Contested Welfare States: Welfare Attitudes in Europe and Beyond 9780804783170

This book explores citizen's attitudes toward welfare policies across European countries and offers critical insigh

184 61 2MB

English Pages 272 [271] Year 2012

Report DMCA / Copyright

DOWNLOAD FILE

Polecaj historie

Contested Welfare States: Welfare Attitudes in Europe and Beyond
 9780804783170

Citation preview

con t est ed w el fa r e stat es

studies

in social

inequality

edi tors David B. Grusky  s t a n f o r d u n i v e r s i t y Paula England  n e w y o r k u n i v e r s i t y edi tor i a l boa r d Hans-Peter Blossfeld Mary C. Brinton Thomas DiPrete Michael Hout Andrew Walder Mary Waters

c on t e s t e d w e l fa r e s t a t e s We l fa re A t t i t u d e s i n E u r o p e a n d B e y o n d

Edited by Stefan Svallfors

sta nford univ ersit y pr ess sta nfor d, califor ni a

Stanford University Press Stanford, California © 2012 by the Board of Trustees of the Leland Stanford Junior University. All rights reserved. No part of this book may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying and recording, or in any information storage or retrieval system without the prior written permission of Stanford University Press. Printed in the United States of America on acid-free, archival-quality paper Library of Congress Cataloging-in-Publication Data   Contested welfare states : welfare attitudes in Europe and beyond / edited by Stefan Svallfors.    pages cm   Includes bibliographical references and index.   ISBN 978-0-8047-8252-4 (cloth : alk. paper)   1.  Public welfare—Europe—Public opinion.  2.  Welfare state—Europe— Public opinion.  3.  Europe—Social policy—Public opinion.  4.  Public opinion—Europe.  I.  Svallfors, Stefan, editor of compilation.   HV238.C656 2012   361.94—dc23 2012009549 Typeset by Newgen in 10/14 Sabon

In memoriam Roger Jowell (1942–2011)

con t en ts

Tables and Figures  xi Acknowledgments  xv Contributors  xvii ch apter one

Welfare States and Welfare Attitudes  1 Stefan Svallfors ch a p t er t wo

Welfare Performance and Welfare Support  25 Wim van Oorschot and Bart Meuleman chapter three

Attitudes Among High-Risk Groups  58 Helena Blomberg, Johanna Kallio, Olli Kangas, Christian Kroll, and Mikko Niemelä ch a p t er fou r

A Normative Approach to Welfare Attitudes  81 Christian Staerklé, Tiina Likki, and Régis Scheidegger chapter five

Unraveling Working-Class Welfare Chauvinism  119 Jan Mewes and Steffen Mau

ix

x

Contents

ch ap ter six

Age, Class, and Attitudes Toward Government Responsibilities  158 Stefan Svallfors, Joakim Kulin, and Annette Schnabel chapter seven

Framing Theory, Welfare Attitudes, and the United States Case  193 Clem Brooks ch a p t er eigh t

Welfare Attitudes in Context  222 Stefan Svallfors Index  241

ta bl es a n d figu r es

Tables 1.1. Welfare index 0–10, by country  14 2.1. Question wordings for welfare support and perceived performance items  31 2.2. Results of multilevel analyses for support for government intervention in favor of the elderly (N = 38,478) and the unemployed (N = 37,963)  39 2.3. Results of multilevel analyses for perceived standard of living of the elderly (N = 38,478) and the unemployed (N = 37,963)  43 3.1. Perceived future risk by high-risk group (percentage)  68 3.2. Evaluation of country’s social benefits and services by high-risk group (percentage)  69 3.3. Welfare attitudes toward the responsibilities of government by high-risk group, mean values  71 3.4. Path-analysis for high-risk groups in different welfare regimes  73 4.1. Model of lay conceptions of social order  87 4.2. Five measures of conceptions of social order, with descriptive statistics for overall sample  95 4.3. Multilevel analysis: Level 1 main effects of social positions and conceptions on welfare attitudes  100 4.4. Multilevel analysis: Cross-level interactions between conceptions and level 2 indicators on welfare attitudes  106 5.1. Support for the inclusion of immigrants: Country-specific response patterns (percentage of respondents within countries)  137

xi

xii

Tables and Figures

5.2. Determinants of welfare chauvinism (multinomial logistic multilevel models)  140 5.3. Macro-level determinants of supporting welfare chauvinism  143 5.4. Predicted mean probabilities to support welfare chauvinism  147 5.5. Predicted mean probabilities for a range of combinations with regard to key covariates  147 6.1. Class, age, gender, and attitudes toward income redistribution for Finland, 2008  171 6.2. Attitudes toward welfare policies in 26 European countries  172 6.3. Class and age differences in attitudes toward income redistribution for Denmark, Sweden, Poland, and Slovakia, 2002−2008 (standard deviations of coefficients (K))  179 7.1. Experiments with target groups: Social benefits and services  204 7.2. Experiments with crisis/class frames: [Not] government’s responsibility  206 Figures 1.1. A conceptual framework for analyzing attitudes toward welfare policies  21 2.1. Aggregate relationship between perceived standard of living (x-axis) and role of government (y-axis) for the elderly  35 2.2. Aggregate relationship between perceived standard of living (x-axis) and role of government (y-axis) for the unemployed  37 2.3. Country-specific effects of perceived standard of living of pensioners on welfare support in favor of the elderly, with 95% confidence intervals  47 2.4. Country-specific effects of perceived standard of living of the unemployed on welfare support in favor of the unemployed, with 95% confidence intervals  48 3.1. Conceptual model of the study  60 3.2. The studied paths in different welfare regimes compared to the rest of the population  72 4.1. Mediation model of welfare attitudes  93



