dominant social science perspective and analysis of discrimination effects we must first turn toward the law, to discern how its interests, aims, and values mattered for its relevant contribution to our concerns— a definition of discrimination.

4 Defining, Finding, and Remedying Discrimination: Dominant Legal Perspectives

T

he legal system operates with par ticu lar logics. A logic drawn from the legal system may or may not be transferred usefully to other spheres of activity. For example, one may apply a logic drawn from contract law to the relationship between adult children and aging parents. Were one to do so, it would be an open question as to whether the transference is helpful or harmful to our understanding of childparent relations. Quite possibly, some or perhaps even all aspects of the child-parent relation would fail to fit the logic of contract law. This observation is germane as we consider the phenomenon of discrimination. Much of the contestation around discrimination has occurred in debates about proposed legislation, executive orders, and court rulings. This contestation has rendered discrimination a legal issue, making it easy for legal reasoning to influence views and understandings of discrimination beyond the statehouse and courtroom. Therefore, though our analysis concerns social science theory, research, and, ultimately, societal understanding, one cannot progress far in that enterprise without considering how the legal system conceives of discrimination and responds to an allegation of discrimination. Yet, the utility of legal reasoning for understanding discrimination is not necessarily obvious. Discrimination is certainly a legal issue, but is it only a legal issue? If discrimination is not only a legal issue, can the legal

Dominant Legal Perspectives / 87

definition of discrimination fully define the sociological phenomenon of discrimination, including any translegal aspects of the phenomenon? In order to answer these questions, we must first describe how the legal system conceives of discrimination, and how it responds to an allegation of discrimination. By analyzing this conception and response, we may identify key features of the courts’ treatment of the phenomenon at issue. Then it will be possible to assess whether the courts’ reasoning and approach provides all we need reference when we consider the phenomenon of discrimination. Of course, by the courts’ reasoning I mean to signify the accepted legal definition, the definition and apparatus that constitutes the official, dominant perspective in the law. Before beginning this section of the work, however, it is worth noting that it should be obvious that law cannot be sufficient to define discrimination. Laws vacillate. For example, on September 15, 1935, the Reichstag unanimously passed the Nuremberg laws, which restricted the recognized rights of Jews in Germany (Dawidowicz 1975: 84). Passage of these laws made the differential treatment of persons legal. If the claim is that law makes and unmakes discrimination, then one would be able to conclude that, for the most part, the Jews in Germany did not experience discrimination in 1936, because their maltreatment had been made legal. I submit, however, that if humans define a phenomenon, the act of defining the phenomenon does not make the definition appropriate; instead, definitions need be consistent with the internal logic and operation of the phenomenon, or incoherence will prevail. And, because law may sometimes define phenomena by attending less to the internal logic of the phenomenon than to the configurations of power in the environment, it is a mistake to expect the law to define phenomena with translegal coherence. These observations are obvious in the abstract. The challenge seems to come in applying the lesson in specific situations, especially situations drenched in the interests and power relations that form the context in which the situation has meaning and material implications. This challenge might appear to be a concern of jurists alone, serving to identify a tendency that complicates their efforts to dispense justice. Yet, as I show in Chapter 6, the dominant legal definition of discrimination has greatly influenced the definition of discrimination social scientists routinely use, making this a matter for our concern as well. Thus, the larger question for us is whether the legal definition of discrimination in the United States, forged in the context of existing power relations, eventuates in an adequate definition for social scientific research and public understanding of the phenomenon, or a flawed definition that hinders sociological research, public understanding, and, ultimately, the development of an inclusive, feasible resolution to the scourge of discrimination.

88 / Chapter 4

To pursue these questions, in this chapter I commence by outlining discrimination as defined in the law, as well as describing the ways in which plaintiffs have been allowed to establish the existence of discrimination. Afterward, I describe key decisions on the way to this definition, as well as emergent distinctions between legally defined discrimination and other social phenomena, while I trace their implications for the unfolding era of contested prejudice. Next, using the existing dominant legal definition of discrimination, I analyze one important arena wherein which discrimination might occur. This analysis provides a shared example of how the concept of discrimination, as defined in dominant legal discourse, operates. Finally, I close by noting the implications of this definition and the understanding it makes possible.

Discrimination before the Law For an issue to come before the courts, someone must regard some action or arrangement as a violation of the law. In pursuing a resolution to the dispute, persons joust over what happened—the physical and mental actions and arrangements—as well as the interpretation of what happened. That is, disputants debate whether someone did or did not act in a par ticular way, or whether some arrangement does or does not exist. At the same time, disputants debate the meaning of the act, nonact, or arrangement, offering alternative understandings of the (sometimes same) occurrence or arrangement. In adjudicating these disputes, the courts seek to render justice in the moment, and also to signal to others what does and does not violate the law. It is this latter aim that inexorably pulls the courts into offering a working definition of discrimination. By publicizing a definition, the court may aid persons and the wider society to construct and maintain arrangements and acts in accord with the law. Yet, if there is a disjuncture between the legal definition and the sociological phenomenon, then persons may be in accord with the law, and yet discrimination still may be operative. We have not yet established such a disjuncture. The first order of business for considering whether there is a gap between legal definition and sociological phenomenon is to look more closely at the legal definition, the courts’ processes, and the ways in which the courts’ aims, values, and interests are enacted through both.

Discrimination as a Legal Concept: Sociohistorical Observations Above I reference the Nuremberg laws, but though Germany furnishes one of the better-known examples of legislators legalizing discrimination, it is certainly not the only such example. Laws that restricted women’s property ownership, or that segregated African Americans, or that interred the Japa-

Dominant Legal Perspectives / 89

nese, or that dispossessed Native American peoples, or that disenfranchised Jews (e.g., Nathans 2004)—examples abound in the United States. Indeed, the original U.S. Constitution legalizes discrimination to the extent it legalized slavery, women’s disenfranchisement, and the dispossession of Native American peoples. As the era of contested prejudice took hold, elites sought to reduce the most blatantly discriminatory legal structures. To accomplish this aim, elites proposed Constitutional amendments (e.g., the Anti-Poll Tax Amendment, which became the Twenty-Fourth Amendment; the Equal Rights Amendment, which was never ratified) while offering legislative acts and executive orders within the context of this elaborated Constitution so as to carve out rights for blacks and women. By proceeding in this way, they endeavored to avoid delegitimating broader national structures, while satisfying both domestic demands and international concerns.

Discrimination as a Legal Concept: Contemporary Observations Perhaps owing to this history, discrimination law is scattered through several legislative acts, judicial rulings, and executive orders (Blank, Dabady, and Citro 2004). However, generally the law has come to observe two par ticular variants of discrimination: (1) disparate treatment discrimination, and (2) disparate impact discrimination. An example of disparate treatment discrimination would be to refuse to hire black job applicants while hiring equally qualified white applicants. This is the most straightforward kind of discrimination—a person is rejected because of his or her race. Another type of discrimination defi ned by law is disparate impact discrimination. This is slightly more complex. Consider the example of a fi re department. If a fire department were to require that all firefighters be 5′10″ or taller, this is possibly disparate impact discrimination in that, owing to women’s shorter average height compared to men, the rule will eliminate far more women than men. This kind of rule would be allowed, and deemed nondiscriminatory, if the fire department could show that being 5′10″ or taller is materially related to being able to perform the job—i.e., if the fire department could show that height of 5′10″ or more is a bona fide occupational qualification (BFOQ). The department would need to show that this requirement is intrinsic to the task, not simply a function of the par ticu lar kinds of arrangements the department has adopted to accomplish the job. So, for example, if the reason the firefighter must be 5′10″ or taller is that the department does not have any firefighter uniforms in smaller sizes, or that the department only has ladders that persons with long legs can climb owing

90 / Chapter 4

to the wide separation between the rungs, then this would not establish a BFOQ. The fire department would be expected to obtain a broader range of uniform sizes, for the clothing available is a trivial concern. And the department would have to obtain ladders with less separation between rungs; indeed, ladders with wide separations pose major problems, for how could many of the persons the department might need to rescue hope to climb down the ladder should the requirement to do so arise? The example is meant to indicate that the disparate impact factor does not, in principle, become a BFOQ just on the basis of assertion. And, in the absence of evidence indicating the necessity of the 5′10″ height restriction, the restriction would be deemed disparate impact discrimination. Thus, both disparate treatment and disparate impact perspectives on discrimination disallow the exclusion of persons on the basis of an irrelevant category; the list of irrelevant categories depends upon the sphere of activity before the court. Where the defi nitions of discrimination differ is in the mechanism used to exclude persons. In disparate treatment discrimination the category itself is the object of focus. In disparate impact discrimination gatekeepers may have maliciously or inadvertently selected a screening mechanism that is not defensible on the basis of the activities required of those who pass the screen, and tends to exclude substantially more of one category than another. In short, the screen may be correlated with persons’ categories enough that it somewhat substitutes for knowing the category of the person, allowing gatekeepers to (consciously or unconsciously) exclude most or all persons of certain categories. Because moving to screening systems that preserved the status quo was an early response of some companies when antidiscrimination laws came into effect, the courts identified the practice and invalidated it (Griggs v. Duke Power Company (401 U.S. 424 (1971))). Hiring is not the only province for disparate treatment or disparate impact discrimination. One can admit students in a manner that entails disparate treatment discrimination or in a manner that entails disparate impact discrimination. One can promote workers in a way that can implicate either or both types of discrimination. One can let an apartment to a renter using approaches that are discriminatory under either or both definitions. In short, one can find the possibility of disparate treatment and disparate impact discrimination in a wide variety of spheres.

Discrimination Litigation: Selected Examples and Implications Despite these fairly straightforward definitions of discrimination, the rules for finding discrimination can be involved, depending on the par ticular stat-

Dominant Legal Perspectives / 91

utory basis of the claim, whether the claim occurs in the context of public sector or private sector activity, whether the claim is raised in federal court or state court, and more. The rules can induce a pattern of shifting burdens as the case moves forward. For example, in disparate impact cases, plaintiffs must first show that they are disadvantaged (e.g., they have not been hired, promoted, or admitted). If the plaintiffs are successful, then the burden shifts to the defendants to show that there are legitimate nondiscriminatory reasons for the disadvantage. If the defendants are successful, then the burden shifts to the plaintiffs to show that the reasons articulated by the defendants are a pretext, are not the real reasons, or are not the only reasons because discrimination is another reason (i.e., there are mixed motives; Ryan 2003; Paetzold and Willborn 2002). Hence, the rules of engagement in the legal sphere can inaugurate a pattern we might label legal ping-pong, where the burden of proof is parried back and forth as claim and counterclaim are almost ritualistically invoked and adjudicated. The above describes the contemporary situation, but rules have changed over time and continue to change in the contested terrain of the legislature and the courts. Given this complexity, and the focus of this work, I only provide general outlines relevant to the issues under consideration here. And, the key issue under consideration here is: How, in general, does the court conceive of and act around discrimination in the contemporary period? Closer consideration of what occurs during litigation may aid this effort.

Legal Practice Plaintiffs and defendants may present any or all of three general types of evidence to establish or rebut a claim of discrimination. They may present anecdotal evidence (i.e., evidence of discriminatory statements or exculpatory statements), comparative evidence (i.e., evidence comparing the treatment of similar situated persons who are not members of the protected class), or statistical evidence (i.e., comparative evidence based on large numbers of cases, which preclude a case-by-case comparison; Paetzold and Willborn 2002). An example of comparative evidence is provided by Frank v. United Airlines (216 F.3d 845 (9th Cir. 2000)). In that case plaintiffs claimed that United Airlines’ weight requirement system was discriminatory. Up until 1994, United Airlines required male and female flight attendants to stay under a maximum weight requirement. These maximum weights differed. Attendants failing to stay below the maximum weight limits could be suspended or, ultimately, fired. The maximum weights were drawn, for the most part, from separate weight by age tables for males and females. The male weights came from tables published by the Metropolitan Life Insurance Company.

92 / Chapter 4

The female weights came from a table of maximum weights promulgated by Continental Airlines. However, the use of different maximum weights was not itself deemed discriminatory. Plaintiffs observed that the Metropolitan Life tables were actually weight by age by frame tables, and the court agreed that the maximum weights for men were drawn from the tables for men with large frames, while the maximum weights for women were consistent with the tables for women with medium frames. As no bona fide occupational qualification for this discrepancy could be shown, the maximum weights were ruled discriminatory. In this case it was simple to directly compare the weight tables for males and females. Often, however, to make the appropriate comparisons, plaintiffs and defendants have relied on social scientific statistical analyses of institutional admissions and corporate hiring, promotion, and salary decisions to establish or rebut a claim of discrimination. In the last few decades multiple regression analyses have come to be accepted as evidence by the courts, with the Supreme Court validating the use of regression in Bazemore v. Friday (106 S. Ct. 3000 (1986)). In Bazemore v. Friday, plaintiffs presented an analysis of wages for blacks and whites in the North Carolina Agricultural Extension ser vice. Their analysis indicated that black workers were paid some $330 (1974-era dollars) less per year than were whites, after education, tenure, job title, and job performance were controlled. Several legal issues were resolved in the Supreme Court decision, including whether it is legal or illegal to pass prior discriminatory inequalities that were formerly regarded as legal through to a period after which statutory changes made the inequalities illegal. The Court ruled such prior illegalities did not absolve actors of their responsibility for wiping away the discriminatory inequality. In addition, the Supreme Court ruled that a regression model did not need to include all factors that could be determinants of wages in order for the plaintiff to meet their evidentiary burden in proving discrimination.

Implications of Legal Concepts and Practice One could continue to explore the case law on discrimination. Many of the cases are fascinating, such that it is easy to become lost in the details of cases. But, without considering additional cases, the broad outlines of the process are clear: Plaintiffs complain of the actions of identified actors who, by virtue of the identification occurring before the court, become defendants defending themselves against the accusation. Analyzing this structure may reveal its possible implications for our understanding. Through this structure, the courts establish par ticular acts as discriminatory and par ticu lar perpetrators of those acts as discriminators. Indeed, it

Dominant Legal Perspectives / 93

is this latter designation that drives much of the development and operation of the law. The key ramification of this structure is that, in U.S. legal doctrine, punishing the violators of a law is at least as important as compensating the victims of crime, even when punitive damages are not at issue. Defendants who only pay restitution are not worse off than they initially were, but they are worse off than they would have been had they prevailed in the case. And defendants who must cease certain business operations, while not necessarily formally paying punitive damages, do incur a cost. Thus, it is punishment of a sort to require a defendant to pay restitution or alter standard operating procedures. Of course, if anything, the general trend in U.S. policy the last few decades has been toward strengthening the role of the punitive imperative. The punitive trend is clear in the criminal courts. The advent of three-strikes laws, the declining role of judicial discretion in sentencing, the spread of victims’ rights laws allowing victims’ voices to be heard at sentencing, the return of executions—all suggest a heightening of the punitive imperative. Given the strength of this movement toward punitive responses to crime, it is plausible to surmise the increasingly punitive impulse is not confined to one side of the court house. The more recent punitive thrust is not directly germane to the punitive imperative in antidiscrimination law, however. The punitive imperative in antidiscrimination law predates the late twentieth century upsurge in punitive sanctions; indeed, the punitive imperative is embedded in the very structure of antidiscrimination law. Thus, the punitive imperative of antidiscrimination law may not ebb even if punitive measures decline in popularity or use in general. The punitive imperative is embedded in dominant legal reasoning around discrimination, because in discrimination law the punishment of one is the reward for the other—awards for successful plaintiffs, even awards lacking a formally “punitive” component, come at the expense of unsuccessful defendants, and the award for successful defendants is freedom to continue the behavior that led plaintiffs to sue in the first place. Hence, the punitive imperative is embedded in the very structure of the processes of antidiscrimination law, made visible in the reality of plaintiffs gaining for every dollar defendants must pay (after court costs and legal fees, depending on the judgment). It is not far off the mark to note that to win is to punish. In this environment, when persons go before the court to claim that they have been victims of discrimination, the court is interested both in whether they were victims of discrimination and in who perpetrated the discrimination. In the process of making their case, the alleged victims need to show not only that they are victims of discrimination, but also that they were victimized as a result of the defendants’ actions. Should plaintiffs be unable to show that the

94 / Chapter 4

parties named in the suit are guilty, plaintiffs will not obtain compensation even if they successfully establish the existence of discrimination. Recall the legal ping-pong pattern of argument. Cases begin with plaintiffs attempting to show they are disadvantaged. If they do so, defendants need to explain away this disadvantage; showing the disadvantage flows from something other than the actions of the defendant can be an effective strategy. If the defendant provides a reason for the disadvantage, it is then up to the plaintiffs to show that this reason is not the real or important reason and that the defendant is at fault. Hence, even if plaintiffs convince the court that they are victims of discrimination, failing in the last analysis will mean this defendant will not be held liable for the discrimination. And, when the defendant is not held liable, there is no relief for the plaintiffs.

The Legal Sphere and the Era of Contested Prejudice How did the punitive imperative ascend, and the particular definition of and approach to discrimination become accepted? The courts played a key role in the emergence of an era of contested prejudice, and in the ascendance of the punitive focus. In a complex process, court rulings came to install a particular definition of discrimination as legitimate and to secure the punitive imperative as a bedrock part of the adjudication process, all in a context constituted by existing interests, power relations, and the courts’ own interests, aims, and values. One can point to key cases in that process. Some cases opened the door to the possibility of class-focused remedies in a way that might obviate the need to identify specific perpetrators. The court struck down the constitutionality of the separate-but-equal doctrine in Brown, rendering dual systems of education unconstitutional. In ordering a remedy, the court required lower courts to consider particularities of desegregation proposals in light of Brown as the process of dismantling dual systems of education proceeded. At that moment there was a possibility that class-focused remedies, without regard to the direct cause of the harm, might gain ascendance. Yet, by the mid-1970s, the idea of a class-focused remedy had no traction whatsoever. Indeed, the defi nition of discrimination itself became more and more constrained, along with the prospects for remedy. In Milliken v. Bradley (418 U.S. 717 (1974)), the Court ruled essentially that only the specific jurisdiction directly responsible for the alleged discrimination could be held accountable, disallowing cross-district desegregation plans as a means to desegregate urban districts whose whites had moved to the suburbs. In this ruling, the Court ignored the role of the state of Michigan in facilitating segregation in the first place. In 1978, the Court further circumscribed matters, ruling in University of California Regents v. Bakke (438 U.S. 265 (1978)) that

Dominant Legal Perspectives / 95

policymakers cannot construct generally compensatory plans; only those who can show they are direct victims of discrimination can be compensated. Finally, in 1990, in Missouri v. Jenkins (495 U.S. 33 (1990)), the Court shielded one social sphere from another, ruling the state could not address segregation produced in one arena (in this case residential segregation caused by white flight) by policies in another area (in this case schooling). With this ruling, the prospects for discrimination law had finally been fully contained. Two observations need be made at this juncture. First, a cottage industry has developed concerning whether courts instigate change, merely signal it, or are irrelevant (e.g., Grossman 1970; Levine and Becker 1970; Kemp, Carp, and Brady 1978; Gaskins 1982; Rosenberg 1991; Mishler and Sheehan 1993; Garrow 1994; Mondak 1994). Our concern, however, does not turn on the ultimate cause of the narrowing definition of discrimination. Supreme Court decisions mark the state of dominant legal reasoning, regardless of whether the Court’s decisions were causal or spurious. And, the pattern we observe is such that it ultimately resulted in a narrowly framed understanding of discrimination that prohibits full class-focused remedy and, instead, demands the identification of specific discriminators in specific spheres of action. Second, the development of the legal definition of discrimination appears largely concerned with African Americans, and inordinately focused on school desegregation. This historical reality creates intriguing contemporary circumstances that, once considered, reveal part of the reason why the era of contested prejudice is an interstitial moment always at risk of being eclipsed by the ascendance of condoned exploitation. For example, one reading of sexual harassment could see it as discrimination. Yet, of late, the law has developed differently around sexual harassment. In discrimination law, plaintiffs must show the par ticular defendants are directly responsible for their disadvantage. The contrast between this state of affairs and the ruling in Burlington Industries, Inc. v. Ellerth (524 U.S. 742 (1998)) is striking. Justice Clarence Thomas, dissenting, and joined in his dissent by Justice Antonin Scalia, put the matter directly: The Court today manufactures a rule that employers are vicariously liable if supervisors create a sexually hostile work environment, subject to an affirmative defense that the Court barely attempts to define. This rule applies even if the employer has a policy against sexual harassment, the employee knows about that policy, and the employee never informs anyone in a position of authority about the supervisor’s conduct. As a result, employer liability under Title VII is judged by different standards depending upon whether a sexually or racially hostile work environment is alleged. (Burlington Industries, Inc. v. Ellerth (524 U.S. 742 (1998), Thomas, C. dissenting))

96 / Chapter 4

Thomas and Scalia accurately assert that the liability standard should be the same regardless of whether race or sex discrimination is alleged. They resolve the discrepancy by claiming that “An employer should be liable if, and only if, the plaintiff proves that the employer was negligent in permitting the supervisor’s conduct to occur” (Burlington Industries, Inc. v. Ellerth (524 U.S. 742 (1998), Thomas, C. dissenting)). Thus, Thomas and Scalia prefer to use one standard, the standard that operates for discrimination law in general and that requires one to prove the employer’s culpability. Prior to Burlington Industries, Janet Dine and Bob Watt (1995) argued that U.S. courts were moving to separate sexual harassment law from key aspects of discrimination law. Indeed, they ultimately contend that disparate treatment is not relevant to sexual harassment, because sexual harassment is at base bullying behavior. The claim that sexual harassment is bullying is, of course, true, and thus their recognition of a coherent development in dominant legal reasoning is persuasive. Yet, persuasive as it is, reconsidering the commentary reveals just how much the violent and coercive basis of discrimination is denied by dominant legal reasoning, for the commentary presupposes that discrimination is not at base bullying behavior. The separation of race and sex discrimination, of course, threatens to bring the era of contested prejudice to a reactionary end. Though many more developments might need to occur to fully restore the regime of condoned exploitation, separating logically inseparable phenomena creates divergent interests and therefore may activate power dynamics that might eventuate in just that outcome. Recall that advance for one large group often occurred in concert with retreat for another. If that process can be run in reverse, then the separation of race- and sex-based litigation developed in Burlington Industries, for all its promise for extending rights and transcending the punitive imperative, also poses a danger. Whether the process that heralded the dawn of the era of contested prejudice can be run in reverse remains an empirical and theoretical question without a clear answer, but it is certainly a question of some social scientific and public policy import. Still, the claim that race and sex discrimination can be treated differently by the courts should not be overstated. In the main, to prove an allegation of sex discrimination requires one meet the same burden one would need to meet to prove an allegation of race discrimination. The important implication of the distinction in standards of liability that Burlington Industries erects in the case of the par ticular type of discrimination termed sexual harassment is its clear indication that the dominant legal defi nition of discrimination is not coaxed out of the logic inherent in the par ticular phenomenon under consideration. Indeed, considering the divergence in treatment of race and sex discrimination under Title VII, it becomes apparent that even though many may regard the existing dominant defi nition of discrimination as “ob-

Dominant Legal Perspectives / 97

vious,” the dominant definition was not fated necessarily to develop as it did. Instead, as one might expect, it is a result of specific decisions judicial actors made in an unfolding context of power relations. Despite the emergent exception of sexual harassment law, in discrimination law we find consistent operation of the punitive imperative, which allows “doing nothing wrong” to be a successful defense. Ironically, the ultimate result of the punitive imperative has been to shield any entity not implicated directly in the specific, identifiable actions involved in constructing the litigated state of affairs, while severely limiting the plaintiff class and the remedies available. This is a far cry from the outcome one might have expected on that May morning in 1954 when the court handed down its ruling in Brown. Yet, the decisions that followed secured a particularly narrow definition of discrimination. It remains an open question as to whether this definition is or is not compatible with a coherent social scientific analytic agenda.

Human Capital Investment and Redemption through the Lens of Legally Defined Discrimination What are the implications of the dominant legal understanding of discrimination for concrete efforts to understand the operation of discrimination in any given instance or arena? Certainly, the full set of abstract principles by which the legal field adjudicates disputes is complex. Par ticular principles apply in par ticular contexts, depending on the jurisdiction, the arena within which the dispute occurs, and more. Yet, the key principle that concerns us—the punitive imperative—applies generally in the area of discrimination. And it is that principle that we may consider as we assess the possibility of discrimination in a par ticu lar, illustrative sphere of social life, the process of human capital investment, and the process by which that investment is redeemed for pecuniary or other income. Before proceeding to an analysis of how the punitive imperative and the dominant legal definition of discrimination allow us to interpret the process as discriminatory or not, I must first trace the outlines of the human capital investment/redemption process. Then, we may turn the lens of the legal definition of discrimination and the punitive imperative on that process.

The Process of Human Capital Investment and Redemption There are many ways to conceive of children’s movement from infancy into adulthood, from dependency on others to relative independence. One

98 / Chapter 4

important framework with which analysts have analyzed this process is provided by human capital theory (e.g., Becker [1964] 1993). Human capital investments are outlays that increase the productivity of the human being by increasing the person’s skill, knowledge, or capability of being economically productive. Investments can be made in many ways—improving health outcomes, migrating, taking courses in school, obtaining on-the-job training—all these and more constitute means of making human capital investments. Human capital investments can be made at any age, but the earlier an investment is made, the greater the opportunity to make later investments that build on the earlier ones, and the longer the period during which returns on the investment can be obtained. One task that families and others accomplish is to invest in the human capital of younger members of the household. These investments are made in many ways. Parents and other relatives who read to young children prior to and in the early years of formal schooling are making human capital investments. Parents who ensure the child receives nutritional meals and appropriate vaccines are not only taking care of the child’s health; such parents are also making human capital investments, facilitating the child’s ability to live and earn without the challenge of debilitating illness. Parents who take their child to museums are not only sharing a recreational and emotional experience they and their child may remember for a lifetime; such parents are also making human capital investments by broadening their child’s exposure to the world. Indeed, it would be difficult to identify all the ways that parents may make human capital investments in a child. Human capital investments intensify when the child enters formal schooling. Indeed, one way to regard the school is as an engine of human capital investment. Students undergo formal training, learning in many domains of knowledge, including mathematics, language, science, social studies, under the tutelage of school personnel usually licensed for that purpose. They also learn outside the classroom—on field trips, yes, but also in the lunchroom, on the playground, and in many other locations. Students learn in interaction with peers in formal venues, such as clubs and teams, and in informal venues as well. All these opportunities facilitate human capital investment. Investments can continue after the end of school, either in informal educational moments, or in formal postschool training. So, for example, some persons receive formal on-the-job training, which is another site of human capital investment. Others receive no formal training at work, but learn through experience; this, too, is a form of human capital investment. Of course, to have access to these par ticular opportunities, one must obtain a job or become profitably self-employed. And, mostly, after this occurs, persons shift to obtaining dividends based on their existing stock of human

Dominant Legal Perspectives / 99

capital investment. The shift may be partial or complete, but, even if partial, it becomes more complete over time, as long as there is no major disruption in the process. In a sense, during their working lives persons cash in their human capital investment, an investment that also tends to depreciate over time. Human capital accumulation is an important factor in the wealth of nations (e.g., Schultz 1961). And, inequality in human capital accumulation, or differential access to profitable places for obtaining human capital or drawing dividends from one’s human capital, are important potential causes of existing individual-level inequality in economic outcomes such as earnings (e.g., Mincer 1970; Sakamoto and Chen 1991). Owing to the important role of human capital, discovering a role for discrimination in human capital accumulation or redemption would have important implications. Yet, to make such a discovery, one requires a theory of discrimination that defines discrimination. The courts supply one such definition.

Discrimination in Human Capital Processes: The Dominant Legal Perspective Using the dominant legal perspective, it is fairly clear where one may find discrimination occurring concerning human capital. Wherever one fi nds a decision of whether to make a human capital investment in a person, a decision to allow a person to reap dividends from his or her human capital, or a decision as to the rate of dividend payment on the human capital, discrimination may occur. Disparate treatment or disparate impact discrimination may occur at school, if teachers bar some students from demanding courses based on a limiting view of persons. In the job acquisition process, we may observe discrimination operating as well. Discrimination may even occur in the family, in either form. For example, if parents decide to teach their twin boy how to count at age four, but decide to not teach their twin girl how to count at age four, and assuming equal ability and interest, the parents will have committed disparate treatment discrimination. Or, if parents decide to dole out praise only on the basis of mathematics knowledge, the parents may have committed disparate impact discrimination in the sense that boys tend to do better in math whereas girls tend to do better in language, at least at this juncture in the United States. Though analytically we may regard the parents as potentially discriminating, the bulk of the attention goes elsewhere. And, with that attention, under dominant legal reasoning the task is to identify specific acts of discrimination. Notably, under the dominant legal perspective, to establish

100 / Chapter 4

discrimination the particular decision and the decision-maker—individual or corporate entity—must be identified. So, for example, applicants denied admission to a college or university may sue that educational institution. They cannot sue other colleges or universities, nor may they sue the applicants who were admitted, for either response would be nonsensical under the dominant legal perspective. In principle, this does not present a problem. Plaintiffs may seek out sector after sector, sphere after sphere, in which they believe they have received discriminatory treatment, and litigate each single sector or sphere separately, naming specific defendants for each sphere. What this means, of course, is that discrimination as a phenomenon is carved into pieces in the law; each piece must be litigated more or less separately, and each piece must be traced to identifiable actors. And, why must par ticular actors, discriminators, be identified? Because the remedies must come from those actors in the form of restitution, changed behavior on the part of discriminators, or both. Thus, absent some identifiable (corporate or individual) person to blame there is no source of remedy, and thus no case to bring.

Concluding Remarks In dominant legal reasoning, discrimination can occur owing to disparate treatment or disparate impact. Regardless of which type of discrimination is at issue, the courts narrow the population responsible for compensating the wronged to those intimately connected to the conduct of a specific instance of the wrong. If it proves impossible to identify perpetrators then, regrettably, there is nothing the courts may do. If the beneficiaries of the wrong extend far beyond the set of persons who might have perpetrated the transgression, there is nothing the courts may do. In this way, U.S. legal doctrine repudiates discrimination, narrowly defined, even as it erects procedures whose material effect is to absolve the wider community and the institutions of society from any serious possibility of being seen to be complicit in discrimination or of being ordered to pay damages to targets of discrimination. This is not the only possible way to construct antidiscrimination law. In sexual harassment law, we observe an incipient effort to erect a firewall between doctrine centered on the courts’ logic of discrimination and sexual harassment jurisprudence. In the former, the punitive imperative is essential; in the latter, the punitive imperative is not determinative. If no other observation does, the comparison between sexual harassment law—a form of discrimination—and antidiscrimination law should tell us that the legal understanding of discrimination is not based in the actual logic of the phenomenon.

Dominant Legal Perspectives / 101

The comparison leads to the conclusion that when humans define a phenomenon, their work is not yet done. Analysts must consider whether the definition is consistent with the internal logic of the phenomenon, and reject definitions that fail this test. This is an obvious fact in the natural sciences; asserting an entity is a bacterium does not make it one, and proceeding as if it is a bacterium when it is not will lead to misunderstanding and eventually may prove fatal. This fact is no less true for the social sciences. Yet, because the subject matter of social science is often in part the very interests and power dynamics that can warp definitions beyond all coherence, social science sometimes may encounter great difficulty in administering this test the more beset by power dynamics the phenomenon at issue may be. Yet, definitions always are constructed in a field of power relations. And, one may contend, for all its flaws, the dominant definition of discrimination has worked, in that it has reduced egregiously harmful conduct, and in that revolution has not destroyed the Union to date. Thus, this society seems relatively stable if one takes a long, historic view of matters, and that stability can be traced, at least in a small way, to the legal system and its working definitions and procedures. Clearly, one might argue, a functioning society requires some kind of legal rulings. And the rulings must be rendered in as timely a fashion as possible. Thus, it is unlikely that any set of rulings in any arena of the law consistently satisfies some translegal criterion of coherence or “fairness,” bound up as the courts are with adjudicating disputes in real time, disputes that flow at least in part from different interpretations of the same reality, in an environment where the motivation for misrepresentation is not trivial. These are serious claims. Certainly, functioning legal systems may not always “get” the story right, may not always rule correctly. Yet, at base, these observations go beyond this point, implying not only that it may be inappropriate to expose dominant legal reasoning to critical review from an extralegal standpoint, but also that even if dominant legal reasoning fails such a test, it may be the best for which we can hope given the necessary constraints of any legitimate legal process. In other words, there may be no viable alternative way in which to approach discrimination within the legal sphere. This is a conclusive rejoinder, if apt. Though the example of sexual harassment law suggests there are other ways to proceed concerning discrimination, to fully assess whether the rejoinder is appropriate we must bring forth a coherent alternative legal perspective on discrimination for comparison. Bringing forth an alternative implicitly raises the question: Must social science use the dominant legal definition of discrimination as a touchstone, or might an alternative definition, rejected by the courts for its own perhaps legitimate reasons, be more appropriate?

102 / Chapter 4

With an alternative framework we will have in hand resources sufficient to allow us to assess the strengths and limitations of dominant legal reasoning, and to consider whether such reasoning can exhaustively and accurately define discrimination as a phenomenon for social scientific research and broader societal understanding. The road to that effort goes through introducing and considering critical legal theories on discrimination, a task to which we now turn.

5 Defining, Finding, and Remedying Discrimination: Critical Legal Perspectives and the Critique of the Dominant Legal View

M

ajor aspects of dominant legal reasoning concerning discrimination have been described. Further, a key feature of that reasoning—the punitive imperative—has been identified, and its implications for prospects of redress have been traced. Ultimately, the punitive imperative, in concert with the vision of discrimination as an act one individual commits against another, has created a situation in which the set of entities regarded as potentially liable for the discrimination so defined is severely constrained, and the likelihood of redress is vanishingly small. The legal rules of order that constrain the source of redress to only those who committed illegal acts may be necessary for a functioning, stable social order; few of us would like to pay for circumstances we did not create nor knowingly sustain, and certainly not on the basis of our demographic characteristics. But, one could make the alternative case. In a democracy, adults are responsible for knowing enough to take part in national decision-making, if only as voters or jurors; citizens, being formally empowered to direct the state, are responsible for the results that follow. Therefore, ignorance may be construed as a luxury one indulges at one’s own risk, and actions that are not opposed successfully may be seen as actions for which one remains liable. Consequently, though it is clear that few would like to pay for acts or arrangements they opposed or of which they are ignorant, some logics suggest that ignorance or opposition

104 / Chapter 5

may be no defense against liability for transgressions from which one benefits, or even actions from which one may or may not benefit. Doubtless, bystanders to acts do not take kindly to being identified with perpetrators of those acts, and do not want to be held accountable for arrangements they may not have consciously constructed. Fortunately for such bystanders, U.S. legal doctrine around discrimination generally does not recognize responsibility on the part of bystanders. Yet, this dominant legal reasoning has been criticized severely, producing resources that enable us to realize that we do not need to accept uncritically the dominant legal framing of matters. Scholars working in the traditions of critical race theory, critical legal theory, and feminist theory seriously question whether justice can ever be produced within the logic of dominant legal reasoning. These observations culminate in casting some of the contradictions inherent in contemporary legal practice in bold relief. And if contradictions beset the law around discrimination, imagine how they may hound social scientific understanding of discrimination. We will do more than imagine; what follows will demonstrate how legal perspectives resonate with social science views and weaken social science research. Note, however, that this chapter, a preliminary to that endeavor, is longer and more involved than the previous one. This is necessary because the material I treat here is not commonly brought into conversation with social scientific analyses of discrimination (e.g., Blank, Dabady, and Citro 2004). Thus, establishing its relevance requires a more painstaking approach. Yet, introducing this material ultimately will pay large dividends for our enterprise. To proceed, I first sketch a series of critiques of accepted legal approaches, drawing on critical legal studies and other perspectives on antidiscrimination law. Then, I elaborate the understanding of human capital processes developed in Chapter 4. Using this elaborated description, and the critical framework developed here, I reanalyze human capital processes to determine how the concept of discrimination, as defined in critical legal discourse, might operate. The chapter closes with a comparative assessment of dominant and critical legal perspectives on discrimination.

Critical Legal Theory: An Alternative Perspective on Discrimination Critical legal theory brings two key aspects into the consideration of discrimination: history and social structure. Thus, critical legal theory, unlike the dominant legal perspective, sees persons as embedded in a social structure built through a less-than-liberatory history. If that history has aspects of discrimination, and those aspects have not been addressed fully, then of course the present result, even if a taken-for-granted state of affairs, entails

Critical Legal Perspectives and the Critique of the Dominant Legal View / 105

discrimination. Hence, to assess whether discrimination exists, one must consider the source of current configurations of advantage, and one must do so in a way that takes seriously the extent to which persons are social, rather than atomistic, individuals. Consequently, a serious introduction to this perspective takes us into considering the history of the very existence of race in the United States, the definition of the concept of property, and the ways in which gender is rendered a property relation.

Historical Continuities and Race One line of thought critical of dominant legal reasoning connects existing white advantage to white privilege that was erected centuries ago and continues to exist at the turn to the twenty-first century. Cheryl I. Harris (1993) draws on historical research to argue that race as a concept was altered greatly by the development of racism—the division of persons on the basis of asserted “racial” characteristics—and whiteness became property in this process. Early in the Euro American experience, both Eu ropeans and Africans were bound to servitude. But, though the reasons are contested, historians agree that, over time, the rights of Africans began to be curtailed; blacks became isolated in slavery, and slavery became transformed into a permanent status inherited by one’s progeny. These developments had the effect of securing much-needed labor, while simultaneously partially weakening the likelihood that lower- class Eu rope ans would join with Africans to assert greater control in the Americas over and against wealthier Euro Americans. Developments in the concept of race occurred in the context of Euro American interaction with both Africans and Native Americans (Takaki [1979] 1990; Richter 2001). At the same time that Euro Americans were pushing Africans into more and more severe bondage, Euro Americans were dispossessing Native Americans of their land and wealth. Notably, both processes were legalized (i.e., rendered within the realm of admissible acts by Euro American law). And, central to this legal rendering were developing understandings of property and its inviolability in the Americas and in Euro American jurisprudence. Harris (1993) contends that a strain between logics of the rights of human beings on the one hand—slaves and dispossessed “savages” among them— and logics of the rightlessness of property coupled with the rights of property owners on the other, played itself out during this period. Over time, through legislation, Constitutional development, court challenge, and more, the outlines of property were drawn; at the same time, the outlines of whiteness as property also were constructed. Of course, understanding a characteristic of a person as a possession is hardly a straightforward enterprise, especially as

106 / Chapter 5

the outlines of property were and continue to be based in many different criteria. Thus, before we can take up the claim that whiteness is property, we must elaborate our understanding of property in the United States.

Constructing Property in the United States The commonsense understanding of property as an object (e.g., a building) does not jibe with more theoretico-legal perspectives, which see property in more complex ways. The theoretical perspectives, according to Harris, “range from first possessor rules, to creation of value, to Lockean labor theory, to personality theory, to utilitarian theory. However disparate, these formulations of property clearly illustrate the extent to which property rights and interests embrace much more than land and personalty” (1993: 1725). Functional criteria also define the class of possible properties expansively; Harris states, “The liberal view of property is that it includes the exclusive rights of possession, use, and disposition. Its attributes are the right to transfer or alienability, the right to use and enjoyment, and the right to exclude others” (1993: 1731; notes omitted). In many of these frameworks and perspectives, some of which appear contradictory to others, it is possible to describe whiteness as property. Consider just two of the many ways of defining property. Analysts have defined property using the concept of first possession. For example, in intellectual property law courts assume the creator of an intellectual product, as first possessor, has a protectable property interest in the product. Such rights may be bargained away (e.g., technological workers may trade their rights to the future financial remuneration that may flow from any inventions they create to a company that provides a wage, working materials, and more). But the first possessor, the actual inventor, must transfer this interest, and many corporations attempt to structure their contracts with employees so that this transfer occurs. Clearly, therefore, the state recognizes an interest that inheres in being the first possessor. Alternatively, property can be defi ned so as to give great weight to the expectations it affords. As Jeremy Bentham maintained, “Property is nothing but the basis of expectation, consist[ing] in an established expectation, in the persuasion of being able to draw such and such advantage from the thing possessed” (quoted in Harris 1993: 1729). Adding that only some expectations rise to the level of property, Harris (1993) contends that the major question, therefore, is what expectations are protected by law and therefore become definable as property? For some expectations, the state answers the question affirmatively.

Critical Legal Perspectives and the Critique of the Dominant Legal View / 107

Discerning Whiteness as Property in the United States Using either of these two perspectives on property, it is possible to demonstrate how one may make the case that whiteness is property. I hasten to add, similar cases can be made with other perspectives on property as well, so matters do not turn on only these two perspectives. But, these two perspectives can serve to illustrate how one may make the case that whiteness is property. We have seen that first possession is, from one point of view, a key feature of property rights. It also has been claimed to be the origin of property in the United States (e.g., Epstein 1979). Yet, during the course of U.S. history, first possession was irrelevant before the courts if the first possessors were Native American. As Joseph William Singer writes: [I]f we focus our attention on the relation between race and property, we notice that American Indians were left out of this traditional story of origins [of property in U.S. law]. Often they are not mentioned at all in legal scholarship about property, in property casebooks, or in cases about property rights. This omission in discussions of the principle of first possession implies that the land which the settlers occupied was vacant. When American Indians are mentioned, their situation and history is often distorted and false. Many non-Indians believe that American Indian nations did not occupy most of the land in the United States; that it really was vacant. Or they assume that these nations voluntarily ceded their lands in mutually advantageous treaties. The truth is that, although the United States generally protected the possessory interests of non-Indians, it has often run roughshod over the possessory interests of American Indians. It is simply not true that most of the land in the United States was “vacant.” Although the land may not have been built up, almost no land fell outside the recognized areas of tribal sovereignty. The settlers did not enter wilderness lands unclaimed by anyone. Rather, they invaded Indian lands. They took the land away from the first possessors, and in most cases, were backed up by the coercive power of the United States military. The history of original acquisition of property in land in the United States therefore is not based on the principle of first possession, but on the very opposite: the seizure of property from first possessors by force. The real first possessors of the land were deemed to be outsiders, savages. Their humanity was questioned, their possessory rights denied, their just claims rendered invisible. (1991: 718–719)

108 / Chapter 5

Taking seriously this history, one cannot maintain that first possession defined property. One can, however, maintain that first white possession did define a property interest. In this sense, therefore, whiteness becomes property, because only possession by someone who is white will allow the entity possessed to be regarded as property such that full ownership rights will be recognized and supported. Singer (1991) notes that this is not merely a historical oddity one might expect as a fledgling nation flexed its muscle. Singer observes that: In 1955, for example, one year after the decision in Brown v. Board of Education, the Supreme Court ruled that Congress was constitutionally free to seize the property of American Indian nations without compensation if the United States had not previously recognized tribal rights to the land by treaty or statute (Tee-Hit-Ton Indians v. United States (348 U.S. 272 (1955))). As recently as 1980, the Supreme Court held that Congress has no constitutional obligation to pay just compensation for property seized from American Indian nations if it is, in good faith, as it sees fit, in the best interests of the tribe, and exchanges that property for property of “equivalent value.” In the non-Indian context, the Supreme Court is not satisfied with equivalent value; it requires fair market value for the property (see, e.g., United States v. 50 Acres of Land (469 U.S. 24 (1984))). Nor would the Supreme Court ask whether or not the government acted in “good faith.” The government’s motivations are simply irrelevant in the non-Indian context; compensation is required if a property right has been invaded, period. (1991: 719–720) These are but a few of the many examples one might present to suggest that full ownership is denied at least some nonwhite groups, making whiteness a potentially constitutive element for property-owning in the United States. In this sense, therefore, whiteness can be conceived as property. With respect to expectations, a different logic with respect to property, Harris offers courts’ behavior as evidence that whiteness became protected as an expectation-based property interest in the United States. Harris (1993) points to the long line of cases disputing the categorization of a person as nonwhite. In Johnson v. Board of Education of Wilson County (82 S.E. 832, 833–835 (1914)), the court ruled that children of an allegedly pure white man and a woman less than one-eighth black could not attend a white school. Other cases reveal how whiteness is property, discriminatorily maintained, by showing the lengths to which individuals will go to protect themselves from being called nonwhite. In Bowen v. Independent Publishing Co. (96 S.E.

Critical Legal Perspectives and the Critique of the Dominant Legal View / 109

2d 564, 565 (S.C. 1957)), a case decided long before affirmative action policy as a redress made African American ancestry a possible positive in some circumstances, the court ruled that to allege a white person actually was black was to cause injury. The injury exists in part because of the expectations of those who are regarded as white, expectations of being able to conduct their affairs with the full weight of the United States government behind them, and with the fullest possible autonomy consistent with a society of free persons. In short, the historic construction and contemporary maintenance of whiteness as property is apparent in legal analyses. Harris summarizes the point: “When the law recognizes, either explicitly or implicitly, the settled expectations of whites built on the privileges and benefits produced by white supremacy, it acknowledges and reinforces a property interest in whiteness that reproduces Black subordination” (1993: 1731). Alan David Freeman (1978) argues that antidiscrimination law as enacted actually legitimizes discrimination by individualizing discrimination. One way in which this occurs is that antidiscrimination law, and the law as an institution, ignores the history of whole communities’ support for, acquiescence to, and gain from discrimination, and thus ignores the existence of historic and perhaps contemporary beneficiary bystanders. Harris’s (1993) analyses of Plessy v. Ferguson (163 U.S. 537 (1896)), Brown v. Board of Education I (347 U.S. 483 (1954)) and II (349 U.S. 294 (1955)), and Milliken v. Bradley (418 U.S. 717 (1974)) also make a case that white privilege is maintained in legal reasoning even when par ticu lar practices are overturned. One mechanism for this maintained privilege—maintained inequality—is that the settled expectations of whites qua whites are not violated, even when those expectations flow from past transgressions easily documented in the historical record. In the contemporary period, a key strategy that tends to accept the settled expectations of whites as inviolable is to narrow the scope of discussion and remedy, ignoring the basis of white privilege in historically constructed racial subordination. Consistent with recognizing this basis of disadvantage, and impatient with the pace of redress, Freeman argues that the individuals who were instrumental in establishing any par ticu lar part of the social conditions are not key; what is important is to implement the fi x for discrimination, that is, change the social conditions. The law, therefore, should not attempt to partition discrimination into pieces and to identify at least one perpetrator for each piece. Instead, the social conditions—joblessness, poverty, homelessness, addiction, and, ultimately, hopelessness—that lie in the wake of discrimination must be addressed as a whole. A wholesale effort to address these conditions necessarily would broaden the population legitimately expected to compensate wronged parties. Freeman sees no other alternative, for as

110 / Chapter 5

long as the law requires one to find narrow instances of disparate treatment or disparate impact discrimination before requiring a remedy, the victims of discrimination will continue to pay costs of discrimination.

Gender and Legal Discrimination: Maleness as Property? Some scholars contend that much of the contemporary discussion of discrimination and its discontents is driven by understandings, accurate or not, of how race matters or does not matter in society, and whether other dimensions of discrimination match the example of blacks. John D. Skrentny (2002) argues, for example, that federal government inclusion of some groups as “minorities” depended on whether federal elites as well as bureaucrats regarded the groups’ situation as analogous to the situation of blacks. Skrentny (2002) finds that when Congress acted to provide a statutory basis for ameliorative action in the late 1960s and early 1970s, it referred to the concept of minorities. When regulators attempted to operationalize this concept, minorities became defined on the basis of paper forms that predated the legislation. Officeholders pressed previously used forms and documents into ser vice in an effort to avoid opening up Pandora’s box, making only slight modifications where avoidance seemed impossible. Yet, key actors read legislation such as the Civil Rights Act as concerning blacks primarily, and regarded the inclusion of women in par ticular as illegitimate, especially given the legislative history of their inclusion (see Skrentny 2002: 96–100; Burstein [1985] 1998: 22–23). Other groups were added relatively easily to the minority category (e.g., Chicanos), regardless of whether group leaders requested inclusion. After inclusion, however, representatives of some groups petitioned the Equal Employment Opportunity Commission (EEOC) to treat them on a par with blacks (Skrentny 2002). Some groups, such as Latino/as, were able to more easily make the analogy to blacks than were others, most particularly women. As women continued to face difficulties making this analogy in the early 1970s, political mobilization of elite women began to pave the way for women to receive affirmative action benefits. Skrentny (2002) argues that, though struggle was necessary, the trail blazed by blacks greatly eased the success of these later efforts. Hence, it took women no more than ten years to obtain affirmative action status, while it had taken blacks decades. Still other groups, such as Poles and Italians, never were able to make the minority group transition. These kinds of observations suggest it is unwise to draw sharp distinctions between race and sex discrimination. In the public mind there is a constantly occurring, indeterminate cross-pollination of perspectives and defi nitions. This state of affairs implies that while differences do exist

Critical Legal Perspectives and the Critique of the Dominant Legal View / 111

depending upon the dimension of and basis for discrimination, the broad outlines likely are similar; the legal system works similarly with respect to race and sex discrimination, and the critiques of dominant legal reasoning, while substantively different, are structurally similar. More evidence of this state of affairs is provided by feminist analyses of the law. Mary Joe Frug (1985) analyzes an important contracts casebook (i.e., a book of excerpted cases, editorial comment, and selected questions used to teach the contracts course in law school). Frug reveals how the selection of cases, the presence and absence of women, and the diction of the editors subtly and not-so-subtly reinforces assumptions of women’s weakness. Frug provides thought-provoking insights. Yet, reconsidering some of Frug’s examples reveals that one can make a case quite different from her focus, and quite like the case for whiteness as property, with males being the privileged group and maleness as property being the result. One example Frug provides is of a case, decided before 1942 in Kentucky, wherein which a female party is not the real party. In Crenshaw v. Williams (191 Ky. 559, 231 S.W. 45 (1921)) a woman, Mrs. Williams, seeks to sell land inherited from her father. However, in the case her husband serves as the party of record because, according to Kentucky law prior to 1942, married women were unable to transfer land without their husbands’ consent. Noting that “Nothing in the casebook explains the problem of incapacity that state law imposed on Mrs. Williams, and other women, at the time of the lawsuit” (1985: 1088), Frug contends that the omitted information makes the discussion of the case convey a limited view of women’s abilities while also indicating to any who know the past that the record of the past, as it pertains to and involves women, is not relevant or important. Yet, the evidence Frug places before us via her discussion of Crenshaw v. Williams, read in the context of Harris (1993), easily leads to the observation that men had at the time of the litigation a property interest in maleness. The argument one would construct here is similar to the argument Harris constructs for whiteness as property. With respect to maleness as property, married men had an expectation of being in control of property nominally owned by their wives. In the example of married women’s property, though women may inherit from their fathers, their husbands—by virtue of being their husbands—will have obtained rights to the same property. Women cannot transfer it without their husbands’ consent, which of course means that it is not truly and completely theirs—they are not owners of property in the dispositive sense. In contrast, men were able to transfer land and engage in other commerce without their wives’ consent. Hence, the right to stand before the law, in the marketplace, and in society on one’s own and in pursuit of one’s own desires was reserved for men. Given the centrality of autonomy to the realization of one’s hopes, even if those hopes entail voluntarily surrendering

112 / Chapter 5

some of that autonomy (via marriage, perhaps), it is indicative of the law’s respect for men, and disrespect for women, that it protected and reinforced the autonomy of one while hampering and delegitimating the autonomy of the other. To the extent this state of affairs handed men advantages, and denied those advantages to women, maleness is a property interest, for with it the doors to an expectation of freedom are opened, in direct analogy to the observations made in the whiteness as property discussion. Lest one believe this property interest has eroded to the point of irrelevance, consider the asymmetric treatment Frug (1985) points out in footnote 90. Frug speculates on whether one may regard some aspects of reproductive law as discriminatory, and writes: One could argue that there is a discriminatory tilt in this area in favor of men. Thus, male spouses with sterility problems are protected against unwanted parental responsibility by statutes requiring their written consent to the artificial insemination procedures [their wives undergo], even though these requirements may make parenthood somewhat more difficult for women whose husbands are sterile. Similarly, male donors who wish to sell their sperm and avoid any further responsibility for children who are conceived are permitted to use contracts for those purposes, while women have not been permitted to contract for the use of their wombs and the sale of their ova. However, these materials can also be understood as discriminating against sterile married men or men who are married to infertile women, for the decisions make siring and fathering children of their own difficult for these men. This reversed understanding of the discriminatory tilt of the decisions is one of the reasons they would be interesting to add to a contracts casebook; the results in these cases cannot be explained by our traditional ideas of what women and men are like. (1985: 1093; notes omitted) Frug points to an intriguing asymmetry in the law—men can sell sperm for reproductive purposes but, at least in many jurisdictions, women cannot sell ova or rent their wombs for reproductive ser vices. Contracts to sell sperm for reproduction are enforceable, but contracts to sell ova or rent wombs for reproduction are not. Frug helpfully treats this asymmetry in terms that might be familiar to anyone who has been exposed to debates concerning affirmative action, for she poses the predicament as an issue of the rights of different parties to obtain the parenting experience. Her discussion implies a fundamental zero-sum aspect in reproductive matters given late-twentiethcentury technology. Though Frug articulates an interesting perspective, it is one that is dissonant with a “property as expectation” perspective.

Critical Legal Perspectives and the Critique of the Dominant Legal View / 113

From the property as expectation perspective, men are accorded property rights vis-à-vis their personal role in reproduction while women generally are not accorded property (as expectation) rights in their personal role in reproduction. Men may assign those rights to others (e.g., by selling their sperm for reproduction); women may not assign those rights to others, for any contract to do so is unenforceable; thus, women’s rights are denied existence— what they have to offer becomes nonproperty. Further, men may enjoin women (e.g., their wives) from pursuing a contract with other men (e.g., sperm donors) around reproduction. This gives the husband (male) power to decide whether the wife (female) will enter into a contract with the sperm donor (another male). In contrast, the woman has no power to decide whether she would like to enter into a contract for her reproductive ser vices, and no power to prevent her husband (male) from entering a contract to provide reproductive ser vices. Again, the autonomy of the male is affirmed in the context of the constraints posed by reproductive technology, while the autonomy of the woman is undermined. Human reproduction at present takes resources from two parties, one male and one female; thus, the asymmetry is striking. One is left with the distinct impression that male interests in reproduction are protected directly to the extent possible given existing reproductive technology, and women’s interests are not protected directly. Instead, women’s interests may be protected ancillarily when they are consistent with men’s fullest possible property rights in the reproductive context. In other words, women and men interact around reproductive technology in a manner that resembles the interest convergence dilemma Derrick A. Bell Jr. (1980) identifies in the case of race in the United States. Again, maleness appears as property, conferring rights of transferability and use, and thus affirming male expectations in ways that resemble whiteness as property, intellectual property, and many other properties before the law. Of course, as I write the legal terrain of reproductive rights is undergoing upheaval. One developing area concerns children whose biological fathers were anonymous sperm donors. In some cases, children successfully are breaking their biological fathers’ anonymity contracts by their own investigations (Stein 2005). In other cases, fathers seek ways to break the contracts that secured their anonymity (Leahy 2005). In still other situations, donors seek to affirm their lack of financial responsibility for their biological children (e.g., Ferguson v. McKiernan 855 A.2d 121 (Pa. Super. 2004)). These and other upheavals are occurring because human reproduction occupies a complex political, social, symbolic, and technological terrain, involving adults in various categories, including biological parents, social parents, and the familial relations of each; the medical establishment; the state; the fetus; the infant; the child as a child; the child as an adult sometime in the future; and

114 / Chapter 5

perhaps others as well. As each involved entity also has separate interests and concerns, one should not expect reproductive issues to be wholly determined by gender relations in the society. Still, the relative power of men and women as such in society is a factor in the development of recognized reproductive rights. Further, this is not to say that there are no circumstances wherein which women have an advantage before the law. But it is to say that male domination is maintained through their par ticu lar and historic and contemporary advantages. Now that men have such advantages—accrued in a more domineering era, treated as the normal state of affairs, backed by the state, and contested at great cost and risk—maleness is property.

Preliminary Summary Remarks on Critical Legal Perspectives and Discrimination Notably, each of these claims—whiteness as property, maleness as property— is not about arcane or isolated cases that come before the Supreme Court. That is, though they may be manifest at the highest court in the land, they actually concern the day-to-day taken-for-granted standard operating procedures of institutions throughout society. Rarely may those operations actually draw the attention of the courts, in part because of the settled nature of those standard operating procedures and the expectations upon which they rely. Yet, the expectations these observations identify structure much of the understanding of what is sensible and what is not, both before the law and within society. These expectations are maintained through processes that sediment race and gender disadvantage (e.g., Oliver and Shapiro 1995). Consequently, it is possible to find such property rights operating within many of the practices and mechanisms that maintain and reproduce society. Given the critical legal perspective, detailed knowledge of day-to-day practice is essential to any effort to understand the structures that maintain the status quo. Reconsidering human capital processes in some depth, therefore, while threatening to take us far afield, will actually illuminate the contrast between antidiscrimination law and the operation of the courts on the one hand and the claims of critical legal scholars on the other. Of course, critical legal perspectives do not deny that disparate treatment and disparate impact discrimination occur. Yet, critical legal perspectives go beyond this framing to identify discrimination as also entailing a wide range of standard operating procedures and arrangements that ultimately are based in a limiting view of persons. Thus, in order to bring critical legal perspectives to bear on human capital processes, it is necessary to elaborate our understanding of human capital processes themselves, to excavate some of the

Critical Legal Perspectives and the Critique of the Dominant Legal View / 115

standard operating procedures that underlie the accumulation and redemption of human capital. Necessarily the analysis must be specific, ignoring a great deal of additional important detail to microscopically analyze key moments in a par ticular arena. Any such analysis cannot be unproblematically generalized to suggest the overall impact of such processes. However, even a single illustration strengthens the case that such processes operate regularly as standard procedure, and thus regularly activate and reinforce race and sex discrimination.

Elaborating Human Capital Processes: The Example of Standardized Testing for College and Postcollege Admissions In order to see how discrimination may be inherent in standard operating procedures, look closely at how standardized tests are constructed, and consider the use of standardized tests in college and postcollege admissions. Nothing could be more defensible than a test of academic achievement to help determine whether a student should be admitted to a college or university program. Yet, as I show, procedures underlying this practice make it a discriminatory process at its core. Commonly, analysts have evaluated the extent to which existing tests are biased on the basis of race or sex, and have developed insightful ways of discussing the possibility of bias (e.g., Jencks 1998). Yet, existing evidence suggests that the case for statistical bias in testing is weak in many cases (e.g., Schmidt and Hunter 1999). Note, however, the issue I raise concerns not statistical bias but, instead, whether the approaches and assumptions inherent in some test construction strategies are by definition discriminatory. This may be difficult for some analysts to fathom. It seems as though whenever a nonpsychologist enters into discussion of the value and limits of existing standardized tests, psychometricians and their supporters respond by mishearing the positions offered and asserting the validity of the tests (e.g., contrast Fischer et al. 1996 and the review by Gottfredson [1997a]). But, psychologists and psychometricians no more own the measurement of cognitive skill than political scientists own the measurement of democracy or sociologists own the measurement of socioeconomic status. Basic principles of measurement apply to every arena in which social measurement occurs, meaning that those not specializing in the subject matter at issue may still turn an evaluative eye to the measurement operations. Indeed, it is useful for those who are not subject-matter specialists, especially those who also are experienced in measurement, to evaluate the measurement operations behind our assessment of important social phenomena, to assure that problematic

116 / Chapter 5

assumptions are not embedded in the measurement process. And, as we shall see, when we peer closely into the process of test construction, there is a great deal about which one may raise critical concerns. So, to be clear, I want to emphasize that my aim is to look inside the black box of test construction and assess whether some procedures within that box are discriminatory. If so, the resulting product is discriminatory, regardless of how much statistical or any other analysis can be marshaled to seemingly defend the product. Hence, the aim here is not to assess the possibility of statistical bias; the aim here is, instead, to consider the possibility that standard operating procedures of test construction entail the differential treatment of persons.

Key Procedures in Standardized Test Construction In order to construct a test such as the SAT and the Graduate Record Exam (GRE) General Test item writers draft a set of candidate questions (CQs) and administer them to a test-taking population. Logistically, administering the CQs to a test-taking population is quite easy to do. For example, for the SAT the test administrators inform all test-takers that one of the sections they receive will contain CQs. But test-takers are not told which section will contain the CQs, though they are told that CQs are not used in the calculation of their scores. After the test administration, analysts evaluate how the CQs performed. Key aspects of this evaluation include the differential item functioning (DIF) analysis and the point biserial analysis. With these analyses, researchers investigate which types of students answered the CQs correctly. Test-makers are not shy about stating the reason for this analysis—test-makers seek to drop any CQ for which low-scorers on the full test matched or out-performed the high-scorers on the full test, while also minimizing the statistical biases that testing critics decry. These two aims are different. The second aim entails minimizing statistical biases by race, sex, region, and so forth. A CQ is deemed statistically biased if two groups of test-takers who performed the same on the overall test show different levels of performance on the CQ. So, if boys and girls at the 90th percentile are equally likely to answer a CQ correctly, then the question does not display gender bias for students at the 90th percentile. If this result is replicated for comparisons at all other levels of measured achievement, the item does not appear biased on the basis of gender. Test designers also use DIF and biserial analyses to investigate whether test-takers who obtained low scores on the existing test were more likely to answer a CQ correctly than were test-takers who obtained high scores on the existing test, or if the two groups were equally likely to answer the candidate question correctly. If they find either unexpected pattern, then the candidate

Critical Legal Perspectives and the Critique of the Dominant Legal View / 117

question is revised or rejected because it is said to fail to differentiate effectively between high and low scorers. But this conclusion is faulty, for the question so eliminated might very well differentiate between high and low scorers. However, its differentiation points in an unexpected (and, to some, unacceptable) direction, suggesting the low-scorers are better than formerly believed, and the high-scorers are worse than formerly believed. The response to this result is to reject the CQ for actual test use, and it is this response that is discriminatory. Let us be straight about what has occurred. Recall that we lack any independent measure of student achievement, uncontaminated by teacher attitudes (e.g., see Chapter 2) or psychometric assumptions. Thus, we have no way to make fine distinctions between students’ levels of knowledge. The test designers resolve this dilemma by using the current test they are attempting to revise as a criterion, as if it is a solid indicator of students’ performance. But note, the current test, their indicator of student achievement, was built with the same process of test construction we are evaluating here; thus, if the process of test construction we are evaluating here is suspect, then so is the criterion. The criterion is suspect because if a student is low scoring on the existing test, it means in part that the student did not have an opportunity to answer former CQs the student might have answered better in comparison to highscoring students because those CQs were rejected last year. Hence, this year’s low-scorer is produced, in part, by prior decisions of the test-makers about which questions to delete from the test. This means that some low-scoring students have their scores lowered to an as-yet-unknown degree by prior choices about which particular questions should be included on the test. The enterprise is self confirming, because the process of deciding which par ticu lar questions should be on the test works to remove questions lowscorers answer correctly when high-scorers answer the questions incorrectly. Thus, low-scorers, who are defined as such on the basis of the prior reasoning of the test-makers, reasoning that is embedded in the current test, may answer a CQ correctly that low-scorers, compared to high-scorers (defined as such on the basis of the prior reasoning of the test-makers), should not be able to answer. Thus, year after year, the test becomes maintained in a way that stacks the deck against “low-scorers,” by deleting questions on which they out-perform high scorers (defi ned as such on the basis of the prior reasoning of the test-makers). The absurdity of this procedure sometimes slips into public view. William C. Kidder and Jay Rosner relate one such incident: [An applicant], Chris Laucks, took the LSAT [Law School Admission Test] in 1981 when it included math problems similar to those appearing on the SAT or GRE today. On a par ticular geometry problem,

118 / Chapter 5

ETS [Educational Testing Ser vice] mistakenly omitted a right angle marker. With the marker, one answer would have been mathematically correct, but in the absence of the marker, a different answer was mathematically correct. Unfortunately, Laucks picked the answer he knew to be mathematically correct instead of the answer he suspected ETS wanted. After Mr. Laucks received his LSAT score, with this geometry question marked incorrect, he wrote a complaint to the Law School Admission Council (LSAC) and attached a flawless mathematical proof of his answer. Oddly enough, LSAC confirmed that Laucks was correct, but it would not give him credit for his correct answer. Strict adherence to correlation requirements accounts for LSAC’s peculiar stance. In its view this item was defective, because in pre-testing “high ability” students picked the other, incorrect answer. Thus, to give Laucks points for this question (and to penalize those picking the other answer) would violate a sacrosanct principle of test reliability. Accordingly, ETS wrote a letter to test-takers explaining why the item was pulled from scoring (rather than crediting Laucks and others who picked (D) as the correct answer), in which it stated: As was noted above, the credited response to the question was (C). Statistical results from a trial administration of the question indicated that the question, with (C) as the answer, was functioning as intended. If the question had been keyed (D) in the trial administration, the statistics would have shown that it did not function properly, and it would not have been used in the LSAT. (2002: 159–160) Such examples certainly undermine the idea that the LSAT, SAT, GRE, GMAT, and other such tests are neutral indicators of competence; indeed, the example suggests that identification of competence is neither the sole nor arguably even the central aim of test construction. Were competence identification the central aim, then awarding credit for the truly correct answer when a typographical error intrudes to produce an alternative (and available) correct answer would seem appropriate. The example, therefore, suggests that standardized tests are less the result of processes designed to measure competence, and more the result of processes that can serve to legitimate presumptions of competence and status quo hierarchies of power. Further, as the test construction criteria are implicated in every effort to evaluate a test, in every effort to rank persons’ competence, the test design process works masterfully to prevent any evidence of its incomplete connection to competence from emerging. Note also that this procedure also undermines the assessment of statistical bias by race, sex, region, and so forth. To the extent the criterion of

Critical Legal Perspectives and the Critique of the Dominant Legal View / 119

achievement is the previous test, a test designed with the same problematic procedures, the criterion is a flawed indicator of real achievement. Hence, comparing the performance of girls and boys or different racial groups on a CQ after equating girls and boys or different racial groups on achievement on the prior test is to conduct a suspect comparison. Consider two sets of students who actually have the same average level of achievement. Yet, owing to prior biases in test construction, one set of students, group Y, averages five points higher on the one-hundred-point test than the other, group X. For simplicity, assume the test is normally distributed. With such a test, if we take all the group-X and -Y students who score eighty or higher on the test, and call those the adept students, because of biases in the test that mask the performance of group-X students, the group-X adept students actually are better than the group-Y adept students. Thus, when we find a question that “adept” students in both groups are equally likely to answer correctly, what we probably have found is a question that actually is biased against students in group X. The reason is that if the less-talented group-Y students who are labeled adept are as likely to answer the question as their more talented group-X peers, then the question is likely biased against group-X. But, because the test construction procedure uses the biased current test as a criterion, we would never be able to learn that group Y is being harmed by questions that appear unbiased. Lest this seem a fanciful example, consider that the SAT routinely underpredicts the actual performance of girls in college (e.g., Leonard and Jiang 1999), just what one would expect to observe for groups equal in achievement were the criterion biased owing to the question selection process. What is the effect of this procedure, especially on patterns of achievement and the stability of those patterns? To answer this question, consider the simple simulation presented in Figures 5.1 and 5.2. The simulation indicates that approximately halving the use of the discriminatory procedure of question selection will lead the mean scores for the original high and low scorers to converge. The results also suggest that reducing use of the discriminatory procedure likely would lead to score convergence by race, suggesting there likely is disparate impact discrimination with respect to identifiable classes of persons as well. The simulation is somewhat crude, and serves only to identify the logical long-run implications of test-construction procedures. The simulation assumes a fifty-item test, scored from zero to fifty, and the need to replace two questions per year. Further, it is assumed that CQs posed for the test fall into two categories: (1) those that top students are 60 percent likely to answer correctly, while those at the bottom are 40 percent likely to answer correctly, and (2) those that top students are 40 percent likely to answer correctly, while those at the bottom are 60 percent likely to answer correctly. Middle-scorers

50 Mean Test Score 20 30 40 10 0

5

10

15 Year

20

Original Low Scorers Original High Scorers

30

Overall Mean

Mean SAT score overall and by prior achievement.

0

10

Mean Test Score 20 30

40

50

Fig 5.1

25

0

5

10

15 Year Black

Fig 5.2

Mean SAT score by race.

20 White

25

30

Critical Legal Perspectives and the Critique of the Dominant Legal View / 121

on the existing test are 50 percent likely to answer any CQ correctly. Point biserial correlations range from −1 to 1, with positive values indicating the question more likely is answered correctly by high-scoring students than by low-scoring students. Negative values indicate the opposite pattern. The point biserial correlation for CQ1 in this simulation is .145, while the point biserial correlation for CQ2 is −.12. The result for CQ1 is within the range of point biserial correlations reported in published ETS reports (e.g., Breland, Kukich, and Hemat 2001), while the CQ2 point biserial is slightly more negative than the most negative point biserial reported by Hunter Breland, Karen Kukich, and Lisa Hemat. Note, however, that the most negative point biserial they report, -.08, may follow from using items for their research that already have been winnowed by test designers using the same assumptions and expectations they bring to the point biserial and DIF analyses. If so, then one should expect point biserial results to tend to be only slightly negative in general. For this simulation, it is assumed that in year zero test-makers decide to, in future years, replace two questions per year, and, instead of pulling new questions only from category one, they now will add only one new question each year from category one, and one new question each year from category two. Note that the addition of a category-one CQ tends to preserve the prior distribution, while the addition of a category-two CQ would tend to alter the prior distribution. Other assumptions, such as the assumption that over twenty-five years the true distribution of the measured characteristic (e.g., knowledge) does not change, are made to isolate the “item pool” effect. Figure 5.1 reveals what would happen to the mean scores, the mean for those who were originally top scorers, and the mean for those who were originally low scorers. It is apparent that if one adds one question per year that would have been rejected on the basis of the point biserial criterion, mean scores will converge until, eventually, the categories reverse. Figure 5.2 reveals what would happen to black-white test score differences. Convergence is apparent there as well, though not as pronounced. These figures suggest that one reason for the stable patterns analysts have observed is the deletion of items whose “defect” was that low-scorers outperformed their higherscoring peers. With respect to race, at the outset the mean black-white difference was approximately 12.3 points on the fifty-point test. The test had a standard deviation of 12.98; hence, the gap at the beginning of the simulation was 95 percent as large as the standard deviation. By year twenty-five, the black-white difference is only approximately 4.69 points on the fi fty-point test, which translates into only 36.1 percent of the original standard deviation. Further, the actual year twenty-five standard deviation is 15.93, suggesting that variance does not need to decline if the question-inclusion algorithm changes. In one understanding of testing, declining variance presents a problem for

122 / Chapter 5

test-score use by making it difficult to discern differences between students. The simulation suggests declining variance is not a necessary outcome of accepting CQs with negative point biserial correlations. The black-white gap in year twenty-five is expected to be less than three tenths of the year twenty-five standard deviation. By any benchmark, then, it appears that one would expect test-score convergence if the discriminatory means of test construction were reduced, and if some other discriminatory mechanism were not substituted in its place. Does this mean that the only factor preventing black-white test score convergence is test construction? Of course not. The simulation is for illustration purposes, and thus it attempts to isolate this par ticu lar factor to observe its implications for test scores. To do so required a ceteris paribus assumption. Yet, the illustration should reveal that when analysts select test questions based on a desire to preserve prior test-score distributions, we should not be surprised that ending this practice will change test-score distributions. And, as current psychometrician use of the point biserial and DIF procedures may eliminate good questions whose inclusion would be justifiable on the basis of test content, we should expect the test to be biased even as a criterion for avoiding direct race and gender bias (see Appendix B for more discussion of the simulation). Of course, the simulation only confirms what should be obvious—if a candidate question is selected for future use because the people who answered other questions correctly also answered the candidate question correctly, then muting that practice eventually will erode the differences between high and low performers to the extent that difference had been maintained by this practice. Further, the simulation is not the only evidence suggesting that test construction entails a mix of assumption and technically involved manipulation that end up preserving the patterns that have been observed in the past. Kidder and Rosner (2002) bring a legal eye to these practices, nicely identifying several moments in the test-construction process—from item writing, to pretesting, to analyzing items, to selecting items, to administering the final product—all conducted by individuals who, despite their commitment to race and gender fairness, employ logics that work in ways consistent with the par ticular moment described in some detail here. Which kinds of students will be more likely than their ostensibly more talented peers to answer a set of questions correctly? Prime candidates for correctly answering questions other allegedly better students answer incorrectly and answering incorrectly other questions those allegedly better students answer correctly are persons whose cultural frames make the pattern of resonance differ from the “norm” in the society in which they live, persons who learned the material in an order the test-makers do not expect, or persons who learned the material in an order different than is common. These

Critical Legal Perspectives and the Critique of the Dominant Legal View / 123

possibilities exist because cultural resonance can make a difference in how a student connects one concept to another, and because the order in which concepts and operations were learned can alter the degree to which a question is easy or hard. As African Americans experience subtle cultural differences from mainstream society that might produce different cultural resonances, and are taught in locations largely segregated from other students, it is easy to recognize that prospects for a different pattern of learning may be real. Further, as African Americans make up a comparatively small proportion of test-takers, a pattern of coverage that might be common for them easily could be uncommon in the wider society. Thus, African Americans are one set of sociodemographically identifiable “prime candidates” who may display an “unexpected” pattern of success on CQs compared to their ostensible level of performance. Similar observations may be made for girls. Girls and boys play separately and differently. Barrie Thorne (1993) nicely analyzes the patterns of play, the various and complex sources of gender-segregated activities, their heightening at school and lessening in the neighborhood, and the extent to which a dualistic understanding of separate cultures illuminates matters or obscures continuity and complexity. Because some gender segregation is at best a constrained choice of the child, and children are by definition impressionable, it is impossible to discern whether any of the gender segregation is truly by choice. By choice or not, segregation opens the door to the possibility of different patterns of correct and incorrect answers. Surely, women are as common as men in general, and thus the implications for test design may usually be small on average. But, small average differences can produce large differences in the upper tail of the distribution, greatly diminishing women’s opportunities for scholarships. Further, for some fields in which standardized tests may be important to the allocation of opportunity, such as physics, chemistry, and engineering, female test-takers may be relatively rare, and thus more likely to be penalized by the assumptions activated by psychometric processes of test construction. This psychometric procedure is problematic for many policy questions, but with specific reference to the issue of discrimination the problems should be clear. Any procedure that rejects a question that students on the bottom of the prior test-score distribution are more likely to answer correctly than those on the top simply because those on the bottom of the prior test-score distribution were more likely to answer the question correctly than those on the top is, by definition, discriminatory. The procedure is discriminatory because it trumpets or disregards achievement simply by virtue of who accomplished the achievement. It would be as if a basketball league had decided to award two points for every shot made from a spot on the floor from which Michael Jordan shoots successfully, and to award zero points for every shot

124 / Chapter 5

made from a spot on the floor from which Michael Jordan shoots unsuccessfully. Michael Jordan was surely a talented athlete, but this approach still would have been patently unfair. Both our obviously unfair basketball rules and the test-construction process that operates on the same principle entail disparate treatment discrimination.

Test Construction: Justifications, Motivations, Alternatives One might wonder why any institution would continue to use such procedures when the discriminatory nature of the procedure is abundantly clear. One reason is that standardized testing differs greatly from other forms of testing. The contrast with testing in the classroom is striking and informative. When I devise a test for students in my class, I use procedures the vast majority of teachers at every level follow. I consider the material of the subject that I have covered in the class, weigh the relative importance of topics and facts and skills, and devise a test whose aim is to allow students to demonstrate proficiency in what I have regarded as the most important aspects of the course. This process—prioritizing the goals of the course, and devising tests that emphasize those goals—is a textual, qualitative, linguistic, judgmental one. It would be possible to adopt a similar approach for college admissions testing. The standardized-test construction methods I have described are used in constructing norm-referenced tests. The vast majority of high stakes tests, such as the SAT, the GRE, and tests for special-education placement, are norm-referenced tests. Yet, there are alternatives. The tests used as part of the National Assessment of Educational Progress (NAEP) trend assessment are criterion-referenced tests. The norm-referenced/ criterion-referenced distinction is not solid, but these two methods do tend to emphasize different procedures. To construct criterion-referenced tests, item writers define the content domain. They then construct candidate items, draw on the judgments of experts, and in this manner determine which concepts and questions are likely to be more or less difficult. Judgments of difficulty are made in order to increase the chance that the test will sample from the full range of the content or skill area. In addition, analysts often establish benchmarks that link different levels of test performance to explicit, real-world competencies. Though analysts may use student performance on pretests in the construction of these benchmarks, ideally the benchmarks serve as anchors that do not shift just because student performance changes over time. An example of such benchmarking is provided in the NAEP trend assessment for mathematics, which identifies

Critical Legal Perspectives and the Critique of the Dominant Legal View / 125

five levels of mathematics competence: (1) Simple Arithmetic Facts, (2) Beginning Skills and Understandings, (3) Numerical Operations and Beginning Problem Solving, (4) Moderately Complex Procedures and Reasoning, and (5) Multistep Problem Solving and Algebra (Educational Testing Ser vice 1997). Clearly, to some, a potential disadvantage of criterion-referenced tests is that they require test-makers to state explicitly just what is being tested. Given that there is no clear consensus on what constitutes proper preparation for college or graduate school entry, it might be difficult to quickly construct a criterion-referenced test to measure preparation for college or graduate school. However, over time one might be able to develop a consensus. And, such a consensus might greatly facilitate students’ successful preparation for college-level work by communicating to them in an explicit manner what is required. Using such methods, either nationally or within subsets of institutions, would mean that when it came time to identify students capable of the work of an institution, it would not be necessary to ascertain which questions worked for which types of students, and eliminate questions from the test on that basis. Instead, the goals of the institution, coupled with its desired level of prior skills, would be sufficient justification for the questions on the test. In short, the institution would be protected from both a disparate treatment claim and a disparate impact claim. Protection from the disparate treatment claim follows from the rejection of DIF and point biserial analyses for selecting questions. Protection from the disparate impact claim follows from the basis of the questions in the stated aims of the institution, which would make any appeal to disparate impact fail if the item(s) can be shown to be related to appropriate institutional goals. In other words, if the test is designed only to indicate readiness for college-level work, and no other distributional/statistical criteria are used to select items, then the test likely would satisfy the logic of a BFOQ. But this is not the procedure followed in the usual construction of standardized tests. There has been no national conversation as to what is required of college matriculants, nor what is expected of college graduates. This means that test-makers cannot turn to content aims or institutional goals in order to differentiate good questions from bad ones, except in indisputable cases (e.g., it is clearly not appropriate to ask questions that require the student to have accurate information about string theory and proofs or disproofs of the theory of the holographic universe [e.g., Beckenstein 2003] on the mathematics section of the SAT). This state of affairs creates a potential legitimation crisis, for test-makers, test-takers, and postsecondary institutions alike, and it is this legitimation crisis that the discriminatory procedures for test construction resolve.

126 / Chapter 5

Institutions’ legitimacy is under threat if the means of admission to the institution come to be seen as arbitrary rather than rooted in some principles of value. In the United States, a claim to merit is a defensible value; standardized tests, accepted as a legitimate measure of preparation in part because of the stability of test-score distributions, support this meritocratic claim. Successful test-takers can gain the feeling of having a legitimate claim on the resources to which they have received access. Scoring well on a standardized test that is purported to fairly indicate the rankings of individuals can allow those who did well to feel better about their social locations. Equally important, those who do poorly on standardized tests have their claim to merit undermined; institutions respond to them differently, and they may come to see themselves as less-than-capable persons. This aspect of the process individualizes both success and failure, and thus legitimates the station of both. And, of course, test-makers cannot headline the process or trumpet the likelihood that the items on tests are the result of a winnowing process that discriminates against people who think outside the box, or in differently ordered ways, or simply did not answer other questions correctly. Were this prospect to become widely recognized, the entire enterprise of test construction could be under serious threat. The discriminatory test-construction procedures are essential to solving this legitimation challenge. To solve this challenge, questions that produce results at variance with expectations are revised or deleted, and doing so reinforces both results and expectations over time. When an entire test administration produces unexpected results, and resulting score distributions differ appreciably from prior distributions, then suspicions of cheating and adjustments in scores are the likely response of test-makers. If one doubts this implication, consider the experience of the students of Jaime Escalante. These barrio students worked hard to attain high scores on an Advanced Placement calculus exam. The response of the testing company was to suspect them of cheating and demand they take the test again under the watchful eyes of test company personnel; when they did retake the test, they scored as well as before (Jesness 2002). Resolution of the legitimation challenge is essential, and this resolution works because the procedure makes the test results look similar from year to year, reinforcing expectations in the process. By eliminating for consideration any CQs that do not reproduce the existing distribution of scores and scorers, the somewhat (though not completely) arbitrary nature of that distribution of scorers is masked. Test-makers often point to predictive validity to support their products, but the predictive validity claim is both weak, concerning, for example, only first year grades, and seriously contested by rigorous analyses (e.g., Lempert, Chambers, and Adams 2000). Alternatively, test-makers

Critical Legal Perspectives and the Critique of the Dominant Legal View / 127

can tell the public that the tests are reliable—they are reliable, by construction, by eliminating potentially valid questions that would alter the rankings of test-takers and thereby reduce test-retest reliability. But, eliminating such questions because those who answer the questions correctly did not answer other questions correctly is clearly discriminatory. Were a classroom teacher to announce that he or she planned to grade the final exam by considering the midterm grades, and planned to eliminate from the final exam grade calculation any question that those who did poorly on the midterm answer correctly if those who did well on the midterm answer the question incorrectly, that teacher would likely be fired—or worse. Yet, that is essentially what standardized test-makers do year after year, with much higher stakes in play.

Self-Contradictory Logic and the Psychometric Tradition Eliminating items on the basis of test-score DIF comparison and point biserial analyses is a taken-for-granted approach for test-makers. Sometimes the logic invoked would be comical, if the substance did not concern a serious matter. For example, in a useful paper titled “Ensuring the Fairness of Licensing Tests,” Michael Zieky of the Educational Testing Ser vice describes several important procedures for constructing a fair test. At one point he counsels test-makers to “avoid arbitrary and inconsistent decisions about test content by following a documented set of rules to determine which aspects of the job are represented in the test. With respect to fairness, be careful to require only knowledge that is required of beginners and not knowledge that is usually acquired on the job” (Zieky 2002: 4; references omitted). After advising the appointment of a Fairness Advisory Group composed of a diverse set of persons, he writes: “Unless the ability to decode difficult text is an important aspect of the job, make sure that items are as easy to read as possible, particularly if the test taking population includes people who are not native speakers of English. Avoid low frequency words unless they are important for entry-level licenses. Avoid convoluted syntax in items and stimulus materials. The fairness advisory group can be useful in spotting items with excessive linguistic demands” (Zieky 2002: 6). Yet, later, Zieky proposes just the kind of DIF with test-score equating discussed herein. He contends that DIF should be used in a fairness review of the test, writing: It is better to use an empirical measure of fairness called differential item functioning (DIF) in which test takers in different groups are matched in terms of relevant knowledge and skill before their

128 / Chapter 5

performances on items is compared. The matching is most commonly done on the basis of test scores. The scores are obviously related to what the test is measuring, and (if you do your job well) the scores can be shown to be reliable, valid, and obtained under standardized conditions. (Zieky 2002: 8; emphasis added) Zieky goes on to indicate that DIF cannot tell you a question is unfair, and that judgment is needed. But, one is struck by the contradiction between the first two excerpts and the last. The first two excerpts take constructing a test as a problem that encompasses the writing and selection of items. The last excerpt suggests one evaluates the functioning of question number one by treating all the other questions (questions two through the end of the test) on the proposed test as a criterion to classify test-takers. After that is done, question number one is tossed back into the mix and it, along with questions three through the end of the test are used as a criterion for evaluating item two. This process continues for each and every item. Of course, this makes no sense if many or perhaps most of the items are problematic—finding that question four does well when questions one through three and five to the end of the test are used as the criterion tells us little if questions one, two, three, five, ten, twelve, and perhaps more also have problems. This problem is not addressed by using as a criterion last year’s test, which was constructed with the same problem-riddled procedures. Yet, in Zieky’s advice it should be clear: The procedure is so taken for granted that obvious logical errors pass without comment. It is certainly the case that in technical reports there is sometimes awareness of such contradictions, but the counsel for constructing tests, and the process itself, has failed to face this contradiction squarely. Of course, test technology has not stood still; the par ticu lar problems identified above provide the point of departure for new developments that deepen the problems. Thus, test technology presents a moving target; one dimension through which it is currently moving involves the supplanting of paper-and-pencil test forms with computerized testing. The change to computerized testing doubtless offers wonderful possibilities, but it also entails myriad opportunities to discriminate, as well as a subtle shift in the meaning of “fairness.” The possibility of computerized discrimination becomes real with the advent of adaptive testing, a type of test most easily administered via computer. Under adaptive testing a test-taker sits at a computer terminal and is given a few questions to answer. Answers instantly are recorded and analyzed somewhat deterministically. As the person moves through the test, the test is adapted so that the next questions the person receives are near their (so far) estimated level of competence. In other words, if a person misses a few allegedly easy questions earlier in the test, the test program will assume there is

Critical Legal Perspectives and the Critique of the Dominant Legal View / 129

little need to ask the person harder questions later on. Therefore, adaptive testing transforms a test into something akin to a tournament, where correct answers win students the right to harder questions, but a few wrong answers can push the test program to zero in on just where in the middle or the bottom the student appears to be located. Note the subtle change in the definition of fairness. Formerly, testing organizations administering large-scale tests, such as the ACT, the SAT, and the GRE, constructed several versions of the test every year. Several versions were needed because of the large number of examinees and the possibility of cheating. Year after year, however, test organizations assured the public that the particular version of the test a test-taker received did not matter—the forms were alleged to be equivalent. Thus, test organizations labored under the idea that, to ensure equal opportunity, every form needed to be equivalent; every test-taker deserved an equal shot to answer an equal number of questions at the various levels of presumed difficulty. Anything else would be unfair. With the advent of computerized adaptive testing, however, test organizations seek to administer tests that are designed to be formally nonequivalent after the first set of questions has been answered, such that test-takers clearly do not have an equal chance to answer questions throughout the possible levels of difficulty. It is as if, once a student fails to answer a set of allegedly easy questions correctly, he or she forfeits any right to equal opportunity, any right to be presented with the alleged more difficult questions, questions that are given higher value in the scoring algorithm. In this way, the problems with the point biserial and DIF analyses have been brought forward into the testing room itself. The assumption underlying adaptive testing is that students who fail to answer questions the test-makers regard as easy (or questions other allegedly better students answer correctly) need not be given much of a chance to redeem themselves. The spread of adaptive testing is based on this dubious assumption. In the process, fairness, formerly interpreted as an equal opportunity to answer every level of question, is now implicitly defined as not being presented with questions the testing algorithm implies one cannot answer correctly. Nothing could be more defensible than a test of academic achievement to at least partly determine whether a student should be admitted to a college or university program. But existing procedures for devising high stakes tests operate to sediment disadvantage in cognitive rankings in a manner similar to how Melvin L. Oliver and Thomas M. Shapiro (1995) have shown that social policies can sediment disadvantage in the case of financial wealth. And the continued use of such techniques furnish but one of many examples one might provide of how property interests, defined as expectations or in other ways, make par ticular standard operating procedures “reasonable” in ways that sustain existing race and gender disadvantage.

130 / Chapter 5

Adjudicating Discrimination in Human Capital Processes: Dominant and Critical Approaches We now have in hand two different legal perspectives on discrimination. The existence of an alternative indicates that the dominant legal perspective is not the only conceivable arrangement. The alternative critical legal discourse allows us to see more clearly specific features of the dominant understanding of the law. And, in doing so, we can more easily discern a key internal contradiction in that perspective. Human capital processes serve as our illustrative case. In order to enter many colleges, graduate schools, law schools, and medical schools, and in order to obtain licenses or even be hired for many jobs, applicants must sit for standardized examinations. The College Board reported that over 1.4 million members of the high school class of 2005 took the SAT, a more than 4 percent increase over the previous year (College Board 2005), and thousands more students took the ACT. Additional thousands take the GRE, GMAT, MCAT, and LSAT. Hundreds of colleges and universities use results from these tests to evaluate applicants. Additionally, tests for hiring and licensing are administered to thousands of job applicants each year, and used by employers in decision-making. Because of the role of standardized tests in college and university admissions as well as hiring, testing plays an important part in determining whether one will be able to make high-level human capital investments in school or in on-the-job training. And, because of the role of standardized tests in licensing and hiring, testing plays an important part in determining whether one will be able to obtain significant returns on whatever human capital investments one has made. Hence, tests and their construction appear intimately connected to success and opportunity in the United States.

Critical Legal Rules of Order vs. Dominant Legal Rules of Order If test-construction procedures ever become litigated, the approach of dominant legal reasoning to the case is clear, even if the outcome may not be. Dominant legal reasoning will focus on the par ticu lar practice, and will ask whether the use of DIF and point biserial analyses in the manner described above is disparate treatment discrimination, disparate impact discrimination, both, or neither. Should the courts rule the act of using such standardized tests discriminatory, dominant legal reasoning will become consumed with determining responsibility for this state of affairs, so that restitution

Critical Legal Perspectives and the Critique of the Dominant Legal View / 131

can be narrowly extracted. But, even before determining whether the behavior is discriminatory, the courts will constrain the class of litigants, both plaintiffs and defendants, such that the defendants in the case will be some set of test-construction companies and perhaps a college, university, or employer who used the test in their applicant review process, and the plaintiffs will be limited to those who applied to the college, university, or employer, and took the test under investigation. Should the case be decided in favor of the plaintiffs, they will be compensated from the specific defendants in the case. And, should the plaintiffs win, use of the par ticular technique will decline, perhaps to zero, though the actual force of the decision will depend on the level of the court that decides the case and the par ticular content of any decision the court renders. The courts’ procedures may turn on whether litigants claim disparate treatment or disparate impact, but the logic of the courts’ approach—the belief that discrimination is what one entity does to others—will be the same. It is clear that disparate treatment discrimination can be defined as what one person or corporate entity does to one or more others. However, in dominant legal reasoning, disparate impact discrimination is also what one entity does to others because the only possibility for remedy arises when some entity is held responsible for using the mechanism that is alleged to have a disparate impact. No general class defined solely by race can sue to gain broad compensation for, for example, the construction and use of standardized, discriminatory tests in the allocation of opportunity in the United States. Only single or corporate actors can be sued, and only they will pay restitution to a set of defendants far smaller than the set of African Americans and women who, over the last several decades, have had their opportunities greatly affected, and likely limited, owing to the use of such tests and the promulgation of the expectations of achievement that they nurture. Throughout the process, the courts act on the basis of an asocial understanding of discrimination. This asocial understanding denies the way in which the phenomenon is connected to common taken-for-granted procedures and assumptions, and remains blind to the way in which common taken-for-granted procedures work and are maintained. Elsewhere I maintain that there is no malicious cabal at the center of the effort to use DIF in test construction (Lucas 2000). I reiterate that claim. Indeed, were there such a cabal, perhaps the courts’ perspective would be appropriate; all that would be needed to rectify matters would be to identify the conspirators, dismantle the arrangements they conspired to produce, and draw vast punitive and compensatory damages from them to allocate to women, blacks, and other victims of their designs. But the social world is not so simple. What we have seen is that test questions and test results that do not reproduce the current patterns of power

132 / Chapter 5

become suspect. The apparatus developed and implemented to root out such questions is time consuming and expensive, meaning that essentially a great deal of effort goes into assuring that the tests support the reproduction of existing patterns of power. The manipulation this process entails is to assure that results conform to expectations, rather than to consciously manipulate public opinion or par ticular outcomes for individuals. Yet, recall that the whiteness as property and maleness as property perspectives suggest that the property interest is based in part in an expectation of the relative power of people of color vis-à-vis whites and women vis-à-vis men. Thus, the key assumptions both in psychometric testing procedures and in the vision of whiteness and maleness as property concern the expectation of who will perform (and thus who will be allowed to perform) and who will not. Recall, also, that testing provided a shared example. Testing is not the only area in which expectations enter in a determinative way. For example, as the Wall Street Journal reported, white parents in upper-middle-class suburbs from the silicon valley town of Cupertino, California, to the New York City suburb of Tenafly, New Jersey, are withdrawing their children from schools with large numbers of Asian students. In such schools, whites are rare in Advanced Placement and honors courses, and join Latino/as and blacks as in the lower-level courses. Citing stultifying academic pressure and cultural exclusion, white parents take their children elsewhere—to schools with the same amount of academic pressure, but more whites (Hwang 2005). One way to understand this response is to point to the expectation-based property interest in whiteness that leads whites to expect to always be on top, such that their failure to be on top “proves” the institution has failed. Seen in this light, pretesting standardized test questions to assure they reinforce settled expectations of the dominant to dominate takes on a different hue. And, consistent with this observation, and contrary to the courts’ asocial understanding, what maintains standard discriminatory operating procedures is a set of expectations about how power should be allocated, who is deserving, and a belief that the present configurations of success and failure are so legitimate, so “natural,” that even a single question on an exam that calls that legitimacy into question must be excised from the process. Thus, because these factors are invisible to the dominant legal perspective, even if the particular psychometric principles are successfully challenged in the courts, the reparative response will be arranged in ways that minimally alter the real distribution of power, a distribution that currently matches the settled expectations as to the racial order and the gendered configuration of opportunity. In contrast to dominant legal reasoning, critical legal scholarship draws our attention to the full facts and relations implicated in the discriminatory arrangement. Critical legal perspectives make visible what the court will not do; the court will not invalidate or otherwise respond in a compensatory

Critical Legal Perspectives and the Critique of the Dominant Legal View / 133

fashion to the larger history of testing, a history that includes each and every one of the millions of decisions that have been made in part on the basis of such tests since their use began. The critical legal perspective should make clear to us that when it comes to the obvious discriminatory process of test construction, each and every one of the millions of admission and hiring decisions for which such information was used are implicated in the injustice. Two points need be made in this connection. First, obviously it may be impossible at this juncture to trace those relations with any confidence. But, it is dominant legal reasoning that is interested in attempting to trace relations of blame and injury; critical legal perspectives are focused on solutions, on resolving the injury, not on a punitive imperative. Second, the critical legal claim is not to assert that some, many, or even most decisions would have been altered were tests constructed differently and had fully fair systems been in place. But, it is to say that to the extent the legitimacy of the decisions and allocated opportunities rest on the discriminatory test-construction procedures, the decisions are illegitimate. Yet, under dominant legal reasoning and procedures, the vast majority of the individuals disadvantaged by the injustice will never receive a single dime toward the compensation they deserve owing to this clearly discriminatory mechanism.

A Core Contradiction of Dominant Legal Reasoning Recall that dominant legal reasoning recognizes disparate treatment and disparate impact discrimination. However, to establish the occurrence of either, plaintiffs must identify the perpetrators of discriminatory acts. Only when such perpetrators are identified may the legal apparatus swing into action, and recompense the victims of discrimination out of the pockets of the specific perpetrators. When these two characteristics—the focus on punishing transgressors, and the illegitimacy of group rights—are applied to the area of discrimination, the result is in some ways self-contradictory. The contradiction follows in that, owing to discrimination logic as developed within the law, the legal system will punish when persons are treated as members of a group (e.g., a racial group). The doctrinal aim, therefore, is to punish anyone who treats others as groups and therefore discriminates. However, ironically, U.S. courts also refuse to treat individuals as members of groups, even if that is the only way to provide remedy to all members of groups, some of whom were treated only as members of their group. Both the legal rules of order and the punitive imperative from which they partially flow are ironic because discrimination is necessarily about treatment

134 / Chapter 5

on the basis of putative membership in certain categories. The violation is that one has been treated as a member of an irrelevant category. If one person or a small subset was treated in this manner, all members of the target group were implicated, even if they happened to not be present to receive that discriminatory treatment or encounter that arrangement directly. That is, even students who do not take a college entrance exam are harmed by the discriminatory standard operating procedures that construct the test, for those procedures are part of an edifice that constructs assumptions of who can and cannot succeed, an edifice they have had to confront as young children, adolescents, and young adults. The existence of that edifice may have determined in part whether they would sit for a college entrance examination, whether their support network (e.g., teachers) would expect them to do so and succeed, and more. Hence, membership in the group actually is sufficient to entail disadvantage owing to discrimination even if one does not meet the courts’ standard of explicitly, directly, experiencing the arrangement. Because U.S. legal doctrine does not recognize group rights, membership in the putative group is insufficient grounds to seek redress before the courts. That is, students who were denied admission to college may sue. But, parents of third-graders, despite knowing their children navigate a minefield of low expectations planted and nurtured in part by discriminatory college admissions test-construction procedures, may not. There are two notable exceptions to the nonexistence of group rights. One case concerns Native American tribes. Many Native American tribes negotiated treaties with the U.S. federal government that recognized Native American sovereignty. Though these treaties were negotiated under duress, and for most of the intervening centuries were violated episodically by U.S. government actions at will, the treaties provide an alternative model for the treatment of racialized minorities in the United States. As of the 1970s, the sovereignty became recognized, though what the recognition has meant remains a matter of contestation. Still, it is noteworthy that in U.S. law, Native American tribes are par ticular kinds of groups—sovereign nations. A second case concerns the corporate form. Corporations are regarded as persons, and as such enjoy many of the rights of individuals. Hence, members of a corporation can have a collective interest that the courts will enforce as if it were the interest of a single individual. These examples suggest that even in U.S. legal reasoning, with its individualistic logic, the formalization of group rights is not unheard of in U.S. law. However, the dominant thread cuts against group rights. One place where the dominant legal vision of group rights is articulated is in statutes and rulings concerning class-action litigation. Notably, a key moment in any class-action lawsuit occurs when the court is asked to certify the existence of the class. A plaintiff’s mere assertion that they are a member

Critical Legal Perspectives and the Critique of the Dominant Legal View / 135

of a class is not sufficient to establish that the plaintiff is a member of a class. Indeed, the rules stipulate that belonging to a par ticular race or sex is not sufficient grounds for class certification (Federal Rules of Civil Procedure Rule 23(a) 1966). One consequence of the rule that membership in a sex or race group is insufficient to establish that the plaintiff represents a class is that certifying a class that is overly broad increases the chance that the class will encompass persons with possibly opposed interests; in such situations it is possible that the plaintiffs who are present will fail to represent the interests of absent others. Indeed, present plaintiffs might trade away the rights or goods of absent members of the class in order to secure rights and goods for themselves. One explicit aim of the rule is to preserve the ability of absent others to act on their own interests in the matter. Hence, this rule has a protective rationale, protecting the ability of those not present against the poor representation of their interests. But the rule also makes it impossible for groups based on race or sex to exist before the law, even when the members of a group share the same interest, as they might vis-à-vis discriminatory procedures of test construction as well as other taken-for-granted operating procedures. Thus, there is virtually no ability for groups that would be defined solely on the basis of race or sex to identify themselves as a group, even though the conditions that create their mutual interest may have been erected so as to treat each as a member of a group in a disenfranchising manner. Of course, this implies both that the group cannot obtain full redress for the transgression, and that it is also impossible to define a residual group, specified as persons who are not members of the plaintiffs’ race or sex, as a defendant class. The primary consequence of this situation is that, owing to this rule and the logic against group rights in general, courts often are unable to compensate persons whose race or sex group membership led them to be wronged (e.g., the third-graders harmed by college admissions test-construction procedures). This occurs because the court certifies a class of plaintiffs, narrowly defined. In the same way, there are mechanisms to certify that a specific set of defendants compose a class, but there is no mechanism for certifying a general class of defendants. Defendants, thus, are narrowly defi ned within the legal proceeding. This narrow definition has many implications, two of which need be noted here. First, plaintiffs may ask the court to certify their existence as a group, their class status, but to do so plaintiffs cannot appeal only to their race/sex membership. Instead, plaintiffs must also identify class-creating or classmaintaining acts perpetrated by par ticular identified defendants; most important, the simple existence of race is not regarded as class-creating or class-maintaining.

136 / Chapter 5

Second, the plaintiffs must name defendants. Defendants likely will argue that they did nothing wrong. Hence, doing nothing wrong is a defense against the charge, as one might expect in U.S. legal proceedings. However, this structure transforms the entire set of bystanders into innocents, outside the realm of law, outside the set of persons who must directly compensate the plaintiffs for the wrong done to them. Consequently, unless the court can identify the par ticu lar individuals who committed a par ticu lar wrong, the wrong is never repaired, even if all agree the wrong occurred. This state of affairs remains true even if a group of nonallied individuals acted as if colluding in a way that creates or acquiesces in the creation or maintenance of circumstances that produced or sustained the wrong. In response, critical legal theorists document that the very creation of a racialized and gendered hierarchy created the classes that have the grievances— blacks (and other persons of color) and women (of whatever race/ethnicity). Hence, the acts that created white privilege and maleness as property created the class and the associated wide-ranging grievance, and all who come thereafter must orient to rectifying that situation, regardless of their personal connection to any par ticular explicit acts or benefits.

Critical Legal Remedy: An Impossible Proposition? As we see, the critical legal analysis of human capital processes de-emphasizes the punitive imperative in favor of the solution imperative: Solve the problem of the disadvantaged. In this perspective, there need be no specific “defendants.” Critical legal scholarship redefines the characters in play, such that the actors with whom we are familiar owing to dominant legal reasoning are not the actors involved in critical legal discourse. As we have seen, U.S. legal doctrine focuses on targets of discrimination and perpetrators, and rules out the possibility of drawing compensatory damages directly from bystanders to discrimination, bystanders who by their exclusion from the targeted group may very well be beneficiary bystanders to discriminatory acts. Such bystanders are, by definition, not members of the group that was the target of discrimination. Given that bystanders are by definition not members of the targeted group, they form a residual group constituted by virtue of their nontargeted status. Yet the courts do not recognize the status of “beneficiary bystander” and thus these persons have escaped (to date) any direct responsibility for compensating the victims of discrimination. Critical legal scholarship does not only assert that both bystanders and perpetrators need to compensate targets. No, critical legal scholarship broadens the assessment of the phenomenon to allow us to see how even expectations that are not explicitly racialized or gendered are reinforced by day-to-day

Critical Legal Perspectives and the Critique of the Dominant Legal View / 137

technical processes that systematically reject any evidence that violates settled expectations of whites and men to dominate. Thus, to the dominant legal tradition’s duo of discriminatory mechanisms, critical perspectives add the taken-for-granted standard operating procedures that maintain the disadvantage about which targets of discrimination are concerned. The critical tradition traces these operating procedures, and the assumptions to which they correspond, back historically, and they project their operation into the future unless some direct action against their operation is taken. Hence, the critical perspective is not focused on identifying par ticular perpetrators (Freeman 1978). To the critical legal perspective, in an environment suffused with discrimination, identifying par ticu lar perpetrators is grossly insufficient for recognizing the wrong and perhaps addressing it. And, pointing to the historic construction of race and gender hierarchies, as well as contemporary taken-for-granted structures of discrimination, the critical legal tradition makes a persuasive case that the environment is indeed suffused with discrimination. In such an environment, dominant legal reasoning is woefully inadequate. Consequently, critical legal rules of order around discrimination would dispense with the punitive imperative and the need to identify a par ticular perpetrator for each discriminatory outcome. This response is necessary because discrimination is not simply an act one person may do to another; discrimination is also the conditions that provide limited expectations for some and boundless expectations for others. This aspect of discrimination cannot be addressed by searching for perpetrators of acts, for it concerns taken-for-granted assumptions about the proper allocation of power in society. Given the ubiquitous nature of such assumptions, to ignore their existence is to greatly narrow the phenomenon under consideration. However, once this critical perspective is appreciated, the prospects for remedy are widened. Certainly, this is not the place for serious consideration of remedy, for we have only concerned ourselves with defining the phenomenon, not with assessing its impact. Yet, because the discussion, however persuasive in the abstract, may appear divorced from really existing legal traditions in the United States, a brief nod to the possibility of a remedy consistent with critical legal reasoning is required at this juncture. Indeed, we may affirm not only that remedy is possible within the critical legal perspective, but, more important, the dominant legal tradition already has acquiesced in remedies consistent with a solution-focused approach in other areas. As one example, consider the savings and loan (S&L) bailout of 1989. Without going into the details, the cost of the federal bailout has been estimated to be as high as $1.4 trillion (Hill 1990: 24). The government entity charged with administering the bailout reported that “fraud and potentially criminal conduct contributed to the failure in 41 percent” of the cases

138 / Chapter 5

(Resolution Trust Corporation 1990: 9). Of course, this means that 59 percent of the failed thrifts had no documented criminal conduct as of their bailout, as of 1990. In other words, in principle the bailout was severed from the prospect of punishment. To be sure, civil recovery efforts and criminal enforcement efforts continued. Yet, the financial returns to those recovery efforts pale in comparison to the cost of the bailout (Pontell and Calavita 1993). And, as the 41 percent figure suggests, many if not most of the cases involved never would face criminal liability. Thus, whether or not a thrift was bailed out by the government (i.e., whether or not shareholders and communities were protected from the liability that would be expected to attach to involvement in banking) was not contingent on a showing of criminal conduct or civilly liable behavior. Of course, one reason the bailout took the form that it did is that, at base, key elites believed that failing to bail out the industry would damage the economy and that their own political fortunes might be affected negatively were that to occur. Owing to the political connectedness and wealth of the potential defendants, prioritizing successful prosecutions and thereby delaying bailouts might allow damaging countermobilization against sitting political elites. The ensuing battle might leave the economy substantially harmed, and those in power deposed or at least weakened. Thus, instead, elites opted for a solution imperative, subordinating (but not abandoning) the punitive aspect in the process. I hasten to add, I am not maintaining that the S&L crisis and discrimination are similar in all respects. The cases have important differences. This is not the place to develop whatever insights the S&L bailout may provide for antidiscrimination public policy. The point of the example here is only to indicate that the current instantiation of antidiscrimination law is not the only possible one given U.S. legal procedures; that there are models of a different approach that have been adopted in other domains, and that these approaches have more in common with the critical legal perspective appreciation for the complexity of the social world than does the current instantiation of antidiscrimination law. And, seen in this way, the case can be made that the major factor blocking a solution-based response to discrimination in the United States may not be predicted economic dislocation or intractable administrative difficulties but, instead, a lack of leadership and a dearth of political will. Notably, this is consistent with what one might expect in the era of contested prejudice.