Tables and Figures

xiii

4.2. Normative beliefs mediate the link between social position and welfare attitudes  101 4.3. Predicting government responsibility as a function of conceptions of social order across five welfare regimes  104 4.4. Cross-level interactions on government responsibility  107 6.1. Attitudes toward income redistribution policies  171 6.2. Income inequality and attitudes toward income redistribution policies  177 6.3. Old-age dependency ratio and attitudes toward labor market policies  178 7.1. Negative perceptions of strain, laziness, and poverty  208 7.2. Negative perceptions of migration and costs  209 7.3. Not government’s responsibility: Jobs, health care, and the old  210 7.4. Not government’s responsibility: Unemployed, child care, and family leave  211 For additional data, please see the Web Appendix tables and figures by following the link in the listing for Contested Welfare States on our website: http://www.sup.org.

ack now l edgm e n ts

The research on which this book is based has received support from numerous sources and individuals. It was conducted in the research project “Welfare Attitudes in a Changing Europe,” organized by the European Science Foundation under their program “Human Values, Institutions and Behaviour” (HumVIB). This included financial support from a large number of national research councils: The Swedish Research Council for Social and Working Life Research (FAS 2007-2288); the Dutch Research Council (NWO 460-08-161); the Academy of Finland, (129153); the Swiss National Science Foundation (105513-121166); the German Research Foundation (MA 3282/3-1); and the U.S. National Science Foundation (SES‐ 0830917). Our project was also supported by the Rockefeller Foundation, which provided space, meals, and accommodations at their wonderful Bellagio Centre for the conference “Welfare Attitudes in a Changing Europe,” which was held on November 23–26, 2010. At this conference, our chapter drafts were thoroughly vetted by three excellent commentators: Staffan Kumlin, Edeltraud Roller, and Michael Shalev. Our manuscript benefited immensely from their constructive criticisms, and we very much enjoyed our discussions. The “tired, self-destructive tone” that Staffan spotted (and lauded) in one of the chapters was certainly not something that characterized their comments! Annica Brännlund, Christin Muth, and Julia Schnakenberg supplied practical assistance in various stages of the book project. Umeå University supported the editor through their Baltic scholarship, allowing him to concentrate on full-time research for most of the project’s time.

xv

xvi

Acknowledgments

Finally, the Department of Political and Social Sciences at the European University Institute provided the editor with valuable support for finalizing the manuscript through one of their Fernand Braudel Fellowships. Our saddest moment in this project came in the final manuscript stage. Roger Jowell—founder of the European Social Survey and before that the International Social Survey Program and the British Social Attitudes ­series—passed away on Christmas Day 2011. His importance for our project was huge, from making the European Social Survey possible in the first place to providing hands-on advice in the different stages of designing the Welfare Attitudes module. We will always remember his extraordinary craftsmanship, great spirit, and good humor. To his memory—one of true intellectual leadership and great companionship—we dedicate this volume. Stefan Svallfors Umeå, January 2012

con t r ibu tors

Stefan Svallfors is a professor in sociology and Baltic Scholar at Umeå University. His current research mainly concerns attitudes to welfare policies and redistribution in Western countries. He is involved in several major attitude research programs: the International Social Survey Program, the European Social Survey, and the Swedish Welfare State Surveys. Among his publications, we find three previous volumes at Stanford University Press: Analyzing Inequality (2005), The Moral Economy of Class (2006), and The Political Sociology of the Welfare State (2007). Helena Blomberg is a professor of social policy at the Swedish School of Social Science, University of Helsinki. She is chairing the theme group The Value Systems and Legitimacy in the Nordic Welfare State Model, within the Nordic Centre of Excellence: The Nordic Welfare State—Historical Foundations and Future Challenges. Her major fields of research concern popular and street-level bureaucrats’ attitudes toward welfare policies in a comparative perspective and policy changes in the welfare state. Clem Brooks is James H. Rudy Professor of Sociology at Indiana University Bloomington. His research is on voter choice, attitude formation, and comparative public policy. Current projects include a study of the U.S. public’s responses to the 2008 economic crisis and survey experiments into attitude formation on counterterrorism policy and health policy. Among Brooks’s publications are Social Cleavages and Political Change (1999), Why Welfare States Persist (2007), and a new manuscript, Whose Rights? (2012). Johanna Kallio is a postdoctoral researcher in social policy and social work at the Swedish School of Social Science, University of Helsinki. Her main research areas are social policy attitudes and social perceptions among citizens, municipal decision makers, and street-level bureaucrats. Olli Kangas is a research professor at the Social Insurance Institution of Finland. He specializes in comparative welfare studies, including analyses of the xvii

xviii

Contributors

i­nstitutional setups of welfare programs, the political and structural factors behind the development of welfare states, and the consequences of social policy in terms of income distribution, poverty, legitimacy, and collective action. His papers have appeared in many leading social policy journals. Christian Kroll is a senior lecturer in social policy at the Swedish School of Social Science, University of Helsinki. His current research focuses on the legitimization and legitimacy of welfare state reforms and on attitude formation among high-risk groups and welfare state professionals in comparative perspective. Joakim Kulin is a postdoctoral researcher at the Department of Sociology at Umeå University. In his Ph.D. thesis “Values and Welfare State Attitudes,” he analyzes the relationship between human values and attitudes toward welfare state redistributive policies and how macro-level factors moderate this relationship across national contexts. Tiina Likki is a Ph.D. candidate in social psychology at the University of Lausanne. Her research interests lie in political attitudes, redistribution, and power relations between dominant and subordinate groups. Her earlier research at the University of Helsinki focused on adaptation of immigrants. Steffen Mau is a professor of political sociology and comparative social research at the University of Bremen. His main research areas include comparative welfare state research, social inequality, European integration, and transnationalism. He is the author of The Moral Economy of Welfare States—Britain and Germany Compared (2003), Social Transnationalism—Lifeworlds Beyond the Nation-State (2010), and European Societies—Mapping Structure and Change (2010; with Roland Verwiebe). Bart Meuleman is an assistant professor at the Centre for Sociological Research, University of Leuven. His research mainly focuses on cross-national comparisons of attitude and value patterns, such as welfare state legitimacy, anti-immigrant sentiment, religious involvement, and basic human values. He has a special interest in research methodology for comparative research and comparability of survey measurements. Jan Mewes is currently COFAS Marie Curie Fellow at Umeå University. From 2005 to 2011, he was researcher with the Chair of Political Sociology and Comparative Analysis of Contemporary Societies at the University of Bremen. His main research interests include migration, social stratification, public attitudes, social capital, and comparative welfare state research. He is the author of Ungleiche Netzwerke-Vernetzte Ungleichheit (2010).



Contributors

xix

Mikko Niemelä is a senior researcher at the Social Insurance Institution of Finland and a professor in social and health policy at the University of Tampere. His current research focuses on mechanisms of institutional change of welfare states, public perceptions of the causes of poverty, and cross-national differences in public attitudes toward the role of government. Wim van Oorschot is a professor of sociology at Tilburg University. He is an honorary president of the Network for European Social Policy Analysis, initiator and director of the European Data Centre for Work and Welfare, and Honorary Professor at the Centre for Comparative Welfare Research of Aalborg University. His research and teaching regard the sociological and cultural analysis of welfare states and social policy. Previous books include Culture and Welfare State: Values and Social Policies in Comparative Perspective (Edward Elgar) and Disability Policies in European Countries (Kluwer). Régis Scheidegger is a doctor in social sciences from Lausanne University. His current research concerns political attitudes, ideologies, and social representations as a function of academic socialization. He is involved in research on the link between subjective political competence and social representations of the social order. Annette Schnabel is an associate professor in sociology at Umeå University. Her research interests cover the questions of why people join social movements; how they form their national, religious, and gender identities; and how emotions and rational choices interact. The impact of political and social institutions and their formation form another focal point of interest. The research on these issues encompasses empirical analyses, as well as their impact on social theory. Recent books include Emergenz (with Jens Greve; Frankfurt a.M.: Suhrkamp, 2010). Christian Staerklé is an associate professor of social psychology at the University of Lausanne and a docent of social psychology at the University of Helsinki. His research is based on both survey and experimental methods and concerns political attitudes, intergroup relationships, and issues of social justice from the perspective of social representations theory. His papers have appeared in major social psychology journals.

con t est ed w el fa r e stat es

chapter one

Welfare States and Welfare Attitudes Stefan Svallfors

The welfare state may in many ways be seen as a particular trademark of the European social model. An extensive set of social and institutional actors provide protection against negative consequences of common life-course risks—for example, by offering economic support in periods of hardship or by ensuring access to care and services. Based on collective responsibility and financing, welfare policies define a set of social rights, meet common vulnerabilities, and address needs for protection from market uncertainties. However, over recent decades, European welfare states have undergone profound restructuring and recalibration. This is a result of economic and political pressures and of adaptive processes to new contingencies due to demographic changes, international migration and economic competition, and persistent unemployment. New forms of risk, taking different shapes across welfare states, have grown out of precarious and insecure life courses. Rising concerns about welfare state sustainability and the slowly growing “Europeanization” of welfare policies are also common challenges across the continent. Increased provider pluralism and new forms of public management, as well as new forms of policies and state intervention, in particular regarding activation policies, work–family reconciliation, and gender policies, are attempts to adjust to these challenges. In parallel with institutional and structural developments, collective beliefs and representations about welfare and justice have also evolved, sometimes reflecting exclusion and perceived lack of deservingness of groups of beneficiaries—in particular, members of minority groups. In all of these challenges and processes of change, the attitudes and orientations of the public are important, although often neglected, factors that must be taken 1