Concluding Remarks The foregoing has delved into dominant and critical legal perspectives on discrimination. Yet, our epistemological project concerns social science, not

Critical Legal Perspectives and the Critique of the Dominant Legal View / 139

the law. Hence, one may wonder, why devote time to cases, law review articles, and a legalistic analysis of how different conceptions of discrimination might respond to our illustrative case concerning human capital acquisition? Is this not a gratuitous distraction from our task? The reason for this seeming digression is that the courts have played a major role in constructing the societal terrain upon which any definition of discrimination must be planted, a terrain that if undisturbed will nurture any definition already consonant with dominant legal reasoning, but leave any alternative, no matter how robust, withering in the sun. To prevent the discussion of social scientific perspectives that awaits us in the chapters that follow from simply reproducing the standard, frustrating, go-nowhere exchanges between persons locked into positions whose underpinnings remain unexamined, we needed to turn up the taken-for-granted grounds of the dominant legal perspective, plowing through it to unearth the rich possibilities and resources that lie deeply embedded in legal debates. Once those resources are made visible, dominant legal reasoning can no longer be seen as the only conceivable framework for considering discrimination a legal issue. Once this happens, the monopoly of dominant legal reasoning on presumed rationality is broken, and thus dominant legal defi nitions can no longer serve as a final trump card in any social scientific debate concerning the appropriate way to define and study discrimination. Afterward, instead of defending a definition of discrimination in the last instance essentially on the basis of its acceptance by dominant elites, social scientists must consider any definition in terms of its resonance with the social phenomenon at issue, eschewing its connection to existing relations of power. In order to realize this analytic advance, however, we needed to disconnect implicit, perhaps unhelpful assumptions taken from the law from our social scientific analysis. And, to accomplish this aim, it was necessary to address the law on its own terms before turning to social science directly. Turning to the law, we find that the legal system focuses its attention on par ticu lar identifiable acts defined as discriminatory, concerns itself with finding and punishing the perpetrator(s) of these par ticular acts, and thereby narrows the category of persons who will have to compensate wronged parties to only those individuals who were demonstrably engaged in the individual discriminatory actions. This state of affairs has not gone unnoticed within the legal field. Considering the critical legal research on discrimination and the courts reveals a useful vantage point from which to survey the day-to-day workings of the legal system. Critical legal scholars argue that antidiscrimination law as enacted actually legitimizes discrimination by focusing on acts of individuals rather than on either the structural factors underlying discrimination or the conditions that need redress. Freeman contends that the social conditions that result

140 / Chapter 5

from being regarded as a group member rather than as an individual are discriminatory and that only through removal of those social conditions can discrimination be delegitimized. But antidiscrimination law enacts an asocial perspective, and treats discrimination as only the behavior of individuals, narrowly defined. Other scholars use the language of property rights to describe a situation that is sometimes subtly, and sometimes not so subtly, biased against blacks, women, and others. Through Freeman’s vision, as well as that of Harris, Frug, Singer, and others, one can discern the existence of beneficiary bystanders and more. Their observations become especially relevant in at least two ways as we turn our attention to the social scientific analysis of discrimination and its effects. Most clearly, and concerning content, critical legal observations provide resources we may use to resuscitate the social in the social scientific analysis of discrimination. We will draw on these and other resources as we pursue this aim in Chapter 7. Yet, the second way in which critical legal scholarship is important as we consider how social scientists have studied discrimination is in the way in which it is positioned in relation to dominant legal reasoning. The critical legal scholars suggest that the assumptions rendering antidiscrimination law impotent are deeply embedded in the construction of the entire legal enterprise and much of the operation of society, so much so that if one stands squarely inside standard operating procedures and arrangements whose dimensions are constructed and legitimated through dominant legal discourse, alternatives may appear literally irrational. Indeed, rationality becomes defi ned by what is seen as normal, even if what is seen as normal is antithetical to human flourishing. The reign of normalcy in the courts is joined to the reign of normalcy in mundane institutionalized processes; in the same way that the courts’ neglect of beneficiary bystanders may seem rational, the illustrative institutionalized practices of psychometric test construction may seem eminently rational, and any critic of those practices may appear irrational, if one ignores the logical implications (i.e., inherent contradictions of normal psychometric practice). This, of course, is also what has happened in the social scientific study of discrimination, as we shall see in Chapter 6; deviations from standard social scientific understandings of discrimination seem virtually irrational to dominant social science perspectives. However, once one moves beyond the narrow asocial understanding of discrimination employed by dominant legal reasoning, the social phenomenon of discrimination can be more fully glimpsed. And, once the phenomenon becomes visible, the institutionalized, normal operating procedures become

Critical Legal Perspectives and the Critique of the Dominant Legal View / 141

questionable, and bystanders, beneficiary or otherwise, become decidedly relevant to our concerns. Yet, the courts have neglected these observations. The courts’ reluctance to cast the net broadly on discrimination is thoroughly consistent with the historical narrative presented in Chapter 3. Recall that a key factor in the decline of the regime of condoned exploitation was that elites changed their minds about the usefulness of blacks and women, for their own reasons, reasons based in their perceptions of the needs of geopolitical competition and internal economic development. It was decidedly not the case that the interests of women and blacks drove the development of policy. Further, it was not exploitation that was repudiated; one justification for exploitive arrangements fell from favor, not the practice itself. In the same way, even during the era of contested prejudice, the court seems to rarely if ever take seriously the interests of blacks and women to escape exploitation unless their escape jibes with a narrow repudiation of prejudice and narrowly defined discrimination in a par ticu lar instance or arena. Seen in this way, even the law’s response to discrimination, championed by some, appears grudgingly provided and severely constrained. Certainly, different legal actors act differently, with some supporting women’s and blacks’ rights, while others work against them. But, the institution of the law appears, at best, negligent with respect to women and blacks. Given that reality, and the distribution of power, it is unlikely the critical perspective will come to rule in U.S. courts anytime soon. Still, one might be concerned at the ultimate outcome of the instigation of social upheaval by courts intent on consistently repudiating whiteness and maleness as property. Reasonable people may disagree about the ramifications of what, at present, appears to be an unlikely possibility. Yet, once we leave the legal area and move to social science research, to take up the issue of whether discrimination exists and what are its effects, many of the contentious aspects that characterize the legal arena—the adversarial structure, the zero-sum nature of the contest, and more—ebb away. One might expect social science, therefore, to adopt a more encompassing perspective on discrimination, freed as it is from the need to daily adjudicate specific allegations of discrimination, allegations that can be painful for plaintiffs, defendants, and others as well. Yet, we do not find social scientists liberated by their opportunity to consider discrimination broadly. No, we find that dominant legal logic actually pervades the social scientific analysis of discrimination. As I will show (in Chapters 6 and 7), this is neither a useful nor necessary transference of legal logic to social scientific work, and thus it is not utility or necessity that leads social scientists to conceptualize discrimination in a manner similar to that

142 / Chapter 5

operating in the courthouses throughout the land. Though the issue of motivation may be interesting, actually the key question for us now is: How do social scientists use dominant legal logics in their research on discrimination? To address this question, we need delve into the way in which social scientists study discrimination in the United States, a task we commence in the chapter that follows.

6 Defining Discrimination Effects: An Asocial Scientific Method

C

ritical perspectives on antidiscrimination law critique the dominant legal perspective on discrimination. Critical legal perspectives also implicitly critique dominant social scientific thinking on discrimination. The criticisms of the law become criticisms of dominant social scientific thinking because social science perspectives are startlingly similar to the courts.’ The social scientific adoption of a perspective similar to the courts’ is intriguing. Even though social science arguably has played a major role in key legal decisions on discrimination, the court of social science is not a court of law. Unlike judges, juries, and attorneys, social scientists rarely are interested in identifying by name the par ticu lar person or persons affected by a social phenomenon. Instead, social scientists focus on broader categories of persons—racial groups, types of parents, genders, age cohorts, and so on. Despite this clear difference between the law and social science, in this chapter I show that social scientists have adopted approaches to studying discrimination that are more consistent with the courts’ focus on individuals and individual acts than with critical scholars’ relatively more social perspective on discrimination. Further, I detail how adopting an approach more consistent with legal reasoning than with social investigation has unnecessarily hobbled social scientific efforts to understand the ramifications of discrimination.

144 / Chapter 6

At this juncture it is important to distinguish two types of discrimination research. One type of study seeks to document the existence of discriminatory behaviors or prejudicial attitudes. A classic study in this tradition is the work of Gunnar Myrdal ([1944] 1962). More recent work in this vein includes historical, sociolog ical, anthropological, psychological, political scientific, economic, and literary analyses that document the existence of past and present racism, sexism, and discrimination on the basis of either or both (e.g., Goggin 1983 and 1992; Kirschenman and Neckerman 1991; Menchaca 1993; Fiske 2000; Kuklinski, Cobb, and Gilens 1997; Abel 1999; Goldin and Rouse 2000; Feagin and Vera 1995; Bonilla-Silva 2001). Such studies chronicle the prior and continuing existence—the ontology— of race and sex discrimination. This kind of research has long been important, and only grows more important day by day as those cohorts of blacks who experienced the Jim Crow South, and those cohorts of women who were legally and brazenly denied the opportunities for which they were more than qualified as well as the equality that was theirs by birth, die off, and the pain of their travails fades from public memory. As time passes, the sociolog ical and political importance of these studies will only increase. However, a second type of research seeks to determine whether the attainments of blacks, women, and other groups are affected by discrimination. In other words, these studies seek to answer questions of the following form: How much does the average woman lose by virtue of discrimination? This second set, which I term discrimination effects studies, is the focus of my concern. Like studies of the ontology of discrimination, discrimination effects studies also address questions that are sociologically and politically important. With respect to politics, if discrimination effects are high, then we might be less likely to blame the downtrodden member of a targeted group for their plight, and we might be more likely to construct policies to restore the losses and end the processes of discrimination. On the other hand, if there are no measurable effects of discrimination, then we must search elsewhere to explain the disadvantageous outcomes some groups seem prone to experience. And, if the effects of discrimination are so small they cannot be measured, we are unlikely to introduce and expand policies and efforts to eradicate discrimination. Discrimination effects studies usually are embedded in par ticular fields, such as urban studies or labor economics, and thus are, at base, efforts to understand the unequal outcomes of blacks and whites, and women and men, in the goods that are the focus of those fields. Hence, the primary aim often is not to understand discrimination but, instead, to understand inequality in housing, earnings, education, or other outcomes. In these analyses,

Defining Discrimination Effects: An Asocial Scientific Method / 145

discrimination is treated as one possibly important factor in explaining how individuals obtain par ticu lar goods. Embedding discrimination effects studies in research endeavoring to understand the variable distribution of goods in society has a long history. However, problems are created by the strategies followed in discrimination effects studies, strategies that have much more in common with the courts’ view of discrimination than with more social perspectives on discrimination. As I will show, it is the courts’ view of discrimination, arguably appropriate for legal adjudications, that stands in the way of a successful effort to estimate the effects of discrimination. To make this case I outline Gary Becker’s quite creative and undoubtedly foundational analytic effort to estimate discrimination effects. I then detail several criticisms of Becker’s approach, an approach that has received sharp criticism over the fifty years since its formulation. Pointed criticisms of Becker’s theory have led to major developments in the study of discrimination effects. Indeed, the debate concerning Becker’s theory of discrimination could be an exemplar for how scholarly disagreement can strengthen both the framework analysts use and our understanding of society. The criticisms usually voiced have concerned the economic assumptions of Becker’s theory. However, additional criticisms are different, for they are implied by the critique of dominant legal reasoning. The criticisms of Becker’s framework implied by critical race theory, critical legal studies, and feminist legal scholarship present the most fundamental challenge to the line of reasoning and research social scientists have pursued in their efforts to understand the effects of discrimination. After relating those criticisms, I briefly detail some of the efforts social scientists have devised to move beyond Becker’s formulation. As I do so, I show how these efforts to transcend Becker’s approach do not address the foundational problems the critical perspectives highlight. Afterward, I show how contemporary empirical research, even though often self-avowedly embedded in Becker’s approach, actually ignores the strengths of Becker’s formulation and draws on the weakest aspects of his analysis. Finally, the chapter concludes with summary remarks as entrée to a more social perspective on discrimination. These many disparate sections are actually in ser vice to only two distinct aims: (1) to describe leading contemporary approaches to studying discrimination effects and (2) to critique those approaches using the resources provided by critical scholars’ cogent comments on existing legal reasoning concerning discrimination. Accomplishing these aims provides the foundation for the chapter that follows, in which a more social approach to the social scientific study of discrimination effects is introduced. Thus, this chapter culminates in one essential point: The foundational basis of social scientific

146 / Chapter 6

research on discrimination effects is distinctly asocial and, because of this feature, necessarily prone to failure. To demonstrate why failure is inevitable, we must focus much of our attention on Becker’s path-breaking theory of discrimination. However, before turning to Becker, it pays to reiterate the problems mainstream social science identifies as the key challenges that studies of discrimination effects must address.

Three Problems of Discrimination Effects Research Mainstream analyses recognize three main problems that hound discrimination effects research. First, discrimination is often illegal. Therefore, key actors—discriminators—tend to try to conceal their unlawful behavior. Because the behavior is usually unseen, estimating its effects is extremely difficult. Second, discrimination may occur anywhere, anytime. Every possible interaction, as well as every possible noninteraction, could involve discrimination. The infinite number of possibly discriminatory moments makes it difficult to identify just where discrimination occurs, making it even more difficult to identify the effects of discrimination. Third, basic epistemological observations imply that it may be impossible to estimate effects of discrimination. In order to make a causal inference, analysts either need access to two instances—a case in which a unit is exposed to a treatment and a case where the same unit is not exposed to a treatment—or analysts need some way to leverage the different instances they have to assess the causal effect. However, comparing the experience of men and women does not identify the effect of discrimination, because changing a person from one sex to another would be to fundamentally change the unit. Because the units are not and cannot be made comparable, one cannot use such comparisons to infer the effects of discrimination. These are mainstream social scientific observations concerning the challenge of estimating discrimination effects (e.g., Blank, Dabady, and Citro 2004). These critical observations are, of course, articulated within the context established in large part by Becker’s theoretical analysis of discrimination and the research that has followed in its wake. In other words, mainstream social science conceives of discrimination in a par ticular manner; given that conception, these three problems will undermine the effort to estimate effects of discrimination. Of course, if one adopts a different conception of discrimination, one result may be that the problems that hound mainstream efforts may be resolved or rendered irrelevant. To begin to assess whether an ancillary benefit of a changed conception of discrimination is an effective

Defining Discrimination Effects: An Asocial Scientific Method / 147

response to these problems, we need reconsider Becker’s framework. How have social scientists conceived of discrimination after Becker, and to what extent does that conception draw on the strengths of Becker’s approach?

Defining Discrimination Effects: An Asocial Scientific Method Becker’s dissertation, published in a slim volume in 1957, virtually inaugurated economists’ study of discrimination. One could write volumes tracing the implications and controversies of this theory alone. However, the basic claims of the theory, and the important features of it for our purposes, may be succinctly described.

Becker’s Theory of a Taste for Discrimination: A Brief Introduction Becker’s initial insight was elegant and powerful. The foundational claim of Becker’s ([1957] 1971) germinal analysis is that those with a taste for discrimination must be willing to pay monetarily, either directly or indirectly, in order to associate with some kinds of persons instead of others. Becker’s analysis of employer discrimination illustrates the theory. Becker posits the existence of white capital and black capital. Capital owners determine which laborers they will allow to work with their capital. The capitalists will pay laborers a wage. Assume that black and white labor is equally productive. If so, the wage should be same. However, capitalists with a taste for discrimination will pay white workers x, but pay black workers x − d, where d is the discrimination coefficient; the larger d is, the greater the capitalist’s taste for discrimination. The reason discriminatory capitalists pay black workers x − d is that the discriminator incurs a cost to associate with blacks, and that cost is d. By subtracting that cost from the wage paid blacks, the net cost to the discriminator is the same regardless of the race of the worker. However, this coefficient implies that black workers will be paid less for the same work when employed by discriminatory capitalists. Though this coefficient is different for every employer, the coefficient also can be identified in the aggregate. The ability to obtain aggregate results allowed Becker to combine his theory and some verifiable and fairly tenable assumptions about the relative amounts of labor and capital in the hands of blacks and whites to estimate the impact of discrimination on the earnings of blacks. Becker estimates that, in the absence of discrimination, black per-capita income would be two thirds of white income. However, under a maximum discrimination regime, defined as occurring when blacks are prohibited from working with white

148 / Chapter 6

capital, black per-capita income would be 39 percent of white income. Becker also estimates that under no discrimination, white per-capita income would be slightly larger than under a regime of maximum racial discrimination. In 1957, black per-capita income was between the two extremes estimated, leading Becker to conclude that the loss “seems substantial and important, although a far cry from the loss assumed in some discussions” (29). The above analysis pertained to employers, but Becker goes beyond employer discrimination to delineate several other forms of discrimination— including employee discrimination, consumer discrimination, government discrimination, and market discrimination—by using a simplified division of the economy into sectors to identify actors who may be involved in discrimination.

Features of the Taste Theory of Discrimination There are several noteworthy features of Becker’s theory. First, in order to identify the discrimination coefficient, Becker does not need to determine employers’ motives. Because assessing motive can be notoriously difficult, Becker regards this ability to ignore employers’ motives to be an important advance in the study of discrimination. Another important feature of Becker’s analysis is that, based on his analysis, and because of the concentration of capital in the hands of whites, black and white incomes would not necessarily be equal even in the absence of discrimination. This specific result revealed a more general implication of his analysis: Inequality might exist even in the absence of discrimination. Further, the inequality could exist even if the group members are equally productive; for example, equally productive persons of different groups could have access to different amounts of capital, and this differential access could lead to inequalities in earnings or other economic outcomes. Moreover, Becker distinguishes between the amount of discrimination practiced and the effects of discrimination experienced. This is important in that the number of discriminators, their amounts of capital, and the size of their taste for discrimination are all relevant for the effects of discrimination on the livelihoods of the targets of discrimination. Becker’s framework implies that the effects of discrimination are different in a world in which large employers have a small taste for discrimination and small employers have a large taste for discrimination, compared to a world where large employers have a large taste for discrimination and small employers have a small taste for discrimination. Of course, this means that Becker distinguishes between the amount of discrimination—the number of discriminators and the total intensity of discrimination—and the effect of discrimination, the loss in wages owing to discrimination.

Defining Discrimination Effects: An Asocial Scientific Method / 149

In addition, Becker’s formulation implies the possibility that the targets of discrimination may not be the only ones to pay the monetary costs of discrimination. In the example Becker devises, owing to the concentration of capital in the hands of whites, the effect of discrimination on whites would not be large. Notably, however, white incomes also were reduced by discrimination against blacks ([1957] 1971: 27–29). Three conclusions clearly flow from Becker’s framework: (1) equality may not prevail even in the absence of discrimination, (2) targeted groups are not the only ones to pay costs of discrimination, and (3) the incidence of discrimination and the effect of discrimination are not the same. These conclusions are reached by erecting a method that ignores the motivation of actors, and focuses on isolated spheres within which action occurs. Becker’s formulation has remained an important theoretical approach to discrimination research (e.g., Badgett 1995; Cavalluzzo and Cavalluzzo 1998; Ragan and Tremblay 1988). The three conclusions Becker reaches surely are correct, and constitute major contributions to our understanding of discrimination. The critical legal perspective, however, raises serious questions as to the way taste theories of discrimination reach these powerful conclusions.

A Social Critique of Becker’s Framework Critical legal theories imply two problems with the taste theory of discrimination, and, coupled with a third problem, suggest the taste theory is beyond repair. First, taste theory attempts to identify the par ticu lar economic sources of economic discrimination. This effort is misguided, for it tends to truncate the range of discriminatory behavior, and turns attention toward individual actors acting alone in fundamentally independent roles. Second, the insightful claim at the center of the theory works for the economy, but severely limits the general application of the theory. And third, the framework uses a definition of discrimination that is hopelessly ambiguous.

A Truncated Range of Discrimination Recall that each employer has a discrimination coefficient. For nondiscriminators this coefficient is zero; for employers with a taste for discrimination the coefficient is greater than zero. Becker cannot measure employers’ tastes directly; instead, he uses their hiring behavior to estimate their taste for discrimination. This approach implies that a focus on the behavior of individual employers acting alone as employers is sufficient to identify the effect of discrimination on wages. It implies that each employer makes an independent decision

150 / Chapter 6

based on their own taste, irrespective of the wider environment in which their decisions are activated. It should be noted that Becker includes an analysis of consumer discrimination. In that analysis he points to how the taste for discrimination of nonemployers and nonemployees may lead a capitalist to behave in ways that are discriminatory. This feature of Becker’s analysis, however, creates the second problem mainstream studies of discrimination have encountered. Under the atomistic conception of discrimination, it makes perfect sense to isolate separate spheres within which discrimination may occur. However, because discrimination can occur anytime anywhere, an infinite number of loci must be studied before one may be confident about one’s estimate of the effect of discrimination. After every study, findings concerning the locus of focus for potential discrimination remain equivocal because, even if apparent discrimination effects are observed, alternative loci whose appropriate analysis might explain away the putative discrimination effects observed in the focal locus remain virtually as numerous as before. This predicament is unavoidable once one takes the first step down the atomistic path. Becker took that step fifty years ago, and one implication of that step is revealed by considering the treatment of noneconomic roles in discrimination. When Becker ignores noncustomer nonemployees of a firm or industry, he ignores the vast social web in which all social action occurs. The employer who lacked a taste for discrimination, employed only those who also had no taste for discrimination, and had only customers who lacked a taste for discrimination would still live in a world with friends, relatives, fellow church members, neighbors, and so on—in other words, such an employer still would be embedded in a rich set of connections, intimate and distant, that constitute the social world. When analysts ignore that web, as they do when they follow the atomistic logic, they essentially search for the particular node or nodes of the web that matter, as if each node can be understood as autonomous, as if all that matters is confined to the nodes. However, searching for the particular node or nodes that matter is unnecessary if one conceives of the phenomenon not as occurring at a subset of nodes but, instead, as emanating from a social web. In such a conception, the web that connects the nodes has its own characteristics that matter, and studying those characteristics becomes essential to any successful effort to understand the phenomena of interest. And, of course, webs, social connections, require maintenance, and maintenance may mean that employers (nodes) and others with no taste for discrimination still behave in a discriminatory manner. Thus, wider social connections potentially play a role in what persons will do. This is the first and perhaps easiest to recognize sense in which the taste theory has problems, problems with major negative implications for discerning discrimination effects.

Defining Discrimination Effects: An Asocial Scientific Method / 151

A taste theorist could of course argue that the determinants of employers’ taste for discrimination are exogenous, and the taste theory of discrimination takes tastes as given. Though this would be a correct response to the criticism, it also would be a response that basically accepts the validity of the criticism without recognizing the damning power of the critical point. For to accept that a factor in the model is determined exogenously is also to accept that there are determinants of the exogenous factor (i.e., it is not a randomly produced factor). Once we accept that tastes are exogenous, then we must accept that there are determinants of taste; once we accept that proposition, then the door is opened to the vast set of other relations that affect persons’ behavior. And once that door is opened, it becomes necessary to look through it to the wide array of norms, values, standard operating procedures, coercive and punitive actions, historic allocations of differential power, and takenfor-granted assumptions that buttress and sustain discrimination. Once the larger context truly is recognized, it becomes difficult to sustain the claim that one can estimate effects of discrimination on wages by considering individual employers acting alone, or even aggregating across individual employers as such. What is true of wages is true of other outcomes as well; one cannot estimate the effects of discrimination on an outcome by focusing only on the proximate actors involved in the production of the outcome. The taste theory truncates the range of discrimination by focusing on the economic behavior of actors as employers, employees, consumers, and the like, and attempting to estimate the effects of discrimination that can be traced to these roles. There is an interesting similarity between this approach and that of the legal system. Both Becker and the courts focus on discrete acts (e.g., the hiring of a person to fill a position, the offer for pay), and both attempt to isolate the actors; Becker focuses on the role of the actor (e.g., employer, employee), whereas the legal system attempts to identify the actor by name. Though persons are never only employers and employees, Becker’s approach makes no room for these other roles in the analysis. The courts also make no room for an employer’s action as a parent, church member, voter, consumer, and so on in adjudicating an accusation of discrimination. While the courts may need to narrow the focus to specific identifiable acts of persons occupying par ticular legally recognized roles, social science need not do so—it is the rare sphere of social science where such microscopic focus pertains. Moreover, it is clear persons’ discriminatory behavior is not confi ned necessarily to the economic realm; persons are employers and also siblings, spouses, cousins, church members, drivers, witnesses, customers, voters, taxpayers, friends, neighbors, landlords, and more. Even if we accept that taste is exogenous to the economy, as persons act within these other spheres their taste is formed, acted upon, and acted on, just as the taste of others is as

152 / Chapter 6

well. Seen in this way, the taste theory essentially attempts to identify the effect of discrimination traced to a set of employers qua employers, while ignoring the effect of discrimination produced by those same people when they act in other capacities, say as voters, neighbors, or witnesses. It is this feature that makes the theory at least as narrow in its focus as that of the dominant legal perspective. In sum, the first problem critical scholars imply can be easily conveyed. Becker claims that the maximum amount of economic discrimination occurs when blacks (or some other group facing discrimination) are not allowed to work with white (or some other dominant group’s) capital. Yet, it should be uncontroversial and readily apparent that far more brutal forms of discrimination exist, including violent coercive physical control (e.g., slavery, rape), capricious application of civil protections (e.g., lynching, unprosecuted domestic violence), and policies of mass murder (e.g., genocide, infanticide). Though Becker certainly is correct that these are not purely economic forms of discrimination, it also is true that the economy is not merely a bystander to these more brutal forms of discrimination. For example, before being murdered the Jews of Nazi Germany were the enslaved laborers of German industry, and their confiscated property supplied the economy with both raw material and finished goods. Thus, while acknowledging that it is impossible to ever pay sufficient damages to fully compensate for the stain of the Holocaust, recognize that damages can be part of a process of societal repair. That recognition makes sense of ongoing efforts to seek redress for Holocaust-era atrocities, regardless of how long it has been since the atrocities occurred (e.g., Bazyler 2003; Menzel v. List (49 Misc. 2d 300, 267 N.Y.S. 2d 804 (S. Ct. 1966)); Bodner v. Banque Paribas (114 F. Supp. 2d 117 (EDNY 2000))). Thus, it should be clear from the content of the cases seeking damages that the economy was implicated in the Holocaust. The identification of banks, manufacturers, insurance firms, art houses, and more as defendants in the cases reveals the interconnected nature of economic and noneconomic spheres. The example of the Holocaust should teach us: Any conception of discrimination that fails to recognize the unavoidable commingling of economic and noneconomic aspects in the phenomenon has to be understood to be at best a distorted rendering of the phenomenon. Thus, while Becker may be correct to claim that denying one group of laborers access to the capital of another group of persons is a most severe form of purely economic discrimination, it is to completely misunderstand discrimination and misestimate its effects to ignore the role of the economy in other more brutal forms of discrimination, and to ignore the role of other more brutal forms of discrimination in the economy. Indeed, without economic support, those more brutal forms of discrimination could never occur; the trains could not have moved Holocaust victims to the death camps

Defining Discrimination Effects: An Asocial Scientific Method / 153

had the conductors of the trains, or at least someone in their stead, not been paid. Hence, the truncated range of discriminatory behavior is a serious problem for the taste theory of discrimination.

A Theory with Limited Generality The taste theory focuses on the economy, and I contend that a narrow focus on the economy is a problem. One solution would be to extend the taste theory to other spheres. However, the second problem with the theory is that it cannot be extended to noneconomic spheres. The claim that economic discriminators behave as if they are willing to forego income is insightful. However, one may ask, what is it that discriminators in other arenas must be willing to forego as the price of discrimination? Without an answer to that question, one cannot use Becker’s approach to calibrate the effect of, say, voting-rights restrictions on the well-being of targets and nontargets of discrimination. Becker downplays the importance of noneconomic factors in discrimination, at least in the case of the United States, though it is curious that he accepts the possible importance of a noneconomic factor (governmental action) against nonwhites in 1950s’ South Africa, but on the previous page asserts that such noneconomic forms of discrimination are secondary in the United States (30–31). At any rate, others have argued persuasively that noneconomic forms of discrimination, such as voting restrictions, are important because they may hinder persons from altering their economic situation (e.g., Thurow 1969). Thus, it is apparent that more work is needed before Becker’s approach can be used to estimate the role of noneconomic forms of discrimination. At present, the framework remains analytically ungeneralizable. Given that the focus on the economy has been shown above to likely misestimate the effects of discrimination, and given the inability to extend the theory to estimate discrimination in other spheres, the approach seems irreparable.

Ambiguous Definition of Discrimination Critical scholarship suggests that a third problem with Becker’s framework is that it provides a circular definition of discrimination. Under Becker’s definition, a par ticu lar result is labeled discrimination because the result resembles what he would expect to be the result of a discriminatory behavior. For example, the gap in wages between coworkers reflects the discrimination coefficient because he expects a discriminator to pay targets of discrimination less to compensate for the distaste of employing them. However, Becker acknowledges that the observed wage gap could reflect very different social processes. For example, some workers may be related to the employer, and

154 / Chapter 6

the other workers may not be related; the employer may pay relatives more out of “positive” discrimination (i.e., nepotism). Given that we never obtain independent evidence that the reason for the wage difference is “negative” discrimination, we cannot adjudicate between the two explanations. Becker defines the gap as discrimination, and argues that the inability to distinguish between discrimination and nepotism usually is irrelevant because empirically it does not matter whether the gap is the result of positive or negative discrimination (Becker [1957] 1971: 160). Becker is incorrect; with a better theory on discrimination, it would be possible to disentangle “positive” discrimination and negative discrimination. And, for understanding the social world, as opposed to understanding economic trade abstracted from its social context, the distinction is of paramount importance. The inability of Becker’s framework to distinguish between positive and negative discrimination has a telling implication. Becker’s framework cannot distinguish between mate selection and employment discrimination. Both marriage and employment discrimination require one to decide with whom one wants to associate, both decisions may entail the loss of some potential income for those not selected (e.g., the tangible gifts a wealthier spouse might bestow), and both may lessen the standing of the spurned. In essence, Becker does not see the inability of his framework to distinguish mate selection and employment discrimination as a problem. Alternatively, one may regard the indistinguishability of discrimination and mate selection as a sign of a severely flawed framework. The indistinguishability of discrimination and nepotism in Becker’s theory flows directly from the reluctance to assess persons’ motivations. Once motivations are ruled off limits, it becomes impossible to trace a difference in treatment to any set of reasons. For the targets of discrimination, this is an unfortunate state of affairs, for it will mean that any difference that could be traced to discrimination if the effort were made will be less secure because the effort is not made, and other unobserved explanations are always possible alternative reasons for any observed difference. Interestingly, this is exactly what has occurred in the empirical literature; analysts find differences between groups, and other analysts question whether the difference is owing to discrimination or owing to some unobserved alternative reason. We shall return to this tendency in the empirical literature, but one reason for this tendency is the foundational confusion as to how to define discrimination at the center of the taste theory of discrimination. Though Becker’s Economics of Discrimination purports to define a discrimination coefficient d that reflects employers’ taste for discrimination, it really defines difference coefficient d* that reflects employers’ taste for paying equally productive workers unequally. The motivation for the wage differ-

Defining Discrimination Effects: An Asocial Scientific Method / 155

ences remain unknown and unknowable within the boundaries set by the theory. It should be noted that Becker proposes a general theory, but attends most directly to the case of blacks and whites. And, at the time of his research, analysts may have been quite willing to regard differences between the wages of blacks and whites as purely or largely the result of an employers’ taste for discrimination. However, in an era of contested prejudice, and in an era of much more developed understanding of the social world, analysts are unwilling to accept such attributions uncritically. And, it should be noted that, even in 1957, many analysts may have been quite unwilling to accept that the gender wage gap was based purely in discrimination. Regardless of the historical context of the work, it is clear that few analysts now will accept the assertion at the center of Becker’s framework. The coefficient d is unobserved; all we can obtain is coefficient d*, which marks a difference, but the basis for the difference remains unknown and hotly contested. Ignorance as to the cause of difference ultimately must translate into ambiguity about the effect of discrimination, for eventually scholars will be unwilling to accept assertions about unobserved phenomena. However, to the critical scholar and others seeking a nontautological approach to estimating discrimination effects, theories erected on ignorance as to the cause of differences between targets and nontargets of discrimination are necessarily fundamentally erroneous theories.

A Cumulative Critical Legal Theory Critique of Becker’s Framework Becker’s theoretical analysis of discrimination is five decades old. Analytic economists have long since moved beyond his approach. Yet, Becker remains important because, implicitly, social scientists have accepted frameworks for studying discrimination effects that defi ne discrimination as problematically as Becker does. As Becker’s taste theory of discrimination provides an articulate, coherent, foundational exposition, his work serves as an important touchstone. And despite the research of the analytic economists, some aspects of Becker’s work, such as the focus on narrow spheres of activity of isolated individuals, remains important. A quick read of a broad spectrum of economics and sociology will find several debates concerning whether the gap between groups is, for example, owing to prelabor market discrimination or labor market discrimination, often treated as a debate as to whether it is employers who are discriminating or someone else (e.g., Farkas and Vicknair 1996; Maume, Cancio, and Evans 1996). But, Freeman’s response is that, from the point of view of the targets of discrimination, it does not matter who is discriminating; employers as

156 / Chapter 6

employers, or employers as voters, neighbors, or any other social role a person may occupy—stolen opportunities are lost opportunities, regardless of the social role the thief occupied during the theft. These observations lead to the conclusion that to the extent social science has drawn on particularly problematic aspects of Becker’s approach, social science has adopted a perspective on discrimination almost as narrow as that of the legal system. Where the court seeks to identify perpetrators by name, social scientists attempt to separate out types of discrimination, such as labor market discrimination, credit market discrimination, housing market discrimination, and so on by attending to the role the discriminator occupied while engaging in the discriminatory act. Even though this strategy seems social, it makes sense only under a highly individualized perspective on discrimination. The individualized, atomistic, perspective sees discrimination as only the act of an employer, realtor, teacher, or doctor; thus, discrimination occurs in separate instances of discriminatory behavior of identifiable actors, and therefore in principle it is possible to sensibly discuss discrimination as if the phenomenon is confined to par ticular spheres. But, the possibility and historic fact of slavery, domestic violence, and genocide should tell us that discrimination is never confi ned to par ticular spheres. These spheres always interact in complex ways to form solidified kinds of social relations. A more effective analysis of discrimination effects will attend to those social relations.

Beyond Becker but Not Critical Legal Studies: Analytic Economists’ Response to Becker’s Approach Analytic economists have engaged Becker’s approach, and offered several alterations and alternatives. One motivation for continued probing of Becker’s framework was the apparently slow pace of an expected decline in discrimination. Some critics reason that Becker’s theory implies that discrimination will, like the state in some Marxist analysis, wither away. They argue that because employers differ in their taste for discrimination, discriminators will have an extra element in the production process (discrimination) that does not increase output, and thus will have either higher costs or lower profits. Thus, moderately discriminatory firms will hire targets of discrimination, paying them more than strongly discriminatory firms would, but still less than the wage the targets should earn owing to their productivity. Because moderately discriminatory fi rms have less of a drag on productivity compared to strongly discriminatory fi rms, moderately discriminatory firms should accumulate surplus capital to expand at a faster

Defining Discrimination Effects: An Asocial Scientific Method / 157

rate, and, therefore, should expand faster than strongly discriminatory firms, eventually driving such firms to either reduce their discrimination coefficients or move toward bankruptcy. The process should continue, as moderately discriminatory firms face disadvantages in comparison to weakly discriminatory firms. As the process continues, moderately discriminatory firms also should be forced to reduce their discrimination coefficients or declare bankruptcy. Eventually, only nondiscriminatory firms should remain. Note that in this reading, the taste theory implies long-run change in the aggregate, as discriminating employers lose out in market competition, not necessarily change in the behavior, attitude, or taste of any individuals. Hence, no one’s taste need change in this formulation; instead, access to resources (i.e., capital) to act on tastes may change, as discriminating firms lose out in the market competition. Becker (1996) maintains that the taste theory of discrimination he articulated in 1957 did not imply that discrimination would ebb away. He argues that the amount of discrimination experienced will depend on many factors, including the distribution of discrimination coefficients and the production functions of the various firms. Whether Becker or the critics are correct, more recent neo-taste theories of discrimination, which maintain many features of Becker’s approach, address this issue. Other theories, motivated in part by the apparent failure of discrimination to disappear, have suggested entirely different mechanisms that may produce discriminatory behavior.

Neo-Taste Theories of Discrimination? Regardless of whether the original theory implies discrimination should fade away, scholars’ view that it did have that implication motivated their efforts to resolve what appeared to be a discrepancy between theory and reality. Matthew S. Goldberg (1982) supplies one such effort by restating Becker’s theory in terms of a preference for white workers. Goldberg shows that a preference for white workers, what we might call “racial nepotism,” is not the same as antipathy for black workers. One of the implications of the nepotism model is that non-nepotistic firms generally cannot buy out nepotistic firms, and nepotistic firms generally cannot buy out non-nepotistic firms. Hence, Goldberg’s taste theory predicts the long-run intransigence of nepotism without relaxing the assumptions of perfect information or competition. George J. Borjas and Stephen G. Bronas (1989) present an equilibrium search model that relaxes some of the assumptions of Becker’s theory. Their model is distinguished by its relaxation of the “complete information” assumption. On the basis of their theoretical analysis of consumer discrimination, they make several predictions: (1) high-ability whites will be more likely to enter

158 / Chapter 6

self-employment, (2) high-ability blacks will be more likely to enter salaried employment, (3) self-employed minorities will have a compressed earnings distribution relative to whites, and more. Their empirical analysis shows evidence consistent with the expectations of the theory, suggesting that the model may accurately depict consumer discrimination and its impact on self-employment. These predictions are intriguing, and the empirical evidence has major implications for our understanding of selection into types of employment. These approaches within the taste-theory framework offer insights one must consider seriously. However, note that these efforts still adopt the atomistic understanding of discrimination, in which discrimination effects are carved into distinct pieces owing to the social location of the specific discriminator. Further, because the theories seem to stipulate the existence of whites who incur some extra cost to interacting with nonwhites, or some extra benefit to interacting with whites, as sufficient to identify discrimination effects on economic outcomes, they follow Becker’s lead in suggesting a minimal or nonex istent role for extraeconomic factors, implicitly truncating the range of discrimination. Hence, the basic problems identified by the critical legal perspectives on discrimination remain, problems that culminate in an inability to develop defensible estimates of discrimination effects.

Statistical Discrimination Kenneth J. Arrow (1973) and Dennis J. Aigner and Glen G. Cain (1977) also relax the “complete information” assumption to provide an alternative understanding of discrimination. In their formulation information gathering is costly, and thus employers may lack the resources to evaluate each potential worker individually. Therefore, employers may use easily observed characteristics, such as race or gender, to indicate the expected productivity of different workers. One can consider at least two scenarios. In one scenario, employers make an incorrect estimate of differences in group-specific average productivity, and proceed to hire on the basis of their assessment. The improvement of scenario one over taste discrimination is difficult to discern, because scenario one begs the question of how it could be that incorrect assessments of group-specific averages would remain uncorrected as employers observe actual employees in action and eventually learn of the error in the estimate. We would expect incorrect assessments of group-specific productivity to be corrected over time. In order to address why incorrect assessments are not corrected over time, it would seem one would need to consider extraeconomic factors, such as community segregation, race-segregated friendship networks, ingrained stereotypes, and other factors that could impede information flow

Defining Discrimination Effects: An Asocial Scientific Method / 159

across groups. As these kinds of noneconomic factors are ignored under scenario one, and also by taste theories of discrimination, scenario one encounters problems similar to those encountered by taste theories of discrimination. The second scenario is more clearly distinct from taste theories of discrimination, for it asserts that assessments may be accurate given the limits of the information to which the employer may have access. There are many ways to set up the scenario; in one formulation, a testing instrument may be less reliable for one group than for another. Thus, the employer discounts the information for members of the group with less reliable test results. In response to the lower quality of information, the employer lowers the wage he or she is willing to pay a member of that group; essentially, the employer is penalizing any worker of the targeted group owing to the employer’s accurate understanding of the average reliability of the measuring instrument for members of that group. Another way to motivate the scenario is to posit risk-averse employers. To the extent a test predicting future productivity is less reliable for one group, risk-averse employers will be less likely to employ members of that group, or will be likely to pay them less to compensate for the risk the employer takes in hiring them. When statistical discrimination is described in terms of a screening mechanism it resembles disparate impact discrimination in some respects. Though the connection is not exact, here again we see an affi nity between dominant legal reasoning and social scientific frameworks for understanding discrimination. However, scenario two essentially brackets the social processes that instigate the development of differential predictions of productivity and that maintain their use. Scenario two begs the question of why indicators with low reliability for some groups remain in use, while low reliability indicators for other groups become an object of sufficient concern that other indicators are selected. As the employers select the means of evaluation, they are implicated at a more fundamental level in the maintenance of the conditions that allow statistical discrimination to exist. Seen in this way, statistical discrimination only pushes the question back in time; perhaps on the day of the interview statistical discrimination is determinative, but before that day the employer selected a standard operating procedure. Employers generally will not accept means of evaluation at variance with their understanding of the world, yet, under statistical discrimination, they are willing to accept means of evaluation that are unreliable for the subordinate members of society. This suggests that employers may have views of subordinate members of society that allow them to accept rather than reject the relatively unreliable indicators; such views would seem to go beyond the phenomenon available for

160 / Chapter 6

inspection under the perspective of statistical discrimination. Thus, statistical discrimination sees only the final part of the process and thus misses key social dimensions of discrimination. To be sure, the concept and possibility of statistical discrimination is important. It relaxes the assumption of perfect information costlessly obtained. Like Goldberg (1982), statistical discrimination theory implies that discrimination need never end. Hence, it would be a mistake to ignore this important alteration in the theoretical field. Despite the importance of the development of statistical discrimination as a theory, statistical discrimination responds no more effectively to the critiques that flow from critical legal perspectives than do taste theories of discrimination. Statistical discrimination theories implicitly truncate the range within which discrimination occurs by leaving processes that would explain the continuation of incorrect assessments of productivity unanalyzed. Both scenarios remain atomistic, bracketing the social processes that instigate the development of differential predictions of productivity in the first place. Further, statistical discrimination, like many taste theories of discrimination, uses a circular definition of discrimination; the phenomenon is identified by its likely effects. Though motives are referenced by the theory, the motives concern attitudes toward risk, not discriminatory intent. The theory begs the questions of why businesspersons who regarded signals from applicants as differentially reliable would rest with such differentially reliable signals, and whether they would do so if the less-reliable signal pertained to the superordinate rather than the subordinate members of society. Thus, in light of the critical legal perspective’s critique of the law, nothing of consequence for the empirical investigation of discrimination effects appears to change in moving from taste to statistical theories of discrimination.

Marxist Theories of Discrimination Social scientists usually have not turned to Marxist theories to understand discrimination. But, Marxist theories are oft-regarded as at odds with mainstream theories. Do Marxist theories resolve the problems implied by the critical legal critique of dominant legal reasoning? With respect to whether the effects of discrimination are confined to the ostensible targets of discrimination, Marxists have long claimed that discrimination is a means of dividing and conquering the working class (e.g., Reich 1978). Marxists contend that discrimination has negative effects on each member of the working class, regardless of whether a specific member is part of a targeted group or not. Thus, an important difference between Becker and the Marxists is that Becker claims that in the case of antiblack discrimination, white capital

Defining Discrimination Effects: An Asocial Scientific Method / 161

loses, white workers gain, black workers lose, and black capital gains. In contrast, Marxists claim that white workers lose and white capital gains by antiblack discrimination (e.g., Reich 1978). Some empirical research has been conducted to assess whether race discrimination is associated with lower earnings for some whites (e.g., Semyonov and Cohen 1990), but, owing to problems with the measurement of discrimination, findings remain unclear. Regardless of the findings, however, the theory adopts the same atomistic approach, carving discrimination into specific types just as taste theorists do. Marxists still seem to see discrimination as occurring in par ticu lar spheres, rather than as transcending the sphere in which the outcome is observed. Thus, despite the substantively different predictions, the epistemological basis of the taste theory of discrimination and the Marxist theory of discrimination are for all intents and purposes the same.

A Word about Collective Models A few economists have questioned the sufficiency of mainstream analyses of discrimination. For example, Lester C. Thurow (1969) and Glenn C. Loury (1981; 1998) in various ways point toward social capital, the embedding of persons in networks through which norms and expectations flow. Thurow sees social capital as maintaining discrimination, whereas Loury (1998) contends that social capital may maintain racial inequality long after marketbased race discrimination declines in importance. There are some affinities with my approach and the discussions of social capital. More recently, however, Loury (2002) has developed an intriguing position on racial inequality built on the concept of racial stigma. Drawing on resources from Erving Goffman as well as research on cognition, Loury argues that racial subordination has taken on a self-reinforcing character that must be interrupted if ameliorative social change is to occur. It is not possible at this juncture to delve into the similarities and differences between my perspective and Loury’s, for I have yet to fully present my view. However, what has gone before does provide some resources to make three important observations. First, our concerns differ. Loury is concerned with racial inequality, whatever its basis; I am concerned with discrimination, whatever its basis. This distinction means that I seek to produce general resources for studying discrimination, not resources targeted only to the experience of African Americans. This has implications for how we proceed; for example, though stigma may be important for understanding much of the African American experience in the United States, it may not be accurate to refer to gender stigma to understand gender inequality in the United States.

162 / Chapter 6

Second, the emphasis on stigma is promising. However, my emphasis differs; I emphasize material structures even as I expect those structures to have some cognitive connections. Neither Loury nor I fully explore the connection between the material and the cognitive, but my preliminary sense is that such an exploration would work fertile ground. Yet, and third, Loury maintains that when historically oppressed, stigmatized groups mobilize and develop solidarity, to others their actions will resemble discrimination. He then goes on to contend that such actions are discrimination. I respectfully disagree; there is no easy isomorphism between the actions of the dominant in constructing and maintaining the structures of discrimination and the actions of the subordinate to struggle against their predicament. Thus, there are some affinities between my view and that of Loury and others who adopt a more collective understanding of discrimination. Yet they remain mere affinities, in part because much of the theorizing that has occurred to date has not broken sufficiently with the foundational commitments of mainstream discrimination research to atomistic analyses of distinct spheres, an approach greatly resembling the dominant legal approach to discrimination. As those commitments have failed to produce even one consensus estimate of discrimination effects in the United States over the last five decades, I believe that until that foundation is rejected we will fail to effectively estimate the effects of discrimination.

Beyond Taste Theories of Discrimination: Concluding Remarks Analytic economists constructed effective responses to Becker’s framework. However, despite many important modifications, the theories in play in the debate concerning discrimination leave the atomistic ethos at the center of Becker’s formulation untouched. The many efforts to theorize discrimination actually illustrate the ubiquitous presence of central atomistic assumptions. Even Marxist formulations, self-avowedly critical of mainstream approaches, fall prey to the atomistic assumptions at the core of mainstream analyses of discrimination. Though some economists are looking in the direction of nonmarket arenas as having import for discrimination, those perspectives do not reject the foundational assumptions of mainstream discrimination research. Even so, consider that after dividing economic research on discrimination into mainstream approaches of taste and statistical discrimination on the one hand, and collective models that assume collusion of groups through various legal or illegal acts on the other, Joseph G. Altonji and Rebecca M. Blank conclude that “Almost all of the theoretical work by economists has been within a

Defining Discrimination Effects: An Asocial Scientific Method / 163

competitive framework,” as opposed to the collective framework (1999: 3168). Thus, it appears that the implications of efforts, such as those of Thurow and Loury, accepting of foundational mainstream assumptions though they may be, have yet to significantly influence the mainstream understanding of discrimination in the social sciences in general, or economics in par ticular. Throughout, a strategy of identifying the locations of discrimination in what reduces to an atomistic manner has served to focus the empirical investigation of discrimination for at least half a century. What has been the result of following this strategy for the social scientific estimation of the effects of discrimination?

Finding Discrimination Effects in the Dark: The Empirical and Epistemological Legacy of Becker’s Analysis In analyzing discrimination, Becker relies on empirical data to some degree, but mostly he traces out the implications of different theoretical relations within the economy. Theoreticians have continued to refine the economic theory of discrimination. But empirical researchers also have drawn on the theory, often only informally, in their efforts to analyze empirical data bearing on the effects of discrimination. It is the work of the empirical social scientists that has documented the facts of race and gender inequality. Three incredibly useful insights flow from Becker’s work: (1) the recognition that, even in the absence of discrimination, equality between groups may not prevail; (2) the recognition that the targeted groups are not the only ones to pay costs of discrimination; and (3) the recognition that the amount of discrimination and the experience of discrimination are not the same. Notably, not one of these three insights has had a sustained impact on empirical analyses of discrimination. Instead, empirical analysts have focused on and sought to preserve two of Becker’s contributions that are of great value under a narrowly atomistic perspective on discrimination—namely, the ability to ignore the motivations of actors, and the ability to trace out the fine details that create discrimination effects, even to the point of using methods whose logic and operations contradict all three of Becker’s conclusions. All contemporary efforts to estimate the effect of discrimination are variations of three strategies, two of which—residual attribution and audit methodology—purportedly allow one to find effects of discrimination without direct observation of discrimination. A third strategy is to measure exposure to discrimination. All three strategies easily lend themselves to isolating the effect of narrowly defined types of discrimination, such as labor market discrimination, credit market discrimination, and the like.

164 / Chapter 6

Before continuing, we must note that all research proceeds from a ground of accepted facts, maintained assumptions, and theoretical commitments that foreclose some avenues of inquiry while opening others. The best research occurs when its accepted facts, assumptions, and commitments breathe life into the research operations and methods. Thus, in conducting the best research possible, analysts have used the three strategies listed above in ways consistent with their theoretical understanding of the phenomenon of interest. Consequently, analysts working within an atomistic perspective have used these strategies in ways consistent with the assumptions of an atomistic perspective. Given an atomistic perspective, good research will (1) ignore the motivations of actors (because these are virtually impossible to measure), and (2) focus on par ticular isolatable spheres of discrimination à la Becker (because discrimination is essentially the isolatable acts of individual actors). Researchers have used all three strategies in an effort to identify the effects of discrimination, given a definition of discrimination that is fundamentally atomistic. As we shall see, despite these efforts, consensus estimates of the effects of discrimination remain elusive.

Residual Attribution Residual attribution is the practice of asserting that the difference between groups is the result of discrimination. One virtue of residual attribution is that it may easily be used to investigate a wide variety of outcomes. If one wants to know the effect of sex discrimination on law school admissions, compare the proportion of men and the proportion of women admitted to law schools. Under residual attribution, the difference is the discrimination effect. A second virtue of residual attribution is that analysts do not have to determine whether disparate outcomes occurred by virtue of disparate treatment that was motivated by a person’s desire to discriminate. Instead, analysts need only find a race or gender difference to infer the effect of discrimination. Note that residual attribution is different from Becker’s approach in important ways. Residual attribution asserts that equality would prevail absent discrimination, denies that nontargets could be hurt by discrimination, and fuses the amount and effect of discrimination into one parameter. Hence, residual attribution denies all three of Becker’s conclusions about discrimination effects. Notably, Becker identifies the effect of discrimination using methods that avoid using the discriminators as an implicit standard of performance, whereas common residual attribution analyses treat the gap between dominant and subordinate groups as the effect of discrimination.