2

Svallfors

into account. Established viewpoints, normative expectations, concepts of justice, and similar perceptions are often very hard to change, and in this way, attitudes often function as a counterweight to abrupt policy changes. Policy reformers need to deal with normative orientations and expectations that have been established by previous politics and policies, and this often hinders or derails policy changes. Conversely, existing attitudes may be a resource and part of the opportunity structure for actors bent on challenging the institutionalized status quo and effectuating political change. Attitudes toward the welfare state and other public institutions should be seen as central components of social order, governance, and legitimacy of modern societies. They tell us something about whether or not existing social arrangements are legitimate. Are they accepted only because people see no alternatives or think that action is futile, or are they normatively grounded? Are institutions considered to be fundamentally just or not? And research into the antecedents and consequences of these attitudes also asks us to judge public policies not only by their distributive effects or by their economic efficiency but by their normative effects on mass publics. In this book, we analyze and report results from a comprehensive research program on citizens’ attitudes toward welfare policies across European countries. We also offer a novel comparison with the case of the United States, putting into further perspective the potential regional distinctiveness of the European context as a whole. We are mainly occupied with normative orientations toward the (re)distribution of resources and life chances and toward public policies aimed at ameliorating adverse conditions. We summarize these under the heading “welfare attitudes.” Our research focuses on the interplay among individual welfare attitudes and behaviors, institutional contexts, and structural variables. We hope to provide essential input to the interdisciplinary field of comparative studies of welfare state attitudes and to offer critical insights into the public legitimacy of welfare state reform. Along with the societal and political context, the dominant welfare state theories also have changed considerably over time. The 1970s and 1980s saw a plethora of treatises on “welfare state crisis” (O’Connor, 1974; Offe, 1984), “legitimation crisis” (Habermas, 1975), and “government overload” (Crozier et al., 1975). All of these arguments suggested that a highly problematic relationship existed between the workings of government and the legitimacy of the state. Either citizens asked for too much government, in a



Welfare States and Welfare Attitudes

3

vicious spiral of insatiable demands, or for less government, as expressed in increasing alienation from government and its bureaucratic intrusions into everyday life. As discussed in the next section, many of these arguments seem—with the benefit of hindsight—widely overblown. Following the publication of the landmark study “Three Worlds of Welfare Capitalism” (Esping-Andersen, 1990) and other publications from the “power resource” school of welfare state research (for example, Korpi, 1989; Huber and Stephens, 2001a; Korpi and Palme, 2003), the 1990s and early 2000s were much dominated by the debate on welfare regimes. What were the antecedents and effects of different institutional configurations of the welfare state (Korpi and Palme, 1998)? Did welfare states in fact come in “worlds” or “families of nations” (Castles and Mitchell, 1992; Castles, 1993; Scruggs and Allan, 2006; 2008)? Here a much more harmonious relationship between welfare states and the attitudes among their citizens transpired. As suggested by the political labeling of Esping-Andersen’s three welfare regimes (liberal, conservative, and social democratic), welfare regimes were seen as rooted in distinct political values and ideologies. Furthermore, existing welfare regimes tended to mold welfare attitudes among their citizens in distinctive ways, making for some degree of congruence between welfare state institutions and opinions among mass publics (Svallfors, 1997; 2003). In the last few years, concerns have been raised that the increasing ethnic and cultural diversity of Europe constitutes a threat against welfare state legitimacy (Alesina and Glaeser, 2004; Banting and Kymlicka, 2006; Boeri et al., 2002). National welfare states were predicated on maintaining clear borders between the population that should be covered by welfare state protection (and asked to finance it) and those who were outsiders. With increasing international migration, both within Europe and from elsewhere, borders have become much more permeable, and the distinction between those who “belong” and those who do not is now less clear-cut. The ethnic fragmentation of the United States and its impediments for the establishment of a universal welfare state have been held up as the new problematic future for European welfare states (Alesina and Glaeser, 2004). In this volume, we pick up elements from all three strands of research, but we apply them within our comparative framework, focusing on ­welfare attitudes as key lynchpins of institutional legitimacy. We conduct our analyses in a context of “permanent austerity” (Pierson, 2001), in which the

4

Svallfors

demographic situation of European countries particularly makes welfare state sustainability a problematic issue. Even if the “crisis and overload” literature was not written in the same climate, many of the themes about demands on and evaluations of welfare policies are relevant also today. Furthermore, we rely to some extent on the literature about welfare regimes for framing our own inquiries. But we are less concerned with comparing levels or group patterns of attitudes across regimes—as argued below, these are exercises with diminishing returns—than on testing to what extent associations and processes play out differently in different welfare regimes. To some extent, we are also involved in transcending the regime framework into more general tests of the influence and interactions of different macro features of the countries we compare. Finally, a novelty of the book concerns the analysis of the relationship between diversity and welfare attitudes. By investigating the effects of new forms of diversity among both target populations and the populations at large, it touches on the social consequences of various forms of migratory flows in European societies. t h e c u r r e n t s tat e of t h e a r t A substantial number of previous studies have analyzed how institutional and cultural factors impinge on the formation of attitudes toward the welfare state in different contexts (for a selection of recent book-length examples, see Cook and Barrett, 1992; Roller, 1992; Jenssen and Martinussen, 1994; Borre and Scarbrough, 1995; Svallfors, 1996; Gilens, 1999; Svallfors and Taylor-Gooby, 1999; Andress et al., 2001; Arriba et al., 2006; Larsen, 2006; Brooks and Manza, 2007; Svallfors, 2007). From the 1990s onward, a growing number of such studies have used a cross-national and comparative approach (see, for example, Svallfors, 1997; 2003; 2007; Bonoli, 2000; Van Oorschot and Halman, 2000; Andress and Heien, 2001; Mau, 2003; Jæger, 2006; Larsen, 2006; Van Oorschot, 2006; Brooks and Manza, 2007; Larsen, 2008). In order to place these contributions, and the ones from this book, in context and perspective, the history of analyses of welfare attitudes needs to be recapitulated. This field was for a long time marred by a lack of data. While reasonably good data had been available for some time when it came to issues such as social mobility, income distribution, and economic indica-



Welfare States and Welfare Attitudes

5

tors of all sorts, that was not the case (even by the late 1980s) when it came to comparing attitudes across countries. This situation is now completely changed with the establishment and growth of data production collaborations such as the European Social Survey, the International Social Survey Program, and the World Values Survey. The field is now “data-rich,” but advanced analyses, explanations, and interpretations lag behind. We hope to make a substantial contribution in this respect through this book. Although election studies and other general surveys had occasionally investigated welfare attitudes from the 1950s onward, more extensive and systematic research did not take hold until the 1970s. It took both the maturation of extensive welfare states and their political questioning in the wake of the economic problems of the 1970s to make public opinion about the welfare state a salient research issue. This first generation of scholars in the field had to make do with compiling, reanalyzing, and comparing national surveys on welfare attitudes—something that made conclusions fragile. One of the earliest attempts to compile and reanalyze existing surveys was made by Richard Coughlin (1979; 1980). Coughlin compared attitude data from eight rich countries using existing national surveys. Another influential series of reanalyses of existing survey data were carried out by Peter Taylor-Gooby (1982; 1983; 1985), who used British surveys to describe patterns and pinpoint ambivalences in welfare attitudes. These first attempts at probing the patterns of welfare attitudes were soon followed by a wave of national surveys, in which different aspects of attitudes toward welfare were investigated. In the 1980s, national surveys were conducted and analyzed in a host of advanced capitalist countries, including the United States (Cook and Barrett, 1992), Germany (Roller, 1992), Sweden (Svallfors, 1989), and Britain (Saunders, 1990). Although conclusions emanated from different data sets and various comparisons, and differed in substance and emphasis, a set of common key findings from these early works may be summarized in five short points: 1. Overall, attitudes were strongly supportive of an encompassing welfare state. In contrast to sweeping statements in the public debate about generational processes leading younger generations away from support for welfare policies or about rising resistance against bureaucratic-­administrative intrusions, the early research in general showed the welfare state to be quite popular. Encompassing welfare policies, which are collectively financed

6

Svallfors

and publicly organized, proved to have overall support from the citizens of the advanced capitalist economies. 2. At the same time, a clear difference in support for universal and selective programs was found. Universal encompassing programs such as pensions and health care received strong support, while more targeted or selective programs such as unemployment benefits and social assistance received much lower support. This pattern essentially applied in most, if not all, advanced capitalist democracies. 3. A clear difference was also documented between general and specific support for the welfare state. General support, in the form of attitudes toward objects such as “the public sector” or “social reforms,” proved to be more dependent on changes in the public discourse and general ideological dispositions, and public support was therefore more volatile at this level. Specific support for concrete welfare policy programs, on the other hand, was shown to be more stable because it was rooted in everyday life experiences. 4. The clear support for welfare policies coexisted with considerable ambivalence regarding several aspects of welfare policies. Quite widespread suspicions about welfare abuse and cheating, for example, and concerns about bureaucracy and inefficiencies in the public sector were important qualifications of the overall support for the redistributive and risk-reducing aspects of welfare policies. 5. The early research also confirmed that class and “class-related” factors (such as income and education) were the most important determinants behind welfare attitudes—in contrast to widespread arguments about sector-related cleavages as the new main factors behind welfare attitudes (Dunleavy, 1980; Saunders, 1986: Ch. 8). Although these first-generation analyses were severely restrained by the nonexistence of truly comparative data, they formed an important backdrop to later developments in the field. Simply by making welfare attitudes a topic for systematic social scientific research, instead of the object of political and speculative projections, they laid the ground for subsequent extensions and improvements. Important single-country studies continued to flourish over the course of the 1990s and 2000s, often focusing on change in welfare attitudes over time—and often finding very little change. (For a selection of analyses based on single-country surveys, see Jenssen and