Defining Discrimination Effects: An Asocial Scientific Method / 165

Despite all these differences, one affinity between the approaches is that both attempt to avoid attending to the motivations of actors. Several analysts have noted that one glaring weakness of residual attribution is that it forces one to assign all of the difference to one asserted and unmeasured factor (e.g., Killingsworth 1985; Pettigrew 1973). The obvious solution for this problem is to remove the effects of as many other factors as possible, and to regard only the net difference between groups as the effect of discrimination. So, for example, one might compare only students who have the same college grades and Law School Admission Test (LSAT) scores; setting aside the critique of testing for a moment, such controls can be introduced statistically, and the remaining difference between men and women can then be regarded as the effect of discrimination. A statistical technique often used for these analyses in the labor market literature are Duncan decompositions (e.g., Duncan 1968), often called Oaxaca (1973) decompositions. Of course, owing to the problems with tests such as the LSAT described in Chapter 5, this approach might easily misestimate the differences between the groups. Any given residual attribution may become more believable as more and more possibly confounding factors are removed statistically, but, then again, some analysts contend that statistical control can obscure more than it reveals (e.g., Lieberson 1985: 120–151). However, even if one were to believe that adding more explanatory factors may move one closer to an estimate of the effect of discrimination, doing so never provides a clear estimate of the discrimination effect because an unknown number of additional factors with unknown magnitudes of effects always remain. A more subtle weakness of residual attribution is that it becomes problematic for outcomes in which blacks and women are advantaged. In contrast to conventional wisdom, blacks and women are not disadvantaged on every outcome. Women have maintained a gross advantage in infant mortality and life expectancy for decades (e.g., Nathanson 1984); blacks have experienced a net advantage in high school graduation for over twenty years (e.g., Lucas 1996; Bauman 1998). If residual attribution is used to infer the effect of discrimination for outcomes in which blacks and women are disadvantaged, then there is no good reason to stop doing so for outcomes in which blacks and women are advantaged. However, that would imply that analysts would have to infer that whites paid the cost of (antiwhite? antiblack?) discrimination in lower rates of high school graduation compared to blacks, and men paid the cost of (antimale? antifemale?) discrimination in their lower life expectancies compared to women. A still more subtle problem with residual attribution is that it cannot be used to analyze the role of discrimination in maintaining equality. If two groups are equal on some outcome, the analyst committed to residual attribution

166 / Chapter 6

must infer that discrimination has no effect on the outcome. Yet, it may be that one group attains its achievements while struggling against discrimination, while the other does as well as it does by virtue of receiving benefits of discrimination against the first group. It would be incorrect, under those circumstances, to infer that discrimination does not matter. But, under residual attribution, one has no other recourse. There are several political implications of these subtle problems as well. One implication is that, under the reasoning inherent in residual attribution, black and female success must signify the irrelevance of discrimination. Under the residual attribution strategy, once blacks and women equal whites and men, one should withdraw aggressive ameliorative action, because obviously discrimination must no longer matter. Thus, residual attribution reasoning may undercut incipient black and female success by providing the “evidence” that “proves” that ameliorative policies are no longer needed. What is even more problematic is that persistent inequality has two interpretations. One may view persistent inequality either as evidence of continuing discrimination or as evidence of the uselessness of ameliorative action. As some argue for even stronger ameliorative measures to make up the remaining gap, others can argue that the gap “proves” the inferiority of the group under discussion. Under residual attribution reasoning, there is no way to adjudicate these causal claims, because anyone may attribute the residual to any factor they so desire. Finally, and most important, note that the residual attribution strategy implies that the dominant group sets the standard of performance. This assumption of standard-setting performance on the part of the dominant group is required because under residual attribution the effect of discrimination is estimated to be the difference between the performance of dominant and subordinate groups. In actuality, subordinate groups might, absent discrimination, outperform dominant groups or, absent discrimination, both groups might exceed the current performance of the dominant group. But residual attribution reasoning implies that this is not possible, necessarily misestimating the effect of discrimination to the extent that this reasoning is incorrect. Notably, Becker did not make this assumption, but residual attribution analyses have ignored this strength of Becker’s taste theory of discrimination. The larger implication of this latter observation is that residual attribution strategies inadvertently erect a limiting view of persons at the center of the effort to discover and remedy the effects of discrimination. To some, such a view of persons might itself be regarded as at least (perhaps inadvertently) discriminatory, if not racist or sexist. Hence, either for political or epistemological reasons, residual attribution must be considered a faulty approach. Social scientists have not generally attended to the view of persons at the center of residual attribution reasoning, but they have come to agree that the

Defining Discrimination Effects: An Asocial Scientific Method / 167

residual attribution strategy is indefensible. The logic leading to this rejection of residual attribution reasoning is that so many unobserved factors could explain an observed difference, that it is indefensible to assert that unmeasured discrimination is the basis for the difference. In essence, then, these critics imply that if we could measure every factor that does matter and introduce all relevant factors except discrimination into an appropriate statistical model, residual attribution would work. These critics contend, however, that in practice we are unable to make such measurements. Thus, residual attribution approaches are indefensible. Notwithstanding the correctness of this critique, it ignores that the very focus on differences is inappropriate, for using differences to identify discrimination effects implies that dominant groups set the pace, and no other group or conditions could produce better performance. Unless one believes that discriminators and beneficiary bystanders necessarily set the standard of performance under conditions of discrimination, the residual attribution strategy is flawed in principle, not simply in practice. This latter point implies that evidence on inequality is irrelevant to the discussion of discrimination effects, though such evidence could be relevant to a discussion of egalitarian policies and to public debate on how to allocate resources in society. Unfortunately, many social commentators on many sides of important issues concerning the status of blacks and women have not highlighted this fallacy of residual attribution. Thus, much of what is offered as evidence in discussions of the importance or unimportance of discrimination is based on residual attribution reasoning. Hence earlier, when I claimed that the vast majority of evidence used to ascertain the effect of antiblack discrimination on the success of blacks and the effect of antiwoman discrimination on the success of women is, in a word, irrelevant, I was not exaggerating. The vast majority of what is accepted as evidence in public discussion is, indeed, irrelevant, because it is based on residual attribution reasoning. And there are many good grounds for rejecting residual attribution reasoning—most notably, the implicit limiting view of persons at the core of the approach. Even so, it is of course important to study possible differences in the experience of different sexes, racial/ethnic groups, and more (e.g., Lucas and Gamoran 2002; Lucas and Berends 2007), even though such studies cannot be used to infer effects of discrimination.

Audit Methodology A more promising approach for estimating the effect of discrimination also allows one to ignore motivation, and also is based in the individualized perspective on discrimination. In an audit study, two persons (testers) of different races or sexes are provided with similar profiles (e.g., résumés or housing

168 / Chapter 6

needs), and sent sequentially to the same establishments (e.g., employers or real estate agents). Some mathematical function of the difference between the outcomes the auditors experience is then regarded as the effect of discrimination. An advantage of the audit study is that one need not attend to the motivations of real estate agents and employers in order to infer a discrimination effect. And, the method is potentially effective for identifying how much is lost owing to discrimination in certain spheres. For example, the audit methodology was originally used to document housing discrimination (Wienk, Reid, Simonson, and Eggers 1979). In his study of the Boston housing market, John Yinger (1986) finds that, on average, 40 percent more rental options and 36 percent more sales options were mentioned to whites in comparison to blacks. Not only were more possibilities mentioned, but also whites were invited to inspect 57 percent more rental units and 34 percent more sales units than were blacks. Yinger concludes that on average a black person would need to visit eight housing agents in order to receive as many invitations to inspect rental units as a white person receives from visiting five agents. Thus, the effect of discrimination is to raise search costs for blacks, and the housing audit reveals the magnitude of the cost increase. Though many examples concern the rental market, the value of the audit methodology arguably is much clearer for the real estate market. For home purchases there is great clarity as to the appropriate criterion for the seller to consummate the transaction. Unlike rentals, which establish an ongoing relationship, in the real estate market the seller is leaving the house and probably the neighborhood, and thus appears to have no reason to prefer any particular buyer over another. The only legitimate economic criteria for home sales are price and ability to pay, and these features can be manipulated in the audit study. Any market within which such clarity reigns is a possible site for the use of the audit methodology. The audit methodology has been extended to studies of buying a car (e.g., Ayres and Siegelman 1995), obtaining a taxi (e.g., Ridley, Bayton, and Outtz 1989), obtaining home-owner’s insurance (e.g., Squires and Velez 1987), obtaining a job (e.g., Cross, Kenney, Mell, and Zimmerman 1990; Bertrand and Mullainathan 2003), and more. The audit methodology is consistent with the individualized perspective that sees discrimination as a series of isolated incidences. Further, the apparent transferability of the method to new spheres is a major advantage. However, the audit methodology is not without notable technical problems. First, the correct way to treat the information from an audit is not clear. Some analysts argue that the difference between the treatment of, say, a female and a male job candidate is a measure of discrimination. However, if both candidates interview at the same ten employers, the female applicant

Defining Discrimination Effects: An Asocial Scientific Method / 169

obtains offers from the first five employers and no offers from the second five, and the male applicant obtains offers from the second five employers and no offers from the first five, it is not clear whether antifemale, antimale, both, or no discrimination has occurred. James J. Heckman and Peter Siegelman (1993) compare different ways of summing audit study data and show that the different ways have different implications for the conclusions one would reach. Second, Heckman (1998) argues that audit studies mismeasure discrimination effects because tester characteristics do not match group-specific distributions. For example, equating black and white testers on observable relevant characteristics while, on average, blacks and whites are not equal on those characteristics, compares relatively unspectacular white candidates with blacks near the top of blacks’ distribution. Such studies misestimate “average” candidates’ experiences. A third criticism is that people actually voluntarily sort themselves; this voluntary sorting makes the audit study results an inaccurate measure of the effects of discrimination (Heckman 1998). In other words, if women know that some employers are biased against them, they apply elsewhere. If so, then the gap in treatment revealed by the study overstates the actual gap in treatment given that women will not apply at some firms biased against them. A fourth criticism is that testers, aware of the study aims, may unwittingly elicit responses expected by the researchers. And, a fifth criticism is that the standard audit study design pairs each tester to one match. By failing to vary the composition of the pairs, the design makes it impossible to assess and statistically adjust for the likelihood that results may be driven by the behavior of a single tester. These criticisms raise knotty problems. Still, perhaps the most troubling problem for the audit methodology is that in many situations it does not escape the problems that proved fatal to residual attribution. Heckman and Siegelman (1993) argue that, at least for labor market research, one cannot easily match on relevant criteria because there is insufficient information about what different types of employers regard as relevant criteria. Moreover, it is difficult to control for possibly extraneous factors and for experimenter contamination. Even auditors trained to behave exactly the same may behave in subtly different ways linked to the characteristic understudy, and the different ways of behaving may be relevant to employers. In short, even after considerable work has been done to gather the data, researchers still cannot rule out alternative explanations for any differences in treatment that emerge from the analysis. Thus, at the end of the audit study, one may be forced to simply assert that the gap equals the effect of discrimination. Even the apparently clear case of home sales is not, upon closer inspection, clear. Home-owners who are selling likely are leaving the neighborhood,

170 / Chapter 6

but the real estate agent likely is remaining behind. The agent’s future income stream may depend, at least in part, on selling homes to clients acceptable to those who are remaining in the neighborhood. Of course, one could limit the study to homes for sale by owner (FSBO). Yet, generalizing on the basis of such an analysis might prove difficult. Further, such a limitation is of no use if the aim is to identify purely economic discrimination. The limitation fails in this regard for two reasons. First, the owner may have friends in the neighborhood, and may be concerned about their interest. If selling to members of groups deemed inappropriate might lower property values, it becomes unclear whether the owner’s behavior signals positive discrimination for one’s friends, negative discrimination against some groups, or some other social phenomenon. Second, selection bias (Berk 1983) may undermine such an audit study. If owners who do not want to sell to certain groups are more likely to sell their homes themselves, then the process of selection into the pool of FSBO homes is nonrandom and directly related to the issues under study in the audit research. Given the individualistic logic of isolating discrimination into separate spheres, such a situation would make it impossible to assess the effects of discrimination, for one could not trace the discrimination to the appropriate part of the process. Regardless of the logical foundation for the study, one would not be able to generalize the results to the wider population, or even the population of FSBO sales, a common result of selection bias (e.g., Berk 1983). Similar problems hound efforts to conduct audit studies of hiring. Hence, for many outcomes audit methodology falls prey to the same problems that stalk residual attribution. If the criteria for selection can be clearly identified and competing motives eliminated, the audit methodology may be of use. But the challenge of meeting this standard is considerably more daunting than it appears at first blush. Audit studies and residual attribution share a basic similarity because neither approach relies upon explicit indicators of discrimination exposure. Instead, both call on the analyst to interpret unequal outcomes as evidence of discrimination (i.e., both claim to allow one to ignore the reasons for the behavior and still infer the effect of discrimination). Moreover, as practiced each tends to view discrimination as only the separable acts of individuals. Thus, these approaches are consistent with Becker’s perspective, though, unlike Becker, both presume equality would prevail in the absence of discrimination.

Measuring Discrimination with Behavioral Indicators Perhaps the most promising approach to finding discrimination effects is the third; in this approach one attempts to measure discrimination exposure

Defining Discrimination Effects: An Asocial Scientific Method / 171

and relate those measurements to outcomes. There are complexities with this approach that may make it work well or, alternatively, fail miserably. The effort to measure discrimination directly is an underutilized approach that may be effective, if adopted and if followed with a social understanding of the phenomenon of discrimination. However, as this is the least commonly used approach, is part of the approach I adopt, and effective adoption of the approach depends crucially on the theory of discrimination with which the approach is pursued, I defer major discussion of the details of this approach until I outline the theory of discrimination, a task undertaken in the chapter that follows. Suffice it to say, heretofore the analyses that have come close to measuring discrimination directly have been conceived as doing so within a narrow sphere of activity. So, for example, Jonathan S. Leonard (1984) uses an indicator of employee discrimination litigation to analyze the impact of laws and policies on blacks’ and women’s employment. Though this work is somewhat illuminating, the work conceives of discrimination narrowly, and thus is harmed by the same problems that haunt the other approaches. Ideally, measurement strategies match the theories that make them coherent. Thus, it should not surprise that analysts adopt approaches to measurement that match the understanding that has been the legacy of the dominant social scientific perspective on discrimination. Hence, despite the obvious observation that there are contending approaches to studying discrimination effects, from one point of view there is little important difference between them. Though these approaches differ, they share assumptions traceable to the line of reasoning established by the empirical research translation of the framework Becker establishes. Therefore, each of these approaches has problems that flow from their shared, truncated, conception of discrimination.

Burgeoning Problems of Mainstream Discrimination Effects Research Mainstream social science sees three problems with the efforts to estimate discrimination effects. First, discriminators tend to hide their discrimination, making it difficult to discern effects. Second, there are an infinite number of possible moments of discrimination. Third, basic epistemology implies that one cannot make a causal inference of discrimination. Juxtaposing critical legal theories with mainstream social scientific conceptions of discrimination reveals that these problems are functions of the conception of discrimination that has been utilized by mainstream social scientific research. First, Becker attempts to render a person’s motivations irrelevant. His effort may have been instigated as a response to the developing illegality of discrimination in the United States, a development that threatened

172 / Chapter 6

to send discriminatory behavior deep underground. Yet, ignoring motivations only seems of value when we conceive of discrimination individualistically; when individual actions are seen as the sum total of discrimination, it makes sense to expect discriminators to hide their behavior when that behavior becomes illegal and to expect their concealment strategies to be so effective that one will be unable to measure discrimination. If one conceives of discrimination as only the acts of individuals, and individuals hide their discriminatory behavior, then of course there will be no social (i.e., extraindividual) source one may use to discern the phenomenon of discrimination. However, if a broader definition of discrimination is adopted, one more in line with the critical legal scholars’ approach, then it is well-nigh impossible for discrimination as a phenomenon to remain hidden, because the very fabric of the environment will exude discrimination. Hence, it is the individualist understanding of discrimination that makes individuals’ concealing of their actions so important that the entire approach to studying the phenomenon is constructed to address this problem. Yet, the solution is so damaging to the effort to estimate discrimination effects that it becomes impossible to distinguish employment discrimination and mate selection! Surely, this is a solution that is worse than the problem the solution is designed to solve. Even a rudimentary check of the literature reveals that scholars have been successful in developing alternative ways of explaining any observed inequality. They have proceeded in this fashion because the atomistic conception underlying their approach suggests that the path to deeper understanding is through detailed study of each possible locus of discrimination and some kind of eventual aggregation of those results, ideally (and eventually) in the same study so that complex feedback mechanisms will be discernible. Thus, it is the atomistic conception of discrimination that gives the illusion of sensibility or even sophistication to the fool’s errand of cata loguing the infinite number of spheres within which discrimination may occur, as well as the decidedly large number of roles a discriminator might occupy when he or she discriminates. Note, however, the resemblance between this search for roles and spheres in the social sciences, and the search for discriminators to blame and actions to punish in the courts. In contrast, if a par ticular kind of holistic approach is adopted instead, the focus shifts from blaming or absolving par ticu lar sets of actors to assessing the overall effect of discrimination on par ticular outcomes. Finally, the problem articulated in the context of Rubin’s Model, a problem that appears insurmountable, also is revealed to be a mere phantom, evaporating into nothingness once a thoroughly social conception of discrimination is adopted. Yet, to fully make this case depends upon a fuller description of the definition of discrimination than has been provided

Defining Discrimination Effects: An Asocial Scientific Method / 173

through considering only the critiques of mainstream legal reasoning and social science. Hence, in Chapter 8 I take up the solution to the dilemma posed by Rubin’s Model. For now, however, it should be clear that the problems that have haunted mainstream research stem directly from the problematic, narrowly individualistic, atomistic conception of discrimination that social science has utilized. Reject that conception in favor of one more consistent with the operation of the phenomenon in the social world, and those apparently rock solid problems sublimate into thin air. .

Concluding Remarks The most promising and illuminating implications of Becker’s approach are that equality need not prevail in the absence of discrimination, all groups may suffer when discrimination occurs, and subordinate groups actually might exceed the performance of dominant groups. None of these implications have influenced the empirical analysis of discrimination. Instead, the most influential feature of Becker’s approach has been its ability to ignore the motivations of actors. Unfortunately, analysts have continued to use approaches that allow motivation to be ignored, but certainly by the 1980s few social scientists would be persuaded by explanations that depend on asserting the content of unobserved factors. Hence, on the one hand, the approach still has currency; on the other hand, the currency does not purchase the confidence of those who did not conduct the analysis. The analytic economists who have proposed alternatives have usefully introduced some real-world complexity into the discussion. Though some of these theories differ greatly from Becker’s approach, they fall prey to the same maladies to the extent they share an unwillingness to attend to the motivations of actors, a reluctance to systematically introduce into the chosen sphere of study operative aspects of discrimination that lie outside that arena, or both. The focus on isolated spheres of action has made the social scientific frameworks for the study of discrimination effects structurally similar to the legal frameworks adopted for the adjudication of disputes concerning alleged discrimination. Clearly, current legal opinion requires identification of parties to the conflict. But social scientists have adopted an approach that essentially requires them to identify parties to conflict as well, and this requirement eventuates in the truncation of the admissible range of discrimination effects. This result likely serves to limit the kinds of effects that can be estimated. For all these reasons, therefore, existing strategies and theories for estimating effects of discrimination are seriously flawed. Thus, one may ask, how can one measure discrimination and assess its effects? The answer to that question depends upon how one regards the social realm, and how one

174 / Chapter 6

places discrimination within it. Hence, in order to construct a more accurate and more epistemologically appropriate perspective on discrimination, it is necessary to suggest some useful ways to view the social sphere. In the process, we may reconsider anew just what discrimination is and set the stage for devising a measurement strategy that takes seriously the social character of discrimination.

7 Discrimination as a (Damaged) Social Relation

A

postulate of contemporary research on discrimination is that discrimination is what one person does to another. In contrast, I propose that discrimination is the act of social individuals, but not only that; discrimination also entails the matrix of norms, values, and public support mechanisms that are erected by social persons, norms, values, and mechanisms whose existence outlives the par ticu lar momentary acts that give birth to them. The more extreme that matrix is, the more discriminatory the environment. And, because these matrices are broad in expanse and long lived, they touch those who were never exposed to their architects, channeling the behavior of even those who disavow the tenets upon which they were built. Indeed, the norms, values, and support mechanisms of discrimination transform a set of what could have been fluid relationships between individuals into a relation between types of persons (i.e., a social relation) that, because of its inherent character, is a damaged social relation. If discrimination is not only individuals’ actions but also the humanly created intersubjective and structural factors that facilitate discriminatory behaviors, as well as the customs and norms that make such behavior accepted, then the individualized perspective and the analysis strategies consistent with it are incomplete, and a more deeply social framework and research strategy is needed. The individualistic framework is a hindrance to understanding; once the individualistic framework is set aside, it becomes possible to erect a framework

176 / Chapter 7

for discrimination research more consonant with the phenomenon of sociological and public policy interest, and to devise research that effectively estimates effects of discrimination. It is quite possible that as we set out to develop this final part of the positive agenda, we begin to enter the most treacherous social scientific terrain. This effort may be experienced as unsettling, for it will at times require pulling aspects of various positions and approaches from their currently perceived moorings, to use in fashioning a different approach, even, perhaps, a different paradigm. This is the place of most danger for miscommunication because this is the place where, if one stands firmly within the dominant social scientific approach, it may appear that all that is happening is that the familiar is being twisted into incoherence. But that is not the case; the aim is to reconfigure the terrain in light of an understanding of discrimination that is more consistent with how it operates in the real world. Consequently, it probably will aid communication greatly if familiar associations and expectations of what follows what, and which concepts belong together, are set aside. Whether this aids acceptance of what is offered here is not my concern; my concern is only that rejection of the position, if it be rejected, be based on the actuality of the position rather than a misconveyance of it on my part or an attempt to evaluate it from what may be incommensurable ground. From where we are now, the terrain ahead calls for drawing together a contrasting, preliminary, theoretical definition of discrimination. In order to reveal the decidedly more useful characteristics of this definition, I contrast this definition with that of Gary Becker. After contrasting the defi nitions that are in play, I offer a working definition of discrimination for social scientific research; I elaborate the definition through discussion of selected social phenomena. Next, the definition of discrimination enters into conversation with two contemporaneously important selected philosophical systems. The aim of those reflections is to indicate that defining discrimination as herein is consistent with and may have ramifications for a variety of philosophical perspectives bearing on human flourishing. To head out on this journey, we begin with Gordon Allport, a comparatively neglected pioneer for research on the effects of discrimination.

Allport as a Starting Point The Nature of Prejudice, Allport’s ([1954] 1979) classic treatment of prejudice and discrimination, provides a definition of discrimination. Allport’s definition is actually the United Nations’ definition; discrimination is defined as “any conduct based on distinction made on grounds of natural or social categories, which have no relation either to individual capacities or merits, or to the concrete behavior of the individual person” ([1954] 1979: 52).

Discrimination as a (Damaged) Social Relation / 177

Immediately, we face a paradigmatic challenge. The usual analysis of Allport might now turn to distinguishing prejudice and discrimination, considering the role of attitudes, and essentially becoming a social-psychological treatise. Valuable though such a discussion may be in general or for relating Allport’s oeuvre, it is not the best approach for our efforts to construct a different basis for discrimination research. Thus, this provides a visible example of how fifty years of accepted associations can stand in our way; we do not need to make many of the distinctions that have occupied mainstream considerations of Allport to pursue our project, even though those distinctions may have value in and of themselves or in other points in any analysis, including the present enterprise. Our project puts the pieces together differently, and at this point in the discussion many of Allport’s other contributions need not concern us. What does concern us is that Allport identifies five possible moments that, taken together, identify the range of the phenomenon: antilocution, avoidance, discrimination, physical attack, and extermination. Some of Allport’s labels are not helpful, but the content those labels reference is. Allport reserves the label discrimination for what he terms “detrimental distinctions of an active sort” ([1954] 1979: 14). The content with which he fi lls the term discrimination includes preventing the targets “from certain types of employment, from residential housing, political rights, educational or recreational opportunities, churches, hospitals, or from some other social privilege. Segregation is an institutionalized form of discrimination, enforced legally or by common custom” ([1954] 1979: 15). Defi ned in this manner, it is possible to regard Allportian discrimination as one point on a continuum of constraining, abusive, intergroup interaction. Yet, constraining, abusive, intergroup interaction sounds a great deal like discrimination, a continuum of discrimination that embraces the full range of the phenomenon. If so, some other more specific label is needed for the point on the continuum that Allport designates as discrimination; perhaps the term exclusion will serve. At the same time, there is a question concerning the first point on the continuum, antilocution. In reference to this point, note that Allport contrasts attitudes and behavior, and further observes that it is almost a given that if persons harbor intolerant attitudes, those attitudes invariably are expressed in some action. Thus, one fi nds attitudes undergirding every point on the continuum. At the same time, antilocution, the fi rst point on the continuum, is properly conceived of as an action, not as an attitude. And, if the general defi nition of discrimination refers in part to any conduct, then antilocution is discrimination. Modified in these ways, what Allport provides is a perspective that takes in the full range of discrimination. We need this resource as we move forward

178 / Chapter 7

in order to assure we do not fall victim to the same maladies that render mainstream approaches unsuccessful. Allport’s definition was attacked by Becker as hopelessly vague, but, despite our clarifications and adjustments, and in contrast to Becker, Allport’s definition is coherent, and is actually anchored in the phenomenon of interest. This anchoring allows Allport to avoid many of the problems Becker would introduce with his framework three years later.

Allport and Becker: A Preliminary Comparison A brief comparative inspection of (our modified) Allport and Becker definitions is illuminating. First, it is recognized that Becker’s framework truncates the range of discriminatory behavior; in contrast, Allport specifically identifies five broad classes of antagonism, ranging from speech through segregation to extermination. Thus, Allport reflects the logical range of the phenomenon, rather than only that portion manifest within territory circumscribed by a particular academic field. This is an unsurprising result, as Allport attends keenly to the actual phenomenon of interest, while Becker does not. Second, Allport’s broader perspective on discrimination makes his framework applicable to both economic and noneconomic spheres. Becker’s framework appears applicable only to the economy. As we have seen, the dangers of this limitation become apparent when Becker asserts that noneconomic forms of discrimination were unimportant in the United States at the time of his writing in the 1950s. Given the protests in Little Rock in 1957, during which hundreds of white citizens attempted to prevent a few black children from enrolling in all-white schools, coupled with the international attention these and similar events garnered both before and after (e.g., Dudziak 2000), Becker’s claim of the unimportance of noneconomic forms of discrimination is difficult to believe. This assertion, argued against the well-documented historical backdrop, suggests that analysts may easily deny that which their analytic apparatus cannot see; Becker’s framework does not allow one to see noneconomic discrimination, so noneconomic discrimination is presumed to be unimportant, if it exists at all. Allport, however, offers a perspective that allows one to treat any domain of human life wherein which discrimination may exist. Third, Becker’s framework requires one to treat all social evaluation as discriminatory. One way this is phrased is that Becker cannot distinguish between “positive” discrimination and negative discrimination. Under Becker’s original formulation, discriminating in favor of one’s own race, for example, is no different than discriminating against those of another race. Yet, because Allport defines discrimination specifically, as behavior that is based

Discrimination as a (Damaged) Social Relation / 179

on distinctions that have no relation to individual capacities or merits, one can distinguish between discrimination and other forms of social selection; one can distinguish between so-called positive and negative discrimination. Under positive discrimination, for example, one should not observe any of Allport’s five broad classes of antagonism that range from speech through to extermination; if I love my cousins and hire them to work in my store, it does not mean that I am antagonistic toward every other person who is not my cousin. Only negative discrimination, true discrimination, might be characterized by such antagonism. Finally, unlike Becker, Allport provides grounds upon which the social relational aspect of discrimination may be affi rmed. By referencing categories, Allport implies that no one person is ever discriminated against as a par ticu lar individual; classes of persons are discriminated against, not individuals. Thus, whereas any distinction between any two or more individuals could be treated under Becker’s taste theory of discrimination, Allport’s definition of discrimination fundamentally concerns a relation between categories of persons. A solidified relation between categories of persons is a social relation. Discrimination is a social relation, but it is not by any means the only social relation. Even in self-avowedly individualistic societies, such as the United States, social relations abound. Social relations are central to human life; that is, social relations are fundamental building blocks of human interaction. Thus, there are many different relations between types of persons: student-teacher, employer-employee, husband-wife, sibling-sibling, doctorpatient, and more. In order to identify the character of a relation one must attend to the motivations inhering in the structuring of the social relation. Once a social relation is structured, even individuals who do not support the motivation behind the structuring of the social relation still will find themselves embedded in the social relation, obtaining benefits and incurring costs that follow in part from the damaged or undamaged nature of the social relation.

A Working Definition of Discrimination Allport provides resources for a perspective on discrimination that is decidedly at odds with the mainstream of social science analysis, a mainstream that has much in common with Becker’s approach. Only one phrase need be added to the Allportian framework to turn his definition into the one I adopt. My working definition is that discrimination entails distinction made on grounds of natural or social categories, which have no relation either to individual capacities or merits, or to the concrete behavior of the individual person, but is, instead, based in a limiting view of some types of persons.

180 / Chapter 7

In order to distinguish discrimination and nondiscrimination, one must, as Allport does, take seriously the social character of discrimination. A definitive characteristic of discrimination is that discriminators target categories of persons, not other individuals per se. This feature is essential, for it makes discrimination a social relation. At times it is difficult to realize that the categorical nature of discrimination is an essential component of the phenomenon. It is the targeting of categories that allows destructive acts to coalesce into a set of solidified social relations, buttressed by theories of inferiority and superiority, and possibly codified by legal statutes and rulings. That is not to say that individuals never suffer condemnation as individuals. But individual condemnation can be extremely painful and wrong without being discrimination. By definition, individuals as such cannot be discriminated against, for when an individual is condemned, no set of norms becomes erected over time, and no set of laws is written to speak directly about the unique features of that individual. Today, it seems that the charge “discrimination” is levied often because the pain of the struggle against discrimination has imbued the term, in some quarters, with the symbolic status of a trump card. But its repeated and, I would argue, often inappropriate use has distracted from the legitimate charges that can be made, and confused the distinction between a legitimate charge and a rhetorical maneuver. In our definition, discrimination is not simple one- on- one condemnation; instead, discrimination also involves the norms and laws that have come to codify the social relations of discrimination. These norms and laws emerge because of the categorical nature of discrimination.

Illustrative Implications of Discrimination as a Social Relation Taking seriously the categorical and relational character of discrimination can aid us in re-evaluating our understanding of social phenomena. This re-evaluation, however, substantially increases the challenge we face, steepening the terrain we traverse. This re-evaluation will call on us to treat phenomena formerly seen as dissimilar as the same, and to treat phenomena formerly seen as essentially the same as distinct. In this manner we will rearrange the categories of social experience in a way that may be disorienting. But disorientation does not indicate our approach is in error, for what has gone before—the consideration of dominant legal reasoning and its similarity with social scientific approaches, the critique provided by critical legal frameworks, the mainstream social science failure to produce a single consensus estimate of discrimination effects in any domain—all these developments and more suggest that, actually, we definitely need to reorient our

Discrimination as a (Damaged) Social Relation / 181

understanding of the social world, head off in another analytic direction, and reject the dominant social scientific perspective on discrimination. In that connection, critical legal scholars have suggested that taken-forgranted arrangements often are discriminatory. Therefore, it is possible that many daily occurrences entail discrimination, and we need re-cognize these phenomena as discrimination, just as we might need to recognize some alleged discriminatory arrangements as nondiscrimination. One key example of such a state of affairs concerns the issue of gender-specific socialization. Many researchers see socialization as one explanation for gender differences in occupational aspiration and attainment (e.g., Cole 1986). For example, some argue that observed male-female differences in occupational outcomes are the result of differential socialization rather than discrimination (e.g., Parsons 1958). Indeed, some researchers claim that women have different preference structures or utility functions than men, need to and want to balance family needs more than men, and consequently trade earnings for other characteristics of jobs such as flexible schedules (e.g., Killingsworth 1985). Though research shows job flexibility does not account for occupational sex-segregation (Glass and Camarigg 1992), the idea remains that men and women behave differently and have different desires, and that these differences flow from differences in socialization. From what we have described so far, however, it should be clear that one person’s socialization is another person’s discrimination. One may easily reveal the discriminatory core of differential socialization. If a school were to socialize girls and boys differently, one could show the socialization was discriminatory by documenting the basis of differential treatment in limiting attitudes about the capability of or appropriate roles for boys, girls, men, and women. Of course, the subjunctive “were” is incorrect, for the evidence indicates that schools do socialize boys and girls differently. As Valerie E. Lee, Helen M. Marks, and Tina Byrd (1994) reveal, discriminatory gender socialization occurs in schools. They observe about the same amount and degree of sexist socialization in coeducational and single-sex secondary schools, from subtle forms to egregious examples. Schools are public institutions (even when privately controlled owing to, for example, compulsory schooling laws), and thus the equality ideal clearly pertains. Differential sex-role socialization based on limiting views of persons violates that ideal and is discrimination. For the working definition of discrimination, therefore, I contend that it is erroneous to explain differences in adult outcomes as the effects of differential socialization rather than discrimination. If the differences in socialization are based in a limiting view of persons, then differential socialization is discrimination, even if the limiting view of persons is invoked unreflectively or unwittingly, and even if the socializers are private citizens (or parents) rather

182 / Chapter 7

than public entities. That we observe differential sex-role socialization—sex discrimination—in schools, and the behaviors match those we observe in some families, simply indicates that families may discriminate on the basis of sex just as schools appear to. We should not change the label because the behavior is pervasive; instead, we need soberly and scientifically acknowledge the pervasiveness of the behavior. Two siblings reared by the same parents but socialized differently simply on the basis of ideas about the capacities or legitimate roles of persons of different sexes are two siblings who encountered discrimination of the most damaging kind—that dispensed by one’s guardians when one is virtually defenseless. Given the fi ndings on parents and gender prejudice I present in Chapter 2, it is likely that such discrimination is common. Seen in this way, when sophisticated analyses point to socialization as a cause of gender inequality, they unwittingly fail to deny that discrimination matters but, instead, simply posit that discrimination has an impact at an earlier moment, likely during childhood. Note that it also is possible for differential treatment to occur without being discriminatory. Two siblings may be reared differently because parents may have learned how to rear children with the older child. They wisely apply the new knowledge when rearing the younger child. Thus, the younger child obtains different, and most likely better, parenting. Though from Becker’s perspective it might appear to be discrimination, this is not discrimination, for it does not flow from a limiting view of older or younger siblings. Note, also, that this means that one cannot simply observe the effect or circumstance—differences in child-rearing—and know its cause. Instead, one must theorize the phenomenon and attend to motivation. Several alternative bases of differential treatment exist. For example, basic age-earnings profi les imply that younger children may benefit from their parents’ greater affluence at the time the younger child reaches critical and costly transitions, such as college entry. This basis of differential treatment also is not discrimination, though it might breed sibling socioeconomic inequality. Or, parents may treat two children differently because they have recognized the disparate dispositions of each child. In this case, differential treatment may be exactly what is needed to secure nondiscrimination. Thus, differential treatment is not necessarily discrimination. Differential treatment is discriminatory when based in a limiting view of persons. In the same vein, when equal treatment is based in a limiting view of persons, equal treatment is discriminatory. For example, if all students are expected to learn in the same exact way, and those who learn differently are considered inferior, then students who learn differently experience discrimination if provisions are not made for their alternative learning needs. The

Discrimination as a (Damaged) Social Relation / 183

limiting view of persons in this example is that there is only one way to attain an important good—knowledge—and anyone needing some other learning approach is deficient. Discriminatory equal treatment is possible in principle and also likely to occur, especially in large organizations such as schools, and may resemble conventionally defi ned disparate impact discrimination in some respects. Still, given norms of equality, a claim of discrimination on the basis of equal treatment is likely to be expected to meet a higher standard of proof (in the court of public opinion, if not elsewhere) than the more common discrimination claims concerning unequal treatment. These observations suggest that widespread views of what is and what is not discrimination can be usefully reconsidered. On the basis of our definition of discrimination, one need regard gender-specific socialization as discrimination. I contend that differences in socialization for boys and girls are early forms of discrimination, forms that may have profound effects because they aid children in circumscribing their own aspirations and conduct. Were this kind of behavior identified on the basis of race, one would likely fi nd little disagreement as to the discriminatory character of the behavior. And one would fi nd widespread agreement as to the possibility of subtle and cumulating destructive effects of early differences in socialization. Our attitude with respect to gender-specific socialization must be no different.

Damaged and Undamaged Social Relations I draw on Allport to develop the definition of discrimination, and I characterize discrimination as a social relation. But I do not stop there; I also, perhaps judgmentally, describe discrimination as a damaged social relation. Is this claim sustainable? The first order of business is to discern the analytic difference between damaged and undamaged social relations. Then, it will be possible to consider in more detail just how discrimination is a damaged social relation. Before turning to either of these tasks, however, we need attend to an issue that may cause some discomfort. Some may accept the claim that certain social relations are damaged, but feel the claim has no place in a social scientific treatise. To these persons, the claim cannot be sustained, at least not social scientifically; after all, they rightly observe, a particular social relation can be characterized as damaged only from the vantage point of particular philosophical systems or life experiences. And science, they might argue, is not about an individual’s beliefs or individual life experiences; it is, instead, about what systematic inquiry can reveal. But this view mischaracterizes scientific inquiry. It ignores the easily established fact: All questions pursued by scientific inquiry, and all statements

184 / Chapter 7

made by scientists, are motivated in part by their values. This does not mean scientists are free to say or write whatever they please; ideally, within the constraints of a field defined as a dialogic community, there exist norms of practice and judgment. Scientists make mistakes, so error is not eradicated by the dialogic process. But, within the shifting boundaries of the relevant community, a conversation develops the agenda of research, the accepted (for now) answers to questions, and the standards of proof. Within such communities, scholars’ work is motivated by their individual and collective values; indeed, the dialogic communities themselves could be termed value communities in that those engaged in dialogue likely share some fundamental values. For example, psychologists do not study mental health only for the pure joy of studying it—there is also a value they attach to knowledge (i.e., they believe knowledge is better than ignorance), and knowledge of mental illness and health is better than ignorance of those matters. Economists study macroeconomic possibilities for many reasons, but one core reason is that they believe wealth, or at least sufficiency, is better than poverty. Physicists probe the nature of matter and energy in the universe in homage to the species’ quest to understand our place in the cosmos—the desire for this knowledge, too, is a value. Thus, asserting that the interest in whether social relations are damaged or not is rooted in values, and eschewing any judgment one might make for this reason, is not to become more scientific—all science is motivated by someone’s values. Accordingly, to characterize the relations one has uncovered honestly and directly as dysfunctional for human flourishing is to state, at the same time, a value commitment to human flourishing. There is, I submit, nothing nonscientific about being direct about one’s values. Indeed, hiding one’s values violates another hallmark of science and, indeed, liberal democracies; potential critics need access to as much information as possible about the research, including the possible known motivations that may have pushed the research awry. When scholars hide their motivations, they alienate others, including other citizens and voters, from full ability to evaluate the work. This cannot be in ser vice to the realization of full human potential or even basic respect. So, with the above position as a maintained assumption about scientific inquiry, what might social science regard as a damaged social relation? And, prior to addressing this question, what exactly defines an undamaged social relation?

Undamaged Social Relations Social relations are not inherently damaging. Indeed, a large number of social relations are salubrious supports to human flourishing.

Discrimination as a (Damaged) Social Relation / 185

Consider the spousal relation. Clearly, norms, values, and support mechanisms of the spousal relation exist, and these transcend the individual parties to any par ticu lar marriage. Norms include the expectation that spouses will live in close proximity, barring extenuating circumstances (e.g., separation owing to war or, more mundanely, job or career imperatives). Values include the expectation of sexual fidelity; the value is reflected in a backhanded way by the tendency to call marriages lacking an expectation of sexual fidelity open marriages. The modifier suggests that the expected (i.e., unstated) value is monogamy. Interestingly, we do not typically refer to closed marriages, the analogous appropriate label for marriages with a stated commitment to sexual fidelity. Support mechanisms include licensing by the state, which provides and enforces certain rights (e.g., rights to privileged communication; for example, one spouse cannot be forced to testify against the other in the courts) and confers certain obligations (e.g., the necessity of obtaining a divorce before marrying a second spouse [i.e., the obligation to not engage in bigamy]), as well as ceremonial rituals enacted by religious entities. Certainly, this matrix is more complex than I have sketched here. And, certainly, the matrix differs from place to place, culture to culture, and epoch to epoch. Even though there is no universal timeless matrix of norms, values, and support mechanisms, individual members of a given society are no less trapped, in a sense, by the way in which the social relation of marriage is structured in their par ticular society. Though individual couples may “do something different,” they do so in a world that defines the social relation in a par ticular manner, a manner different from their preferred style of interaction. The couple must navigate that world, even though the couple may not share the same vision of the social relation that intersubjectively pertains in their environment. Spouses are not the only persons who navigate that world; nonmarried persons also navigate that world, with the same spousal relation as a given. Hence all parties in the society are affected by the spousal relation, its norms, implied values, and support mechanisms. One is not “protected” from the character of the spousal relation by refusing to marry. In other words, though one may reduce the implications, there is no way to escape the implications of the particular spousal relation of the context in which one is located. Considering a similar social relation—the cohabiter relation—is revelatory. In many respects the cohabiter relation and the spousal relation are similar. But cohabiters, spouses, and nonspouse noncohabiter adults know there is a difference, a difference reflected in the likely behavior of persons under the following thought experiment. Imagine randomly sampling cohabiting couples and married couples, and tossing a coin for each couple. Depending upon the outcome of each coin toss, assign couples to marriage or

186 / Chapter 7

cohabitation. Thus, some married couples most likely would be required to divorce (or annul their marriage) and live together, and some cohabiting couples most likely would be required to marry. That the vast majority of both cohabiters and spouses likely would be unwilling to switch their statuses if required to do so by the toss of a coin suggests that the cohabiter social relation and the spousal social relation are not the same. One point of this observation is to indicate that social phenomena that superficially resemble other social phenomena may not actually be the same. Cohabiting and marrying are not, at this sociohistorical moment, the same. A second point of this observation is to identify a means to determining if two social relations are the same—if persons are indifferent between them, then they may be, for all intents and purposes, the same. However, if persons are not indifferent between the social relations, either by preferring one location within the social relation, or by preferring one social relation to another ostensibly identical social relation, then it is likely that the two social relations, or the different locations within the social relation, are not the same. Examples concretize the claim that these social relations are not the same. For example, a long-time cohabiting couple may fi nd it difficult at times to arrange for the rights that come as a matter of course to newlyweds. One cohabiter may fi nd their rights nonex istent should the other fall prey to a protracted illness; the newlyweds will be less likely to have this problem. Once the spousal relation is seen as a social relation, it becomes clear that those who maintain that allowing gay marriage will change the definition of marriage, ultimately affecting heterosexuals, are correct. Yet, fully appreciating the social relational reality also leads to at least two questions. First, does the state need to support “marriage” to accomplish the aims of the state? Second, if the state did alter its relation to the institution of marriage, would the changes for marriage be positive, negative, or simply neutral? It is unclear just what the final tally of changes would show; at present there is far more heat than light on this question. However, the role of the state in the marital relation in the West harkens back to the developing churchstate relation. Thus, further development, toward a new social relational form—basically a separation of the state interest and whatever church interest there may be—is not some violation of the historic trajectory. If the state were to license adults for civil unions that would carry the obligations and protections the state affords, while marriages became the province of religious bodies and would offer no obligations or protections other than those the private religious bodies were able to require or enforce, this might accurately reflect the complexities of a developing social relational reality. No religious institution would ever have to allow gay marriage, racial intermarriage, or interfaith

Discrimination as a (Damaged) Social Relation / 187

marriages, nor would a religious order have to allow any means of marital dissolution. The state, however, would allow any of those types of civil unions, as well as procedures for union dissolution. Some couples might opt for one or the other type of arrangement, or both, or neither. But, the marital social relation would be shorn of its state connection, and would find support in the religious institutions that are concerned directly with its definition, while the civil union social relation would be rendered unto the state. This separation would recognize distinct religious interests and state interests. Such an explicit reconfiguration of the social relational space could reflect the multifarious lenses through which a diverse society views the current spousal social relation. It is fair to say that marriage potentially is supportive of human flourishing, given a par ticular matrix of norms, values, and support mechanisms that heighten voluntaristic entry into the institution and maintain individual rights and protections within it. This social relation, therefore, can be conceived as a supportive social relation, supportive for every party to the relation. We might make similar observations for other exemplary social relations (e.g., parent-child, teacher-student, doctor-patient).

Damaged Social Relations: Existence and Transformation In contrast, damaged social relations are not supportive to every party to the relation. Instead, one set of parties to the social relation has its existing capacities or potential stifled. Note that in a damaged social relation, the stunting of the capacities or potential of one set of parties is not a function of the par ticular individual persons involved but, instead, a definitive part of the social relation. A parent may emotionally, physically, or sexually abuse his or her child. When this occurs it does not mean that the societal parent-child relation is itself a damaged social relation in principle; instead, the abusing parent has violated the trust, norms, support mechanisms, and even the basic logic of the social relation—parents nurture children toward or perhaps even far into adulthood. Hence, the existence of abusive parents does not transform the archetypal parent-child relation into a damaged social relation. The salubrious nature of the parent-child social relation survives the existence of par ticular abusive parents; only if such abuse becomes a defi nitive feature of the abstract relation is the abstract relation altered. With salubrious social relations, poorly behaving parties do not taint the abstract social relation. At the same time, the good intentions of parties to a damaged social relation cannot transform the damaged social relation into a salubrious one in the short term. Thus, if the damage of a damaged social

188 / Chapter 7

relation is attenuated by the behavior of individuals in a par ticu lar instance, the essential character of the relation remains; similarly, if a par ticular salubrious social relation happens to turn bad because of the behavior of a par ticu lar individual, the essential character of the abstract social relation remains positive. That is how social relations work. The social relational character greatly constrains the actors; individuals can make some differences, but to make deep and lasting differences they must erect different social relations. Working within the constraint of a given social relation will be, as expected, constraining. No one person as party to the social relation can change the definitive character of the relation. Still, individual acts can make a difference in the lives of people, if not the social relations in which they are embedded; and, over time, even this may matter for the transformation of damaged social relations, perhaps into salubrious ones. But note that individuals acting within the social relation will only slowly change the social relation, over decades or centuries, and that route to change always will be at serious risk of being completely undone by the reassertion of the fundamentally damaged social relation.

Discrimination as a Social Relation, Discrimination as a Damaged Social Relation Discrimination is a social relation in that it involves types of persons. And, it is a damaged social relation in that it involves a limiting view of types of persons and constraints on action that entail a stunting of the capacities of some parties to the relation. As we observed in the cohabiter/spousal social relations example, two social phenomena may resemble each other in superficial aspects, but that does not mean that they are the same. It is clear from the claim that discrimination is a social relation defined by the emergence of norms, values, and a matrix of support mechanisms that some claims concerning discrimination do not cohere. For example, the concept of “reverse discrimination” is a contradiction in terms. Considering the idea of reverse discrimination, the idea does not signify a social relation—it is not indicative of solidified patterns of interaction between types of persons. Assuming it is a social relation for the moment, we can ask whether subordinate members in the reverse discrimination social relation would be willing to trade their subordinate status under reverse discrimination for subordinate status under discrimination unmodified. That it is unlikely that those accusing others of reverse discrimination would be willing to make this trade should suggest that the social relations of reverse discrimination lack some defi nitive, particularly

Discrimination as a (Damaged) Social Relation / 189

onerous features that defi ne discrimination as a social relation—taking reverse discrimination, erroneously, as a social relation for this par ticu lar example only. But, in actuality, there can be no such thing as reverse discrimination because for discrimination to exist there must be a matrix of norms and traditions that render the treatment a continuing social relation. Whenever one must use the modifier reverse, it indicates that definitive aspects of the concept of discrimination—the denigration of the subordinate category of the relation, potential exposure to the full range of manifestations Allport identifies (e.g., from speech to genocide)—are not present. Absent that defi nitive matrix, one may have some other kind of mistreatment, but one cannot have discrimination. Thus, by definition, one cannot have reverse discrimination. This, of course, raises a knotty question: Are blacks and women the only possible targets of discrimination as defined herein? Of course not. The way to determine whether discrimination is involved is to ascertain whether salient dimensions of power relations render the alleged target of discrimination subordinate. If so, and if the subordination is not a momentary encounter but, instead, a solidified aspect of social relations, then discrimination is possible. Illustratively, given the U.S. context, it is easy to see that Jews and other religious minorities, those whose ancestors hailed from Asia, some white ethnic groups, and gays and lesbians are historic and contemporary targets of discrimination. Two important issues need attention here. One concerns the status of everyone who is outside the categories listed above, or any set of categories that satisfy the requirement described above for the social relation to be characterized as discrimination. Another issue concerns the possibility of targets of discrimination engaging in discrimination. Taking the second issue first, it should be clear that women and blacks can engage as discriminators. The status of a person is no bar to their being discriminatory toward others. Indeed, to the extent that blacks support the solidified social relation of antiblack discrimination, perhaps by holding black job applicants to a higher (or lower) standard than they would nonblacks, then those blacks are discriminators against blacks. And, of course, blacks can engage in discrimination against women and other groups located in subordinate positions that can give rise to discrimination. Thus, it should be clear—many categories can be targets of discrimination, and persons who are in targeted categories can themselves discriminate against members of other categories or even their own category. Still, note that key to discrimination is the taken-for-granted structures that channel action; action so channeled is discrimination regardless of the demographic category of a given actor.

190 / Chapter 7

As for the first issue, what of those not included in the categories above, those whose social location by definition does not allow a matrix of norms, values, and support mechanisms to emerge? This is where the limitations of our use of language and the mythologized understanding of our history create problems. It is obvious that many persons outside the groups named above, or even outside any defensible list of groups that can be targets of discrimination, suffer denigration and worse at times. But, this denigration need not be discrimination. This presents a challenge—if the denigration of some is discrimination but the denigration of others is not, and if reverse discrimination is a contradiction in terms, then what label(s) can be used to refer to the mistreatment of persons whose categories make it impossible for them to be targets of discrimination? The challenge we encounter here would seem to have two regrettable sources: (1) the legacy of the inception of the era of contested prejudice, and (2) the lack of coherent, widely shared languages for articulating grievances in the United States. The first source is traceable to the individualistic ethos through which advances were made, advances that signaled the end of the regime of condoned exploitation but that stalled at the era of contested prejudice. The individualistic ethos, embodied in the matrix of laws that purported to make discrimination illegal, has left us able to discern only with great difficulty the social relational character of discrimination. Social scientific analyses that have adopted the same individualistic ethos have not helped to resolve this par ticular problem, but social science is not the source of the problem. The second source of the problem is that it is difficult to articulate a grievance when one lacks a language within which the articulation of the grievance makes sense. In the United States, intermittent progress against race and sex discrimination appears to have given the language of discrimination some visible power. Other logics and languages—perhaps of class, of human rights, of basic respect—have not found wide exposure or thoroughgoing, widely recognized success. The noteworthy result of this predicament is that persons may feel uneasiness or anxiety. In such a state—but absent a framework within which the trouble can be accurately identified, named, and addressed—persons may seize on the only widely available language within which it appears possible to articulate a grievance: the language of discrimination. When such responses occur, they hinder our understanding of the phenomenon, and hinder successful response to grievances that very well may be legitimate. Still, the lack of a language of grievance should not mean that one term, discrimination, becomes required to carry all of the weight for any grievance one may have. Discrimination is a par ticular social phenomenon, a social relation with a par ticular character that pertains between types of

Discrimination as a (Damaged) Social Relation / 191

persons. There are other grievous phenomena, and those phenomena must be named and those appellations used in seeking redress.

Distinguishing Damaged and Undamaged Social Relations: Illustrative Western Philosophical Systems and the Case of Discrimination Certainly, my definition of damaged social relation finds articulation within a par ticular social, cultural, and philosophical context. That context provides several unstated and oft-unreflectively accepted commitments as to what is essential for human flourishing and what human flourishing actually entails. It would take us far afield to attempt to delve into every potential commitment exhaustively at this juncture, even though the web of commitments undergirds my distinction between damaged and undamaged social relations. In addition, I certainly have not attempted to survey the full range of damaged social relations, nor have I drawn on the full range of social relations in general. And, here, it would go far beyond the scope of this inquiry to attempt a full discussion of the plethora of coherent philosophical systems and the placement of damaged social relations in general, or discrimination as a damaged social relation in par ticular, within those systems. Yet, it is possible and potentially useful to offer some observations concerning the damaged social relation of discrimination in light of some selected systems. These observations may deepen the provisional sense of how one might distinguish damaged and undamaged social relations, and suggest how discrimination so defined may be understood within selected philosophical perspectives while, at the same time, providing resources with which to interrogate some of the claims these perspectives articulate. The set of possible frameworks within which these concerns could be pursued is daunting. Hence, I will begin, here, with only two—Rawls and Marx. The Rawlsian perspective is a useful framework to consider owing to its status as an important response to utilitarianism. Utilitarianism allows societies to trade away the rights or well-being of some for the welfare of others, and thus is a framework that would seem unable to reject any social relation as damaged, if that social relation results in greater good for some. Rawls rebuts this perspective, but does his rebuttal have any implications for our understanding of discrimination? Marxism is both an academic/philosophical perspective and a set of claims upon which political action has been taken. If we consider some of the claims of Marxists, can we discern a ground for regarding discrimination as a damaged social relation? And, are there any implications for Marxism if we

192 / Chapter 7

define discrimination as a damaged social relation; must Marxists take this definition seriously for their own intellectual project?