Welfare States and Welfare Attitudes

7

Martinussen, 1994; Svallfors, 1996; Van Oorschot, 1998; Blomberg-Kroll, 1999; the individual chapters in Svallfors and Taylor-Gooby, 1999; Andress et al., 2001; Arriba et al., 2006; Staerklé et al., 2007.) But crucially, from the 1990s onward, research on welfare attitudes also took an explicitly comparative turn. The precondition of this was the establishment, consolidation, and increasing sophistication of comparative data sets. Data collected (in order of appearance) through the Eurobarometer, the European (later World) Values Surveys, the International Social Survey Program and the European Social Survey have formed the base for systematic research about the political and institutional impact on welfare attitudes and about the interplay between institutional and attitudinal change. Perhaps the most important single enterprise in the first round of systematic comparison of welfare attitudes was the Beliefs in Government (BiG) project, headed by Max Kaase and Kenneth Newton (for a summary, see Kaase and Newton, 1995). In their five volumes, the BiG research team aimed to take stock of what could be learned from the first rounds of genuinely comparative attitude research. In particular, the volume on the “Scope of Government” added a host of valuable analyses of citizens’ orientations toward government activities and spending (Borre and Scarbrough, 1995). In making a distinction between attitudes toward the range of government activities and toward the degree of government involvement, these analyses questioned the “government overload” hypothesis, in which voters’ insatiable requests for ever more government intervention overloaded government budgets and capacities. It was clearly not the case that voters asked for further and further extensions of government responsibilities; on the whole, such demands were quite stable over time. The comparative analyses in the BiG project were still severely restrained by data access. As readily admitted by the editors, many aspects of comparative welfare attitudes could simply not be tapped with the existing data. Time series were still too short and the range of countries still too restricted to allow more wide-ranging conclusions. Furthermore, one could note that the BiG project was still way short of establishing any coherent theoretical framework with which to understand the pattern of attitudes toward state intervention and redistribution. The many interesting analyses it included were not driven by an integrated set of analytical concepts and perspectives.

8

Svallfors

What gradually emerged as somewhat of a master frame for comparative inquiries was the “worlds of welfare” categorization famously introduced by Esping-Andersen (1990). A number of studies analyzed whether attitude patterns and conflict patterns corresponded to the typology he suggested and what might explain instances of noncorrespondence. Pioneering studies in this regard were Svallfors (1993) and Svallfors (1997), which compared attitudes to redistribution in different Western countries, using Esping-Andersen’s worlds of welfare as a frame for country selection and analysis. These studies were followed by many others (Bean and Papadakis, 1998; Evans, 1998; Edlund, 1999a; 1999b; Matheson and Wearing, 1999; Bonoli, 2000; Gelissen, 2000; Andress and Heien, 2001; Arts and Gelissen, 2001; Svallfors, 2003; Jæger, 2006; Larsen, 2006; 2008). The main findings of this “comparing-attitudes-in-regimes” industry are not completely clear-cut, since both conceptual and empirical problems beset the analyses (see following). There appears to be agreement, however, on some findings: We do find substantial differences among countries in overall public support for the welfare state, corresponding roughly to welfare policy commitment. Support for equality, redistribution, and state intervention is strongest in the social democratic regime, weaker in the conservative regime, and weakest in the liberal regime. However, we do not find any clear regime-clustering of countries. Differences and similarities among countries show interpretable patterns, but they are too complex to be summarized as “worlds of welfare attitudes.” Furthermore, there are general similarities across countries in the impact of different social cleavages: Categorical differences along class, gender, or labor market status lines show similar patterns across welfare regimes. Where interesting differences among countries in the magnitude of categorical attitude differences were found, they did not at all follow the model suggested in the closing chapter of Esping-Andersen’s treatise. Instead, they seem to follow the historical articulation of particular social cleavages in different contexts. For example, class differences were especially pronounced in Sweden and some other northwestern countries, reflecting the comparatively high salience of distributive and class-related issues in the political programs and practices of these countries. Even though much has been learned about comparative welfare attitudes from the comparisons of welfare regimes, this whole line of inquiry has currently reached an impasse. At present, we may expect little new to



Welfare States and Welfare Attitudes

9

be learned from yet another comparison of how welfare attitudes differ in “x worlds of welfare capitalism.” In reaching this impasse, a number of conceptual and empirical problems with the analytical framework have emerged. One of these problems is specifically related to the regime concept itself: What are “welfare regimes,” really? Are they country clusters, or are they ideal types that countries approximate more or less well? Are regimes constellations of institutions, constellations of distributive outcomes, or constellations of political actors and political institutions? None of this is entirely clear from reading Esping-Andersen’s founding monograph, and different authors have taken different directions in trying to indicate and compare welfare regimes (see Castles and Mitchell, 1992; Korpi and Palme, 1998; Huber and Stephens, 2001b). If a loose fit is found between country clusters and some particular outcome (such as, in this case, welfare attitudes), is this an indication of a loose coupling of institutions and attitudes or of the fact that countries do not “fit” the welfare regime (see Scruggs and Allan, 2008)? Another more general problem is how welfare state support should best be measured in relation to welfare regimes. Studies vary widely in the kind of indicators they use for attitudinal support, often without much explicit argument as to why a particular measure was chosen. Analysts vary as to whether they choose manifest summary measures composed of multiple individual indicators (for example, Svallfors, 2003), some kind of latent constructs (for example, Andress and Heien, 2001), or some kind of “global” single-item measure (for example, Jæger, 2006). They also differ as to whether they try to measure support for specific welfare policies or attitudes toward (re)distribution in more general terms. All in all, this particular “dependent variable problem” has made comparisons between different analyses hard and cumulation painstakingly slow. To sum up, all these national and comparative studies have clearly contributed to our understanding of the formation of, causes behind, and changes in welfare attitudes. Yet, both data limitations and analytical shortcomings apply to existing analyses. This book yields added value because it includes analyses that (1) combine individual-level attitudinal data with country-level data on the institutional setup of welfare policies; (2) are more firmly based in welfare state theory; (3) more clearly specify the relevant dependent variables that may reveal variations among different institutional

10

Svallfors

and cultural environments; and yet (4) use data that are comparable across a larger number of countries, compared to earlier surveys. a n a ly t ic a l f r a m e wor k Our book is based in comparative analysis, in which actors located in spatially different settings are compared, and elements of the environment in which they are located are used to explain differences and similarities that arise. Such comparisons have often taken the form of comparing countries, the different actions and outcomes that occur within these settings, and the aspects of national settings that may explain why they occur. Important aspects of distributive processes, such as wage and income distribution or access to the labor market, are structured by national institutions of various kinds. Furthermore, countries differ in their basic characteristics such as the age structure of the population or their economic structure. Our aim in this book is to use such macro-factors to study both their main effects for welfare attitudes and their impact on micro-relations of various kinds. We argue that welfare attitudes constitute a key aspect of the moral economy, in which conceptions of the mutual rights and obligations in a society are condensed (Kohli, 1987; Mau, 2003; Svallfors, 1996). According to Mettler and Soss (2004: 61), public policies and political institutions, as well as other aspects of the social environment, “influence the ways individuals understand their rights and responsibilities as members of a political community.” The notion of a moral economy pinpoints that people’s notions of social relations are guided by normative ideas of reciprocity, justice, obligation, and responsibility, and not only by narrow self-interest. We bring various kinds of macro-variables into our analyses. One type relates to features of the institutional settings, in particular those related to the programmatic structure of the welfare state. Here we are interested in institutions as configurations—that is, “regimes” (Esping-Andersen, 1990)—but we are also interested in other, more specific features of welfare state institutions, such as their level of social expenditure or their agespending profiles (Lynch, 2006). We also put emphasis on different distributional profiles of countries. This emphasis relates to the level and structure of inequality and stratification in different country contexts: income distributions, levels of poverty,



Welfare States and Welfare Attitudes

11

and the like. But it also pertains to the size and composition of different (potential) claimant groups: the age structure of the population, the number of immigrants and other “risk-groups” in the population, the relative poverty rates in specific population groups, and so on. All these factors are clearly affected by the institutional setup of different countries, but they are hardly to be seen as institutional features themselves. Another type of macro-variable indicates the prevailing discourse on social policy matters in different countries. In some chapters, such political articulation by various political actors (such as parties and trade unions) is seen as potentially affecting attitudes or group differences in such attitudes. In discussing what kind of macro-to-micro effects on values, beliefs, and attitudes among mass publics we could expect, it might be useful to think of these in the same terms that Pierson (1993) and Svallfors (2007: 267–68) use to discuss feedback effects from institutions and public policies. Pierson summarizes such effects under the headings of “resource and incentive” effects and “interpretive” effects. The resource and incentive effects are present when macro-factors affect the material interests of individuals—for example, when differing risks of unemployment and poverty affect the material interests of members in different classes. Interpretive effects are manifested as different degrees of visibility and traceability of social conditions and policies. For example, the size and composition of claimant groups might affect their visibility, and thereby people’s perceptions of the size and nature of different social problems. To this should be added a normative mechanism (Svallfors, 2007: 267–68). A normative mechanism is present where macro-factors such as public policies and discourse provide citizens with a sense of not only what their material interests are and the current state of affairs but also what state of affairs they would like to see. What do institutions and policies tell citizens about what the world should look like? Macro-to-micro effects such as these should not only be expected to affect levels of attitudes in various respects (“main effects”), but they can also be expected to affect categorical differences in attitudes and the mechanisms through which attitudes are affected (“interaction effects”). As shown in the chapters that follow, all of these forms of effects are present and can be examined through comparative analysis.

12

Svallfors

data : t h e e u rop e a n so c i a l su rv e y 2 0 08 The chapters of this book are all based on data from a new module for the European Social Survey 2008, Welfare Attitudes in a Changing Europe, which was granted to a team headed by Svallfors and includes several of the chapter authors. Already in the early stages of preparing this module, we developed a preliminary conceptual model of welfare attitudes (for a summary, see Figure 1.1). The main dependent variables in the model are welfare attitudes, composed of orientations toward welfare state scope and responsibilities, collective financing, different models of welfare, service delivery, and the target groups and receivers of welfare. Another set of dependent variables concern evaluations of the welfare state—for example, in terms of the task performance of the welfare state, the economic consequences of welfare policies, and the moral and social consequences of welfare policies.

Predispositions Trust (institutions, political actors, fellow citizens)

Attitudes toward Welfare state scope and responsibilities Taxation and financing Alternative welfare state models

Risk perception

Service delivery (privatization, choice,

Beliefs

etc.)