Rawls John Rawls ([1971] 1999) proposes that we can identify the principles of justice by positing an original position of equality and ignorance; from this position, individuals work to establish the principles of justice and arrangements that will structure their society. Equality pervades the original position, dissolving the role of power. Ignorance characterizes every actor, wiping away the role of interest. In short, each stands in the original position shorn of both the knowledge of their parochial real-life interests and the power they have in real life to pursue those interests. Under this veil, the community develops a set of principles that will bind them afterward. The question then is asked: What principles and societal arrangements would find support under these conditions? Thus, in the original position one can ask: Which social relations would find support sufficient to be accepted? It is likely that salubrious social relations would find support, for those social relations are beneficial to all parties to the relation. In contrast, damaged social relations, such as discrimination, would be unlikely to generate a consensus, for one would not know whether one would be the target of discrimination or not, and the very social relation is painfully constraining for targets. Certainly, an individual’s willingness to support a damaged social relation may be related to the individual’s taste for risk. Risk-averse persons will be less willing to allow undamaged social relations to come to be. Risk-preferent persons might be comfortable playing the odds. The risk-preferent may become a larger share of the population as the chances of being in the target group decline; that is, as the target group becomes smaller, the population willing to take the risk of allowing the damaged social relation may rise. Yet, at the same time, the gains to the damaged social relation that any given nontarget may obtain likely decline as the target population shrinks, which would lower the motivation for taking the risk. Of course, the degree to which there is a relationship between the size of the targeted group and gains to the nontargets depends on the content of the relation and thus is not estimable absent such information. However, the claim that risk preference may determine whether a social relation will or will not be allowed is trumped in two ways. First, consensus is required; the risk averse could “block” any consensus supporting the existence of damaged social relations such as discrimination. Second, there is no necessary reason to supplement the original position with information as to the odds of obtaining various positions in the society. If that information is

Discrimination as a (Damaged) Social Relation / 193

not provided, the set of persons willing to roll the dice by calling into existence damaging social relations, such as discrimination, likely would be vanishingly small. These observations suggest that one might distinguish damaged and undamaged social relations by asking whether a given social relation would be the basis of consensus in a Rawlsian original position. Social relations that can generate consensus would appear to be salubrious social relations; social relations that would not generate consensus would appear to be damaged social relations. And, by this reasoning, one would conclude that discrimination is a damaged social relation.

Marxism Though a triumphant post-1989 American political polemic has consigned communism to the proverbial dustbin of history, the social sciences continue to take Marxism seriously as an analytic frame. Certainly, there is little reason to question many of the key concepts used in Marxism, such as alienation, exploitation, and class struggle; it is unlikely that any serious commentator, regardless of his or her political persuasion, would deny the contemporary existence of alienation, exploitation, and class-linked interests, any of which may inspire struggle under certain conditions. Indeed, and ironically, at the very time that many of the conditions Marx asserted would be required for the overthrow of capitalism are emerging worldwide, such as the immiseration of working people, and precipitously declining ability of nation-states to restrain capitalists’ initiatives or even tax capital to pay for social insurance for the populace, the most powerful segments of capitalist societies seem oblivious to the unfolding crises. Though this does not imply that Marx’s claims as to how the crises will be resolved will come to fruition, it appears foolish to deny the signs of possible crisis, signs that in many, but not all, respects eerily match many aspects of Marx’s analysis. Alienation Marxists generally regard the means of production, and the social relations connected with those means, as important for understanding society. Classically, Marxists have regarded discrimination as a tool for obscuring the collective interests of the working class, allowing capitalists to divide and conquer. Yet, seeing discrimination as a damaged social relation seems defensible within Marxist reasoning, and doing so makes discrimination a much more central aspect of the phenomena of interest to Marxists. Seen as a damaged social relation, discrimination is transformed from a mere tool of capitalists into a possibly fundamental aspect of capitalist social relations.

194 / Chapter 7

When one attends to Marxist understandings of alienation, it becomes apparent that alienation is a sometimes forgotten part of the core of the basis of class antagonism. Karl Marx contends that men realize themselves in work, in labor. Yet, under capitalism, Marx argues that men are alienated from their labor, from nature, and from each other, and that this alienation is evident in that the laborer’s product turns against him, oppressing him (e.g., Marx [1844] 1977). A laborer in a factory produces the means of his own oppression—with every gun that can be used by functionaries of order to maintain the status quo protection of private capital; with every tractor that can be used to mechanize farming, increase the capital intensity of the act, toss laborers off the farm as superfluous, and more—the worker produces goods that are turned against worker interests. As men are alienated from their work, they are alienated from both themselves and each other as well. Once men are seen as alienated from each other, it is easy to see men’s relation to each other under capitalism as a damaged social relation. And, once men’s relation to each other under capitalism is seen as a damaged social relation, then it is a small step to seeing discrimination, when defined as a damaged social relation, as one manifestation of the damaged social relations characterizing capitalist formations. This characterization applies for many kinds of discrimination, and perhaps most especially for sex discrimination. Interestingly, Friedrich Engels argues that women were the first oppressed class (Engels [1884] 1978), and monogamy was the subjection of one sex by another. In this view, the marital relation is a damaged social relation. Parenthetically, this may very well be the case—as I indicate, analysis of the fundamental aspects of the social relation are required to identify a relation as damaged or not. One must not recoil from such an analysis just because of the label of the relation. It is unlikely that contemporary observers would maintain that, at all times and in all places, marital relations reached the par ticular current ideal of marriage. If past marital relations would be criticized as damaged social relations—and I submit they would be from the vantage point of many present societies—then the point is made: The marital relation can be a damaged social relation and may be a social relation of discrimination. If the marital relation is seen as a damaged social relation, part of the damage inflicted is owing to the alienation between parties to the union. A fundamental misrecognition attends the relation, complicated by ideologies of primacy, subordination, or specialization. Of course, this example also indicates that the experience and meaning of a damaged social relation may vary depending on other societal factors, such as the economic system, the political system, and the cultural context. Yet, alienation would seem to be a possible feature of a damaged social relation, regardless of the society in which the damaged social relation is located.

Discrimination as a (Damaged) Social Relation / 195

Of course, a clarification is in order. Observing that the marital relation can be a damaged social relation does not contradict the earlier observation that marriage is not necessarily discriminatory. The two discussions concerned two different definitions of discrimination and marriage. The current discussion concerns whether marriage as in institution in a society is constructed such that at least one spouse must forego their full human development. In contrast, the previous discussion concerned whether the selection of a mate necessarily discriminates against those not selected. Whether or not marriage is a damaged social relation (i.e., a social relation in which at least one party to the relation is inherently disadvantaged) is a different question than whether a particular person’s decision of whom to marry is discriminatory. Under Becker’s framework, a person who selects a partner to marry is discriminating against all other potential partners. Under the framework proposed here, a person who selects a partner to marry is not discriminating necessarily, even as both may be about to enter a damaged social relation— marriage. Whether or not marriage is a damaged social relation depends on the history and contemporary nature of the institution in a given society, not on whether the two parties to any given marriage conceive of the institution in any par ticular manner. Of course, if the social relation is a damaged one, then the parties may be engaging in a discriminatory social relation, even if they endeavor not to do so. Returning to discrimination, when seen as a manifestation of alienation, discrimination becomes an object of more than passing interest to Marxists. It moves from the periphery, where it appears to be one tool capitalists may use to divide workers, to the center, where it becomes visible as a condition central to creating and maintaining workers’ alienation. Exploitation One key concept in Marxism is exploitation. Analytic Marxists have attempted to explicate this concept so as to give it technical, rather than simply pejorative, content. John E. Roemer (1988: 133) contends that one way to define exploitation is to see an agent as exploited if he or she “would have been better off at the distribution of output associated with an egalitarian initial ownership of the means of production than . . . at the actual distribution. Similarly, an exploiter is one . . . rendered worse off by the change to an egalitarian initial distribution.” This is not to say that the only final nonexploitive output would be equal. Roemer contends that agents may differ in the final observed distribution, but if that final distribution is worse than it would be for someone had the beginning distribution been egalitarian, then the system entails exploitation. Under Roemer’s defi nition, it is possible to evaluate discrimination and regard it as an exploitive social relation. Discrimination essentially entails

196 / Chapter 7

an inegalitarian distribution of power, and an agent can be regarded as exploited under discriminatory social relations if he or she would have been better off had the distribution of power initially been egalitarian. Alternatively, if an agent is better off now than he or she would have been had the initial distribution of power been egalitarian, then the person is a Roemer exploiter under the discriminatory social relation. An implication of Roemer’s framework is that those not directly involved in particular acts of discrimination may still be exploiters within discriminatory social relations. Roemer’s framework implies that the concept of reverse discrimination is a contradiction in terms, akin to the incoherent concept of reverse exploitation. Roemer provides a useful Marxist-inspired, technical definition of exploitation. However, Erik Olin Wright’s definition of exploitation creates problems; as we shall see, the implication is that only some Marxist definitions of exploitation prove sustainable. Wright (2000) defines exploitation as those phenomena that meet three criteria: (1) the material welfare of the exploiters depends on the material deprivation of the exploited, (2) the exploiters’ ability to obtain advantage depends on excluding the exploited from productive resources, and (3) the exploiters thus are able to appropriate the labor effort of the exploited. Wright also defines nonexploitive economic oppression as those oppressive arrangements that do not fall into the category of exploitation as defi ned above. Wright notes: The crucial difference between exploitation and non-exploitative oppression is that in an exploitative relation, the exploiter needs the exploited since [sic] the exploiter depends upon the effort of the exploited. In the case of non-exploitative oppression, the oppressors would be happy if the oppressed simply disappeared. Life would have been much easier for the European settlers in North America if the continent had been uninhabited by people. Genocide is thus always a potential strategy for non-exploitative oppressors. It is not an option in a situation of economic exploitation because exploiters require the labor of the exploited for their material well-being. It is no accident that culturally we have the saying, “the only good Indian is a dead Indian,” but not the saying “the only good worker is a dead worker.” The contrast between South Africa and North America in their treatment of indigenous peoples reflects this difference poignantly: in North America, where the indigenous people were oppressed (by virtue of being coercively displaced from the land) but not exploited, genocide was the basic policy of social control in the face of resistance; in South Africa, where the European settler population heavily

Discrimination as a (Damaged) Social Relation / 197

depended upon African labor for its prosperity, this was not an option. (1994: 40; emphasis in original) Three observations need be made. First, Wright’s elaboration remains unclear in important respects. Second, the effort to distinguish genocide and slavery is misguided, for it fails to appreciate the indistinguishable logic of both policies. Third, Wright’s definition of exploitation privileges par ticular kinds of exploitation and risks trivializing other kinds of exploitation. As a first observation, at times it is unclear whether Wright’s focus is on economic exploitation specifically or exploitation as a general concept. To the extent that this distinction is necessary, however, it suggests that Wright’s analysis has some of the same features that present great difficulty for mainstream social scientific analyses of discrimination. There is little value in definitively identifying economic as opposed to noneconomic exploitation, just as there is little value in identifying economic as opposed to noneconomic discrimination, unless one adopts the atomistic, identify-the-perpetrators approach that characterizes the problematic analyses of discrimination analyzed earlier. Hence, this move presents problems. The second problem is that the attempt to distinguish analytically between genocide and exploitation appears misguided. Wright seems to lose sight of the many cases in which genocide and slavery co-occur, at least in the modern period, including Nazi Germany—the intellectual forebears of South African apartheid—and their enslavement of and genocide against Jews, Roma, homosexuals, communists; and the United States, during which whites enslaved Africans and visited genocide upon indigenous North Americans. These points on the continuum of discrimination frequently co-occur because they are different manifestations of the same limiting view of persons, the same damaged social relation (i.e., discrimination). Wright writes as if the factors that determine which will occur are logical principles that follow from a par ticular system and economic calculus but, in actuality, the factors that determine which will occur are the on-the-ground strategic and tactical decisions of powerful actors choosing not whether but how to exploit those in their midst, choosing which to enslave and which to eradicate. Under Wright’s definition, actors decide on either genocide or slavery based on an efficiency-focused calculation of their economic interests, a calculation that in conditions of labor scarcity regards workers of more value than corpses. Yet, as scholars have documented, during World War II, surely a time of labor scarcity, the Nazis repeatedly ignored explicit pleas of businesspersons to save Jewish craftsmen from the Holocaust (e.g., Browning 1998: 19–25, 136–137). Similarly, tactics, emotion, and a fundamental disregard for human life operating within a political economy in which both

198 / Chapter 7

financial and ideological currencies matter better explain why Native American nations who offered aid to Euro Americans were not spared the genocidal actions of the fledgling nation, and why perhaps one million Africans died during the Middle Passage (Haines, McDonald, and Shlomowitz 2001)—not genocide, but genocide-level numbers inconsistent with economic concerns with efficiency. Genocide and slavery are both discrimination, occurring owing to a limited view of persons. Both constitute Roemer exploitation; both are social arrangements the enslaved or the murdered would have rejected had they been afforded an original egalitarian distribution of power from which to make a choice. Third, Wright encounters these problems because he seems to be attempting to do for exploitation what Becker did for discrimination. In doing so he seems to define as nonexploitation any phenomenon that does not entail economic transfers. Under Wright’s analysis neither the parent-child social relation, nor the student-teacher social relation, nor the doctor-patient social relation, nor any other social relation, neither definitively nor specifically, can entail exploitation, to the extent the social relation at issue is not economic. This definition would lead Wright to define away the possibility of exploitive child abuse. The basis of this denial is the requirement that exploitation entail the exploiters’ appropriation of the labor effort of the exploited. Clearly, child abusers do not always appropriate the labor effort of the children they abuse—the child may be wholly inactive or perhaps even unconscious during the abuse. If we adopt Roemer’s definition and ask would the child have been better under an egalitarian first distribution (in this case, of power), we would answer in the affirmative, and therefore can regard the child-abuser as an exploiter. If one takes Wright’s perspective, however, one is forced to regard child abuse, and many other facially exploitive circumstances, as “something else.” Wright’s position seems to turn on a commitment to privileging labor effort, a classic Marxist aim. It is not possible necessarily to fathom a scholar’s motivation, but one logical reason for privileging labor effort is to believe that doing so will allow the organization of work, and the development of class consciousness among the working class, to be conceived as pivotal factors in sociopolitical change. Thus, the structure of his argument, at least with respect to exploitation, seems consistent with the prophetic aim of some Marxist thought. Yet, the child-abuse example suggests that defining exploitation in this manner is a mistake. The lesson to learn is that any definition of a general oppressive phenomenon, such as discrimination or exploitation, is fl awed if it is based in an a priori judgment of how the general phenomenon will be manifest in a par ticular sociocultural milieu at a par ticular historical

Discrimination as a (Damaged) Social Relation / 199

moment. Wright’s approach would force one to treat discrimination as a noneconomic oppressive relation, introducing an unnecessary complication. In any given instance, discrimination may indeed be a noneconomic oppressive relation, but it is not definitively a noneconomic oppressive relation. Thus, Wright’s analysis ignores the easily established fact that Becker also generally does not acknowledge: Sometimes, and in some places, discrimination may be an economical ly driven factor, whereas at other times and places it may not be. Thus, definitions of phenomena must be applicable broadly, or they risk incoherence as conditions develop. This indeterminacy has major implications. It means that only a holistic analysis can hope to encompass the full range of discrimination and thereby accurately reveal the effects of discrimination; atomistic analyses are so likely to misestimate the effects of discrimination, by missing aspects of the phenomenon rendered invisible by analysts’ categorical schemes, that their conduct should be ruled out of bounds. In sum, contra Wright, there is no reason to rule a priori on the place or role of discrimination in a society. Wright’s view of exploitation requires one to make many such a priori distinctions. In contrast, and returning to discrimination, it would appear that using Roemer’s defi nition of exploitation reveals that the social relation of discrimination is an exploitive social relation. Like those who begin under a Rawlsian veil of ignorance, agents in Roemer’s analysis would be unlikely to accept social arrangements that would allow them to possibly end up worse off than they would have been under an egalitarian first distribution of resources. Is Class Discrimination Possible? This is a difficult but important question. Given Marx’s claims concerning alienation, class relations are damaged social relations. However, it is possible for a social relation to be damaged but not be discrimination. Damaged social relation and discrimination are not synonyms. Recall that the working definition of discrimination is that it entails distinction made on grounds of natural or social categories, which have no relation either to individual capacities or merits, or to the concrete behavior of the individual person, but instead are based in a limiting view of some types of persons. The logic of capitalism asserts the power of financial resources to secure goods and services, and the basic inability of anyone to claim a right to any resource or ser vice absent the ability to pay. To the extent this logic is foundational to any really existing capitalism—and this is not a given, as different types of capitalism moderate this stark principle in various ways — capitalism would seem to require the existence of the damaged social relation of class. However, because the logic of the economic system of capitalism is that

200 / Chapter 7

fi nancial resources are necessary to receive goods and ser vices, the class of a person is, by definition, not an irrelevant category. In other words, under capitalism, class does have a relation to the capacities of persons. In this view it would seem impossible within capitalism to discriminate on the basis of class, because class relations are foundational to capitalism, yet discrimination is defined to involve “distinction made on grounds of natural or social categories, which have no relation either to individual capacities or merits, or to the concrete behavior of the individual person” (as stated earlier in this chapter). As class does have a relation to individual capacities or merits in capitalist society, in capitalist society, class discrimination would seem impossible by definition, even while class distinctions seem inevitable. This conclusion, of course, still would allow the social relation of class to be a damaged social relation, worthy of concerted action against it by those so inclined. The problem with this view, however, is that it violates a foundational principle described earlier. The problem is revealed if we consider the prospect of a state that has a rigid sex hierarchy. In such a state, access to goods, ser vices, authority, and more would be contingent on one’s sex. If so, by the reasoning described above, in such a state, sex discrimination would be impossible, because sex would be as fundamental to that state’s logic as class is to capitalism. Yet, we already have noted that legal mandates, likely imposed and administered in a complex process by privileged elites, cannot be the final arbiter of whether and how a social phenomenon exists. Thus, allowing the logic of capitalism to make class discrimination impossible within a capitalist economy would force us to allow the “logic” of Nazism to make anti-Semitism impossible within the Nazi state, the “logic” of misogyny to make sex discrimination impossible within the misogynist state, and so forth. Because legal mandates cannot be the arbiters of social phenomena, we must reject this line of reasoning. For this reason, therefore, class discrimination must be a logical possibility. And, this implies that the capitalist form actually is erected on a foundation of damaged social relations that may entail a foundation of class discrimination. Summary In sum, alienation, exploitation, and class are key concepts in Marxist thought. Focusing on these concepts reveals that discrimination, defined as a damaged social relation, falls within the core of Marxist concerns. When this is done, discrimination is no longer simply a tool capitalists may use to divide the working class; discrimination, instead, becomes an important manifestation of humanity’s alienation. The damaged social relation of discrimination may be manifest inside or outside the economy, and may entail a variety of exploitive arrangements.

Discrimination as a (Damaged) Social Relation / 201

Other Perspectives As I indicate above, there is no end to the perspectives with which one may usefully dialogue at this juncture. Habermasian positions might draw our attention to the way in which discrimination as a damaged social relation endangers the chance of communicative rationality, makes impossible the mutual recognition of others, and at best installs a politico-legal framework that ignores the multiplicity of cultures and values of those to be governed by it (e.g., Habermas 2000; 2003). A position consistent with Martin Buber ([1923] 1970), who calls on persons to orient to others as ends in and of themselves, might regard discrimination as an essentially means-based relation that perverts human interaction, a relation in which others are simply means and never ends in themselves. A perspective in line with the thought of Adam Smith (Fleischacker 2004: 72–80), who assumes a necessary equality for appropriate economic interaction, might fi nd discrimination an abhorrent social relation, the likes of which no consensus is sufficient to allow. The effort to consider discrimination as a damaged social relation in light of various philosophical systems is worth continuing, for important insights may be unearthed in the effort. But, our considerably more modest purposes have been met by the brief foray into that terrain. Discrimination can be understood as a damaged social relation given at least some existing, more or less coherent philosophical systems. We now may turn to consider how this understanding of discrimination changes the task of the empirical researcher, an issue we take up in Chapter 8.

Concluding Remarks Discrimination is a social relation, a relation between types of persons. And, it is a damaged social relation, a social relation that necessarily entails stunting the development of some parties to the relation. I draw on Allport to develop a working definition of discrimination: Discrimination entails distinction made on grounds of natural or social categories, which have no relation either to individual capacities or merits, or to the concrete behavior of the individual person, but is, instead, based in a limiting view of some types of persons. The working definition of discrimination appears in harmony with the research claims of the critical legal, critical race, and feminist scholarship discussed in Chapter 5. Both the critical perspectives as well as the Allportian perspective understand discrimination as deeply social, as inherently involving types of persons, not individual members of society. These observations give rise to an incontrovertible if possibly controversial fact: As a social relation, it simply is not possible for generally pervasive

202 / Chapter 7

reverse race or sex discrimination to operate. In order for allegedly reverse race or sex discrimination to occur, a social relation would have to come into existence in which dominant groups—whites and men—are locked out of positions of authority and become so rightless in their own society that they become subject to capricious search and seizure, involuntary segregation, and threats of genocide. As it is unlikely white males would commit raciogender suicide as a group, the brunt of these actions would have to be carried out by nonwhites and women. On the basis of race and sex, placing such constraints on white and male freedom is just not practically possible, though it may be possible on the basis of class or other cleavages, an observation that implies the phenomenon would pit white males against white males to some extent. Of course, given the number of people in the United States, I have no doubt that some whites are disempowered in par ticu lar instances by nonwhites, by procedures that resemble the kinds of actions one might regard as discrimination under the individualistic definition of discrimination. And, I have no doubt that some men are disempowered in par ticular instances by women, by procedures that resemble the kinds of actions one might regard as discrimination under the individualistic definition of discrimination. But, whites qua whites and men qua men, to the extent their consciousnesses are grounded in rationality, never seriously entertain the possibility of falling victim to slavery or genocide; never entertain the possibility of the establishment of norms, values, and matrices of interaction that systematically disempower them in nearly every instance of their life, and do so on the basis of their race or sex rather than some other sociodemographic characteristic. These possibilities are inconceivable and, thus, entertaining them appears irrational because any individual instances of denial of opportunity are not discrimination; they involve individuals, not a pervasive social relation. Therefore, when individual men or whites are treated poorly in a par ticu lar instance, it obviously is not positive for human development—no one should be treated poorly. But, as the foregoing discussion of the multiplicity of grounds of exploitation suggests, there are many ways in which poor treatment can arise, and discrimination is not the only basis for poor treatment. Further, when members of dominant groups are treated poorly by subordinate groups, the act should be disavowed. But it is a mistake to regard any treatment that should be disavowed as discrimination. Discrimination is not the only social phenomenon worthy of disavowal, and disavowal does not imply that the social phenomenon is discrimination. Discrimination can only occur when a pervasive social relation structures the interaction and grounds it in a limiting view of persons. Absent the codification of norms—written or unwritten—disparate treatment fails to become discrimination. It fails because it cannot concern

Discrimination as a (Damaged) Social Relation / 203

types of persons unless it transcends the interaction of the individuals immediately involved to tap into or become solidified norms, values, and matrices of supporting mechanisms. Clearly, the centuries of history, briefly referred to at various places herein, indicate that antiblack discrimination and antiwomen discrimination meet this standard. Though discrimination is not the only phenomenon that may be destructive of human flourishing, it is one such phenomenon. I treat just a few of the existing coherent philosophical systems of import in the United States to illustrate that most persons, to the extent they adhere to a coherent philosophical system, would conclude discrimination is a damaged social relation. I am sure one might find a system for which this conclusion does not hold, but the challenge is to find a coherent system for which this conclusion does not hold that otherwise is supportive of the development of all humans to the extent possible within the limits imposed by the material constraints of the environment in which we live. I doubt such a pre-existing system is present, though, given the unbounded creativity of human beings, perhaps in response to these and similar observations one may be constructed. Of course, it is clear that there are some philosophical systems that might admit discrimination as a defensible social relation; utilitarianism furnishes one such example. But utilitarians probably would not deny that discrimination damages at least one party to the relation. If so, then utilitarians are in the position of accepting every claim heretofore developed, only to maintain that the costs (often to others) are affordable (to them), a dubious externalization of costs of a social arrangement if ever there were one. We also can observe lived philosophies by attending to such historical actors as Thomas Jefferson. One might argue that persons such as Jefferson somehow simultaneously maintained the existence of discriminatory structures and the language of justice and liberty. It is possible, therefore, to make the case that such a system did not affirm the damaged nature of the discrimination relation. Yet, juxtaposing the philosophical statements and the lived life of its adherents indicates that the system is, to put the matter charitably, incoherent. The case for incoherence is easy to make; the words and the deeds were not reconciled. Thus, the philosophy did not meet the coherence standards needed to be considered herein. All these observations culminate in the likelihood that discrimination is not a life-affirming social structural phenomenon. This may seem an unnecessary argument to make, but it became necessary owing to the ways in which other theories of discrimination seem to equate the salubrious social relation of marriage and the damaged social relation of discrimination. It should be obvious that discrimination is damaging in its foundations if not in its effects, yet the last half century of social scientific research on discrimination has proceeded as if this may not be the case. Hence, it became extremely

204 / Chapter 7

important to affirm the inherently damaging aspect of discrimination, as well as the social relational character of the phenomenon; after doing both, distinguishing healthful and harmful social relations became an essential element of our task. Finally, as social structures exist for people, not the other way around, one can conclude that a phenomenon such as discrimination, which does not appear to support human flourishing, is damaged in some way. And, as discrimination is a social relation, it is, therefore, a damaged social relation. Further, and most important, the persons damaged by the damaged social relation likely are not only those who are targets of discrimination, not only those in the target category, but any who seek a free and easy interaction with others, who might learn something from someone else, who might gain by a fluid association with other caring human beings. In sum, discrimination is a social relation, a relation between categories of persons. And, it is a damaged social relation, a social relation that necessarily entails stunting the development of some parties to the relation. What does this mean for the effort to estimate effects of discrimination?

8 Epistemological Foundations for Studying Effects of Discrimination as a Social Relation

S

ocial scientists have severely criticized existing methods of estimating discrimination effects. Proposals to solve the problems run the gamut: Glen G. Cain (1991) calls for analysts to make qualitative additions to standard statistical analyses; Charles N. Halaby (1979) suggests analysts relax the standard of proof; Roslyn A. Mickelson (2003) advocates reversing the burden of proof; Barbara R. Bergmann (1974) calls for analyses closer to the site of discriminatory behavior; and Eleanor Engram (1982) contends that every difference should be judged a result of discrimination. Each solution has advantages and disadvantages, but none confronts the disutility of the atomistic framework for the social scientific analysis of discrimination effects while preserving common standards of proof in the social sciences. If the atomistic framework is rejected, however, then none of these fi xes is required, nor is it necessary to fully embrace the critical legal scholars’ attack on the U.S. legal system. Using critical legal studies and Gordon Allport’s classic analysis to conceptualize discrimination as a damaged social relation makes possible strategies of estimation that can avoid ad hoc qualitative judgments, preserve accepted social scientific standards of proof, acknowledge additional differences between people’s backgrounds and experiences, and still deepen our understanding of the effect of discrimination. Taking discrimination as a damaged social relation, how may analysts study discrimination effects? The short answer is, not by attempting to

206 / Chapter 8

insert the new definition into existing practice. Research is a complex edifice, ideally built self-consciously from the ground up, with each plank reinforcing and being reinforced by others. The dominant definition of discrimination joined with and reinforced other research commitments, making certain research operations sensible, while rendering others erroneous or even irrational. Now that we have excavated the foundations of contemporary discrimination research, we not only need to establish a new foundation; if we hope to construct a coherent framework for the analysis of discrimination effects, we also need to configure a superstructure of research operations consonant with the new foundation. To accomplish this aim, we need attend to several interimbricated elements of research: (1) aim of research, (2) focus of inquiry, (3) morphology of measurement, (4) level of analysis, and (5) source of information. Each of these elements must be thought through in the context of the definition of discrimination as a damaged social relation. After these elements are discussed, a question arises: Does self-conscious attention to these elements ultimately aid our efforts to estimate effects of discrimination? To facilitate evaluation of the claim that it does, I first take a page from other work and foreshadow specific measurement decisions. Then, I revisit the dilemmas that have occupied mainstream research to briefly consider to what extent they remain vexing at this juncture. I then close with brief illustrative findings of a study of discrimination effects. But, before such observations can be made, how exactly must we see the aim of research?

Research Aim: Distinguishing Existence and Effects Studies of Discrimination Studies of the existence of discrimination and studies of the effects of discrimination are analytically identifiable research types with decidedly different aims. Gary Becker maintains this distinction when he argues for distinguishing the amount of discrimination and its effects. For Becker, if access to capital is unrestricted by tastes for discrimination, then there is no discrimination; whereas if equal levels of earnings accrue to equally productive labor, there is no effect of discrimination (which might occur if, for example, two groups possessed equivalent amounts of equal-quality capital while at least one group had a taste for discrimination). As Becker suggests, therefore, studying whether there is no discrimination or some discrimination is to search for the existence of discrimination; studying whether the discrimination that exists has no effect or some effect is to search for the effect of discrimination. However, most research on discrimination does not distinguish between the amount and the effect of discrimination. Instead, studies of existence

Effects of Discrimination as a Social Relation / 207

and studies of effects have become fused. But, both Becker’s original formulation and the argument I present here see effects studies and existence studies as completely different—one can fi nd no effect of discrimination but still have evidence of the existence of discrimination. Because the aims are distinct analytically, analysts should indicate which question they are addressing—the question of whether discrimination exists, or the question of what the effects of discrimination are. Accepting this distinction places discrimination on par with most other factors in the social world. In general, a par ticular factor may matter for some outcomes but not others. For example, for some outcomes parents’ earnings matter, while for other outcomes parents’ earnings have no effect. The null effect of parents’ earnings does not imply that parents do not exist, nor does it imply that parents have no earnings. The same reasoning should apply to the phenomenon of discrimination: Null effects do not imply there is no discrimination. However, if one rejects the social definition of discrimination and retains the usual analytic approaches, nodding toward Becker but ignoring his insights at a fundamental level, then studies of discrimination effects become studies of the existence of discrimination. That is, in such studies, finding no effect of discrimination is to find no discrimination. This is a problem. The problem is even more consequential than it may appear at first blush. Fusing the two types of studies implies that the epistemological requirements for research on the existence of discrimination and the requirements for research on the effects of discrimination are the same. This is an error. The two types of studies require different epistemologies, and the use of the same one for both has contributed to the difficulties researchers have encountered.

Studying the Existence of Discrimination Jomills Braddock II and James McPartland (1987) study the recruitment, hiring, and promotion behaviors of employers to discover whether and how structural features of placement processes result in race-linked differences in occupational access and reward. Braddock and McPartland identify three stages of potential employment discrimination: (1) the candidate stage, (2) the entry stage, and (3) the promotion stage. As an example of their findings, Braddock and McPartland document the existence of discrimination at the job-entry stage. In a very different study, Joleen Kirschenman and Kathryn M. Neckerman (1991) investigate the perceptions of employers with respect to race to discover the structures of thinking that undergird hiring practices. As an example of their findings, they document that employers have a rank ordering

208 / Chapter 8

of the demographic characteristics of employees, preferring Latino/a to African American workers. These two studies concur in fi nding discrimination in hiring. But how they go about fi nding the existence of discrimination is different—and I am not referring to the observation that one uses qualitative research techniques while the other uses quantitative research techniques. No, the key difference concerns their vision of what counts as evidence of the existence of discrimination. We must briefly step back from these illustrative studies. When discrimination is defined individualistically, it might seem easy to establish the existence of discrimination. All one needs to do is find an instance wherein which a qualified person is denied a job, or admission to a college, or a seat in a restaurant in favor of a less qualified applicant. However, if one observes a single instance, it can be difficult to eliminate alternative explanations for the outcome. For example, perhaps the student was denied college admission because the college has more than enough tuba players, and the student is without question the thousandth-best tuba player in the pool. Because of such complexities, in order to nail down whether discrimination exists, scholars tend to search for patterning rather than single instances, even when they conceive of discrimination individualistically. Returning to the Braddock and McPartland and Kirschenman and Neckerman analyses, it is clear that pattern-search is exactly what they do. Braddock and McPartland identify the existence of discrimination by assessing whether the patterns in their data resemble what they would expect if discrimination were operating. As they would expect fewer blacks to be interviewed, hired, and promoted in the presence of discrimination, when their data shows fewer blacks being interviewed, hired, and promoted, they conclude that they have found the existence of discrimination. Hence, their approach is an effect-based one; in this case, they conclude a par ticu lar process is discriminatory because the results resemble what they expect to be the effect of a discriminatory behavior. Kirschenman and Neckerman search for the existence of discrimination differently. They use a definition of discriminatory treatment that focuses on discriminatory actions and perspectives themselves; for example, if employers reference race in ranking their preference for employees, that is discriminatory. Hence, when Kirschenman and Neckerman find a pattern of employers claiming to prefer hiring members of one racial/ethnic group over another, they can conclude that they have found evidence for the existence of discrimination. In other words, Kirschenman and Neckerman use a phenomenonbased approach, not an effect-based approach; the approach is phenomenon based because they document the existence of the phenomenon directly, rather than indirectly by observing its presumed effects.

Effects of Discrimination as a Social Relation / 209

Of course, in a sense, all research is effect based. Analysts only apprehend the social or natural world owing to their sensory organs. The phenomenon of interest has an effect on the senses, either “directly” or via some measurement apparatus, and we know the world thereby. Thus, one could say that all research is effect-based. Though this is true, once we accept this fundamental truth, an important distinction between effect-based and phenomenonbased research remains. Yet, as evidenced by Braddock and McPartland on the one hand and Kirschenman and Neckerman on the other, both phenomenon-based and effect-based approaches can be useful for research on the existence of discrimination. Despite their different approaches, these two studies concur in finding the existence of discrimination. And though different readers may find one approach more persuasive than the other, with respect to finding the existence of discrimination, the approaches would seem to have different strengths and weaknesses. A central strength of phenomenon-based approaches for the study of the existence of discrimination is that one may construct a plausible theoretical edifice and then search for the phenomena defi ned within that edifice. If one fi nds the theoretically identified patterns, then, to the extent the theoretical edifice appears plausible and the research operations are solid, acceptance of the existence of the phenomenon is secured. An example in physics would be the theorized existence of a neutrally charged, heavy, elementary particle. To date, existence of the Higgs boson has not yet been confi rmed (e.g., Ananthaswamy 2007), but analysts have theorized its existence (e.g., Higgs 1964). Should a particle with its theorized pattern of characteristics be observed, then the existence of the Higgs boson will be confi rmed. A central strength of effect-based approaches is that as long as one is able to specify what one would expect to be the result of the phenomenon, one necessarily does not need to observe the phenomenon to conclude that it exists. To the extent that one’s expectations are statable and defensible, and the research operations are solid, then the conclusion that the phenomenon exists is secure if one fi nds the effects that one expected to fi nd. Certainly, it may be necessary to have more and more complex specifications structuring the observation mechanism. Yet, effect-based approaches that successfully construct such mechanisms can document the existence of phenomena. An example in physics would be if analysts were to theorize the existence of an entity with a gravitational field so strong that light, the fastest-known phenomenon in the universe, could not escape, and then were to obtain observations reflecting such an effect, then they could conclude that the theorized entity—the black hole—exists. Thus, black holes provide an example of effect-based reasoning in physics. So, at base, particle physics, probing the essence of matter and energy in the universe, is

210 / Chapter 8

both a phenomenon-based and an effect-based science when it comes to ascertaining the existence of entities. Hence, it would appear that both phenomenon-based and effect-based approaches may be appropriate to the study of the existence of phenomena. It is apparent that both phenomenon-based and effect-based approaches may be useful for identifying the existence of discrimination. Research on the existence of discrimination increases our understanding of the contours of discrimination, as well as the foundation of discrimination in prejudice and experience. On the basis of such evidence, many appear to accept the continued existence of discrimination in the United States.

Studying the Effect of Discrimination with Effect-Based Approaches Unfortunately, matters are more complex when one turns to study the effects of discrimination, because effect-based studies of the effects of discrimination are hopelessly tautologous. For studies of effects, the question of interest is: What is the magnitude and direction of the effect of discrimination? Basically, effect-based approaches do not work for studying the effects of discrimination because effect-based approaches essentially require a researcher to ask others to accept his or her estimate of the magnitude and direction of the effects of discrimination because the estimate is what the researcher expected the magnitude and direction of the effects of discrimination to be! Notably, effect-based studies of discrimination effects could never allow one to securely infer that effects of discrimination were zero. The inference would remain insecure because it would be impossible to determine whether the effects of nonzero discrimination were zero, or whether the level of discrimination itself was zero. It is possible, however, to construct a theory of knowledge production for effects studies that will not be tautological, as discussion of the focus of inquiry indicates. But, the point here is that we must ask explicitly: Are we researching the existence of discrimination? Or, are we searching for the effects of discrimination? Both questions are important. But they are not the same.

A Cautionary Counsel: A Cause-Based Approach to Discrimination Research Analysts have become accustomed to studying race and gender inequality, and some have attempted thereby to make progress in understanding discrimination. Lacking indicators of discrimination, but being aware of pervasive race and gender inequality, analysts have used such inequality to focus

Effects of Discrimination as a Social Relation / 211

their work. Focusing on inequality, however, causes several problems, and may foreclose several areas of vital research because it will direct attention only to areas where we observe race or gender inequality. Yet, if discrimination is as powerful as the painful history surrounding it in the United States suggests, its effects may not be confined to areas where we observe race or gender inequality. Of course, we already have indicated that effect-based studies of discrimination effects are tautological. I submit that for studying effects of discrimination, analysts would be wise to eschew the effect-based approach to investigation, and follow a cause-based analytic strategy (i.e., to identify explicitly the phenomenon of interest with some theoretical specificity, and to then search for differences that can be linked directly to that phenomenon). In this chapter, I trace out in more detail what I mean by this cautionary comment, what I believe implementing a strategy sensitive to its implications will entail for discrimination research, and why I contend such a strategy is necessary. Essentially, I call for phenomenon-based studies of the effects of discrimination and a rejection of effect-based approaches. When the study concerns a posited causal factor, defined using a phenomenon-based approach, we might usefully term the research strategy cause-based. Cause-based research is the application of a phenomenon-based posture à la Kirschenman and Neckerman to the study of discrimination effects. In a cause-based approach, the theorist needs to argue that the phenomenon actually exists (e.g., discrimination exists). Studies of the existence of discrimination, such as those of Karen Anderson (1981); D’Ann Campbell (1984); Valerie E. Lee, Helen M. Marks, and Tina Byrd (1994); Joe Feagin (1991); Kirschenman and Neckerman (1991); Feagin and Hernán Vera (1995); Eduardo Bonilla-Silva (2001); and many others can be useful for making this argument. After identifying the phenomenon, cause-based researchers may ask which outcomes are affected by the phenomenon (e.g., are earnings affected by discrimination?). If an outcome is affected by the phenomenon, then the phenomenon is a cause in the social world. Two published studies of other phenomena illustrate the difference between effect-based and cause-based research. In The Black-White Test Score Gap, Christopher Jencks and Meredith Phillips (1998) coordinate a detailed, wide-ranging effect-based study of racial differences in measured achievement. They note that racial differences in test scores exist, and the contributors to the volume endeavor to explain the difference. The strategy of explanation calls on researchers to search for factors that also differ by race, in an effort to introduce enough of those factors into models to eradicate the test-score difference. This fine study is an exemplary collection of effect-based research, effect-based in that the analysts ultimately are interested in explaining

212 / Chapter 8

achievement. Thus, all studies focus on the same dependent variable—the outcome to be produced—and each analyst considers alternative factors that might play some role in the outcome. In Help or Hindrance? The Economic Implications of Immigration for African Americans, Daniel S. Hamermesh and Frank D. Bean (1998) coordinate a detailed, wide-ranging cause-based study of the effects of immigration on the socioeconomic prospects of African Americans. They note that immigration to the United States exists, and the contributors to the volume are interested in ferreting out any positive or negative effects for African Americans that may follow from immigration. Thus, they investigate several outcomes— high school completion, college entry, wages, crime, job quality, and more. This fine study is an exemplary collection of cause-based studies, cause-based in that the analysts ultimately are interested in determining whether the phenomenon of focus—immigration—has any effects in the social world. Thus, all studies focus on the same independent variable, the factor that may have causal force, and each analyst considers different outcomes for which the posited factor might play some role. Clearly, the previous two examples suggest that both approaches can work. However, the effect-based approach is problematic for studies of discrimination for several reasons; in comparison, the cause-based approach has several advantages for discrimination research. Consequently, I contend that, for estimating discrimination effects, analysts need to use the causebased approach, and avoid the effect-based approach. One advantage is that the existence of phenomena cannot plausibly be questioned as easily as effects can be questioned, partly because research on the existence of a phenomenon tends to be more solid. In the case of discrimination, it is comparatively easy to determine whether a posited social fact, such as the historic enslavement of and continued discrimination against blacks in the United States, is true or false. It is comparatively easy to determine whether or not women lost the right to own property upon marrying at par ticular times in certain societies. Whether the enslavement of blacks and the abuse of women in marriage had effects is an empirical question, but that blacks were enslaved and women were (and are) abused is a matter of record (Washington [1787] 1976; Walker 1979). A cause-based research program benefits from the relative difficulty of denying known conditions and incidents. Clearly, it always is possible to deny a phenomenon—some still appear to deny that the earth is more or less round. But, no matter how easy it may be to deny a phenomenon (e.g., the torture of prisoners), it is easier to deny effects (e.g., the psychological ill effects of torture on those same prisoners). One might respond by contending that differences also are relatively easy to notice. Yet, this ease of notice presents two problems. First, tracing differences back to other phenomena, and thereby making them effects of a particular

Effects of Discrimination as a Social Relation / 213

cause, is difficult, a difficulty that contributes to making effect-based approaches vexing to use in some (but not all) areas of research. Jencks and Phillips are able to use an effect-based approach to study racial differences in test scores because the evidence of racial differences in test scores is accepted widely, even though the meaning of those differences is contested (e.g., Schiele 1991; Lucas 2000). Even with the advantage of consensus that the test-score difference exists, and setting aside the claims concerning bias and test construction, however, by the end of their volume we scarcely have begun to understand what underlies the racial difference in test scores. Never are the differences fully or even largely explained, and most of the explanations offered remain reasonable hypotheses, with few of them solidly established as facts or eliminated as incorrect. If indeterminacy prevails around a phenomenon so recognized as the black-white test score gap, what might we expect in an area so lacking in consensus as discrimination research? If we know there is gender inequality in earnings, that fact by itself does virtually nothing to advance our understanding of the possibility of discrimination operating by sex, because an infinite number of other causes might create the difference. Hence, as the example of The Black-White Test Score Gap suggests, pursuing an effect-based strategy essentially commits the researcher to explaining inequality, and thus commits the researcher to a potentially never-ending exploration of the multifarious factors that might produce the outcome. Such a commitment, possibly of value for something so recognized as test-score differences, is not necessary to produce knowledge. The second problem with using inequality to focus research provides an additional reason to pursue cause-based approaches. Cause-based researchers and effect-based researchers behave differently when confronted with null effects. In cause-based research, if the cause appears to have no effect on an outcome, the response is to search for intervening mechanisms that dissipate the effect, or to try to relate the cause to some other effect. Few researchers studying any phenomenon would maintain that the particu lar phenomena they study have effects on every conceivable outcome. Hence, null effects for outcomes are acceptable, and do not raise doubt as to the existence of the possibly causal phenomenon. For example, if immigration does not appear to affect an outcome, then analysts may look to potential mediating variables that dissipate the effect, or to other outcomes for which there might be effects. No one would conclude, on the basis of a finding of no effect of immigration on an outcome, that immigration is not occurring. Similarly, if discrimination does not appear to affect earnings, cause-based researchers could search for mechanisms that dissipate the effect, such as education, marriage, and more, or they may accept that result and begin investigating whether discrimination matters for health or other important outcomes. And, no one should conclude, on the basis of a finding of no effect

214 / Chapter 8

of discrimination on an outcome, that discrimination is not occurring. In contrast, when estimated effects are zero, effect-based researchers, or others for them, question the existence of the phenomenon that is alleged to be causal, at least in the case of discrimination. Thus, pursuing effect-based research of effects of a phenomenon ultimately can weaken the case for the very existence of the phenomenon. Though it is technically dubious to allow an effect-based investigation of the effects of a phenomenon to weaken one’s confidence in the existence of the phenomenon, it is a difficult-to-prevent response. Recall that effect-based approaches also are used in studies of the existence of phenomena. As mainstream discrimination research has fused the question of the existence of discrimination and the effect of discrimination, it makes sense in the dominant framework for studies that find no effect of a phenomenon to be interpreted as indicating the phenomenon does not exist. However, interpreting a finding of no effects of discrimination as evidence of the nonexistence of discrimination, correct in the mainstream analysis of discrimination, is incorrect in a broader epistemological sense. A phenomenon can exist, and not necessarily affect the specific outcome one is studying (while perhaps having other effects). Hence, questioning the existence of a phenomenon because one does not find effects of it in a par ticu lar domain, par for the course in mainstream discrimination research, is an epistemological error. In addition, many effect-based researchers behave as if they seek to understand how outcomes are produced. So, for example, they may want to learn how earnings or cognitive achievement are produced. Any factor that produces or inhibits the production of these outcomes is of interest. But asking how a given outcome is produced is not the same as asking what outcomes are produced by a given phenomenon. The first question makes effects the center of the research, producing a recipe for the outcome, while the second question makes causes the center of the research, producing a profile of the power of the causal phenomenon in social life. The price of pursuing an effect-based strategy to estimate effects of discrimination is high, so high that I believe scholars interested in understanding the effects of discrimination must pursue a cause-based approach. Effect-based research has a systematic blind spot. Because effects may be hidden by intervening processes, the effects of a cause may be larger than observed gross or net differences imply. If researchers use observed differences— inequality—to generate research questions rather than theorized potential causes to generate research questions, and if the effects of one cause are cancelled by some other cause, then neither causal relation may ever be discerned. In the case of research on discrimination at the turn to the twenty-fi rst century, the differences between effect-based and cause-based strategies may

Effects of Discrimination as a Social Relation / 215

be crucially important. The historical existence of discrimination may have fostered responses that reduce the ultimate effects of discrimination on targeted groups. For example, evidence indicates that black women in 1990 exceeded the earnings of white women (e.g., Zalokar 1990: 39), suggesting that one should not expect the targets of a par ticu lar kind of discrimination to always fare worse than nontargets. Indeed, even if targets of discrimination equal nontargets on some outcome, cause-based researchers still may discover effects of discrimination on that outcome. Yet, effect-based researchers likely will turn away from research on discrimination in a domain if equality appears to be the case, on the possibly mistaken assumption that there is nothing to explain. Of course, settled expectations of who will and will not achieve may make these observations appear untenable. But, the point remains, a coefficient of zero for race or sex does not mean that discrimination has no effect, because the zero coefficient could be the result of multiple processes such that discrimination does have an effect that ultimately is cancelled by other factors. However, an effect-based approach will interpret a race or gender coefficient of zero in ways that deny this possibility. As an example, note that the socioeconomic success of some Asian groups likely has undercut the motivation for studying the effects of anti-Asian discrimination on socioeconomic outcomes, yet it is quite possible that Asians’ (and others’) socioeconomic outcomes are affected by anti-Asian discrimination. If so, then discrimination has effects on those who may not appear to be targets of (that par ticu lar type of ) discrimination. More generally, discrimination may be of no interest to effect-based researchers when inequality is low, but discrimination remains of interest for the cause-based agenda even after one can no longer identify social relations of discrimination, for the possibility of lag-effects makes discrimination inherently interesting for the foreseeable future, far beyond the moment of Justice Sandra Day O’Connor’s suggested re-evaluation of affirmative action. These problems resemble those observed for residual attribution in Chapter 6. Indeed, their existence in effect-based studies of discrimination effects suggests that any study of discrimination effects that uses an effect-based strategy ultimately will fall prey to the same problems that render residual attribution indefensible. Thus, analysts seeking to study the effects of discrimination must adopt a cause-based approach. In sum, the differences between cause-based and effect-based approaches are large. An example will serve to indicate the different implications for research. If one sought to follow an effect-based research approach for college attendance, one would look for differences between whites and blacks and, if some were found, one would label the causes of the effects as one chose. One might label the cause discrimination, genetics, or culture.

216 / Chapter 8

A cause-based approach is different. If one wondered about the effects of genes, one would have to argue or show that genes exist, and then start with genes and search for differences in college attendance that can be clearly traced to genes. If one wondered about the effects of culture, one would have to argue or show that culture exists, and then one would have to start with culture and search for differences in college attendance that clearly can be traced to culture. And if one wondered about the effects of discrimination, one would have to argue or show that discrimination exists, and one then would have to start with discrimination and search for differences that can be clearly traced to discrimination. Central to each of these efforts would be the task of identifying variation, in genes, culture, or discrimination, and investigating the connection between that variation and variation in some outcomes of interest (in this case, college attendance) using appropriate groups for comparison. At present, some now wonder about discrimination. Those who so wonder must argue that discrimination exists—and here the studies of the existence of discrimination can be immeasurably useful—and then they must search for a relation between outcomes and discrimination. If others wonder about genetics or cultural tastes, they must follow the same logic—identify the phenomenon, measure it, and then conduct empirical research to discern effects of the theoretically defined cause.

The Morphology of Measurement: Gradational Measures, Relational Measures, and Discrimination as a (Damaged) Social Relation The foregoing argues that analysts need to apply a cause-based approach to discrimination research. Formerly, analysts may have adopted effect-based approaches because of the unavailability of indicators of discrimination. Certainly, the cause-based approach is easiest to implement when one has measures of the potential cause. This raises the key question, of course: How may one measure discrimination? At the outset, to measure discrimination, defined as a damaged social relation, one must deeply appreciate the distinction between gradational and relational phenomena. A great deal of work has concerned the distinction between relational and gradational phenomena. For example, the debate between neo-Weberian and neo-Marxist analysts of socioeconomic status and class is, in part, a debate about whether economic inequality and power in societies is best conceived in gradational or relational terms (Grusky and Sørenson 1998). The content of that debate does not concern us here; the important observation is that no such potentially clarifying debate has occupied discrimination researchers, to the detriment of our understanding of discrimination.

Effects of Discrimination as a Social Relation / 217

Fundamentally gradational phenomena are defined in terms of more or less. In contrast, fundamentally relational phenomena are defi ned in terms of the (social) relations upon which they are based (Wright 1979). Because no one person ever is discriminated against as a par ticular individual—classes of persons are discriminated against, not individuals—discrimination fundamentally is a relation between types of persons, and, as such, is best measured relationally, not gradationally. Gradational measures focus on calibrating higher and lower, more and less. Earnings provide an example of a gradational phenomenon often measured gradationally. Earnings typically reflect amount obtained. Further, for earnings there is virtually universal agreement that $1 million in earnings is greater than $1,000 in earnings, an agreement shared among millionaires and nonmillionaires alike. We have noted that the spousal relation is a social relation. Thus, marital status (single, married, divorced, widowed, cohabiting) is relational, and researchers typically reflect this relational character by not attempting to construct scores that would purport to indicate which persons are more married and which are less married. As a relational phenomenon, discrimination also does not fundamentally concern “amounts,” even as one can recognize more and less discriminatory environments. However, our knowledge is not developed sufficiently—and may never be developed sufficiently—to allow the kind of precise calibration that is routine for gradational indicators such as earnings, test scores, proportion of voters supporting a referendum (e.g., Smith 2007), and more. Yet, if the phenomenon is relational—and we have maintained that it is—gradational precision easily could be misplaced precision. Deeply appreciating the relational character of discrimination entails noting this implication as well.

Level of Analysis for Discrimination as a Social Relation As a social relation, discrimination fundamentally is systemic (inhering in systems), rather than individualistic (occurring only between directly interacting individuals). Thus, discrimination cannot be measured at the individual level alone. Though individuals may perpetrate discriminatory acts, discrimination as a social relation transcends the persons who (knowingly or unknowingly) experience a discriminatory moment directly. This is true for both targets and nontargets of discrimination; parties to the damaged social relation need not be actively oriented to the damaged social relation to be involved unknowingly in its operation and properties. This observation follows directly from the critical legal critique of dominant

218 / Chapter 8

legal reasoning, the discussion of taken-for-granted standard operating procedures that sediment discrimination in institutional processes, and the resources developed in our discussion of social relations. For these reasons, measurement must be at the system level, not the individual level. Reconsidering the spousal relation as an example, if our question concerns who is married and who is not, and what are the effects of occupying one marital status category or another in a given society or societies, the individual-level may be sufficient for measurement. However, if our question concerns what is the character and effect of the social relation (e.g., how the structuring of marriage in society matters for well-being), then systemic indicators of that social relation in different societies or epochs are required. Assuming data were sufficient to the task, one might ask, for example, whether healthy relationships are more likely in jurisdictions allowing gay marriage or in jurisdictions prohibiting gay marriage. Of course, just as we may identify spouses and nonspouses at the individual level, we may identify targets and nontargets of antifemale discrimination at the individual level. Yet, to assess the effect of the social relation per se, as opposed to the effect of occupying one location within the categorical scheme, one needs to measure the character of the social relation relationally, at the system level. All our research on race and gender inequality is based on studying the implication of occupying one category in a categorical scheme, an assessment that requires comparing the outcomes of those who occupy the different categories (e.g., men vs. women). In contrast, as discrimination is a social relation, we need to measure the social relation at the systemic level, and then ascertain its connection to outcomes for those located within discriminatory social relations (men and women in a discriminatory environment) in comparison to those not in discriminatory social relations (men and women not in a discriminatory environment). The counsel to move to the system level contradicts the advice most consistent with the dominant perspective on discrimination research, advice that even many critics often have failed to question. Indeed, Bergmann (1974), frustrated with the slow pace of change, calls for firm-specific research on discrimination to replace large-scale surveys of workers in the United States. Thus, Bergmann argues for moving in the exact opposite direction I suggest. Bergmann seeks to find an observable mechanism of discrimination close to the site of effects. In so doing, Bergmann tacitly seeks to identify individual losers (i.e., people we can show failed to achieve because of discrimination). In order for such a strategy to work, we need to observe actual discrimination, and then watch while its effects are made visible. Thus, Bergmann’s counsel steps toward taking the individualistic tendency to the limit. Following Bergmann’s counsel is problematic for several reasons. First, by using firms one suggests that if we observe discrimination occurring, we

Effects of Discrimination as a Social Relation / 219

easily can identify its effects. However, it is far more difficult to discern effects at the ground-level than the strategy suggests. For example, we might observe the following: A woman makes a comment in conversation, a comment that is ignored; then, five minutes later, a man makes the same comment in the same conversation, and kudos as to his brilliance immediately follow. Such a pattern might be regarded as evidence of antifemale discrimination. But to trace effects on income or likelihood of dismissal to the discriminatory conversation patterns is quite difficult. In order to follow Bergmann’s strategy, we would need to collect data on who was involved in which conversations, how those conversations developed, whether participants differed prior to conversation, what the conversations concerned, and in which conversations the discriminatory pattern emerged. We also would need to consider processes of selection into firms and conversations, because recruitment into both is not random. There are many other factors that would need to be considered in the case of specific firms that do not need to be considered in the approach I outline. As this example suggests, the difficulty of identifying effects of discrimination for individual persons is large, as evidenced by the uncertainty persons report for their own attributions (e.g., Feagin 1991). It certainly is fair to say that Bergmann may not have intended to push a microanalytic focus on the dynamics of conversation. Her counsel may have entailed only looking as closely at processes of recruitment, hiring, and promotion as did Braddock and McPartland (1987). However, the logic of her counsel is for one to move closer and closer to the alleged site of discrimination, a seductive but dangerous logic for efforts to estimate effects because, as one moves in that direction, one simply will replicate existing evidentiary problems on a smaller scale, and thus fail to produce solid estimates of discrimination effects; the failure then will push one to move closer, where the same evidentiary problems will recur, to the same result. This problem is generated in part because the firm-specific focus is an atomistic, individualistic one, and thus fails to address the larger social relational character of discrimination. Not only is it exceedingly difficult to discern effects at the ground level, but also the standard of proof is far too stringent when compared to social science norms. For example, Robert M. Hauser, Shu-Ling Tsai, and William Sewell (1983) deepen our belief in a causal role of significant others’ influence. Yet to do so, they do not have to identify a par ticu lar parent passing on a par ticu lar lesson, amount of money, or iota of encouragement. Discrimination research should not be held to a higher standard of proof than other social phenomena, but Bergmann’s counsel eventuates in doing just that. A meteorological example may prove illuminating. In the wake of the 2005 North American hurricane season, the most active on record to that

220 / Chapter 8

time, many wondered which of the thirteen hurricanes and twenty-six named storms might have been caused by climate change (i.e., global warming). The implication of such a question is that one can identify par ticular hurricanes that would not have come into existence were it not for climate change. This expectation is based in a combination of insufficient sophistication and insufficient humility—the climate does not work that simply, and our knowledge is not that advanced. And, if the only way one would believe climate change were occurring would be if one could identify the specific hurricane that would not have become a hurricane in the absence of global warming, well, then one probably never will be able to accept the hypothesis of climate change. Global warming works on the systems that produce the weather, and that means that the gross number of events will be affected. Even so, we likely will be unable to identify a specific event that would not have occurred had the climate not been altered by global warming (e.g., we likely never will be able to identify the hurricane that would not have occurred absent climate change). Further, we likely will be unable to identify the nonevent that did not occur owing to climate change (e.g., we likely never will be able to identify the gust of wind blowing off the west coast of Africa that did not form a hurricane but would have formed a hurricane absent climate change!). Thus, requiring researchers to identify specific events and nonevents, and tie their specific occurrence or nonoccurrence to climate change, is a too-stringent standard of proof, simultaneously denying the complexity of systems and ignoring our burgeoning ignorance. Bergmann’s effort to move close to the site to study discrimination and find specific instances of discrimination effects is akin to trying to ascertain the effects of global warming, a climactic factor, by identifying the specific hurricane(s) caused by global warming. Yet for hurricanes, we know this standard of proof is too stringent. Our standards of proof for discrimination should not be more stringent than our standards of proof for climate change, for parental effects on children’s adult attainments, and other phenomena in the social, physical, and natural world. There is a place for firm-specific research. The value of firm-specific research, or any research that moves closer to the site of a social phenomenon, is not in establishing a causal connection, but, instead, in elaborating the mechanism through which known effects are generated. However, effects of discrimination are not known. Thus, the state of knowledge is just the opposite of what is needed in order for fi rm-specific research to be most useful. Owing to work that has been conducted on firms, therefore, at present we have a great deal of knowledge about potential mechanisms through which discrimination effects might be generated, and embarrassingly little knowledge about actual effects of discrimination. For this reason it is imperative

Effects of Discrimination as a Social Relation / 221

that analysts move to study discrimination as a damaged social relation using system-level measures of discrimination.

Measurement Implications of the Damaged Nature of Discrimination as a Social Relation As we have noted, the spousal relation is a social relation. But, because it is arguably not a damaged social relation, one usefully may ask the vast majority of adults in the society to place themselves with respect to the social relation. Measurement problems with the social relation are unlikely, as most any adult will be able to report his or her own marital status, perhaps the marital statuses of others, and perhaps even pronounce accurately concerning the deeper aspects of the spousal social relation in their society. To the extent the spousal relation is a damaged social relation—and the case that it is might be made by considering norms of power in marriage, heterosexist exclusions, and norms of childcare—what protects the social relation from the measurement difficulty is the failure of that reality to seep into wide public awareness. In contrast, discrimination has no such positive mask, and is clearly a damaged social relation. As we recognize that a damaged social relation is an exploitive social relation, it is clear that a damaged social relation places some persons in a society in a position of unjust disadvantage and others, knowingly or not, in a position of dominance. Yet, superordinate members of the damaged social relation routinely may deny its damaged social relational character in various ways, including by constructing superficial analogies (e.g., by promulgating incoherent concepts such as reverse discrimination), aggressively attempting to silence discussion of the phenomenon (e.g., by charging that someone is playing the “race” card—an inherently offensive term—at the slightest suggestion that historic and contemporary patterns of racial discrimination might be worth exploring as potentially relevant in a par ticular domain), and more. These illustrative observations strongly suggest that the damaged nature of the social relation may make it difficult to measure the phenomenon, because members of the society for which the measures are needed are implicated in the damaged social relation. To obtain reports from society’s members as to whether the social relation is exploitive is to ask persons to judge themselves to be involved in exploitation, surely an untrustworthy proposition in general, and even more so given the legal reasoning that has dominated the understanding of discrimination in the era of contested prejudice. Avoidance and dissimulation of the issue is not the only problem. Discrimination, being a damaged social relation, is maintained through behaviors,

222 / Chapter 8

processes, norms, and traditions that by their very existence tend to obfuscate the discriminatory character of the relation. Hence, taken-for- granted standard operating procedures, such as the operations constituting psychometric norm-referenced test design, appear eminently logical to their practitioners, and alternatives often appear literally irrational. Consequently, there is an inherent, thoroughgoing error implicated in expecting or requiring superordinate members of the damaged social relation of discrimination to generally, spontaneously, and consistently declare the social relation to be a damaged one. Still, any effort to ask only presumed targets of discrimination to characterize the social relation may build in the finding that the research is supposed to be designed to determine. How, then, can we obtain plausible, defensible measures of damaged social relations? Two resources that open the door to successful measurement under the conditions noted above are (1) the legitimating projects and activities of the superordinate and (2) the existence of witnesses. Here we see more implications of the changed social context for our research on discrimination effects. The decline of the regime of condoned exploitation certainly raised the likelihood that discriminators would attempt to conceal their actions. This presents problems. But, the onset of the era of contested prejudice also means that definitional battles concerning what counts as discrimination will ensue. Those battles and their results provide the social residues needed to discern discrimination at the systemic level. Thus, during the era of contested prejudice, actors pursue two different strategies to facilitate maintaining or instigating discriminatory behavior. One strategy is to hide their actions; we might regard this as the cloaking approach, a criminal act if the action has been deemed illegal. An alternative strategy is to work to make their actions become regarded as legitimate, to legalize the behavior; we might term this strategy the legitimating approach. Consequently, equipped with a specific definition of discrimination, we can discern discrimination even if the par ticular acts or procedures have been regarded as legitimate. Certainly, both cloaking and legitimating approaches have long been used, but the advent of the era of contested prejudice heightened the value of the legitimating efforts. Recall that the era of contested prejudice delegitimated a par ticular explanation for exploitative arrangements. Thus, finding some other explanation for those arrangements, and thereby relegitimating them, becomes especially important, and preferable to attempting a possibly criminal strategy of concealment owing to the chance that cloaking the behavior could be interpreted as an admission of the illegality of the arrangement should the behavior and the cloak behind which it occurred come to light. Hence, if one conceives of discrimination as a matrix of norms, values, and standard operating procedures, and one recognizes that actors seek to

Effects of Discrimination as a Social Relation / 223

legitimate their behavior by having it rendered legal, then those legitimations and legitimation projects create evidence of discriminatory climate. Thus, system-level evidence of the social relation of discrimination is greater than one might expect. In order to use these social residues, however, one must conceive of discrimination as not simply a set of acts deemed illegal; that is, legal acts and arrangements also must be regarded as possibly discriminatory, such that the only criteria relevant for categorizing an arrangement as a social relation of discrimination is the definition of discrimination. As we already have observed, if one recoils from this approach and remains committed to defining discrimination only based on the definitions promulgated by a par ticular court, then one would need to affirm that the Jews of Nazi Germany were not victims of discrimination, because nearly every act against them was rendered legal by the duly constituted authorities in power. In order not to be held captive by the courts and authorities of a given era or society, one must use a theoretically defensible, coherent, definition of discrimination. A theoretically defensible, social relational definition is articulated in Chapter 7, and it provides the kind of grounding needed to measure discrimination and resolve the problem posed by actors’ tendency to conceal their behavior. Indeed, one might go so far as to note that those who use such a coherent definition of the phenomenon, and stand ready to stand in judgment of their own society, are no longer only members of society; instead, one might even regard them as witnesses, willing to testify to the truth of the character of the social relations in which they themselves are embedded. Of course, few maintain that the way to measure discrimination is to ask persons to self report. But some “self” must report; measures of social phenomena do not come into existence on their own. The foregoing indicates the social processes that make evidence of the damaged social relation of discrimination available, and suggests what kind of respondent has standing for providing the raw data that would be needed to measure the damaged social relation of discrimination, relationally, at the system level.

Foreshadowing Measuring Discrimination as a (Damaged) Social Relation I advocate the cause-based strategy for the reasons noted above. Implementing this strategy entails moving to the macrolevel, using relational measures based on witness-provided evidence of the social residues of the legitimating projects and activities of superordinate members to the relation. These moves theoretically are defensible because, as I maintain, discrimination is a classbased relational phenomenon whose effects transcend those who are directly exposed to discriminatory acts.

224 / Chapter 8

This volume, outlining a theoretical approach, does not delve into the gritty details of operationalization. Those concerns are pursued elsewhere, where a slew of data issues can be addressed on the way to investigating the effects of race and sex discrimination on several social and economic outcomes. Even so, some general observations and illustrations can be presented here, as they may help to clarify the direction of the enterprise. I suggest how macrolevel analyses are better than analyses of single firms. The macrolevel also is appropriate because the phenomenon of discrimination concerns persistent patterns of interaction. How will I measure these persistent patterns at the macrolevel? In later work I detail and use the mea surement strategy I propose. In brief, though, I use laws and recorded behaviors to index expected intensity of exposure to the social relations of discrimination. The idea of expected intensity of exposure to discrimination may seem to contradict my contention that discrimination is a relational as opposed to gradational phenomenon. However, the contradiction is only apparent. Certainly, gradations exist in the severity of the social relation. Some epochs and contexts have more extreme versions of discrimination. Thus, there is variation in the intensity of the social relation over time and space. Still, the defi nitive aspect of the phenomenon is not spatial and temporal variation but, instead, the constitution of the phenomenon in the relations between types of persons. I use laws and recorded behaviors because both are good vehicles for identifying persistent relations between types of persons (i.e., social relations). Certainly, many persistent patterns are not turned into law; however, many are. And, while some persistent behaviors are not recorded, some are. By building indicators of discrimination from discriminatory law and behavior, it becomes possible to rely on a clear sign of persistent discrimination that has face validity. Law certainly has been used in various ways before in studies of discrimination (e.g., Leonard 1984). However, often these analyses have utilized an individualistic vision of discrimination. When the defi nition of discrimination is brought into line with the social phenomenon, the way in which the laws are used changes. Suffice it to say here that the understanding of laws in other analyses of discrimination effects has limited greatly the utility of the resource for revealing effects of discrimination. Regardless of the definition, laws allow one to consider domain-specific or broadly defined discrimination. However, a narrow interpretation of law as a measure of discrimination has its own difficulties. Domain-specific measures would be consistent with Becker’s dividing discrimination into types owing to the economic role the discriminator occupied at the moment of his

Effects of Discrimination as a Social Relation / 225

or her discriminatory act. However, this interest is inconsistent with viewing discrimination as a social relational phenomenon. A major implication of the class-based character of discrimination is that disentangling the location of discrimination (e.g., the family or the school, the school or the labor market, the labor market or the credit market, the credit market or the housing market) becomes less important empirically and less meaningful theoretically. Thus, though nothing prevents the construction of narrow measures of discrimination out of legal edicts and documents—one may develop measures of expected exposure to labor market discrimination, expected exposure to housing discrimination, and more—if theory must precede measurement, note that the foregoing provides no theoretical warrant for constructing narrow measures of discrimination out of law. Using a set of laws, but retaining the dominant, atomistic, social scientific perspective on discrimination, likely will culminate in an ad hoc mélange of internally contradictory maintained assumptions and procedures, reproducing many of the problems heretofore mentioned. Of course, laws do not allow one to measure “direct” exposure to discrimination. I use laws to index the intensity of expected exposure to discrimination. The concept of expected exposure allows one to accept the intractable difficulties in measuring direct exposure to discrimination, while still affirming the importance of researching the effects of discrimination. In addition, the idea of exposure allows one to appreciate the distinction between the amount of discrimination and exposure to discrimination. Exposure to discrimination is what happens to those in the sphere of influence of discriminatory social relations, which includes those who come into contact with a specific discriminator and those who do not. This feature is consistent with Becker’s distinction between actor and result.

Mainstream Problems: Reiterated and Resolved The dominant social scientific conception of discrimination and the concomitant approaches to studying discrimination seem to encounter three problems. First, actors tend to conceal their discriminatory behavior, appearing to render direct measurement of discrimination virtually impossible. Second, discrimination is a possibility throughout the social world, exponentially magnifying the number of possible sites to study, as well as the possibilities that further study will reveal that what was thought to be discrimination actually follows from complexities of some other previously unstudied arena. Third, Rubin’s Model, a widely accepted framework for considering causality, allegedly conclusively indicates that analysts never will be able to make the kinds of comparisons needed to establish an effect of discrimination.

226 / Chapter 8

Analysts have been creative in attempting to respond to these challenges. Some offer large-scale analyses of surveys and census data as evidence of discrimination effects (e.g., Turner 1952). But, as we have seen, others contend that effects of discrimination cannot be inferred from gross or net differences in outcomes (e.g., Sowell 1983). Still others defend large-scale analyses but suggest alternative strategies of measure (e.g., Engram 1982) or coefficient (e.g., Coverman 1986) interpretation. Others call for analyses targeted at the site where they expect discrimination to occur (e.g., Bergmann 1974). And still others call for analysts to make qualitative additions to standard statistical analyses (e.g., Cain 1991), to relax the standard of proof (e.g., Halaby 1979), or to reverse the burden of proof (e.g., Mickelson 2003) in the case of discrimination. Clearly, the “something else” one must do for discrimination research is a matter of extended debate, perhaps with no end in sight (e.g., Blank, Dabady, and Citro 2004). Yet, perhaps a social relational understanding of discrimination effectively may address these problems.

Illegality, Concealment Behavior, and Legitimation One difference between discrimination and other phenomena social scientists commonly study is that a great deal of discrimination is illegal. While it is true that one must resolve the problem posed by actors’ tendency to hide their discriminatory behavior, at least when they are aware of investigators’ presence, the importance of and the difficulty with resolving this problem depends greatly on one’s conception of discrimination. If one conceives of discrimination as only the acts of individuals, then one must either develop a means to mea sure concealed behavior, or accept that direct mea sures are impossible. Accepting that direct mea sures are impossible leaves one apparently no recourse other than the problematic research strategies already discussed. But, as I remark earlier, the era of contested prejudice is rife with efforts to conceal discrimination by legitimating it. These efforts provide the social residues witnesses may use to testify as to the character of discriminatory social relations. Thus, the illegal nature of much discrimination poses a formidable challenge, but legality is neither the only nor the most important barrier to isolating the effects of discrimination. Indeed, if criminality were the basis for the difficulty, then many entire fields would encounter similar debilitating difficulties. Yet students of criminology and deviance have been able to persuasively identify the effects of criminal behavior (e.g., Gill and Michaels 1992; Kaestner 1994). This suggests that actors’ concealment effort is not fundamentally the problem. Further, it is unlikely that the criminal character of discrimination is key to the difficulty, because discrimination has not always

Effects of Discrimination as a Social Relation / 227

been illegal, and some discriminatory processes currently are not illegal (e.g., gender-specific socialization).

The Validity of Atomism Denied As I indicate previously, the atomistic perspective on discrimination exponentially increases the loci one must assess. As illustration, consider that to establish that discrimination effects matter for access to credit one must study the credit market. However, after completing this study, it still will be impossible to eliminate the likelihood that what one has observed in the credit market really is based elsewhere, perhaps in the operation of the Realtors engaged in the housing market. Yet, once one studies the housing market, any findings there could be based in unobserved factors traceable to the labor market. Turn our attention to the labor market, and any fi ndings obtained there cannot be secured against the possibility that they are accounted for by the operation of the educational system. Move to study the educational system, and one cannot dismiss the likelihood that processes allocating persons to neighborhoods, perhaps driven by the operation of the credit market, actually underlies the phenomenon. And on and on the merry-go-round goes, as additional sites that might explain away findings observed in any particular site are proposed. This process occurs because analysts have theorized discrimination as occurring in analytically distinct spheres, at the behest of analytically separable, autonomous actors. Theorized in this manner, the only cross-sphere connection occurs by viewing the outputs of one part of the social system as the inputs of another. Hence, even if we find telltale signs of discrimination in any given sphere, the finding of course raises the possibility that the observation occurs by virtue of the operation of some other part of the system, most likely a part of the system whose outputs serve as inputs for the part of the system under study. Of course, one might regard the promise of precision that is made, but never realized, by the atomistic approach as essential for the development of policy to reduce or eliminate discrimination. One might maintain that learning that employers are discriminating, or schools are discriminating, will allow policy-makers to intervene. Two queries are raised directly by this line of reasoning. First, why assume that the locus of discrimination is relevant to the focus of amelioration? Policies of all sorts often address problems in one arena of life by altering another. Thus, identifying the locus of discrimination is not necessary for policymakers. Second, and more pointedly, how will a method of analysis that is inherently unable to secure any consensus estimate of the effect of discrimination ever produce an estimate clear enough to motivate consistent, ameliorative

228 / Chapter 8

policy response? Mainstream approaches seem well designed for hindering an ameliorative policy response, not for facilitating one. The focus on identifying separate loci of discrimination is misplaced, of course, for once one conceives of discrimination in a holistic manner (i.e., accurately), finding an effect of discrimination for a par ticular outcome does not identify which social actors produced the outcome. For example, if we find an effect of discrimination on students’ test scores, that does not mean that teachers or schools discriminated against children. It may mean that parents in discriminatory environments have insufficient resources to support their children’s development. If so, we still will have found an effect of discrimination on students’ test scores. At base, the holistic approach is not concerned with identifying perpetrators; it is, instead, concerned with identifying effects. In contrast, possibly because of its similarity to the courts’ perspective, the atomistic social scientific approach appears to foster a blame/avoid-blame dynamic. Seen in this way, the atomistic approach loses the forest for the trees, and hinders investigation of the total effect of discrimination in the lives of persons. To be sure, under the regime of condoned exploitation, an atomistic definition of discrimination and a social relational definition of discrimination seem scarcely worth distinguishing. Blacks and women were so unable to secure their rights that adopting either definition would lead to illuminating work. Yet, the advent of the era of contested prejudice renders individualistic, atomistic definitions hopelessly confused, for all the reasons discussed already. In this environment, a real difference exists between the social relational definition and the atomistic individualistic defi nitions. It turns out that in the era of contested prejudice, atomistic individualistic approaches cannot distinguish discrimination from any form of social selection. In this context, only a social relational definition can provide the resources needed to allow researchers to study discrimination specifically, in hiring, promotion, and more, and not find themselves unable to distinguish their research from the research of those who study marital selection decisions. Marital selection decisions, of course, are important; they also do not necessarily entail discrimination.

Solving Rubin’s Quandary The empirical research on discrimination has culminated in a growing acceptance that it is difficult to establish firmly the effect of discrimination (e.g., Blank, Dabady, and Citro 2004). One elegant approach to articulating this claim draws on theoretical developments in the methodological literature that purport to describe the conditions for appropriate causal inference (e.g., Holland 1986). This framework, known as Rubin’s Model, has gained

Effects of Discrimination as a Social Relation / 229

increasing acceptance in the social sciences, and threatens to render impossible any effort to estimate the effect of discrimination. Rubin’s Model maintains that in essence there are two instances to which analysts need access—the case in which a unit i is exposed to a treatment, and a case where the same unit i is not exposed to a treatment. To identify the causal effect, we need to observe both cases. The difference in outcomes between these two instances, OiT − OiC, is the causal effect of being exposed to treatment in comparison to the alternative; in experiments, this alternative is often called the control condition. The Polish fi lm Przypadek (1981; English title Blind Chance) by Krzysztof Kiéslowski nicely illustrates the inferential challenge social scientists confront. In this fi lm, the main character attempts to catch a train but encounters challenges along the way. Through the course of the fi lm, we see what different decisions would have produced for the protagonist. The difference or lack of difference in his experience identifies the causal effects of different ways of navigating the obstacles. The problem for causal inference is that we observe only one of the conditions—either we make the train, or we do not. We can see what occurs after one of the conditions pertains, but we cannot see what would have happened had the opposite occurred. Hence, and unfortunately, we appear trapped in one world, and we observe only one instance. With only one instance in hand, we have no access to the outcome that would have followed some alternative chain of events. Thus, we cannot compare the two outcomes to determine the causal effect of the chain of events that did occur. To resolve this problem, analysts invoke more or less plausible assumptions to allow themselves to leverage the one world they observe to make causal inferences about that world. In other words, they try to make comparisons between what they observe (a factual observation) and what they would have observed had the same unit been exposed to the alternative condition (a counterfactual observation). So, for example, a chemist might take a piece of iron from a shelf and pour a solution onto it. If the iron upon which the solution has been poured changes its properties (perhaps becoming weaker), the chemist would infer that the solution poured onto the iron caused the change. There are at least two ways to secure this inference. The chemist might assume that all the other pieces of iron on the shelf would respond the same way; in essence, here she would assume that the pieces of iron essentially are substitutable for each other. Hence, the chemist would make an assumption of unit invariance. She invokes this assumption by comparing the piece of iron upon which the solution was poured to the pieces of iron still on the shelf. The difference between them estimates the causal effect of the solution on iron.

230 / Chapter 8

Alternatively, the chemist could assume that the piece of iron was affected only by the pouring of the solution, and that otherwise the piece of iron before the pouring was exactly the same as the piece of iron after the pouring of the solution, except for changes caused by the solution. If the chemist makes this assumption, she invokes a temporal invariance assumption— the only aspect changing over time is the pouring of the solution onto the piece of iron. What occurs afterward can be regarded as the effect of the pouring of the solution onto the piece of iron. In social science research, the closest approximation to this chemical investigation would be an experiment in which analysts randomly assign persons to treatment and control. In such an experiment, analysts make an assumption that random assignment produces equivalent groups on average, and thus they can give one group the treatment, the other group the control, and then compare the outcomes of the groups to identify the causal effect of treatment. A problem arises, therefore, in the study of discrimination. We cannot randomly assign people to different racial groups or genders. The audit study comes closest to such random assignment. In audit studies, two individuals— arranged to be as similar as possible, but of different categories—are sent to a series of potential employers. Analysts infer that some function of the difference in number of job offers identifies the causal effect of discrimination. However, under Rubin’s Model, this inference is incorrect. Because analysts cannot change the sex, race, native language, or many other attributes of the person without fundamentally changing the person, comparing the job offers received by, for example, males and females does not identify the causal effect of sex discrimination. One cannot invoke the unit invariance assumption, the temporal invariance assumption, or any other plausible identifying assumption. These assumptions cannot be invoked because the attributes that make up the persons are not the same, as there is an ineradicable difference between them. A tag line to this causal inference theory is that there can be no causation without manipulation; therefore, because sex generally cannot be manipulated without fundamentally changing the unit, we cannot identify the effect of changing it by comparing males and females. In short, we cannot compare units that are fundamentally different in an effort to resolve the dilemma of inference in a one-world reality. Thus, the logic of Rubin’s Model imperils the entire study of discrimination effects. However, I submit that it is the individualistic conception of discrimination that forces Rubin’s Model to imply Rubin’s Quandary. The reason analysts find themselves in this cul-de-sac is that they have utilized individualistic definitions of discrimination. Given an individualistic vision of discrimination, they have applied methods that lead them to compare the outcomes of men and women or blacks and whites to see whether women or blacks suffer

Effects of Discrimination as a Social Relation / 231

from discrimination, and what the costs of facing discrimination could be in a given domain. The assumption underlying these comparisons is that blacks are exposed to antiblack discrimination and whites are not; women are exposed to antifemale discrimination and men are not. Comparing blacks and whites and men and women, in this view, is to compare people who have been exposed to discrimination to people who have not been exposed to discrimination. Such comparisons purport to reveal, for example, how a woman would fare if she were a man. These kinds of comparisons call for a traditional yet inappropriate counterfactual, because in reality we do not want to know how the woman would have fared had she been a man; we want to know how the woman would have fared had she not been exposed to discrimination. In other words, the appropriate counterfactual requires one to compare women not exposed to discrimination and women exposed to discrimination, or, at least, women exposed to more extreme discrimination and women exposed to less extreme discrimination. Note that if this reasoning is true, the only way to make the research reflect an appropriate counterfactual is to measure expected or actual exposure to discrimination and to compare the outcomes of people of the same sex or race exposed to more or less discrimination. Using an appropriate counterfactual has, as one might suspect, other positive benefits. Interestingly, use of an appropriate counterfactual would make claims about biological difference irrelevant to the study of discrimination. Indeed, any race- or sex-specific explanation would be “controlled” by using an appropriate counterfactual— comparing women to women means that differences cannot be attributed to some of the women being . . . women. Similarly, “cultural deficit” or “cultural difference” cannot explain why blacks in locations with less discrimination fare better than blacks in places with more discrimination, because heretofore those blacks in different locations generally were theorized as sharing a culture of blackness. Certainly there are local perturbations in any culture, but the theoretical perspective on race and gender inequality has painted with a broad brush, de-emphasizing local facets. For example, John U. Ogbu consistently implicates black culture as one factor causing black students’ low achievement, based on separate studies of black students of the East Coast (Washington, D.C.), the Midwest (Shaker Heights, Ohio), and the West (Stockton, California; Fordham and Ogbu 1986; Ogbu 2003; Ogbu 1991). The consistency of his fi ndings suggests the possibility of a theoretical commitment to black culture as a fairly undifferentiated phenomenon. One may look to Karolyn Tyson (2002), James W. Ainsworth-Darnell and Douglas B. Downey (1998), Philip J. Cook and Jens Ludwig (1998), and Prudence L. Carter (2005) for serious misgivings about Ogbu’s and Signithia Fordham and Ogbu’s analyses, but these works contest Fordham and Ogbu’s claim on other grounds, and do not directly address the idea of a transregional black

232 / Chapter 8

culture. Hence, an appropriate counterfactual that compares blacks with blacks and women with women would control for cultural or biological explanations, because, in the main, those theories, as currently articulated, do not claim that within the United States we should expect to find large differences in the outcomes of blacks or women owing to differences within black culture or regional biological differences for blacks or women. The implication of these observations is that nothing is wrong with Rubin’s Model. The problem is not with the model, an elegant articulation of the challenge of making causal inferences. The problem, instead, is with the dominant social scientific conception of discrimination, which forces Rubin’s Model to imply that discrimination effects cannot generally be ascertained. The dominant social scientific perspective routinely conflates race or sex and discrimination; taking that conflated understanding as true necessarily leads to an intractable problem for those who accept the veracity of Rubin’s Model for causal inference. However, once the individualistic conception of discrimination is rejected in favor of a more social perspective, a perspective that resonates with the actual operation of discrimination as a phenomenon, then Rubin’s Model directs us to compare units exposed to more and less discriminatory social relations. As those social relations vary over time and space, they are clearly manipulable, and thus fall within the class of possible causal phenomena. Once we adopt an appropriate counterfactual for studying the effect of discrimination on women and blacks, it becomes obvious that one also might devise an appropriate counterfactual for studying the effect of antiblack discrimination on whites, and for studying the effect of antifemale discrimination on men. It becomes clear that any group may suffer by virtue of being in an environment within which some people are exposed to discrimination; the extent to which any group has losses associated with discrimination becomes a tractable research question. In short, adopting means that allow one to invoke the appropriate counterfactual push us beyond the current state of research on discrimination. This is a direction we need take to fully investigate the possible effects of discrimination on the outcomes of persons. Intriguingly, it is a direction consistent both with some of the most promising aspects of Becker’s original formulation—aspects the empirical literature has ignored—and it is a direction in line with the more social understanding of discrimination articulated by critical theorists in their critique of dominant legal thought.

Results of an Illustrative Example: 1990 Earnings A study utilizing methods consistent with the theoretical claims espoused herein investigated the effect of race discrimination on yearly earnings of black males, white males, black females, and white females in the United

Effects of Discrimination as a Social Relation / 233

States in 1990. Data limitations made it impossible to study other racial/ethnic groups. The investigation reveals that black males in more discriminatory environments earned 20 percent less than did black males in less discriminatory environments. Interestingly, white males in environments with more antiblack discrimination earned approximately 8 percent less than did their peers in less racially discriminatory environments. Females also earned less owing to racial discrimination, with black females earning approximately 15 percent less, and white females earning approximately 8 percent less in more racially discriminatory environments than in less racially discriminatory environments. How can this be possible? Here, only a suggestion or two will have to suffice. If we understand that discrimination is not only what one person does to another but is, instead, a massive climactic condition of the environment, then we open the door to seeing how that climate affects everyone in the environment (given data limitations that may exist). When we see discrimination in this way, we come to recognize its effects may range from reducing economic efficiency (as neoclassical economists may maintain) to creating competition that saps the power of workers to struggle for wages and better working conditions (as Marxists may maintain). Notably, both the Marxist and neoclassical economics story make sense of the findings, agreeing that discrimination could lower average incomes for everyone in discriminatory environments. Yet, the dominant means of analysis systematically denies such possibilities and hides such patterns. If we want to understand the effects of discrimination, therefore, our methods of analysis must change. At this juncture, of course, one cannot assert exactly what lies behind these particular findings, not to mention explore how other complexities of the analysis deepen our theoretical challenge. But, these results indicate the error with estimating the discrimination effect by comparing targets and nontargets. The illustrative results suggest the standard estimation strategy may easily underestimate the effects of discrimination on targets and erroneously fi x the effects of discrimination on nontargets at zero. As such, the standard methods, systematically producing erroneous results by implying no costs of discrimination for nontargets, harm our understanding of discrimination. Sadly, the harmed understanding of discrimination effects may lower the likelihood of the development of creative responses to the costs of discrimination, while tragically undermining any incipient groundswell of public support for eradicating the damaged social relation of discrimination.

Concluding Remarks The end of our theoretical journey is at hand. In this chapter, I indicate what I believe are the implications for analysts who seek to take seriously the social

234 / Chapter 8

relational character of discrimination. It should be clear that comparisons of whites and blacks or men and women, no matter how simple or complex, never will allow defensible estimation of discrimination effects, as long as the epoch for which the estimation pertains can be characterized as an era of contested prejudice. Yet, even if one does endeavor to construct and use appropriate counterfactuals, and to view discrimination as a damaged social relation, this will come to naught if analysts invoke an effect-based strategy; retain an individualistic, sphere by sphere approach; or otherwise uncritically adopt standard operating procedures of existing research for studying the effects of discrimination. Such strategies will leech away whatever advantages one might have obtained by virtue of conceiving of discrimination holistically as a social relational phenomenon. Though these requirements are quite severe, they are no more demanding than the requirements analysts must meet to effectively study many other phenomena in the social sciences. The alternative, at present, would appear to be to accept the current dilemma as insurmountable. We reached this place, where no estimate of discrimination effects can find acceptance after Becker’s seminal theoretical move stripped motivation from the equation and atomized the conception of discrimination, a move that culminated in consigning discrimination to seemingly perpetual status as an omitted variable in statistical and nonstatistical analyses, a status from which no persuasive claims about effects can be sustained. We need not make that mistake again. If we want to consider and estimate effects of discrimination, we need measure the phenomenon of discrimination. And, if we want to measure the phenomenon of discrimination, we need measure the fullest range of the phenomenon we can, such that our measure taps the phenomenon as what it is—not only as incidents of interpersonal interaction, not only what one person does to another, but also and most importantly, as a social relation, as a matrix of norms, values, and support mechanisms that are erected by social persons, norms, values, and mechanisms whose existence outlives the par ticu lar momentary acts that give birth to them. All parties to the social relation can then be studied to discover the possibly damaging effects of a damaged social relation on all.

9 Theorizing Discrimination in an Era of Contested Prejudice

W

e have come a long way. In considering discrimination we have used survey research, studied historical treatments, weighed dominant legal approaches and alternative perspectives, dissected economic theories, constructed a thoroughgoing defi nition of the phenomenon, placed the definition into dialogue with a few coherent philosophical systems, and elaborated what appear to be epistemological implications of the definition of discrimination we have developed. At this juncture a summary of the claims is now in order. Yet, I am ambivalent about summarizing the view we have reached. It is not news to suggest that discrimination is contentious. Thus, several impediments endanger communication, especially communication of perspectives that diverge from the dominant one. The fi rst few chapters document an asymmetry of experience and point to complex contradictory history, partly in an effort to construct conditions conducive to dialogue and consideration of what was to follow, while also explaining and excusing the earlier use of approaches that no longer effectively document the effects of discrimination. Against the backdrop of those observations, perhaps ambivalence can be understood, for any summary that simply asserts the findings we established over several chapters is at serious risk of appearing nonsensical. Be that as it may, a concluding statement is needed, and a summary that does not retread every step along the terrain upon which it is based

236 / Chapter 9

can serve to bring matters to a cumulative end. It is that task that occupies us now.

The Era of Contested Prejudice We began by observing a divergence in views concerning race and sex discrimination. Some commentators contend that matters have changed so dramatically that race and sex discrimination are not relevant for understanding the fortunes of blacks and women. At the same time, others contend that nothing has changed in the intervening years; the prospects of blacks and women are tenuous, always at risk of being reversed should national fortunes be perceived to require it. We then replicated existing analyses that have revealed precipitous declines in antiblack and antifemale attitudes. Analyses of these attitudes have supported the more optimistic claim of major change in race and gender relations. However, critics have contended that these analyses use outmoded questions that do not tap the mechanisms of contemporary discrimination. Yet, we considered whether one might use fairly straightforward techniques, and some simple assumptions, to estimate the likelihood that a black or a woman would encounter prejudiced authorities in their lives. Despite the exceedingly low incidence of prejudiced persons, blacks and women were exceedingly likely to encounter authorities who doubted their commitment and questioned their legitimacy. This result implies an asymmetry of experience. Observers see declining levels of prejudice against blacks and women, while blacks and women virtually are assured of encountering authorities prejudiced against them. This asymmetry underlies, in part, the different understandings that have been voiced. Persons are less likely to regard a fact as true if it clashes with their personal observations. Because the evidence on the declining levels of prejudice has been disseminated widely, the fact is seared into the consciousnesses of many. Yet, few analyses have attempted to trace the implications of the level of prejudice that remains for blacks and women. When this is done, it becomes apparent that blacks and women virtually are assured of being exposed to authorities prejudiced against them. If there is any sense in which attitudes are connected to behaviors, this cannot be positive for the fortunes of women and blacks. I submit that this asymmetry of experience underlies the vastly different assessments of the realities of race and sex discrimination. This asymmetry has developed owing to the failure of the process that led to whatever progress we have observed. That process did move U.S. society, extinguishing what I call a regime of condoned exploitation. Social protest aided the overthrow of that regime.

Theorizing Discrimination in an Era of Contested Prejudice / 237

However, we did not move into a period of easy and equal relations across the cleavages that formerly divided society into those with rights protected by the full weight of the government, and those whose rights could be jettisoned at any moment. Instead, the end of the regime of condoned exploitation ushered in an era of contested prejudice, a time characterized by discordant tendencies. Some currents signal advance toward equal rights and status; others signal forced retreat back toward a time in which exploitation was brazenly condoned. This era dawned because elites, the drivers in the processes that preserve black and female rightlessness, never waged a full-scale war against discrimination. Hence, exploitation never was repudiated directly; instead, a par ticular explanation for the exploitation became unacceptable. This occurred because change was driven primarily by elites’ desire to solve important national problems—chiefly, labor force needs in the case of women, and geopolitically relevant reputational issues of the nation-state in the case of blacks. This led initially to some accommodation to the demands of blacks and women. However, at the same time, and especially in the case of race, fundamental material infrastructures as well as legal precedents were reconstituted in ways that facilitated racial division, while doing so in a less brazenly obvious manner. I hasten to add that one need not posit some malicious conspiracy at the center of this effort. The series of decisions were not some conspiracy; no, the decisions were the unfolding result of an effort to solve an embarrassing material and symbolic problem that imperiled the nation’s ability to ally with emerging third-world nations in an international struggle with the Soviet Union, as well as economic difficulties linked to labor supply. Hence, with the aim of preserving as much of the existing distribution of power as possible, as decision points were reached the most palatable options were those that did not upset that distribution. Thus, the accommodations that were made, for all their power in improving the lot of blacks and women, were even more important for their role in solving those national challenges. And because the goal was the conservation of the distribution of power, not a redistribution of power, the solutions never culminated in a defi nitive repudiation of the prior regime, nor any thoroughgoing material response to the oppressions of the past. This path has brought us forward, to a moment of easier race relations in some respects, and toward more independence for women in some arenas. Yet, while the structures erected in this process have been heralded by many as important in securing equal rights, they have been just as important in preventing a complete resolution, a social repair, of the injuries inflicted during the regime of condoned exploitation. Hence, we stand in this interstitial moment, between a past of condoned exploitation and a future threatened by

238 / Chapter 9

our incomplete effort to repair the legacy of that past. This interstitial moment is the era of contested prejudice.

Dominant and Critical Legal Perspectives on Discrimination Grasping the implications of this moment is essential for our project. Understanding that crosscurrents exist owing in part to asymmetric experience prepares the way for our work by giving us an explanation for diametrically opposed perspectives that does not require the insincerity of others. This should facilitate our efforts to dialogue as we consider anew the very defi nitions and approaches we use to study discrimination. Even so, we must recognize that we stand in this moment partly because the law, a major means through which justice was pursued, has defi ned discrimination in par ticu lar, truncated ways. Discrimination in dominant legal discourse is an act that some commit and by which others are victimized. But, there is no room in antidiscrimination law for the concept of beneficiary bystanders, and no way in which victims can receive compensation from those who may have gained, did not lose, or who lost only owing to the collateral damage of discrimination. This legal logic is particularly ironic because discrimination is regarded as wrongful treatment because it entails treating people only as members of groups. Hence, discriminators in legal doctrine necessarily construct two groups by their discriminatory act: the target of discrimination, and nontargets (i.e., all others). Presumably, the nontargets are eligible to obtain benefits that the targets cannot obtain. Hence, the benefits the nontargets receive are, from one point of view, ill-received gains—not ill-received necessarily because of some act of the nontarget, but ill-received nevertheless. In some arenas of law, innocence does not absolve one of compensating for bad acts of others (e.g., sexual harassment law, the savings and loan bailout). However, in antidiscrimination law, no such logic has dominated. Hence, targets of discrimination must recover damages only from those who can be shown to have discriminated. This logic greatly reduces the number of persons who can be expected to pay damages, and greatly limits the range of behavior within which discrimination can be identified. Critical legal scholars provide one response, calling us to consider the basis of race and gender inequality in the developing national definition of property. When we do so, we come to a realization that whiteness and maleness can be conceived of as property. In this view, a property right in whiteness and maleness renders discrimination a social relation in which whiteness and maleness are superordinate positions, conferring power in arena after arena for whites qua whites and men qua men that others just do not have.

Theorizing Discrimination in an Era of Contested Prejudice / 239

We can bracket the implications of this vision for the operation of the legal system, for we have not attained sufficient perspective to address this issue. However, seen from the lens of the critical legal tradition, the courts’ narrow view systematically overlooks important, defi nitive aspects of discrimination; surely this conclusion may prove useful in developing a social scientific approach to studying discrimination effects, regardless of the difficulties that may make it infeasible for practical legal institutions. After establishing the existence of an era of contested prejudice, one might move directly to an analysis of social scientific perspectives on discrimination. Yet this would leave us in the same state as the usual discussion, with unexamined preconceptions that determine the degree to which any given claim will seem sensible, preconceptions derived in part from the courts’ power-laden assessment of what does and does not constitute discrimination. In this way, the legal sphere has played a powerful role in determining what an allegedly rational understanding of discrimination may be; given this reality, it is necessary to first delve into the dominant legal construction of a definition of discrimination. By treating the matter on that terrain, it is possible to directly establish that dominant legal reasoning is but one legal perspective, and that it has par ticu lar features that, while useful for the law, may or may not be useful for social science and societal understanding. Only by turning directly to the law, before taking up social science itself, could we deconstruct the taken-for-granted perspective on discrimination that has deeply influenced social scientific analyses. And once we proceed in this manner, we find little reason to suggest social science should privilege Supreme Court assessments of what counts as part of the social phenomenon of discrimination when those assessments are inconsistent with the operation of the phenomenon of discrimination.

Asocial Scientific Approaches to Studying Discrimination Effects With the legal analyses in hand, it becomes possible to coherently propose that, though the courts may have legitimate reasons for foreclosing whole areas of social life and social history from legal adjudication, social scientific analysis of social phenomena is not served by such power-based structuring of its object of study. Yet, social scientists adopted a defi nition of discrimination as narrow as that adopted by the courts, even though social scientists do not have to rule in specific, usually painful disputes between identifiable conflicting parties. Thus, social science has used a definition that has fatal affinities with dominant legal doctrine. Most notably, mainstream social science has defi ned discrimination in an atomistic, individualistic manner, seeking to trace discrimination effects

240 / Chapter 9

back to par ticu lar economic spheres. This approach ends up being blind to many kinds of discrimination, and ignores the way in which discrimination transcends the actors immediately involved. For example, if an employer discriminates by not hiring women, the women who apply to work at that firm are disadvantaged by the discriminatory behavior. Consequently, all else equal, every firm that will hire women will have more women vying for the positions, creating more competition at the nondiscriminating firm(s). Hence, women who never even heard of the discriminating firm may pay a cost for the discrimination of the discriminating fi rm, a cost reflected perhaps in part by lower expected pay owing to both increased competition among women for positions at less discriminating firms and lower likelihood of obtaining a position. Further, in an environment where a firm can decide to not hire women, other arenas of life in that environment very well may be affected. One way this will occur is that employers who discriminate as employers also occupy many other roles. It is quite possible that they will discriminate in their other social roles as well; they may discriminate as landlords, witnesses, or in other roles. Moreover, the conditions that allow discrimination are themselves implicated in the existence and maintenance of the phenomenon. Hence, an environment that allows one kind of discrimination will be unlikely to have simply one kind of discrimination. Thus, the effort to trace discrimination effects back to par ticular spheres, owing to the role occupied by a putative discriminator, is a social scientific error, no matter how worthwhile or necessary, though difficult, this action may be for the judicial system. These elements of the dominant social scientific approach to studying discrimination flow, in part, from Gary Becker’s foundational research. However, though many analyses of discrimination ritualistically nod in Becker’s direction, at a fundamental level the contemporary study of discrimination effects actually bears little resemblance to his line of inquiry. Becker affirms that equality might not exist even in the absence of discrimination, that one needs to distinguish the amount of discrimination and the effect of discrimination, and that nontargets of discrimination might also lose owing to the operation of discrimination. All three of these affirmations are rejected or, more accurately, ignored by much of the contemporary social scientific analysis of discrimination. Yet, these same affirmations can be resurrected if one adopts a definition of discrimination consistent with its actual operation.

Discrimination as a (Damaged) Social Relation Interestingly, once one sees discrimination as a damaged social relation, one is well on the way to restoring Becker’s three, clearly accurate affi rmations to

Theorizing Discrimination in an Era of Contested Prejudice / 241

the core of research on discrimination. Drawing on insights from critical legal studies and Gordon Allport, one can conceive of discrimination as a social relation. Allport teaches us that discrimination runs the gamut from simple antilocution to avoidance, exclusion, physical attack, and extermination. The full range of discrimination requires categories of persons, for categories allow the development of norms, values, and support mechanisms. Thus, discrimination is a social relation, which rules out simple one-on-one condemnation as the exemplar of the phenomenon. Making this determination does not mean that one person cannot harm another as an individual, but it is to say that when such harm occurs absent the edifice that constitutes discrimination, it reflects some other phenomenon, not discrimination. Defining discrimination as a damaged social relation makes it possible to see how some standard operating procedures are discrimination. For example, at the present time, the United States Army bars women from direct combat positions. This bar partly is based on the idea that women will encounter higher levels of danger should they be captured in combat. The view of mainstream social science on whether excluding women from direct combat is discrimination is unclear. Taste theorists might argue that Army brass are indulging some taste in deciding which soldiers to allocate to direct combat. Yet, the inability to give content to the label taste renders taste theories of limited utility. Those who prefer the statistical discrimination framework might contend that the quality of information underlies the exclusion of women from direct combat, and thus might regard the prohibition as discriminatory, though perhaps understandable (but not necessarily excusable). Yet, if the risks that accompany capture are the concern, any serious perusal of the history and contemporary reality of imprisonment will reveal the limitless lengths to which captors may go regardless of the sex of their captive, and thus will indicate that every captured soldier is at equal and extreme risk. Hence, this prohibition seems based not in accurate information that would raise one’s concern but, instead, in a limiting view of women. Thus, under the view that discrimination is a damaged social relation, excluding every female soldier from direct combat is identified as discrimination. That the exclusions persist suggests that dominant legal perspectives either have regarded sex as a bona fide occupational qualification (BFOQ) for combat or have not identified this policy as discriminatory. Taken together, these observations provide one example of how the dominant perspectives on discrimination, and the view that defi nes discrimination as a damaged social relation, may reach different conclusions as to whether a par ticu lar policy is discriminatory. In this case, the dominant perspective appears to defi ne the policy as nondiscriminatory, whereas the social perspective sees the policy as discriminatory.

242 / Chapter 9

At the same time, seeing discrimination as a social relation recategorizes a variety of situations that superficially resemble discrimination under the dominant perspective. The concept of reverse discrimination is incoherent when discrimination is seen clearly, as a social relation. Discrimination involves the acts of individuals, but not only that; it also involves the matrix of norms, values, and public support mechanisms that make acts of discrimination possible. It is impossible for norms, values, and public support mechanisms to be arranged such that the dominant group in society is systematically and continuously disempowered. At the same time, contesting the norms, values, and public support mechanisms that constitute discrimination, and thus working to overturn the privilege of nontargets of discrimination, is not to discriminate. Of course, when blacks or women disadvantage whites or men inappropriately, it is a social phenomenon worthy of study and ameliorative response. But, it is not discrimination. Discrimination is not the only phenomenon that is best understood in social relational terms. Other phenomena entail categories of persons—parent and child, teacher and student, spouse and spouse, and more. Some social relations are so dysfunctional for human flourishing that we may regard them as damaged social relations. In order to identify a social relation as a damaged one, we need to consider some more or less coherent philosophical system from which criteria may be drawn. To fully explore the philosophical perspectives that might be relevant would take at least another volume. Thus, I selected two perspectives, one because of its status as a major response to utilitarianism, the other for its status as an allegedly radical response to mainstream sociology, philosophy, economics, and social orga ni zation. Both Rawlsian philosophy and analytic Marxism provide resources that allow one to delineate damaged and undamaged social relations, suggesting the categories have some congruence with the real world or, at least, with existing coherent philosophical systems.

Discrimination as a (Damaged) Social Relation: Epistemological Imperatives Seeing discrimination as a damaged social relation is consequential. The methods that work for studying or even measuring undamaged social relations do not work for discrimination. However, by examining several elements of research, we are able to construct a new edifice for discrimination research, one based on seeing discrimination as a damaged social relation. Further, we note that if one remains committed to using the individualistic definition of discrimination, then developments in our understanding

Theorizing Discrimination in an Era of Contested Prejudice / 243

of the difficulties of causal inference, crystallized in Rubin’s Model, raise serious doubt as to whether one could ever determine the effect of discrimination. This is so because standard analytic approaches compare men and women or blacks and whites to ascertain the effect of discrimination. But, Rubin’s Model implies that any such comparison is invalid. Proper inference calls for one to compare entities that are essentially the same, save one is exposed to a treatment one wishes to assess, and the other is not so exposed. The invalidity stems in part from the inability to change a person’s sex or race without fundamentally changing the person. Hence, for the study of discrimination, conceived of atomistically and individualistically, Rubin’s Model results in Rubin’s Quandary. However, when discrimination is seen as a social relation, it becomes apparent that men live in environments with more and less antifemale discrimination. By comparing men in more discriminatory environments with men in less discriminatory environments, we may assess the effect of antifemale discrimination on men. And, obviously, we can make a similar comparison of women in different types of environments. Hence, seeing discrimination as a social relation solves the dilemma of estimating discrimination effects that Rubin’s Model appears to imply. We do not delve into the gory details of constructing a par ticular indicator for a par ticu lar study; here, we are concerned with theory. We take the view that theory must precede measurement or only problems will ensue. Even so, we are able to make some general observations concerning the character of the measures that are likely to suffice. And, what we indicate is that when discrimination is conceived as a damaged social relation, the indicators of discrimination cannot and should not be at the individual level. As a social relation, all in the environment are implicated in the social relation of discrimination, a social relation that may be more severe in some times or places, less severe in others. Using this variation, we may assess the relationship between discrimination and outcomes of interest. We illustrate what we might learn with such an approach. Interestingly, we find that black males, white males, black females, and white females all have lower earnings when exposed to antiblack discrimination. This result concretizes the claim—using nontargets as the comparison to discern the effects of discrimination is to distort the on-the-ground reality of discrimination effects. Tragically, the distortion likely is damaging not only to academic knowledge, but also to public support for eradicating the damaging social relation of discrimination.

244 / Chapter 9

Bounding the Scope of the Concept of Discrimination as a (Damaged) Social Relation: A Clarifying, Concluding Step The critical legal scholars and Allport provide resources that make visible the social relational character of discrimination. For social scientists, this perspective offers clear advantages, most notably by opening many new doors for progress on empirical research. Most important, these doors all open onto possibilities to estimate the effect of discrimination in a par ticular sphere of activity, not to determine which corporation or employer may or may not have discriminated against par ticu lar plaintiffs. Because some of the resources are drawn from critical legal scholarship, one might surmise that use of this perspective in the social sciences commits one to see discrimination as a social relation in the legal sphere as well. There may be advantages to seeing discrimination as a social relation in the legal sphere, but weighing those advantages and disadvantages is not the object of the analysis presented herein. What is established here in that regard concerns the most promising way for social analysts to conceive of the phenomenon of discrimination. Given the well-known and serious limitations of existing approaches, the promise of the conception developed here should be clear. However, an entirely different set of concerns need be addressed before one can conclude that discrimination should be seen as a social relation in other spheres of activity, by other actors, including by the legislative, executive, and judicial branches of government. We do not fully address the concerns one must address to answer this question, so we remain silent on this issue. Despite these qualifications, the claims established here are not without consequence for the understanding of discrimination outside the academy. First, given that we know that discrimination is captured poorly by focus on microinteraction between, for example, employers and potential employees, we must be skeptical of estimates based on such individualistic definitions of discrimination. We should expect such estimates to understate the effect of discrimination, if only because major aspects of discrimination will not be included in the analysis. Further, we should be wary of making too much of claims that trumpet or lament change or stability in race and gender gaps, for on the basis of such claims we cannot infer whether we have identified the decline or the intransigence of discrimination. For many reasons we develop throughout, discrimination is a phenomenon that may rise or fall in incidence, and a phenomenon whose effects may rise or fall in magnitude. The patterns of incidence and effects need not be associated, and thus learning the

Theorizing Discrimination in an Era of Contested Prejudice / 245

overall race or gender gap does little to solidly identify either incidence or magnitude. Existing discussions generally have been unable to partition incidence and effect, but it is an important partition to make. In order to make this partition, one must measure the social relations of discrimination, and introduce those measures into appropriate analyses of outcomes of interest. Once that task is done, the effect of discrimination will be revealed.

The Legacy of Goodwill Studying discrimination is difficult. In this context, many analysts have worked hard to maintain an awareness of the existence and possible power of discrimination in the lives of the disadvantaged. At the same time, absent a methodological solution to the problem of how to measure discrimination and estimate its effects, discrimination eventually will be abandoned as a causal consideration. Certainly, discrimination is not the only causal factor in the social world. Still, it is the elephant in the room of any study of social phenomena. It is the equal of socioeconomic status or class in that respect—no self-respecting social scientist would ignore the possibility that his or her results might have been driven, at least in part, by persons’ socioeconomic status or class. Similarly, given the history of the United States, it is likely that many analysts would prefer to be able to routinely include measures of discrimination in their analyses, just as they do for socioeconomic dimensions, if only to clarify the operation of other processes in which they are interested. However, as yet unable to develop a consensus analytic strategy for discovering the effects of discrimination, analysts are in danger of abandoning discrimination as a causal factor. The prospect of abandoning discrimination as a causal factor is distressing particularly because many of the alternatives are neither well measured nor sufficiently theorized (e.g., contrast Herrnstein and Murray 1994 and Fischer et al. 1996, or Fordham and Ogbu 1986 and Cook and Ludwig 1998). Given this possibility, it is extremely important that a generalizable social scientific response to the difficulties with studying discrimination effects be pursued, a response that will allow measures of discrimination exposure to be routinely introduced into statistical models and other analyses. It is tempting to speculate as to why researchers adopted the individualistic perspective. Why did a view that separates discrimination into spheres become the dominant one? After all, Allport’s landmark work predates Becker’s by about three years. Allport’s work is far more empirical than Becker’s, and it is the empirical social scientists who have been at the forefront of discrimination research. Additionally, Allport’s work is more in line with the

246 / Chapter 9

classic that predates it, Gunnar Myrdal’s An American Dilemma, an empirically driven survey of the existence of antiblack discrimination in the United States. One institutional explanation begins by noting that at the founding of the national security state the social sciences were engaged in battles for legitimacy (Kleinman and Solovey 1995). Atomistic strategies of research had proven successful in the natural sciences, and many social scientists thought they might follow that model to the land of legitimacy. Becker’s perspective may have gained currency because it was promulgated within that milieu, and also because economics, a more accepted discipline even at the time, held itself out as a model social science. Atomistic research strategies have a long history in the West; what may have given added force to the atomistic model and urgency to the question of social science legitimacy was the takeoff in federal appropriations for scientific research, coupled with a growing federal role in determining the categories in which research might be funded. A more psychological explanation is that analysts, who were mostly white and male, could not ignore the possibility that their ivory-tower jobs might not have been theirs were it not for the rigging of the race they had won. Under this view, these analysts subconsciously needed to put distance between themselves and the direct operators of discrimination. Viewing discrimination as confined to isolated incidents that at most come together within identifiable spheres may have provided sufficient distance to prevent a paralyzing cognitive dissonance. But a troubling feature of each of these purely speculative possibilities is that they are both at least somewhat accusatory. And, in being accusatory, they ignore an important observation—hundreds of scholars used the best means available to document shameful inequalities and trace those inequalities to discrimination as best as the technology available allowed. Indeed, in the social and political context in which the work began, the dominance of the individualistic perspective could have been based in the palpable acceptance that any difference had to be caused by the discriminatory treatment meted out to blacks and women. Only as the era of contested prejudice dawned may the complexities of the unfolding scene have made that assumption insufficient for scholarly work to proceed. Thus, whatever the reason for the dominance of the individualistic perspective, it must be noted that, for decades, within that perspective, many white scholars and many male scholars devoted their energies to assessing race and gender disadvantage in an effort to identify the effects of discrimination. Questioning the motives of people who apparently conscientiously were attempting to document the results of oppression arguably is shameful itself or, at a minimum, a distraction from the task of erecting a means for studying the phenomenon in changed circumstances.

Theorizing Discrimination in an Era of Contested Prejudice / 247

The vast majority of scholars who have worked in this area appear to have been acting on three beliefs: (1) confidence that there are effects of discrimination, (2) awareness that inconclusive evidence of the effects of potentially powerful causal phenomena constitutes a social scientific failure of massive proportions, and (3) commitment to documenting the effects of the phenomenon so that work can be done to rectify ongoing injustices. So many have sought to study discrimination, seemingly with this troika of beliefs, that discrimination is one of the most studied social phenomena in the United States. Still, the many explicit and implicit searches for discrimination effects have not produced solid evidence of a discrimination effect, because each strategy researchers have used has had insurmountable flaws that ultimately trace back to the shackles their conception of discrimination locked into place. Thus, continued perception of discrimination as an important causal phenomenon has rested more on a reservoir of researcher goodwill than on the kind of social scientific evidence to which we have become accustomed. Researchers have successfully documented the existence of discrimination and prejudicial attitudes (e.g., Kirschenman and Neckerman 1991). But as an explanation of the status of blacks and women, discrimination is tenuous when compared to other factors in the social world whose causal status for various outcomes is consistently documented, such as educational attainment. It is likely that only a small proportion of social scientists completely discount discrimination as a causal fact for the mid–twentieth century through to the contemporary period. Yet, it would be unwise to expect even researchers’ goodwill to forever buttress the discrimination explanation against state-of-the-art investigations of alternative explanations; eventually, all researchers will turn away from explanations they cannot assess, because the process of research is by definition one of assessing explanations. Absent some means to assess the effect of discrimination, discrimination will be well on its way to becoming a neglected factor in analysts’ explanations, even as the existence of discrimination may continue to be accepted as a social fact. Goodwill has not been confined to the research community, for it is goodwill, coupled with evidence of continuing discrimination that has sustained public understanding of the importance of discrimination. But, if policy-makers design policy based on the goodwill of citizens, and if there is no success in documenting the effect of discrimination, then eventually citizens will look elsewhere to understand the plight of the disadvantaged. When they do, the remaining goodwill most likely will dissolve very quickly, as will whatever policies designed to resolve the historic dilemma of America that remain. This is, intriguingly, the definitive predicament of the era of contested prejudice. Thus, the need for a new paradigm that provides an effective

248 / Chapter 9

analysis strategy for discerning discrimination effects—an analysis strategy that accepts that not only blacks and women, but also whites and men, can lose through discrimination—is clear.

Concluding Remarks A necessary condition for any solid analysis is a solid theoretical conception of the phenomena under study. Decades ago, social scientists took what, for research in the era of contested prejudice, turned out to be a wrong conceptual turn in their efforts to assess the effect of discrimination. By measuring the effect of discrimination by whether estimates matched expected effects rather than by the social relational character of the phenomenon, they removed motivation from the social scientific analysis of the phenomenon. At the same time, in an effort to facilitate their analyses, social scientists moved toward more and more fine-grained investigation of discrimination effects, pursuing an atomizing strategy of research. Once social scientists made these moves, the eventual inability of scholars to obtain any estimate of discrimination effects that could find wide, defensible, consensus acceptance was almost a certainty. Indeed, these moves appear ultimately to have made it impossible to trace differences to any motivation of actors, while exponentially increasing the loci one would need to study to establish any connection between putative cause and effect. Thus, the last few decades of discrimination effects analyses suggest that the resulting lack of consensus on discrimination effects may have been a foregone conclusion after the wrong turn was taken some fifty years ago. But research and theorizing continued, providing resources that eventually could be used to return a social understanding of discrimination to the social sciences. The implications of this social restoration are difficult to overstate. Most intriguingly, the social restoration makes it possible to estimate the effects of race and sex discrimination in the United States on targets of discrimination and nontargets alike. Because it allocates attention equally to both targets and nontargets, the social perspective on discrimination is decidedly different—and definitively not zero-sum—in its approach. The benefits of such an approach for our estimates of discrimination effects await us in Just Who Loses?, an empirical investigation based on the perspective developed here. Suffice it to say here that one must theorize discrimination in a manner that includes the supra-individual aspects of the phenomenon, for only with such a conception may we successfully estimate effects of discrimination in the era of contested prejudice. Within the foregoing pages, I attempt to make that case, and in doing so seek both to indicate how mainstream social scientific approaches and conceptions go awry, creating problems that are insur-

Theorizing Discrimination in an Era of Contested Prejudice / 249

mountable within that paradigmatic perspective, and to adumbrate the analytic opportunities that the demonstrably more accurate social relational conception of discrimination makes possible. At this point, we have established the existence of an era of contested prejudice. We have described it, contrasted it with a regime of condoned exploitation, excavated evidence of the asymmetry of experience that haunts the denizens of the period, and suggested the dangers inherent to the era. We have contrasted dominant and critical legal perspectives on discrimination, thereby identifying the punitive imperative as an impediment to a full understanding of the social phenomenon of discrimination, while highlighting the existence of an alternative perspective such that the monopoly dominant legal reasoning has held on the mantle of rationality is broken. We have deconstructed the dominant social scientific understanding of discrimination, revealing its affinities with dominant legal reasoning, while identifying its neglect of key resources provided by Becker’s foundational work, resources more easily taken up by the truly social framework for studying discrimination proposed herein. And, observing that the problems that hound the dominant social scientific approach to studying discrimination effects are so vexing that not a single consensus estimate of the effect has been produced, we have turned away from the dominant approach to develop a truly social framework that sees discrimination as a damaged social relation. We have then moved on to consider it in light of a small set of existing important philosophical systems, and traced how it responds successfully to the challenges that bedevil the dominant social scientific approach. Finally, we illustrated the payoff to this approach by summarizing one set of findings on discrimination and earnings, a set of findings that reveals negative effects of race discrimination on black males, white males, black females, and white females. Amidst this work, many questions remain. What would we find if we more precisely mapped the contours of the era of contested prejudice through more detailed historical-comparative analyses? What would we learn if we considered additional bases of discrimination, and how might that deepen our definition of the phenomenon? What other standard operating procedures might we illustratively analyze to reveal the daily enactment of the damaged social relation of discrimination, and what would we learn about its reach were we to do so? What might other philosophical systems conclude about the effort to distinguish damaged and undamaged social relations, and how might that affect our understanding of the social phenomenon of discrimination? These questions, and others, suggest that additional profitable work may be possible in the theoretical vein we have mined. If that work is pursued, no doubt some of the particulars of my account will require modification. As accuracy is a goal, it would be a positive development if those directions were

250 / Chapter 9

pursued. At the same time, because this volume seriously contests the existing dominant social scientific approach to studying discrimination, a primary outcome of this effort is the prospect of future research using the foundation it provides to conduct empirical study of discrimination effects. Owing to taken-for-granted understandings, our foundation—erected in contradiction to the dominant perspective that has behind it the power of the state, the norms of common interaction, and the commitment of major sectors of the elite—required a volume to explicate. Now, with a coherent, alternative, phenomenon-based conception of discrimination in hand, we not only are enabled to continue to work to distinguish and assess the existence of an era of contested prejudice, the coherence of the framework in light of philosophical systems, or the many other phenomena with which discrimination often has been conflated, but we also are poised to address an important yet simple question: What are the effects of race and sex discrimination in the United States? However, regardless of whether one moves on to that analysis, our considerations here make it harder for discussants to unreflectively invoke the dominant perspective on discrimination in public policy debate as if a more coherent alternative perspective does not exist. For a more coherent perspective, more consonant with the social phenomenon of discrimination, is available, conveyed in the insights collected here, insights that are largely reflections from the gems provided by dozens of scholars working over more than five decades in dedicated ser vice, endeavoring to resolve the theoretical, social scientific, and social policy challenge of discrimination. Ideally, we can take up those resources and work together to turn them to good effect. For truly, only by adopting a perspective in line with the actual operation of discrimination can we ever hope to inclusively transcend the agony and distrust that is both the legacy of the long and painful regime of condoned exploitation, and the debilitating specter that haunts us all in the era of contested prejudice.

Appendix A Commentary on Methods of Data Analysis for Chapter 2

I

n Chapter 2, I use General Social Survey (GSS) data to estimate the proportion of white adults who explain the long-standing disadvantage of blacks as caused by, first, blacks’ inherent inability and, second, lack of motivation. I then use GSS data to estimate the proportion of adults who believe women should stay home and let men run the country, and who believe it is wrong for a woman to work if her spouse can support her. Further, I select those persons who worked in the key occupations of teacher, Realtor, or police officer, and those persons who are parents living with children, and subject them to further analysis. When I do so, I greatly reduce the sample size for the analysis. In a national probability sample of U.S. residents, one can expect to obtain some respondents employed in these large and readily identifiable occupations. However, the number of respondents in these occupations fluctuates across samples and is usually not very large. Tables A.1 and A.2 show the number of cases by occupation for each of the analyses by year. Sparse data produces imprecise estimates, and clearly the data for Realtors and police officers are sparse. Still, the data used provides the best estimates for different occupations currently available. Because of the sparseness of the data available, I use a smoothing technique to better estimate the proportion holding a par ticular position in a given year. Smoothing is analogous to calculating a moving average. The technique used—lowess smoothing (Cleveland 1979)—essentially calculates an

252 / Appendix A

TABLE A.1 NUMBER OF RESPONDENTS BY OCCUPATION FOR RACIAL ATTITUDES ANALYSES DO BLACKS HAVE LOW ABILITY?

DO BLACKS LACK MOTIVATION?

YEAR

ALL

REALTOR

TEACHER

POLICE

ALL

REALTOR

TEACHER

POLICE

1977 1985 1986 1988 1989 1990 1991 1993 1994 1996 1998 2000

1267 1279 1203 787 819 1107 820 881 1565 1521 1422 1403

9 9 7 4 4 14 5 10 11 9 12 16

65 48 55 12 35 38 43 25 75 60 55 74

2 10 8 9 8 5 11 7 24 12 15 10

1256 1275 1198 780 810 1075 808 843 1531 1471 1351 1346

10 9 5 4 4 12 5 10 11 8 11 14

64 46 53 12 33 36 43 23 75 60 51 68

2 9 8 9 8 5 11 7 22 11 13 10

estimated proportion holding a par ticular view for a given year by weighting the proportion obtained in nearby years more heavily than proportions obtained in years more temporally distant. Note that lowess produces a weighted estimate for each year. I graph those estimates as the best figures we could obtain for the proportion of persons holding certain attitudes. However, for the next stage of the analysis, the stage wherein I estimate the likelihood that blacks and women will encounter unsupportive authorities, I return to the observed proportions that form the foundation for the smoothed lines reported earlier. By using observed proportions in the second analysis, I make each analysis more or less equally removed from the observed data. In contrast, had I used the smoothed proportions as the base data to calculate estimated exposure probabilities, the smoothing might play too big a role in the second set of analyses. With the approach used here, the exact same smoothing technique is applied only once in each analysis. However, the second analysis obviously does not simply graph the observed proportions. Instead, I use the observed proportions and the cumulative binomial distribution function to calculate the probabilities of exposure implied by the observed proportions holding par ticu lar attitudes. The cumulative binomial distribution function is: N

Prob(r ≥ 1) =

∑ r!(N − r)! pr q N − r N!

r =1

where N signifies the number of opportunities; r—the number of “successes” in N trials; p—the probability of success in a trial; and q = 1 − p, which

1972 1974 1975 1977 1978 1982 1983 1985 1986 1988 1989 1990 1991 1993 1994 1996 1998

YEAR

TABLE A.2

— 1424 1441 1476 1476 1807 1537 1493 1423 949 958 889 978 1041 1891 1874 1810

ALL

— 26 13 29 19 15 16 16 21 6 16 13 13 7 12 19 14

TEACHER

— 6 8 1 6 6 5 10 8 10 8 5 11 4 21 15 14

POLICE

— 1104 1065 1106 1076 1321 1119 1075 1036 681 696 630 716 760 1353 1352 1304

PARENT

SHOULD WOMEN STAY AT HOME?

— 320 376 370 400 486 418 418 387 268 262 259 262 281 538 522 506

NONPARENT

ALL

1569 1443 1457 1492 1504 1821 1558 1485 1441 960 981 900 993 1054 1924 1904 1833

21 26 13 29 19 15 17 16 21 6 15 14 13 7 12 19 15

TEACHER

6 6 9 1 6 6 5 10 8 10 7 5 11 4 22 15 15

POLICE

1207 1117 1077 1116 1094 1328 1135 1067 1046 693 713 640 730 768 1380 1379 1317

PARENT

SHOULD “SUPPORTED” WOMEN NOT WORK?

NUMBER OF RESPONDENTS BY SOCIAL ROLE FOR GENDER ATTITUDES ANALYSES

362 326 380 376 410 493 423 418 395 267 268 260 263 286 544 525 516

NONPARENT

254 / Appendix A

designates the probability of failure in a trial. Using the cumulative binomial distribution function in this context is to assume that over a series of opportunities one has a probability, p, of being exposed to a person holding a particular attitude. The probability p is treated as stable across the opportunities, and each opportunity is independent of the other opportunities in the series. For some occupations the number of opportunities was set low (e.g., two realtors, two police officers), whereas for others the number was set comparatively high (e.g., twelve or thirty for teachers). Methodologists, myself included when I wear that hat, enjoy debating the merits and validity of different assumptions for a given analysis. However, it is important when evaluating substantive analyses to consider whether the assumption changes the qualitative conclusion. Further, there are two types of assumptions in the analysis in Chapter 2—those statistical assumptions that can be relaxed and still reflect the gist of the question, and those statistical assumptions whose relaxation entails abandoning the question or capitulating to the conclusion of the chapter, namely, that whites and adults who doubt the legitimacy of blacks and females are rare but numerous enough in key authorial positions that blacks and females have a high probability of being exposed to authorities likely to withhold support when they deal with whites and adults, respectively. The first type of assumption is a purely statistical, simplifying assumption; the second operationalizes a normative interest through the statistical decisions of the analysis. In this case the normative interest is that no citizen, prospective home-owner, or child should face the prospect of substandard treatment on the basis of his or her race or sex. Unless one rejects the normative interest, it makes no sense to relax the statistical manifestation of the normative interest, obtain a different fi nding, and then claim that the Chapter 2 analysis is flawed. One assumption that is a purely statistical simplification is the assumption that p will be stable across opportunities for the analyses of teachers and students. Students spend many years in school, and statistics revealed that p changed over time. So, of course, the assumption is false. To calculate an estimate that relaxes that assumption, one would need to identify a year in which school began, information on the number of teachers the student had each year, and the year-specific probabilities of teachers’ holding prejudicial views. This calculation would be straightforward once such information was specified. Some complexity might arise because of the missing years in the data series; however, one could interpolate values for intermediate years using proportions obtained in the surrounding two survey years. Clearly, the analysis in Chapter 2 is somewhat simplified in that it essentially provides an instantaneous estimate of the chance a black or female child who enters first grade in the survey year will encounter unsupportive teachers should teachers’ attitudes remain unchanged throughout the child’s

Commentary on Methods of Data Analysis for Chapter 2 / 255

twelve years of schooling. But, the qualitative conclusion is unlikely to change by introducing additional complexity into the analysis. For example, assuming that a student starts school in 1985 and stays in school through 1996, and interpolating for years missing from the series, the probability of having not a single white teacher who doubts the ability of blacks would be (.9167 × .9273 × .9220 × .9167 × .9143 × .9474 ≈ .446, × .9767 × .92835 × .88 × .9467 × .965 × .9833) the probability that a black student will have not a single white teacher who doubts black students’ ability. Thus, 1 − .446 = .554 is the probability that a black first-grader in 1985 will be exposed to one or more white teachers who doubt the ability of blacks between grades one and twelve. In other words, the qualitative conclusion does not change. The assumption of the stability of p is an example of a purely statistical simplifying assumption. Although it is good to probe such assumptions, for altering them sometimes does change the finding, in this case relaxation of the assumption does not alter the basic finding of the analysis. A second type of assumption is a statistical assumption that actually reflects normative interests. There is certainly value to discussing any and all statistical assumptions to assess whether normative commitments underlie the statistical operations. Indeed, it is likely that many ostensibly innocuous statistical operations actually are based on witting or unwitting normative considerations (MacKenzie 1978; Schweber 2001). As scholars we need be concerned with reducing the set of unwitting implications. Our analyses can only improve by openly discussing what now remains hidden. However, when evaluating and replicating research, methodologically inclined analysts must avoid relaxing statistical assumptions that actually reflect such interests, unless of course the methodologist disagrees with the normative interest. One such assumption in Chapter 2 is the supposition that black students will have a large number of white teachers. Clearly this assumption is false for the majority of black students. However, it is also the case that the normative interest underlying the calculations is that every student, regardless of race, should be teachable by every teacher, regardless of race. If exposure to white teachers serves to systematically increase black students’ exposure to teachers who doubt their abilities, this is a societal problem. Thus, the calculations were designed to determine whether the normative interest—every teacher should have confidence in the ability of every student to learn, regardless of race—was realized and, if not, what the implications would be for black students who might encounter a series of white teachers. Thus, given the large gulf in racial attitudes between blacks and whites, surely the figures in Chapter 2 would change if black teachers were added to

256 / Appendix A

the analysis of whites’ attitudes. But, as noted briefly in Chapter 2, doing so is to imply that the key factor lowering black children’s exposure to unsupportive whites is blacks in positions of authority. Hence, if one changes this aspect of the analysis, one is basically accepting the conclusion of the analysis—blacks dependent on white authorities are almost certain to encounter whites whose attitudes are prejudiced against them. Even so, note that both women and men were included in the analysis for gender, and the results were very similar to those obtained in the analysis of blacks’ exposure. Serious questions should be raised about the environments to which girls are exposed in the classroom given the views of the teachers at whose feet they sit to learn. In sum, it is possible to adjust the numbers by altering the assumptions used to craft the figures presented in Chapter 2. But, to date, only two eventualities transpire when the assumptions are changed. Either the more complex analysis produces numbers that condense to the same qualitative result reported in Chapter 2, or altering the assumption is to retreat from the normative commitments upon which the entire enterprise is based, including, notably, the values of equal opportunity the institutions under study often profess to hold. Hence, either way, the assumptions made in Chapter 2 seem appropriate for an investigation of discrimination.

Appendix B Commentary on Simulation for Chapter 5

I

n Chapter 5, I construct a simulation to assess the impact on mean test scores should psychometricians end the practice of deleting questions that officially designated lower-scoring students are more likely to answer correctly. To construct the simulation, I assumed that for one set of candidate questions students at the top had a 60 percent chance of answering the question correctly, while students at the bottom had a 40 percent chance of answering the question correctly. I also assumed that for a second set of candidate questions, students at the top had a 40 percent chance of answering the question correctly, while students at the bottom had a 60 percent chance of answering the question correctly. I calculated the point biserial correlation as r=

χ1 − χ0 (p) (1 − p) Sx

where p signifies the proportion of persons answering the question cor− desigrectly, Sx designates the standard deviation of the full test score, χ 1 nates the mean test score for those who answer the candidate question − , the mean test score for those who answer the candidate correctly, and χ 0 question incorrectly. The actual point biserial correlations are .1455 and −.1198 for the two sets of candidate questions. Using these pools of questions, I then constructed estimates of test scores for a series of subsequent years. For each year I subtracted from the

258 / Appendix B

test score calculation scores assuming that two questions with positive point biserials had been deleted, and then added back one question with a positive point biserial and one question with a negative point biserial. Clearly this is a very simple simulation. And, as such, it rides on some simplifying assumptions. One assumption is that the pool of test-takers will not change over time. This may appear a major problematic assumption, but it is easily addressed by simply assuming that all students in the population take the test and/or that the populations do not vary systematically from year to year. This could easily occur; for example, a state may require every child to take certain tests in the course of his or her schooling. Thus, this simplifying assumption, seemingly very problematic, actually can be regarded as having no effect, at least for some testing scenarios. A more important assumption is that students’ knowledge level will be stable, an assumption designed to isolate the effect of changing the question. Of course, more complex simulations could be constructed to allow the analyst to discern how the outcomes would change with multiple factors changing. This could be of interest, but it was not the concern that motivated the simulation presented here. A third assumption is that the point biserial correlations will remain stable. This is a more challenging assumption. It could be that as test-constructors relax the requirement to delete all candidate questions with negative point biserial correlations, the number of questions with negative point biserial correlations may decline. In other words, once the test includes some of those items, it is easy to imagine scenarios that would lower the incidence of negative point biserial correlations. First, and most obviously, students who did well on the earlier test may eventually learn to do well on those questions, dissolving any disadvantage they would have formerly encountered. More subtly, as the test begins to reflect questions from the domain of approaches that led to negative point biserial correlations, the criterion—the test—will cease to be as exclusionary as in the past. This will make students who formerly did poorly do better, making it less likely they will fall into the group at the bottom. As this happens, it will eventually become the case that when they do answer a question effectively that others answered incorrectly, it will no longer eventuate in a negative point biserial correlation, for they will have joined other students in the middle or even the top of the distribution. Hence, the process of inclusion has a possibly self-limiting character to how much test scores could actually change simply on the basis of failing to delete questions with negative point biserial correlations. These complexities are worthy of further study before a par ticular program is put in place. But the simulation should drive home the obvious point—the official scores students receive can and are easily affected by the deletion of candidate questions simply because the wrong people tended to answer the questions correctly.

References

Abel, Elizabeth. 1999. Bathroom doors and drinking fountains: Jim Crow’s racial symbolic. Critical Inquiry 25:435–481. Aigner, Dennis J., and Glen G. Cain. 1977. Statistical theories of discrimination in labor markets. Industrial and Labor Relations Review 30:175–187. Ainsworth-Darnell, James W., and Douglas B. Downey. 1998. Assessing the oppositional culture explanation for racial/ethnic differences in school performance. American Sociological Review 63:536–553. Allport, Gordon W. [1954] 1979. The nature of prejudice. Reading, MA: Addison-Wesley Publishing Company, Inc. Altonji, Joseph G., and Rebecca M. Blank. 1999. Race and gender in the labor market. In Handbook of labor economics, volume 3B, ed. Orley Ashenfelter and David Card, 3143–3252. New York: Elsevier Science. American Sociological Association. 2003. Brief of the American Sociological Association, et al., as Amici Curiae in Support of Respondents. Ananthaswamy, Anil. 2007. Higgs Boson: Glimpses of the God particle. New Scientist, March 2, 2007, 8–11. Anderson, Karen. 1981. Wartime women: Sex roles, family relations, and the status of women during World War II. Westport, CT: Greenwood Press. Arrow, Kenneth J. 1973. The theory of discrimination. In Discrimination in labor markets, ed. Orley Ashenfelter and Albert Rees, 3–33. Princeton, NJ: Princeton University Press. Ayres, Ian, and Peter Siegelman. 1995. Race and gender discrimination in bargaining. American Economic Review 85:304–321.

260 / References

Badgett, M. V. Lee. 1995. The wage effects of sexual orientation discrimination. Industrial and Labor Relations Review 48:726–739. Banaji, Mazharin R., and Curtis B. Hardin. 1996. Automatic stereotyping. Psychological Science 7:136–141. Bauman, Kurt. 1998. Schools, markets, and family in the history of African-American education. American Journal of Education 106:500–531. Bayor, Ronald H. 1988. Roads to racial segregation: Atlanta in the twentieth century. Journal of Urban History 15:3–21. Bazyler, Michael J. 2003. Holocaust justice: The battle for restitution in America’s courts. New York, NY: New York University Press. Beckenstein, Jacob D. 2003. Information in the holographic universe. Scientific American 289 (2): 58–65. Becker, Gary. [1957] 1971. The economics of discrimination. 2nd ed. Chicago: The University of Chicago Press. ———. [1964] 1993. Human capital: A theoretical and empirical analysis with special reference to education. 3rd ed. Chicago: The University of Chicago Press. ———. 1996. Accounting for tastes. Cambridge, MA: Harvard University Press. Bell, Derrick. 1992. Faces at the bottom of the well: The permanence of racism. New York: Basic Books. Bell, Derrick A., Jr. 1980. Brown v. board of education and the interest-convergence dilemma. Harvard Law Review 93:518–533. Bell, Robert M., and Phyllis L. Ellickson. 2000. School-based drug prevention: Challenges in designing and analyzing social experiments. In Public policy and statistics: Case studies from RAND, ed. Sally C. Morton and John E. Rolph, 9–28. New York: Springer. Bergmann, Barbara R. 1974. Towards more useful modes of research on discrimination in employment and pay. Sloan Management Review 15 (3): 43–45. Berk, Richard A. 1983. An introduction to sample selection bias in sociological data. American Sociological Review 48:386–398. Bertrand, Marianne, and Sendhil Mullainathan. 2003. Are Emily and Greg more employable than Lakisha and Jamal? A field experiment on labor market discrimination. Working Paper Number W9873, National Bureau of Economic Research, Cambridge, MA. Blank, Rebecca M., Marilyn Dabady, and Constance F. Citro. 2004. Measuring racial discrimination. Washington, DC: National Academies Press. Blee, Kathleen M. 1991. Women of the Klan: Racism and gender in the 1920s. Berkeley, CA: University of California Press. Blight, David W. 2001. Race and reunion: The Civil War in American memory. Cambridge, MA: The Belknap Press of Harvard University Press. Bonacich, Edna. 1976. Advanced capitalism and black/white race relations in the United States: A split labor market interpretation. American Sociolog ical Review 41:34–51. Bonilla-Silva, Eduardo. 2001. White supremacy and racism in the post-civil rights era. Boulder, CO: Lynne Rienner Publishers. Borjas, George J., and Stephen G. Bronas. 1989. Consumer discrimination and selfemployment. Journal of Political Economy 97:581–605.

References / 261

Borstelmann, Thomas. 2001. The Cold War and the color line: American race relations in the global arena. Cambridge, MA: Harvard University Press. Bourdieu, Pierre, and Jean-Claude Passeron. 1977. Reproduction in education, society, and culture. 2nd ed. London, UK: Sage Publications. Braddock, Jomills Henry, II, and James M. McPartland. 1987. How minorities continue to be excluded from equal employment opportunities: Research on labor market and institutional barriers. Journal of Social Issues 43:5–39. Breland, Hunter, Karen Kukich, and Lisa Hemat. 2001. GRE automated-editing-task phase II report: Item analyses, revisions, validity study, and taxonomy development. GRE Professional Board Report No. 98-05P/ETS Research Report 01-12. Princeton, NJ: Educational Testing Service. Brewster, Karin L., and Irene Padavic. 2000. Change in gender-ideology, 1977–1996: The contributions of intracohort change and population turnover. Journal of Marriage and the Family 62:477–487. Browning, Christopher R. 1998. Ordinary men: Reserve police battalion 101 and the final solution in Poland, with a new afterword. New York: Harper Perennial. Buber, Martin. [1923] 1970. I and Thou, trans. Walter Kaufmann. New York: Charles Scribner’s Sons. Burstein, Paul. [1985] 1998. Discrimination, jobs, and politics: The struggle for equal employment opportunity in the United States since the New Deal. Chicago, IL: Chicago University Press. Butler, John S. 1980. Inequality in the military: The black experience. Saratoga, CA: Century Twenty One Publishing. Butters, Robert D. 1993. The real estate industry’s view of audit results: Comments. In Clear and convincing evidence: Measurement of discrimination in America, ed. Michael Fix and Raymond J. Struyk, 153–163. Washington, DC: Urban Institute Press. Cain, Glen G. 1991. The uses and limits of statistical analysis in measuring economic discrimination. In Essays on the economics of discrimination, ed. Emily P. Hoffman, 115–144. Kalamazoo, MI: W. E. Upjohn Institute for Employment Research. Campbell, D’Ann. 1984. Women at war with America: Private lives in a patriotic era. Cambridge, MA: Harvard University Press. Carter, Prudence L. 2005. Keepin’ it real: School success beyond black and white. New York: Oxford University Press. Cavalluzzo, Ken S., and Linda C. Cavalluzzo. 1998. Market structure and discrimination: The case of small businesses. Journal of Money, Credit and Banking 30:771–792. Cherlin, Andrew, and Pamela Barnhouse Walters. 1981. Trends in United States men’s and women’s sex-role attitudes: 1972 to 1978. American Sociological Review 46:453–460. Chevan, Albert. 1989. The growth of home ownership: 1940–1980. Demography 26:249– 266. Cleveland, William S. 1979. Robust locally weighted regression and smoothing scatterplots. Journal of the American Statistical Association 74:829–836. Coase, R. H. 1959. The federal communications commission. Journal of Law and Economics 2:1–40. Cole, Robert E., and Donald R. Deskins, Jr. 1988. Racial factors in site location and employment patterns of Japanese auto firms in America. California Management Review 31 (1): 9–22.

262 / References

Cole, Stephen. 1986. Sex discrimination and admission to medical school, 1929–1984. American Journal of Sociology 92:549–567. Coleman, James S. 1988. Social capital in the creation of human capital. American Journal of Sociology 94:S95–S120. Converse, Jean M., and Stanley Presser. 1986. Survey questions: Handcrafting the standardized questionnaire. Beverly Hills, CA: Sage Publications. Cook, Philip J., and Jens Ludwig. 1998. The burden of “acting white”: Do black adolescents disparage academic achievement? In The black-white test score gap, ed. Christopher Jencks and Meredith Phillips, 375–400. Washington, DC: Brookings Institution Press. Coverman, Shelley. 1986. Occupational segmentation and sex differences in earnings. Research in Social Stratification and Mobility 5:139–172. Cross, Harry, Genevieve M. Kenney, Jane Mell, and Wendy Zimmerman. 1990. Employer hiring practices: Differential treatment of Hispanic and Anglo job seekers. Washington, DC: The Urban Institute Press. Csikszentmihalyi, Mihaly. 1990. Flow: The psychology of optimal experience. New York, NY: Harper & Row. Cunningham, William A., Kristopher J. Preacher, and Mazharin R. Banaji. 2001. Implicit attitude measures: Consistency, stability, and convergent validity. Psychological Science 12:163–170. Daly, John Patrick. 2002. When slavery was called freedom: Evangelicalism, proslavery, and the causes of the Civil War. Lexington, KY: The University Press of Kentucky. Dawidowicz, Lucy S. 1975. The war against the Jews, 1933–1945. New York: Bantam Books. Dine, Janet, and Bob Watt. 1995. Sexual harassment: Moving away from discrimination. Modern Law Review 58:343–363. Dovidio, John F., Nancy Evans, and Richard B. Tyler. 1986. Racial stereotypes: The contents of their cognitive representations. Journal of Experimental Social Psychology 22:22–37. Dudziak, Mary L. 2000. Cold War civil rights: Race and the image of American democracy. Princeton, NJ: Princeton University Press. Duncan, Otis Dudley. 1968. Inheritance of poverty or inheritance of race? In On understanding poverty, ed. Daniel Patrick Moynihan, 85–110. New York: Basic Books. Durkheim, Emile. [1895] 1982. The rules of sociological method, ed. Steven Lukes, trans. W. D. Halls. New York: The Free Press. Durlauf, Steven N. 1999. The case “against” social capital. Focus 20 (3): 1–5. Educational Testing Ser vice. 1997. NAEP 1996 trends in academic progress. Washington, DC: U.S. Department of Education, Office of Educational Research and Improvement. Engels, Friedrich. [1884] 1978. The origin of the family, private property, and the state. In The Marx-Engels reader, 2nd ed., ed. Robert C. Tucker, 734–759. New York: W. W. Norton and Company. Engram, Eleanor. 1982. Methodological problems in the study of black families. In Science, myth, and reality, 139–48. Westport, CT: Greenwood Press. Epstein, Richard A. 1979. Possession as the root of title. Georgia Law Review 13: 1221– 1244.

References / 263

Faludi, Susan. 1991. Backlash: The undeclared war against American women. New York: Random House. Farkas, George, and Kevin Vicknair. 1996. Appropriate tests of racial wage discrimination require controls for cognitive skill: Comment on Cancio, Evans, and Maume. American Sociological Review 61:557–560. Farley, Reynolds, and Walter R. Allen. 1989. The color line and the quality of life in America. New York: Oxford University Press. Feagin, Joe. 1991. The continuing significance of race: Antiblack discrimination in public places. American Sociological Review 56:101–116. Feagin, Joe R., and Hernán Vera. 1995. White racism: The basics. New York: Routledge. Featherman, David L., and Robert M. Hauser. 1976. Sexual inequalities and socioeconomic achievement in the U.S., 1962–1973. American Sociological Review 41:462–483. Federal Rules of Civil Procedure 23(a) 1966. Ferguson, Ronald F. 1998. Can schools narrow the black-white test score gap? In The black-white test score gap, ed. Christopher Jencks and Meredith Phillips, 318–374. Washington, DC: Brookings Institution Press. Ferree, Myra M. 1974. A woman for president? Changing responses 1958–1972. Public Opinion Quarterly 38:390–399. Fischer, Claude, Michael Hout, Martín Sánchez Jankowski, Samuel R. Lucas, Ann Swidler, and Kim Voss. 1996. Inequality by design: Cracking the bell curve myth. Princeton, NJ: Princeton University Press. Fiske, Susan T. 2000. “Stereotyping, prejudice, and discrimination at the seam between centuries: Evolution, culture, mind, and brain. European Journal of Social Psychology 30:299–322. Fleischacker, Samuel. 2004. On Adam Smith’s Wealth of Nations: A philosophical companion. Princeton, NJ: Princeton University Press. Florida, Richard, and Martin Kenney. 1991. Transplanted organizations: The transfer of Japanese industrial organization to the U.S. American Sociological Review 56: 381–398. Foner, Philip S. 1979. Women and the American labor movement: From colonial times to the eve of World War I. New York: The Free Press. Fordham, Signithia, and John U. Ogbu. 1986. Black students’ school success: Coping with the “Burden of ‘Acting White.’ ” Urban Review 18:176–206. Freeman, Alan David. 1978. Legitimizing racial discrimination through antidiscrimination law: A critical review of Supreme Court doctrine. Minnesota Law Review 62: 1049–1119. Freeman, Donald J., Susan M. Brookhart, and William E. Loadman. 1999. Realities of teaching in racially/ethnically diverse schools. Urban Education 34:89–114. Frug, Mary Joe. 1985. Re-reading contracts: A feminist analysis of a contracts casebook. The American University Law Review 34:1065–1140. Garrow, David J. 1994. Hopelessly hollow history: Revisionist devaluing of Brown v. Board of Education. Virginia Law Review 8:151–160. Gaskins, Richard. 1982. Second thoughts on “law as an instrument of social change.” Law and Human Behavior 6:153–168. Gilens, Martin. 1996. “Race coding” and white opposition to welfare. American Political Science Review 90:593–604.

264 / References

Gill, Andrew M., and Robert J. Michaels. 1992. Does drug use lower wages? Industrial and Labor Relations Review 45:419–434. Glass, Jennifer, and Valerie Camarigg. 1992. Gender, parenthood, and job-family compatibility. American Journal of Sociology 98:131–151. Goggin, Jacqueline. 1983. Countering white racist scholarship: Carter G. Woodson and the Journal of Negro History. Journal of Negro History 68:355–375. ———. 1992. Challenging sexual discrimination in the historical profession: Women historians and the American Historical Association, 1890–1940. American Historical Review 97:769–802. Goldberg, Matthew S. 1982. Discrimination, nepotism, and long-run wage differentials. Quarterly Journal of Economics 97:307–319. Goldin, Claudia. 1986. The economic status of women in the early republic: A quantitative approach. Journal of Interdisciplinary History 13:375–404. ———. 1990. Understanding the gender gap: An economic history of American women. New York: Oxford University Press. Goldin, Claudia, and Cecilia Rouse. 2000. Orchestrating impartiality: The impact of “blind” auditions on female musicians. American Economic Review 90:715–741. Gottfredson, Linda S. 1997a. Inequality by design: Cracking the bell curve myth. Personnel Psychology 50:741–746. ———. 1997b. Why g matters: The complexity of everyday life. Intelligence 24:79–132. Greene, Brian. 1999. The elegant universe: Superstrings, hidden dimensions, and the quest for the ultimate theory. New York: Vintage Books. Grossman, Joel B. 1970. The Supreme Court and social change. American Behavioral Scientist 13:535–551. Grusky, David B., and Jesper B. Sørenson. 1998. Can class analysis be salvaged? American Journal of Sociology 103:1187–1234. Habermas, Jürgen. 2000. Remarks on Erhard Denninger’s triad of diversity, security, and solidarity. Constellations 7:522–528. ———. 2003. Intolerance and discrimination. International Journal of Constitutional Law 1:2–12. Haines, Robin, John McDonald, and Ralph Shlomowitz. 2001. Mortality and voyage length in the Middle Passage revisited. Explorations in Economic History 38:503–533. Halaby, Charles N. 1979. Job-specific sex differences in organizational reward attainment: Wage discrimination vs. rank segregation. Social Forces 58:108–127. Hamermesh, Daniel S., and Frank D. Bean. 1998. Help or hindrance? The economic implications of immigration for African Americans. New York: Russell Sage Foundation. Harris, Cheryl I. 1993. Whiteness as property. Harvard Law Review 106:1707–1791. Hauser, Robert M. 1970. Context and consex: A cautionary tale. American Journal of Sociology 75:645–664. Hauser, Robert M., Shu-Ling Tsai, and William Sewell. 1983. A model of stratification with response error in social and psychological variables. Sociology of Education 56: 20–46. Heckman, James J. 1998. Detecting discrimination. Journal of Economic Perspectives 12:101–116. Heckman, James J., and Peter Siegelman. 1993. The urban institute audit studies: Their methods and findings. In Clear and convincing evidence: Measurement of discrimina-

References / 265

tion in America, ed. Michael Fix and Raymond J. Struyk, 187–258. Washington, DC: Urban Institute Press. Herrnstein, Richard J., and Charles Murray. 1994. The bell curve: Intelligence and class structure in American life. New York: The Free Press. Higgs, Peter. 1964. Broken symmetries, massless particles and gauge fields. Physics Letters 12:132–133. Hill, G. Christian. 1990. A never ending story: An introduction to the S&L symposium. Stanford Law and Policy Review 2 (2): 21–24. Hoff, Joan. 1991. Law, gender, and injustice: A legal history of U.S. women. New York: New York University Press. Holland, Paul W. 1986. Statistics and causal inference. Journal of the American Statistical Association 81:945–960. Hoxby, Carolyn M. 1998. Do immigrants crowd disadvantaged natives out of higher education? In Help or hindrance? The economic implications of immigration for AfricanAmericans, ed. Daniel S. Hamermesh and Frank D. Bean, 282–312. New York: Russell Sage Foundation. Hwang, Suein. 2005. The new white flight. Wall Street Journal, November 19, A1. Jacobs, David, and Ronald E. Helms. 1996. Toward a political model of incarceration: A time-series examination of multiple explanations for prison admission rates. American Journal of Sociology 102:323–357. Jaynes, Gerald David, and Robin M. Williams, Jr. 1989. A common destiny: Blacks and American society. Washington, DC: National Academy Press. Jencks, Christopher. 1998. Racial bias in testing. In The black-white test score gap, ed. Christopher Jencks and Meredith Phillips, 55–85. Washington, DC: Brookings Institution Press. Jencks, Christopher, and Meredith Phillips. 1998. The black-white test score gap. Washington, DC: Brookings Institution Press. Jesness, Jerry. 2002. Stand and Deliver revisited. Reason 34 (3): 34–49. Jones, James H. 1993. Bad blood: The Tuskegee syphilis experiment. Rev. ed. New York: The Free Press. Jussim, Lee, and Jacquelynne S. Eccles. 1992. Teacher expectations: II. Construction and reflection of student achievement. Journal of Personality and Social Psychology 63: 947–961. Kaestner, Robert. 1994. The effect of illicit drug use on the labor supply of young adults. Journal of Human Resources 29:126–155. Kalmijn, Matthijs. 1998. Intermarriage and homogamy: Causes, patterns, trends. Annual Review of Sociology 24:395–421. Katznelson, Ira. 1976. Black men, white cities: Race, politics, and migration in the United States, 1900–30, and Britain 1948–68. Chicago: The University of Chicago Press. Kemp, Kathleen A., Robert A. Carp, and David W. Brady. 1978. The Supreme Court and social change: The case of abortion. Western Political Quarterly 31:19–31. Kenealy, Pamela, Neil Frude, and William Shaw. 1991. Teacher expectations as predictors of academic success. Journal of Social Psychology 131:305–306. Kidder, William C., and Jay Rosner. 2002. How the SAT creates “built-in headwinds”: An educational and legal analysis of disparate impact. Santa Clara Law Review 43: 131–211.

266 / References

Killingsworth, Mark R. 1985. The economics of comparable worth: Analytical, empirical, and policy questions. In Comparable worth: New directions for research, ed. Heidi I. Hartmann, 86–115. Washington, DC: National Academy Press. Kirschenman, Joleen, and Kathryn M. Neckerman. 1991. “We’d love to hire them, but . . .”: The meaning of race for employers. In The urban underclass, ed. Christopher Jencks and Paul E. Peterson, 203–232. Washington, DC: The Brookings Institute. Kleinman, Daniel Lee, and Mark Solovey. 1995. Hot science/Cold War: The National Science Foundation after World War II. Radical History Review 63:110–139. Kolb, Kathryn, and Lee Jussim. 1994. Teacher expectations and underachieving gifted children. Roeper Review 17:26–30. Kuhn, Thomas S. [1962] 1970. The structure of scientific revolutions. 2nd ed., enlarged. Chicago: University of Chicago Press. Kuklinski, James H., Michael D. Cobb, and Martin Gilens. 1997. Racial attitudes and the “New South.” Journal of Politics 59:323–349. Lacy, William, and Ernest Middleton. 1981. Are educators racially prejudiced? A crossoccupational comparison of attitudes. Sociological Focus 14:87–95. Leahy, Michael. 2005. Family vacation: Why would Raechel McGhee fly her two beloved children across the country to stay with a man they had never met? Because he is their father. Washington Post Magazine, June 19, W12. Lee, Valerie E., Helen M. Marks, and Tina Byrd. 1994. Sexism in single-sex and coeducational independent secondary school classrooms. Sociology of Education 67: 92–120. Lemann, Nicholas. 1991. The promised land: The great black migration and how it changed America. New York: Alfred A. Knopf. Lempert, Richard O., David L. Chambers, and Terry K. Adams. 2000. Michigan’s minority graduates in practice: The river runs through law schools. Law and Social Inquiry 25:395–505. Leonard, David K., and Jiming Jiang. 1999. Gender bias and the college predictions of the SATs: A cry of despair. Research in Higher Education 40:375–407. Leonard, Jonathan S. 1984. Antidiscrimination or reverse discrimination: The impact of changing demographics, Title VII, and affirmative action on productivity. Journal of Human Resources 19:145–174. Levine, James P., and Theodore L. Becker. 1970. Toward and beyond a theory of Supreme Court impact. American Behavioral Scientist 13:561–573. Lewis, Tom. 1997. Divided highways: Building the interstate highways, transforming American life. New York: Viking. Lieberson, Stanley. 1980. A piece of the pie: Blacks and white immigrants since 1880. Berkeley, CA: University of California Press. ———. 1985. Making it count: The improvement of social research and theory. Berkeley, CA: University of California Press. Lloyd, Cynthia B., and Beth T. Niemi. 1979. The economics of sex differentials. New York: Columbia University Press. Loury, Glenn C. 1981. Intergenerational transfers and the distribution of earnings. Econometrica 49:843–867. ———. 1998. Discrimination in the post–civil rights era: Beyond market interactions. Journal of Economic Perspectives 12:117–126.

References / 267

———. 2002. The anatomy of racial inequality. Cambridge, MA: Harvard University Press. Lucas, Samuel R. 1996. Selective attrition in a newly hostile regime: The case of 1980 sophomores. Social Forces 75:511–533. ———. 2000. Hope, anguish, and the problem of our time: An essay on publication of the Black-White Test Score Gap. Teachers College Record 102:463–475. ———. 2001. Effectively maintained inequality: Education transitions, track mobility, and social background effects. American Journal of Sociology 106:1642–1690. Lucas, Samuel R., and Mark Berends. 2007. Race and track location in U.S. public schools. Research in Social Stratification and Mobility 25:169–187. Lucas, Samuel R., and Adam Gamoran. 2002. Tracking and the achievement gap. In Bridging the achievement gap, ed. John E. Chubb and Tom Loveless, 171–198. Washington, DC: Brookings Institution Press. Lucas, Samuel R., and Marcel Paret. 2005. Law, race, and education in the United States. Annual Review of Law and Social Science 1:203–231. Lucas, Samuel Roundfield. 1999. Tracking inequality: Stratification and mobility in American high schools. New York: Teachers College Press. MacKenzie, Donald. 1978. Statistical theory and social interests: A case study. Social Studies of Science 8:35–83. Mair, Andrew, Richard Florida, and Martin Kenney. 1988. The new geography of automobile production: Japanese transplants in North America. Economic Geography 64: 352–373. Mannheim, Karl. [1928] 1952. The problem of generations. In Essays on the sociology of knowledge, 276–322. London, UK: Routledge and Kegan Paul. Manski, Charles. 1993. Identification problems in the social sciences. Sociological Methodology 23:1–56. ———. 1995. Identification problems in the social sciences. Cambridge, MA: Harvard University Press. Marx, Karl. [1844] 1977. Alienated labor. In Karl Marx: Selected Writings, ed. David McLellan, 77–87. Oxford, UK: Oxford University Press. Mason, Karen Oppenheim, and Yu-Hsia Lu. 1988. Attitudes toward women’s familial roles: Changes in the United States, 1977–1985. Gender & Society 7:39–57. Massey, Douglas S., and Nancy A. Denton. 1993. American apartheid: Segregation and the making of the underclass. Cambridge, MA: Harvard University Press. Maume, David J., Jr., A. Silvio Cancio, and T. David Evans. 1996. Cognitive skills and racial wage inequality: Reply to Farkas and Vicknair. American Sociological Review 61:561–564. McWhorter, John H. 2000. Losing the race: Self-sabotage in black America. New York: The Free Press. Menchaca, Martha. 1993. Chicano Indianism: A historical account of racial repression in the United States. American Ethnologist 20:583–603. Mickelson, Roslyn A. 2003. When are racial disparities in education the result of racial discrimination? A social science perspective. Teachers College Record 105: 1052–1086. Mills, C. Wright. 1959. The sociological imagination. New York, NY: Oxford University Press.

268 / References

Mincer, Jacob. 1970. The distribution of labor incomes: A survey with special reference to the human capital approach. Journal of Economic Literature 8:1–26. Mishler, William, and Reginald S. Sheehan. 1993. The Supreme Court as a countermajoritarian institution? The impact of public opinion on Supreme Court decisions. American Political Science Review 87:87–101. Mondak, Jeffrey J. 1994. Policy legitimacy and the Supreme Court: The sources and contexts of legitimation. Political Research Quarterly 47:675–692. Myrdal, Gunnar. [1944] 1962. An American dilemma: The negro problem and modern democracy, volumes I and II. Evanston: Harper Torchbooks. Nalty, Bernard. 1986. Strength for the fight: A history of black Americans in the military. New York: The Free Press. Nalty, Bernard C., and Morris J. MacGregor. 1981. Blacks in the military: Essential documents. Wilmington, DE: Scholarly Resources, Inc. Nathans, Heather. 2004. A much-maligned people: Jews on and off the stage in the early American republic. Early American Studies 2:310–342. Nathanson, C.A. 1984. Sex differences in mortality. Annual Review of Sociology 10: 191–213. Newman, Louise Michelle. 1999. White women’s rights: The racial origins of feminism in the United States. New York: Oxford University Press. NORC at the University of Chicago. General social survey. http://www.norc.uchicago. edu/projects/gensoc3.asp. Oaxaca, Ronald. 1973. Male-female wage differentials in urban labor markets. International Economic Review 14:693–709. O’Connor, Julia S., Ann Shola Orloff, and Sheila Shaver. 1999. States, markets, and families: Gender, liberalism, and social policy in Australia, Canada, Great Britain, and the United States. New York: Cambridge University Press. Ogbu, John U. 1991. Low school performance as an adaptation: The case of blacks in Stockton, California. In Minority status and schooling: A comparative study of immigrant and involuntary minorities, 249–285. New York: Garland Publishing, Inc. ———. 2003. Black American students in an affluent suburb: A study of academic disengagement. Mahwah, NJ: Lawrence Erlbaum Associates. Oliver, Melvin L., and Thomas M. Shapiro. 1995. Black wealth/white wealth: A new perspective on racial inequality. New York: Routledge. Omi, Michael, and Howard Winant. 1994. Racial formation in the United States: From the 1960s to the 1990s. 2nd ed. New York: Routledge. Overy, Richard. 1997. Russia’s war: Blood on the snow. New York: Penguin Books. Paetzold, Ramona L., and Steven L. Willborn. 2002. The statistics of discrimination: Using statistical evidence in discrimination cases. New York: Thomson West. Palmer, R.R., and Joel Colton. 1978. A history of the modern world. 5th ed. New York: Alfred A. Knopf. Parsons, Talcott. 1958. Social structure and the development of personality: Freud’s contribution to the integration of psychology and sociology. Psychiatry 21:321–340. Pettigrew, Thomas. 1973. Review symposium of Inequality: A Reassessment of the Effect of Family and Schooling in America. American Journal of Sociology 78:1527–1532. Pfeffer, Paula F. 1990. A. Phillip Randolph, pioneer of the civil rights movement. Baton Rouge, LA: Louisiana State University Press.

References / 269

Pontell, Henry N., and Kitty Calavita. 1993. The savings and loan industry. Crime and Justice 18:203–246. Press, Julie E., and Eleanor Townsley. 1998. Wives’ and husbands’ housework reporting: Gender, class, and social desirability. Gender and Society 12:188–218. Pushkareva, Natalia. 1997. Women in Russian history: From the tenth to the twentieth century. Trans. and ed. Eve Levin. Armonk, NY: M. E. Sharpe. Putnam, Robert D. 2000. Bowling alone: The collapse and revival of American community. New York: Simon & Schuster. Quadagno, Jill. 1988. The transformation of old age security: Class and politics in the American welfare state. Chicago: The University of Chicago Press. ———. 1994. The color of welfare: How racism undermined the war on poverty. New York: Oxford University Press. Rabin, Yale. 1973. Highways as a barrier to equal access. The Annals of the American Academy of Political and Social Science 407 (May): 63–77. Ragan, James F., and Carol Horton Tremblay. 1988. Testing for employee discrimination by race and sex. Journal of Human Resources 23:123–137. Randolph, A. Phillip. [1948] 1971. A. Phillip Randolph urges civil disobedience against a Jim Crow army. In Black protest thought in the twentieth century, 2nd ed., ed. August Meier, Elliott Rudwick, and Francis L. Broderick, 274–280. Indianapolis, IN: BobbsMerrill Educational Publishing. Rawls, John. [1971] 1999. A theory of justice. Rev. ed. Cambridge, MA: The Belknap Press of Harvard University Press. Reich, Michael. 1978. Who benefits from racism? The distribution among whites of gains and losses from racial inequality. Journal of Human Resources 13:524–544. Resolution Trust Corporation. 1990. Report on investigations to date: Office of investigations, resolutions, and operations division. December 31. Washington, DC: Resolution Trust Corporation. Rhode, Deborah L. 1989. Justice and gender: Sex discrimination and the law. Cambridge, MA: Harvard University Press. Richter, Daniel K. 2001. Facing east from Indian country: A native history of early America. Cambridge, MA: Harvard University Press. Ridley, Stanley E., James A. Bayton, and Janice H. Outtz. 1989. Taxi service in the District of Columbia: Is it influenced by patron’s race and destination? Paper prepared for the Washington, D.C., Lawyer’s Committee for Civil Rights. Roberts, Paul. 2004. The end of oil: On the edge of a perilous new world. New York: Houghton Mifflin Company. Roemer, John E. 1988. Free to lose: An introduction to Marxist economic philosophy. Cambridge, MA: Harvard University Press. Rosenberg, Gerald N. 1991. The hollow hope: Can courts bring about social change? Chicago, IL: University of Chicago Press. Ryan, James E. 2003. Race discrimination in education: A legal perspective. Teachers College Record 105:1087–1118. Ryan, Mary P. 1975. Womanhood in America. New York: New Viewpoints. Sakamoto, Arthur, and Meichu Chen. 1991. Inequality and attainment in a dual labor market. American Sociological Review 56:295–308. Saracho, Olivia N. 1991. Teacher expectations of students’ performance: A review of the research. Early Child Development and Care 76:27–41.

270 / References

Sawyer, R. Keith. 2002. Durkheim’s dilemma: Toward a sociology of emergence. Sociological Theory 20:227–247. Schiele, Jerome H. 1991. An epistemological perspective on intelligence assessment among African American children. The Journal of Black Psychology 17 (2): 23–36. Schmidt, Frank L., and John E. Hunter. 1999. Bias in standardized educational and employment tests as justification for racial preferences in affirmative action programs. Research in Social Stratification and Mobility 17:285–302. Schultz, Theodore W. 1961. Investment in human capital. American Economic Review 51:1–17. Schuman, Howard, and Cheryl Rieger. 1992. Historical analogies, generational effects, and attitudes toward war. American Sociological Review 57:315–326. Schuman, Howard, Charlotte Steeh, Lawrence Bobo, and Maria Krysan. [1985] 1997. Racial attitudes in America: Trends and interpretations. Cambridge, MA: Harvard University Press. Schweber, Libby. 2001. Manipulation and population statistics in nineteenth-century France and England. Social Research 68:547–582. Sears, David O. 1988. Symbolic racism. In Eliminating racism: Profiles in controversy, ed. Phyllis A. Katz and Dalmas A. Taylor, 53–84. New York: Plenum. Seligman, Martin E. P. 1968. Chronic fear produced by unpredictable shock. Journal of Comparative and Physiological Psychology 66:402–411. Semyonov, Moshe, and Yinon Cohen. 1990. Ethnic discrimination and the income of majority-group workers. American Sociological Review 55:107–114. Singer, Joseph William. 1991. Re-reading property. New England Law Review 26:711– 729. Skrentny, John D. 2002. The minority rights revolution. Cambridge, MA: The Belknap Press of Harvard University Press. Smith, Michael L. 2007. Making direct democracy work: Czech local referendums in regional comparison. In Towards DIY-politics: Participatory and direct democracy at the local level in Europe, ed. Pascal Delwit, Jean-Benoit Pilet, Herwig Reynaert, and Kristof Steyvers, 67–120. Brugge, Belgium: Vanden Broele. Sneider, Allison. 1994. The impact of empire on the North American Woman Suffrage Movement: Suffrage racism in an imperial context. UCLA Historical Journal 14:14–32. Sommers, Christina Hoff. 1994. Who stole feminism? How women have betrayed women. New York: Simon & Schuster. Sowell, Thomas. 1983. The economics of politics and race: An international perspective. New York: William Morrow Company, Inc. Spitz, Vivien. 2005. Doctors from Hell: The horrific account of Nazi experiments on humans. Boulder, CO: Sentient Publications. Spitze, Glenna, and Deborah S. Freedman. 1980. Changing attitudes toward women’s non-family roles: 1938 to 1978. Sociology of Work and Occupations 7:317–335. Spruill, Julia Cherry. 1938. Women’s life and work in the Southern colonies. Chapel Hill, NC: University of North Carolina Press. Squires, Gregory D., and William Velez. 1987. Insurance redlining and the transformation of an urban metropolis. Urban Affairs Quarterly 23:63–83. Steele, Claude M. 1997. A threat in the air: How stereotypes shape intellectual identity and performance. American Psychologist 52:613–629.

References / 271

Stein, Rob. 2005. Found on the web, with DNA: A boy’s father. Washington Post, November 13, A9. Swidler, Ann. 1986. Culture in action: Symbols and strategies. American Sociological Review 51:273–286. Takaki, Ronald. [1979] 1990. Iron cages: Race and culture in 19th century America. New York: Oxford University Press. Terborg-Penn, Rosalyn. 1978. Discrimination against Afro-American women in the women’s movement, 1830–1920. In The Afro-American woman: Struggles and images, ed. Sharon Harley and Rosalyn Terborg-Penn, 17–27. Port Washington, NY: Kennikat Press National University Publications. Thorne, Barrie. 1993. Gender play: Girls and boys in school. New Brunswick, NJ: Rutgers University Press. Thurow, Lester C. 1969. Poverty and discrimination. Washington, DC: The Brookings Institution. Tise, Larry E. 1987. Proslavery: A history of the defense of slavery in America, 1701–1840. Athens, GA: The University of Georgia Press. Tjadan, Patricia, and Nancy Thoennes. 2000. Extent, nature, and consequences of intimate partner violence: Findings from the national violence against women survey. Washington, DC: National Institute of Justice and National Centers for Disease Control. Turner, Margery Austin. 1993. Limits on neighborhood choice: Evidence of racial and ethnic steering in urban housing markets. In Clear and convincing evidence: Measurement of discrimination in America, ed. Michael Fix and Raymond J. Struyk, 117–152. Washington, DC: Urban Institute Press. Turner, Ralph. 1952. Foci of discrimination in the employment of nonwhites. American Journal of Sociology 58:247–256. Tyson, Karolyn. 2002. Weighing in: Elementary-age students and the debate on attitudes toward school among black students. Social Forces 80:1157–1189. United States Department of Commerce, Census Bureau. 2001. Survey of income and program participation users’ guide. 3rd ed. Washington, DC: United States Department of Commerce, Economics and Statistics Administration. United States Department of Labor, Women’s Bureau. 1946. Women workers in ten war production areas and their postwar employment plans. Bulletin 209. Washington, DC: Government Printing Office. Walker, Lenore E. 1979. The battered woman. New York: Harper and Row. Washington, George, and the Members of the Continental Congress. [1787] 1976. The Constitution of the United States. In A documentary history of the United States, expanded and updated bicentennial edition, ed. Richard D. Heffner, 24–42. New York: New American Library. Waters, Mary C. 1990. Ethnic options: Choosing identities in America. Berkeley, CA: University of California Press. Weisman, Leslie Kanes. 1994. Discrimination by design: A feminist critique of the man-made environment. Chicago, IL: University of Illinois Press. Weissbrodt, Sylvia. 1946. Women workers in ten war production areas and their postwar employment plans. Washington, DC: Women’s Bureau, Department of Labor, United States. West, Candace, and Don H. Zimmerman. 1983. Doing gender. Gender and Society 1:125– 151.

272 / References

West, Cornel. 1982. Prophesy deliverance! An Afro-American revolutionary Christianity. Philadelphia, PA: Westminster Press. Wienk, Ronald E., Clifford E. Reid, John C. Simonson, and Frederick J. Eggers. 1979. Measuring racial discrimination in American housing markets: The housing market practices survey. Washington, DC: United States Department of Housing and Urban Development. Wilkie, Jane Riblett. 1993. Changes in U.S. men’s attitudes toward the family provider role, 1972–1989. Gender & Society 7:261–279. Wilson, William Julius. 1978. The declining significance of race: Blacks and changing American institutions. Chicago, IL: The University of Chicago Press. ———. 1987. The truly disadvantaged: The inner city, the underclass, and public policy. Chicago, IL: The University of Chicago Press. ———. 1996. When work disappears: The world of the new urban poor. New York: Alfred A. Knopf. Wright, Erik Olin. 1979. Class structure and income determination. Toronto: Academic Press. ———. 1985. Classes. London: Verso. ———. 1994. Interrogating inequality: Essays on class analysis, socialism, and Marxism. New York: Verso. ———. 2000. Class, exploitation, and economic rents: Reflections on Sørensen’s “sounder basis.” American Journal of Sociology 105:1559–1571. Yinger, John. 1986. Measuring racial discrimination with fair housing audits: Caught in the act. American Economic Review 76:881–893. Zalokar, Nadja. 1990. The economic status of black women: An exploratory investigation. Washington, DC: United States Commission on Civil Rights. Zieky, Michael. 2002. Ensuring the fairness of licensing tests. http://www.nocheating.org/ Media/Research/pdf/FAIRNESS-LICENSING.pdf. Zimmerman, Don H., and Candace West. 1975. Sex roles, interruptions and silences in conversation. In Language and sex: Difference and dominance, ed. Barrie Thorne and Nancy Henley, 105–169. Rowley, MA: Newbury House.

Cases Cited Alexander v. Sandoval ((99-1908) 532 U.S. 275 (2001)) Bazemore v. Friday (106 S. Ct. 3000 (1986)). Bodner v. Banque Paribas (114 F. Supp. 2d 117 (EDNY 2000)) Bowen v. Independent Publishing Co. (96 S.E. 2d 564, 565 (S.C. 1957)) Brown v. Board of Education I (347 U.S. 483 (1954)) Brown v. Board of Education II (349 U.S. 294 (1955)) Burlington Industries, Inc. v. Ellerth (524 U.S. 742 (1998)) Crenshaw v. Williams (191 Ky. 559, 231 S.W. 45 (1921)) Ferguson v. McKiernan (855 A.2d 121 (Pa. Super. 2004)) Frank v. United Airlines (216 F.3d 845 (9th Cir. 2000)) Griggs v. Duke Power Company (401 U.S. 424 (1971)) Grutter v. Bollinger (539 U.S. 306 (2003)) Johnson v. Board of Education of Wilson County (82 S.E. 832, 833–835 (1914))

References / 273

Korematsu v. United States (323 U.S. 214 (1944)) Menzel v. List (49 Misc. 2d 300, 267 N.Y.S. 2d 804 (S. Ct. 1966)) Milliken v. Bradley (418 U.S. 717 (1974)) Missouri v. Jenkins (495 U.S. 33 (1990)) Plessy v. Ferguson (163 U.S. 537 (1896)) Roberts v. City of Boston (59 Mass (5 Cush.) 198 (1850)) Santa Clara County v. Southern Pacific Railroad (118 U.S. 394 (1886)). Swann v. Charlotte-Mecklenburg Board of Education (402 U.S. 1 (1971)) Tee-Hit-Ton Indians v. United States (348 U.S. 272 (1955)) United States v. 50 Acres of Land (469 U.S. 24 (1984)) University of California Regents v. Bakke (438 U.S. 265 (1978))

Index

abuse: child, 187, 198; women in marriage, 212 adaptive testing, 128–129 affi rmative action policy, 109, 110, 215 Africa, South, 153, 196–197 African-Americans. See blacks Aigner, Dennis J., 158 Ainsworth-Darnell, James W., 231 airlines, Frank v. United Airlines, 91–92 alienation, damaged social relation, 193–195, 199, 200 Allport, Gordon, 205, 245–246; defi ning discrimination, 176–180, 189, 201, 241, 244 Altonji, Joseph G., 162–163 An American Dilemma (Myrdal), 246 American Indians. See Native Americans American Revolution, 67 Anderson, Karen, 71–72, 211 Anthony, Susan B., 57 Arrow, Kenneth J., 158 Asians: discrimination targets, 189; school per for mance, 132; socioeconomic success, 215 asocial scientific method, 131, 132, 140–141, 143–174, 239–240 assumptions: standardized testing simulation, 258; statistical, 29–30, 254–256. See also taken-for-granted assumptions

asymmetry of experience, 21–25, 48–52, 53, 79–80, 235–236, 238 atomistic perspective, 163, 164, 172, 173, 205, 246; defi ning discrimination, 239–240; fi rm-specific focus, 219; social relational perspective vs., 227–228. See also individualist perspective Attucks, Crispus, 67 audit methodology, 163–164, 167–170 authorities encountered: by blacks, 28–36, 48, 53, 54, 236, 251–256; by women and girls, 38–47, 48, 53, 54, 236, 251–256 auto workers, 71–72 backlashes, 14–15 Bayor, Ronald H., 62 Bazemore v. Friday, 92 Bean, Frank D., 212 Becker, Gary (Economics of Discrimination/ taste theory), 145–174, 176, 232, 234, 240–241; Allport and, 178–179, 245–246; amount distinguished from effects of discrimination, 148, 206–210, 240; analytic economists’ response to, 156–163, 173; inequal ity without discrimination, 148, 153–155, 161–162, 167, 170, 173, 240; Marxist theory and, 198, 199; mate

276 / Index

Becker, Gary (continued) selection as discriminatory, 195; measures consistent with, 224–225; motivation ignored by, 148, 163, 164, 167, 168, 171–172, 173, 234; and noneconomic forms of discrimination, 150, 152, 153, 178; nontargets’ effects, 153, 164, 240; residual attribution differing from, 164–165, 166 Bell, Derrick, 14, 18 Bell, Derrick A., Jr., 68, 113 Bentham, Jeremy, 106 Bergmann, Barbara R., 205, 218–220 BFOQ (bona fide occupational qualification), 89–90, 125, 241 black holes, 209–210 blacks: black culture, 231–232; blacks’ attitudes toward, 37, 231; Middle Passage, 198. See also blacks in labor force; black-white relations blacks in labor force: agricultural, 76–77, 92; black/white capital, 76–77, 147–149, 160–161; employee discrimination litigation, 92, 171; exploited, 55–56, 76; Latino/as preferred over, 207–208; manufacturing, 60–62, 64–65; taste theories, 147–148, 153–161 black-white relations, 84, 144; capital, 76–77, 147–149, 160–161; comparisons, 119, 230–232, 234, 243; delegitimation of exploitation rhetoric (1948–1970), 62–69, 79, 84; discrimination litigation, 92, 94–95, 96; disparate treatment/disparate impact discrimination, 89, 109–110; environmental level of discrimination, 231, 233; exploitive, 55–58, 72–73, 237; geopolitical implications, 66–69, 72–73, 237; historical, 60–62, 105–106; housing, 28–33, 36, 62–64, 77, 80, 168; income, 92, 147–148, 153–156, 161, 232–233, 243; “lacks willpower” explanation for black socioeconomic disadvantage, 26–27, 32–35, 251–252; legalized discrimination, 88–89, 105–106, 108–109; “low ability” explanation of black socioeconomic disadvantage, 26–32, 251–252; military, 65–68, 69–70; property rights, 105–110, 111, 114, 132, 141; residual attribution and, 165, 166, 167, 215; reverse race discrimination, 202; segregated, 63–69, 95; separate-but-equal, 65, 68, 69, 94; slavery, 89, 105–106, 197, 212; social security, 76–77; socioeconomic disadvantage of blacks, 26–36, 212, 251; South Africa, 153; standardized testing, 120–123, 131, 132,

211–212, 213; survey research, 25; taste theories, 147–148, 153–161; white attributions concerning blacks, 25–36; white authorities blacks may encounter, 28–36, 48, 53, 54, 236, 251–256; whiteness as property, 107–110, 111, 114, 132, 141, 238–239; women’s rights and, 57, 71, 110–111. See also blacks in labor force The Black-White Test Score Gap (Jencks and Phillips), 211–212, 213 Blank, Rebecca M., 162–163 Blight, David W., 58 Bonilla-Silva, Eduardo, 47, 211 Borstelmann, Thomas, 69 Bowen v. Independent Publishing Co., 108–109 Bowling Alone (Putnam), 20 Braddock, Jomills II, 207, 208, 209, 219 Breland, Hunter, 121 Brown v. Board of Education, 68, 94, 108, 109 Buber, Martin, 201 Burlington Industries, Inc. v. Ellerth, 95–96 Burstein, Paul, 75 Byrd, Tina, 181, 211 bystanders, 152; beneficiary, 136, 140, 141; legal responsibility, 104, 136, 141 Cain, Glen G., 158, 205 Campbell, D’Ann, 70–71, 211 capital: black/white, 76–77, 147–149, 160–161. See also human capital theory; income capitalism, and damaged social relations, 193–194, 195, 199–200 Carter, Prudence L., 231 causal inference, 11–12, 146, 171, 228–232, 243. See also Rubin’s Model caused-based approaches, 210–216, 223 character, defended in discrimination discussion, 2–5, 6–7, 51 Cherlin, Andrew, 36 Chevan, Albert, 63 children: abusers of, 187, 198; gender-specific socialization, 42–47, 181–183, 253; of sperm donors, 113–114 cities, racial discrimination, 60–65 Civil Rights Act, 75, 110 civil rights activists, 69; and blacks in military, 66–68; and women’s rights, 75, 110 civil unions, 186–187 Civil War, 72 class: damaged social relation, 193, 194, 199–200. See also socioeconomic status classes of persons, 175, 179, 180, 188, 217, 223, 224–225. See also systemic focus

Index / 277

climate change, 219–220 cloaking behavior, 171–172, 222–223, 225, 226 Coase, R. H., 20 cohabiter relation, 185–186 coherence standards, 5, 201, 239, 250; damaged social relation, 203, 242; defi nition of discrimination, 223, 235; legal reasoning and, 87, 96, 101; measurement strategies, 171; research, 206; social scientific, 97, 101. See also incoherence Cole, Robert E., 64–65 collective models, 161–163 A Common Destiny (Jaynes and Williams), 25 communists, 193, 197 comparisons: differential item functioning (DIF), 125, 127–128, 130–131; factual/ counterfactual observations, 12, 229; gender, 91–92, 116, 119, 146, 218, 230–232, 234, 243; legal evidence, 91–92; race, 119, 230–232, 234, 243; standardized testing, 116–119, 125, 127–128, 130–131; targets/ nontargets, 233, 243 concealment/cloaking behavior, 171–172, 222–223, 225, 226 condoned exploitation. See regime of condoned exploitation conspiracy, 80, 237 Constitution, U.S., 89 consumer discrimination, 150 Cook, Philip J., 231 corporations, individual rights, 134 counterfactual observations, 12, 229, 231–232, 234 court cases, 88; Bazemore v. Friday, 92; Bowen v. Independent Publishing Co., 108–109; Brown v. Board of Education, 68, 94, 108, 109; Burlington Industries, Inc. v. Ellerth, 95–96; Crenshaw v. Williams, 111; discrimination litigation, 90–97, 100, 171; Frank v. United Airlines, 91–92; Johnson v. Board of Education of Wilson County, 108; Korematsu v. United States, 83–84; Milliken v. Bradley, 94, 109; Missouri v. Jenkins, 95; Plessy v. Ferguson, 69, 109; Roberts v. City of Boston, 68; University of California Regents v. Bakke, 94–95. See also legal system Crenshaw v. Williams, 111 crime. See illegality crises, rights disregarded during, 57–58 critical legal perspectives, 104–115, 141, 143, 145, 238–239; and discrimination as social relation, 201, 241, 244; human capital processes, 130–133, 136–138; social science

of discrimination, 140, 149, 155–156, 171–172, 205; systemic rather than individualist discrimination, 217–218; on taken-for-granted arrangements, 181; and taste theories of discrimination, 149, 155–161 Daly, John Patrick, 58 damaged social relation, 179, 217–218; alienation, 193–195, 199, 200; discrimination as, 21, 183–184, 187–206, 221–225, 234, 240–245; marriage, 194–195, 221; measur ing damages, 221–225; not synonymous with discrimination, 199 The Declining Significance of Race (Wilson), 60–62 defendants, defi ning, 135–136 defensiveness, 2–5, 6–7, 51 defi ning discrimination, 2, 7, 9, 10–11, 101, 173; Allport, 176–180, 189, 201, 241, 244; Becker, 145–174, 178; circular, 153–155; history versions, 78–81; inappropriate, 87; individualist, 13, 78–79, 172–176, 208, 242–243; legal, 10–11, 22, 86–90, 99–102, 139, 238; as social relation, 183–184, 188–191, 223, 244; social science, 10–11, 19–21, 78, 101, 139, 239–240; United Nations, 176; working defi nition for social science research, 176, 179–182, 199, 201–204 Denton, Nancy A., 63–64 Deskins, Donald R., Jr., 64–65 Dine, Janet, 96 discrimination, 5–6, 161; change/continuity, 13–18, 47–50, 237; character defended in discussion of, 2–5, 6–7, 51; computerized, 128–129; cross-sphere connection, 227; as damaged social relation, 21, 183–184, 187–206, 221–225, 234, 240–245; damaged social relation not synonymous with, 199; environmental level of, 231, 233, 243; eradication efforts, 54, 81, 144, 233, 243; existence distinguished from effects of, 148, 206–216, 240, 244–245; geopolitical implications, 66–69, 72–73, 237; hidden/ cloaked/concealed, 171–172, 222–223, 225, 226; illegal, 10, 11, 54–55, 146, 171–172, 223, 226–227; infi nite number of possible moments of, 146, 171, 172; language of, 190; legalized, 88–89, 105–106, 108–109; Marxist theories, 160–161, 162, 195–200; positive and negative, 154, 178–179; power relations determining, 189, 196; reconceptualization/paradigm shift, 4–5, 16–18, 21,

278 / Index

discrimination, (continued) 24, 50–51, 146–147, 176, 247–248; reverse, 188–190, 196, 201–202, 242; as social relation, 19–21, 180–184, 187–234, 244; statistical, 158–160, 241; by targets, 189; targets of, 189–190, 192, 217–218, 222, 238; taste theories, 145–174, 241. See also defi ning discrimination; discrimination effects; era of contested prejudice; gender; individualist perspective; labor force; legal system; race; social policy discrimination effects, 2, 9–24, 143–174; audit methodology, 163–164, 167–170; caused-based approaches, 210–216, 223; defi ning, 143–174; effects-based approaches, 148, 206–216, 234; epistemological foundations, 146, 171, 205–234; equality’s relevance to, 1–2, 148, 153–155, 161–162, 167, 170, 173, 210–214, 240; estimating, 146, 171–173, 205–234, 244; existence of discrimination distinguished from, 148, 206–216, 240, 244–245; on nontargets, 153, 164, 190, 215, 217–218, 233, 240, 243, 248; researching, 18, 21, 144–174, 205–234, 246–248; residual attribution, 163–167, 170, 215; Rubin’s Model and, 12–13, 172–173, 225, 228–229, 243. See also Becker, Gary; economics Disney-esque version of history, 78–81 disparate impact, 91; black-white relations, 89, 109–110; human capital theory, 99–100, 114; labor force, 89–90; of standardized testing, 125, 131 disparate treatment: black-white relations, 109–110; failing to become discrimination, 202–203; human capital theory, 99–100, 114; labor force, 89, 90; of standardized testing, 125, 131 disposability, of disenfranchised persons, 56 domestic violence, and police officers’ views of women’s rights, 41–42 dominant legal reasoning, 22, 86–102, 103, 140–145, 238–239; core contradiction, 133–136; defi ning discrimination, 10–11, 86–90, 99–102, 139, 238; remedy, 137; standardized testing, 130–136; taken-for-granted assumptions, 104–105, 137, 139, 217–218, 239, 250. See also disparate impact; disparate treatment dominant social science approach, 22, 51–52, 176, 180–181, 239–241, 250; defi ning discrimination, 10–11, 19, 239–240; dominant legal reasoning pervading, 10–11,

141–142, 239; individualist, 19–20, 218, 232, 245–246; problems with, 225–232. See also atomistic perspective Douglass, Frederick, 57 Downey, Douglas B., 231 DuBois, W.E.B., 67–68 Dudziak, Mary L., 69, 79 Duncan decompositions, 165 earnings. See income economics, 147, 199, 246; Allport’s defi nition going beyond, 178; analytic, 156–163, 173; Asian success, 215; audit methodology, 167–170; capitalism, 193–194, 195, 199–200; exploitation, 55, 58, 68–69, 196–198, 237; neoclassical, 233; underestimating women’s productivity, 73. See also Becker, Gary (Economics of Discrimination/taste theory); labor force; socioeconomic status education. See socialization; standardized testing; teachers empirical research, 6, 7, 163, 201, 244, 245–246, 250 employers: discrimination by, 92, 147–161, 168–169, 171, 207–208, 240; nepotism by, 153–154, 157 Engels, Friedrich, 194 Engram, Eleanor, 205 environmental level of discrimination, 231, 233, 243 epistemology, 6, 7, 22, 24, 52, 138–139; black-white relations, 35–36; discrimination as damaged social relation, 205–206, 221–225, 234, 242–243; estimating effects of discrimination, 146, 171, 205–234 Equal Employment Opportunity Commission (EEOC), 110 equality: discriminatory, 182–183; income, 148, 154–156; original position of ignorance and, 192; persistent inequal ity, 166; relevance to discrimination effects, 1–2, 148, 153–155, 161–162, 167, 170, 173, 210–214, 240 era of contested prejudice, 8–24, 58–59, 235–250; asymmetry of experiences, 51–52, 79–80, 235–236, 238; atomistic defi nitions, 228; cloaking behavior, 222–223, 226; comparisons, 234; from condoned exploitive relations to, 53–85, 141, 190, 222, 236–237; contrasting eras of race and sex oppression, 13–15; delegitimation of exploitation rhetoric (1948–1973) leading to, 62–76, 79–81, 84, 89; legal sphere, 94–97;

Index / 279

legitimating approach, 222–223, 226; mistreatment of nontargets of discrimination, 190; reactionary possibility, 96 Escalante, Jaime, 126 ETS (Educational Testing Ser vice), 118, 127 experiences: asymmetric, 21–25, 48–52, 53, 79–80, 235–236, 238; social scientists’, 50–52 experimentation, on humans, 56 exploitation: damaged social relation, 193, 195–199, 221; defi ning, 195–198, 199; economic, 55, 58, 68–69, 196–198, 237; geopolitical implications, 66–69, 72–73, 237; indirect arguments for, 59; labor force, 55–56, 58, 70–73, 76, 82, 237; Marxist, 195–199; rhetoric denounced but not practices, 58–59, 62–69, 77–78, 79, 81. See also regime of condoned exploitation factual observations, 12, 229 fairness, 101, 127–129 Faludi, Susan, 14–15, 18 family. See children; marriage; parents Feagin, Joe, 51, 211 Federal Housing Administration (FHA), 63–64 federal policies. See social policy Ferguson, Ronald F., 34 fl ight attendants, Frank v. United Airlines, 91–92 Ford auto plants, 71–72 Fordham, Signithia, 231 Frank v. United Airlines, 91–92 Freeman, Alan David, 109–110, 139–140, 155–156 Frug, Mary Joe, 111–112, 140 gender, 2n, 5–6, 256; bona fide occupational qualification (BFOQ), 89–90, 241; change/ continuity in discrimination, 14–15, 47–50; comparisons, 91–92, 116, 119, 146, 218, 230–232, 234, 243; and housework, 48, 73–74; income, 232–233, 243; maleness as property, 110–114, 132, 141, 238–239; ontology, 144; rape/assault, 41–42; reverse sex discrimination, 202; segregation, 123; socialization to, 42–47, 181–183, 253; social scientists, 246; standardized testing, 115–129, 131; state’s rigid sex hierarchy, 200. See also sex; women General Motors, 72 General Social Survey (GSS), 26, 36, 45, 251 genocide, 196–198

geopolitics, 66–69, 72–73, 237 Germany. See Nazi Germany Gilens, Martin, 26 global warming, 219–220 Goffman, Erving, 161 Goldberg, Matthew S., 157, 160 Goldin, Claudia, 73, 74 goodwill, 247–248 gradational measures, 216–217, 224 Graduate Record Exam (GRE), 116, 117–118, 130 Great Depression, 63 Great Migration, 60, 61 Greene, Brian, 17 group rights, illegitimacy of, 133–136 GSS (General Social Survey), 26, 36, 45, 251 Habermasian positions, 201 Halaby, Charles N., 205 Hamermesh, Daniel S., 212 Harris, Cheryl I., 105–106, 108–109, 111, 140 Hauser, Robert M., 219 Heckman, James J., 169 Help or Hindrance? The Economic Implications of Immigration for African Americans (Hamermesh and Bean), 212 Hemat, Lisa, 121 Higgs boson, 209 history: black-white relations, 60–62, 105–106; in critical legal theory, 104–105; Disney-esque and more complex versions, 78–81; major top-to-bottom shifts in social relations, 82–83; misrecognized, 53–54; property, 106–114; romanticized, 59. See also era of contested prejudice; regime of condoned exploitation holistic approach, 172, 199, 228, 234. See also systemic focus Home Owners’ Loan Corporation (HOLC), 63 homosexuals: discrimination targets, 189; gay marriage, 186–187, 218; German genocide and slavery co-occurring vs., 197 housework: over-reported by men, 48; significant economic production, 73–74 housing, 227; audit methodology, 168, 169–170; racialized policy, 62–64, 77, 80. See also Realtors Housing and Urban Development Act, 77 human capital theory, 97–100, 114–115; adjudicating discrimination in, 130–138; elaborating, 115–129. See also labor force; socialization hurricane season, 219–220

280 / Index

ignorance: no defense against liability, 103–104; original position of equality and ignorance, 192 illegality: discrimination, 10, 11, 54–55, 146, 171–172, 223, 226–227; hidden/cloaked/ concealed, 171–172, 222–223, 225, 226 imagination, sociolog ical, 2–5 immigration, African Americans affected by, 212 impression management, 69, 79 incoherence, 87, 176, 203; reverse discrimination, 242; reverse exploitation, 196. See also coherence standards income: black-white, 147–149, 153–156, 161, 232–233; economic discriminators foregoing, 153; gender, 232–233; gradational phenomenon, 217 individualist perspective, 6–7, 10–11, 18–20, 234, 244, 245–246; Allport’s categories of persons vs., 179, 180, 245; corporations’ rights as individuals, 134; defi ning discrimination, 13, 78–79, 172–176, 208, 242–243; Rubin’s Quandary, 230–232, 243; vs. social relational character of discrimination, 190, 219, 243; systemic focus vs., 217–221, 224–225. See also atomistic perspective inequal ity. See equality infrastructural change, racialized, 62–63, 80 isolationism, U.S., 68 Jaynes, Gerald David, 25 Jefferson, Thomas, 203 Jencks, Christopher, 211–212, 213 Jews: discrimination targets, 189; genocide and slavery co-occurring vs., 197; Holocaust, 152–153, 197–198; legality of discriminatory acts, 223 Jim Crow, 144 Johnson administration, 77 Johnson v. Board of Education of Wilson County, 108 Just Who Loses? (Lucas), 7, 248 Katznelson, Ira, 61 Kidder, William C., 117–118, 122 Kiéslowski, Krzysztof, 229 Kirschenman, Joleen, 207–208, 209, 211 knowledge. See epistemology Korean War, 67, 69–70 Korematsu v. United States, 83–84 Kuhn, Thomas S., 5

Kukich, Karen, 121 Ku Klux Klan, 56 labor force, 227; agricultural, 76–77, 92; audit methodology, 168–170; authorities women may encounter, 39–42, 251–256; black/ white capital, 76–77, 147–149, 160–161; damaged social relation, 193–194; discrimination litigation, 91–92, 95–96, 100, 171; disparate impact discrimination, 89–90; disparate treatment discrimination, 89, 90; employer discrimination, 92, 147–161, 168–169, 171, 207–208, 240; existence of discrimination, 207–208; exploitive, 55–56, 58, 70–73, 76, 82, 237; fi rm-specific research vs. large-scale surveys, 218–221; human capital investment, 98–99; income and race, 92, 147–148, 153–156, 161, 232–233; lower rungs, 55–56; manufacturing, 60–62, 64–65, 70–73, 82; slave, 89, 105–106, 196–198, 212; and standardized testing, 130; unions, 70, 71–72, 73; white authorities blacks may encounter, 28–35, 36, 54, 251–256; World War II, 58, 60, 70–73, 82. See also blacks in labor force; women in labor force language, for articulating grievances, 190–191 Latino/as: employers preferring over blacks, 207–208; minority category, 110 laws: Civil Rights Act, 75, 110; Constitution (U.S.), 89; Housing and Urban Development Act, 77; Nuremberg, 87, 88; Selective Training and Ser vice Act, 66; Social Security Act, 76–77. See also legal system Law School Admission Test (LSAT), 130, 165 Lee, Valerie E., 181, 211 legal system, 10, 86–142; anecdotal evidence, 91; comparative evidence, 91–92; defendants, 92, 135–136; defi ning discrimination, 10–11, 22, 86–90, 99–102, 139, 238; logics, 86; measur ing expected intensity of exposure to discrimination, 224–225; normalcy in, 140; not neutral, 85; perpetrator identification, 137; plaintiffs, 92, 136; property rights, 105–114, 132, 140, 141, 212, 238–239; punitive imperative, 93, 94, 97, 103, 133–134, 137; remedy, 137; restitution payments, 93; social science perspectives and, 10–11, 104, 140–145, 149, 155–156, 171–172, 205, 239; statistical evidence, 91. See also court cases; critical legal perspectives; dominant legal reasoning; illegality; laws; rights

Index / 281

legitimating approach, 222–223, 226 Leonard, Jonathan S., 171 Lloyd, Cynthia B., 74 logics, legal, 86 Loury, Glenn C., 161–162, 163 LSAT, 130, 165 Lu, Yu-Hsia, 37–38 Ludwig, Jens, 231 MacGregor, Morris J., 66 maleness: as property, 110–114, 132, 141, 238–239. See also gender manufacturing jobs, 60–62, 64–65, 70–73, 82 Marks, Helen M., 181, 211 marriage: civil unions, 186–187; damaged social relation, 194–195, 221; gay, 186–187, 218; relational phenomenon, 217; salubrious/undamaged social relation, 185–187, 203; as social relation, 185–187, 194–195, 203, 217, 218, 221; women in labor force and, 36–41, 54, 70, 74–75; women’s property rights, 111–112, 212; working women barred from, 74–75 Marxism, 160–161, 162, 191–200, 233, 242; neo-Marxist, 216–217 Mason, Karen Oppenheim, 37–38 Massey, Douglas S., 63–64 McPartland, James, 207, 208, 209, 219 McWhorter, John H., 13–14, 18, 50 measurement, discrimination, 170–171, 206, 216–225, 234, 243; atomistic perspective, 163–164; damages, 221–225; expected intensity of exposure, 224–225, 245; gradational measures, 216, 224; laws used for, 224–225; legitimating projects and activities of the superordinate, 222; level of analysis, 206, 217–221, 224–225, 226, 243; morphology of, 206, 216–217; recorded behaviors used for, 224; relational measures, 216–217, 223, 224; witnesses, 222, 223, 226 men: maleness as property, 110–114, 132, 141, 238–239. See also gender; marriage Mickelson, Roslyn A., 205 Middle Passage, 198 migration, black history in U.S., 60, 61 military: blacks, 65–68, 69–70; bona fide occupational qualification (BFOQ), 241; women, 70–72, 241 Milliken v. Bradley, 94, 109 Mills, C. Wright, 2 “minorities,” 110, 189 misrecognition, history, 53–54 Missouri v. Jenkins, 95

motivation: discriminators’, 148, 163, 164, 167, 168, 171–172, 234, 248; “lacks willpower” explanation for black socioeconomic disadvantage, 26–27, 32–35, 251–252; risk-taking, 192; social scientists’, 184 Myrdal, Gunnar, 18, 144, 246 NAEP (National Assessment of Educational Progress), 124–125 Nalty, Bernard C., 66 National Academy of Sciences, 25 National American Women Suffrage Association convention, 57 National Assessment of Educational Progress (NAEP), 124–125 National Centers for Disease Control (CDC), 41 National Institute of Justice (NIJ), 41 National Opinion Research Center (NORC), 26, 36 Native Americans: genocide, 196, 197, 198; legalized discrimination vs., 89, 105–106, 108; property dispossession, 105–106, 107–108; treaties with U.S. recognizing sovereignty of, 134 The Nature of Prejudice. See Allport, Gordon Nazi Germany: disposability of disenfranchised persons, 56; genocide and slavery co-occurring by, 197; Holocaust, 152–153, 197–198; legality of discrimination, 223; logic of, 200; Nuremberg laws, 87, 88 Neckerman, Kathryn M., 207–208, 209, 211 The Negro Motorist Green Book, 65 neo-Marxist analysts, 216–217 neo-Weberian analysts, 216–217 nepotism, employer, 153–154, 157 Niemi, Beth T., 74 Nixon, Richard, 77 norms: normative interest of statistical analysis, 254–256; social capital, 161; social relations, 175, 180, 183–190, 202–203, 221–223, 234, 241, 242, 250; social science, 19, 184, 219; standardized testing, 122–123, 124, 222 Nuremberg laws, 87, 88 Oaxaca decompositions, 165 O’Connor, Sandra Day, 215 Ogbu, John U., 231 Oliver, Melvin L., 129 operationalization, 224 paradigm shift, 16–18, 24, 50–51, 247–248. See also reconceptualization

282 / Index

parents, 251; abusive, 187; disparate impact discrimination, 99–100; gender socialization by, 42–47, 54, 182, 253; reproductive rights, 112–114; sperm donors, 112, 113–114 Phillips, Meredith, 211–212, 213 physics, 16–17, 209–210 plaintiffs, 92, 136 Plessy v. Ferguson, 69, 109 police officers: authorities women may encounter, 39–42, 251–252, 253; white authorities blacks may encounter, 28, 29–30, 32–33, 251–252 policy. See social policy power: conservation of distribution of, 132, 141, 237; discrimination determined by, 189, 196; equality dissolving role of, 192; exploitive, 196–198; standardized testing reproducing existing patterns of, 131–132 Press, Julie, 48 property, 140; defi ning, 106; history, 106–114; maleness as, 110–114, 132, 141, 238–239; of Native Americans, 105–106, 107–108; slaves as, 105–106; whiteness as, 107–110, 111, 114, 132, 141, 238–239; women’s rights to, 111–112, 212 Przypadek (Kiéslowski), 229 punitive imperative, 93, 94, 97, 103, 133–134, 137 Pushkareva, Natalia, 72 Putnam, Robert, 20 Quadagno, Jill, 76–77 quantum mechanics, 16–17 race, 5–6; change/continuity in discrimination, 13–14, 47, 48–50, 237; comparisons, 119, 230–232, 234, 243; employers ranking by, 207–208; not regarded as class-creating or class-maintaining, 135; ontology, 144; reverse race discrimination, 202; of social scientists, 246; standardized testing and, 115–129, 211–212, 213; stigma, 161–162. See also Asians; blacks; Latino/as; Native Americans; whites Randolph, A. Phillip, 67, 68 random assignment, 12–13, 230 Rawls, John, 191, 192–193, 242 Realtors, 227; audit methodology, 168, 169–170; federal housing policies and, 77; white authorities blacks may encounter, 28–33, 36, 251–252 reconceptualization: of discrimination, 4–5,

18, 21, 50, 146–147, 176. See also paradigm shift Reconstruction (1865–1877), 56, 76 redemption, human capital theory, 97–100 redlining, 63–64 regime of condoned exploitation, 55–59, 81, 82, 83, 141, 250; atomistic or social relational defi nitions, 228; delegitimation of racial exploitation rhetoric (1948–1970), 62–69, 79, 84; delegitimation of sex exploitation rhetoric (1948–1973), 69–76, 79, 84; to era of contested prejudice, 53–85, 141, 190, 222, 236–237; limitations of delegitimation of, 76–78; restoration of, 96 relational measures, 216–217, 223, 224 relativity theory, 16–17 remedy, legal, 137 reproductive rights, 112–114 research: aim, 206–210; best, 164; caused-based, 210–216, 223; discrimination effects, 18, 21, 144–174, 206–234, 246–248; effects-based, 148, 206–216, 234; empirical, 6, 7, 163, 201, 244, 245–246, 250; external challenge, 3–5; focus of inquiry, 206, 210–216; goodwill, 247–248; pattern-searching, 208; phenomenon-based, 148, 206–216, 234; random assignment, 12–13, 230; source of information, 206; working defi nition for, 176, 179–182, 199, 201–204. See also measurement; survey research residual attribution, 163–167, 170, 215 reverse discrimination, 188–190, 196, 201–202, 242 rhetoric: delegitimation of racial exploitation rhetoric (1948–1970), 62–69, 79, 84; delegitimation of sex exploitation rhetoric (1948–1973), 69–76, 79, 84; rhetoric denounced but not practices, 58–59, 62–69, 77–78, 79, 81 rights, 13, 83; of corporations, 134; illegitimacy of group rights, 133–136; reproductive, 112–114. See also black-white relations; discrimination; legal system; property; women’s rights risk, damaged social relations and, 192–193 road construction, racialized, 62–63, 80 Roberts v. City of Boston, 68 Roemer, John E., 195–196, 198, 199 Roma, German genocide and slavery co-occurring vs., 197 romanticization, of history, 59 Roosevelt, Franklin D., 76

Index / 283

Rosner, Jay, 117–118, 122 Rubin’s Model, 11–13, 172–173, 225, 228–232, 243; control condition, 229; counterfactual observations, 12, 229, 231–232; factual observations, 12, 229; temporal invariance, 230; unit invariance, 229, 230. See also causal inference Rubin’s Quandary, 230–232, 243 SAT, 116–120, 130 savings and loan (S&L) bailout (1989), 137–138 Scalia, Antonin, 95–96 Schuman, Howard, 25, 26–27, 32 Sears, David O., 47 segregation: defi ned by Allport, 177; by gender, 123, 181; by race, 63–69, 95 selection bias, audit methodology and, 170 Selective Training and Ser vice Act, 66 separate-but-equal doctrine, 65, 68, 69, 94 Sewell, William, 219 sex, 2n; rape/assault, 41–42. See also gender; homosexuals; marriage sexual harassment, 95–97, 100, 101 Shapiro, Thomas M., 129 sibships, defi ned, 45 Siegelman, Peter, 169 simulation, standardized testing, 119–122, 257–258 Singer, Joseph William, 107–108, 140 Skrentny, John D., 110 slavery, 89, 105–106, 196–198, 212 Sliding Doors, 12 Smith, Adam, 201 smoothing statistical data, 27, 28; lowess, 251–252 socialization: gender-specific, 42–47, 181–183, 253. See also teachers social policy, 2, 7, 79–80, 227–228; affi rmative action, 109, 110, 215; geopolitical implications, 66–69, 72–73, 237; housing, 62–64, 77, 80; impression management, 69, 79; racialized, 26, 72–73; social security, 76–77; treatment of women, 72–73. See also laws social relation, 20–22, 172–173; atomistic perspective and, 227–228; civil unions, 186–187; as classes of persons, 175, 179, 180, 188, 217, 223, 224–225; cohabiter, 185–186; consensus-generating, 192–193; discrimination as, 19–21, 180–184, 187–234, 244; historical major top-to-bottom shifts, 82–83; relational measures, 216–217, 223, 224; salubrious/undamaged, 179, 183–188,

191–201, 203, 242; spousal, 185–187, 194–195, 203, 217, 218, 221. See also black-white relations; damaged social relation; power; women’s rights social scientists, 138–139, 141–142; asocial scientific method, 131, 132, 140–141, 143–174, 239–240; coherence standards, 97, 101; and damaged social relations, 183–184, 205–206; defi ning discrimination, 10–11, 19–21, 78, 101, 139, 239–240; estimating discrimination effects, 146, 171–173, 205–234, 244; experiences, 50–52; goodwill, 247; legal perspectives and, 10–11, 104, 140–145, 149, 155–156, 171–172, 205, 239; Marxism taken seriously by, 193; norms, 19, 184, 219; residual attribution strategy, 166–167, 170, 215; value communities, 184. See also Becker, Gary; dominant social science approach; research; Rubin’s Model Social Security Act, 76–77 society: critical legal theory on structure of, 104–105; variable distribution of goods, 145; web of connections, 150–152. See also social relation; socioeconomic status socioeconomic status, 245; Asian successes, 215; black disadvantage, 26–36, 212, 251–252; “lacks willpower” explanation, 26–27, 32–35; “low ability” explanation, 26–32, 251; neo-Weberian and neo-Marxist analysts, 216–217; nondiscriminatory disadvantage, 182; white attributions concerning blacks, 26–36. See also class; housing; income sociolog ical imagination, 2–5 sociology of knowledge. See epistemology Sommers, Christina Hoff, 14–15, 18 South Africa, 153, 196–197 Soviet Union, 66–67, 68, 72, 237 sperm donors, 112, 113–114 spousal relation. See marriage standardized testing, 115–132; ACT, 130; adaptive, 128–129; bona fide occupational qualification (BFOQ), 125; candidate questions (CQs), 116–117, 119–122; for college and postcollege admissions, 115–132; competence indicators, 118, 124–125; construction of tests, 116–134; criterion-referenced, 124–125; critical legal perspective, 130–133, 136–138; differential item functioning (DIF), 125, 127–128, 130–131; dominant legal perspective, 130–136; ETS (Educational Testing

284 / Index

standardized testing, (continued) Ser vice), 118, 127; fairness, 127–129; GMAT, 130; Graduate Record Exam (GRE), 116, 117–118, 130; legitimation crisis, 125–127; LSAT, 130, 165; MCAT, 130; meritocratic, 126; NAEP, 124–125; norms, 122–123, 124, 222; point biserial analyses, 125, 127, 130–131, 257–258; predictive validity, 126–127; race and, 115–129, 211–212, 213; reproduction of existing patterns of power, 131–132; SAT, 116–120, 130; simulation, 119–122, 257–258; technological innovation, 128 statistical data: assumptions, 29–30, 254–256; legal evidence, 91; smoothing, 27, 28, 251–252. See also survey research statistical discrimination, 158–160, 241 stigma, racial, 161–162 suburbs, housing policies, 62–64, 77, 80 survey research: General Social Survey (GSS), 26, 36, 45, 251; limitations, 25, 47–48; Survey of Income and Program Participation (SIPP), 16; on workers, 218–221 Swidler, Ann, 80 syphilis, Tuskegee Syphilis Experiment, 56 systemic focus, 217–221, 224–225. See also classes of persons; holistic approach taken-for-granted assumptions, 114, 181, 189, 222; dominant legal perspective, 104–105, 137, 139, 217–218, 239, 250; standardized testing, 127, 131, 135 targets: of discrimination, 189–190, 192, 217–218, 222, 238; discrimination effects on nontargets, 153, 164, 190, 215, 217–218, 233, 240, 243, 248. See also bystanders; gender; race taste theories of discrimination, 145–174, 241; neo-taste theory, 157–158. See also Becker, Gary teachers: authorities women and girls may encounter, 39–40, 42, 251–256; disparate impact discrimination, 99; mostly female, 42–43; white authorities blacks may encounter, 28–35, 54, 251–256 Terborg-Penn, Rosalyn, 57 terror, war on, 83–84 tests. See standardized testing Thomas, Clarence, 95–96 Thorne, Barrie, 123 three-fi fths compromise, 57–58, 82 Thurow, Lester C., 161, 163 Tise, Larry E., 58

Townsley, Eleanor, 48 The Traveler’s Guide: Hotels, Apartments, Rooms, Meals, Garage Accommodations, Etc., for Colored Travelers, 65 Truman, Harry S., 67 Truth, Sojourner, 57 Tsai, Shu-Ling, 219 Tuskegee Syphilis Experiment, 56 Tyson, Karolyn, 231 unemployment rate, black, 56 unions, women’s experiences with, 70, 71–72, 73 United Airlines, 91–92 United Auto Workers (UAW), 71–72 United States Public Health Ser vice, 56 University of California Regents v. Bakke, 94–95 USSR, 66–67, 68, 72, 237 utilitarianism, 191, 203, 242 “validity coefficients,” 6 value communities, social scientists as, 184 Vera, Hernán, 211 Veterans Administration (VA) loans, 63 Walters, Pamela Barn house, 36 wars: racialized military, 65–68, 69–70; on terror, 83–84; women in military, 70–72. See also World War II Watt, Bob, 96 welfare, racialized opposition to, 26 West, Candace, 51 whites: income, 232–233, 243; reverse race discrimination, 202; whiteness as property, 107–110, 111, 114, 132, 141, 238–239. See also black-white relations Wilkie, Jane Riblett, 36–37 Williams, Robin M., Jr., 25 Wilson, William Julius, 60–62, 81, 82, 84 witnesses, to damaged social relations, 222, 223, 226 women, 256; abused in marriage, 212; fi rst oppressed class, 194; income, 232–233, 243; military, 70–72, 241. See also women’s roles women in labor force, 251, 253; atomistic perspective, 240; audit methodology on job interviews, 168–169; authorities with limiting views of, 38–47; bona fide occupational qualification (BFOQ), 89–90, 241; discrimination litigation, 91–92, 95–96, 171; exploited, 55–56, 58, 70–73, 82, 237; manufacturing, 60–62, 64–65, 70–73,

Index / 285

82; married, 36–41, 54, 70, 74–75; occupational sex-segregation, 181; teachers, 42–43; unions, 70, 71–72, 73; World War II, 58, 70–73, 82 Women’s Bureau, 71 women’s rights: black-white relations and, 57, 71, 110–111; Civil Rights Act and, 75, 110; discrimination litigation, 91–92, 95–96, 100; disparate impact discrimination, 89–90; law negligent with, 141; legalized discrimination vs., 89; maleness as property and, 110–114, 132, 141, 238–239; National American Women Suffrage Association convention, 57; postsuffrage 1920s, 56; to property, 111–112, 212; reproductive, 112–114; residual attribution and, 165, 166, 167; vs. sexual harassment, 95–97, 100, 101; social science research, 144; submerged implications of judgments of, 36–47; women’s attitudes toward, 37–38, 231. See

also women in labor force; women’s roles women’s roles: authorities women and girls may encounter, 38–47, 48, 53, 54, 236, 251–255; delegitimation of sex exploitation rhetoric (1948–1973), 69–76, 79, 84; exploitive, 55–58, 237; housework, 48, 73–74; socialization to, 42–47, 181–183, 253. See also sex; women in labor force; women’s rights World War I, 67–68 World War II: Japa nese ancestry, 83–84; manufacturing jobs for blacks, 60; women in labor force, 58, 70–73, 82. See also Nazi Germany Wright, Erik Olin, 55, 196–199 Yinger, John, 168 Zieky, Michael, 127–128 Zimmerman, Don H., 51

Samuel Roundfield Lucas is Associate Professor of Sociology at the University of California-Berkeley. He is the author of Tracking Inequality: Stratification and Mobility in American High Schools and a co-author of Inequality by Design: Cracking the Bell Curve Myth.