Social values

Target groups/receivers

Risks and

Institutional

resources

framework

Evaluations of Task performance (personal + general) Economic consequences

Sociodemographic factors

Moral/socialconsequences

Figure 1.1  A conceptual framework for analyzing attitudes toward welfare policies



Welfare States and Welfare Attitudes

13

In a nutshell, the model explains these welfare attitudes as a function of a set of predispositions, including interpersonal and institutional trust, risk and threat perceptions, beliefs about welfare policies, social values, and personal experiences. At the individual level, these predispositions, in turn, are expected to vary as a function of the risks and resources that individuals and groups are differentially exposed to and endowed with. As with all ESS modules, Welfare Attitudes in a Changing Europe went through extensive vetting and subsequent revisions in the process of construction. In light of comments from the ESS survey methodologists and the national coordinators of the participating countries, and based on results from extensive pretests of the module, individual survey questionnaire items were constructed, reconstructed, and sometimes discarded along the way. The same degree of thoroughness, transparency, and preciseness also characterizes the stage of translating the English source questionnaire into all the languages of Europe. This whole process is thoroughly documented, and all documents are publicly available at the ESS website (www.europe ansocialsurvey.org). We are confident that as far as questionnaire construction is concerned, the process behind our survey module truly represents the current state of the art. Whatever shortcomings and problems of comparability are still in the questionnaire, they cannot be blamed on the procedures used to construct it. As a first look at our data, we display findings for what is perhaps the core battery of the module. Here respondents are asked what government’s responsibility should be in respect to the following: 1. . . . ensure a job for everyone who wants one 2. . . . ensure adequate health care for the sick 3. . . . ensure a reasonable standard of living for the old 4. . . . ensure a reasonable standard of living for the unemployed 5. . . . ensure sufficient child care services for working parents 6. . . . provide paid leave from work for people who temporarily have to care for sick family members

Responses are given on an 11-point scale, where 0 means it should not be government’s responsibility at all, and 10 means it should be entirely government’s responsibility. Since all items scale well, it makes sense to summarize them into an overall “Welfare Index.” Table 1.1 displays the results for all of the countries in the survey, on a 0–10 scale.

14

Svallfors ta bl e 1.1 Welfare index 0–10, by country Country

Mean

Standard deviation

N

Belgium Bulgaria Croatia Cyprus Czech Republic Denmark Estonia Finland France Germany Greece Hungary Israel Latvia Netherlands Norway Poland Portugal Romania Russia Slovakia Slovenia Spain Sweden Switzerland Turkey Ukraine United Kingdom Total

7.13 8.32 8.16 8.11 7.34 7.59 7.92 7.84 7.05 7.35 8.65 8.24 8.32 8.83 6.77 7.86 7.73 8.11 7.65 8.31 7.24 7.80 8.30 7.74 6.43 7.90 8.79 7.22 7.82

1.21 1.62 1.75 1.28 1.77 1.18 1.55 1.13 1.41 1.54 1.41 1.45 1.55 1.40 1.13 1.19 1.59 1.54 2.13 1.63 1.64 1.46 1.24 1.29 1.56 1.95 1.52 1.33 1.61

1,740 2,108 1,422 1,173 1,957 1,549 1,566 2,143 2,031 2,658 2,051 1,494 2,394 1,933 1,711 1,534 1,507 2,300 1,944 2,311 1,693 1,250 2,326 1,751 1,680 2,314 1,756 2,270 52,566

A few preliminary observations can be made about Table 1.1. One is that Europeans, on average, ask for a quite wide-ranging government responsibility. The average score on the 0–10 scale is 7.82, which indicates a quite substantial government involvement in the well-being of its residents.1 A second is that index values still vary quite a bit among countries. We find that, on average, people living in Eastern Europe ask for the most expansive government responsibility, followed by people living in Southern Europe. So the most interventionist attitudes seem to be found outside the traditional “three worlds of welfare capitalism.” But great caution is needed when comparing levels of attitudes across countries. No matter how well-designed the questions are, or how much they aim at comparing absolute levels of support, it is always the case that survey questions are answered in the context in which respondents are em-



Welfare States and Welfare Attitudes

15

bedded. So some degree of relativism is probably unavoidable, which makes it unwise to argue that, for example, Scandinavians really want the state to do less for them in an absolute sense than people in Bulgaria or Romania do. This is also one of the reasons why our analyses in this book are more concerned with comparing relations and associations than with comparing levels. These relations concern group differences in attitudes across countries, micro-level associations between risks, values and welfare attitudes, and macro–micro interactions that affect welfare attitudes. ou r c on t r i bu t ion s The various chapters tap into our empirical model in different ways. Some of them focus on the explanatory mechanisms that link social categories with welfare attitudes and how these are affected by macro-factors. Others focus mainly on what affects variations in welfare state evaluations more broadly. Still others focus on specific mechanisms, such as the level and forms of ethnic diversity, in the forming of welfare attitudes. Each chapter is driven by a quite specific research question, but we all use the same data set and basic framework. In combination, the chapters provide the most comprehensive picture of European welfare attitudes yet achieved. In Chapter 2, the first empirical chapter, Wim van Oorschot and Bart Meuleman analyze the relationship between (perceived) welfare performance and people’s support for welfare state responsibility. The chapter raises the following questions: 1. How do Europeans in various countries perceive the present performance of their welfare state regarding the standard of living of elderly people and unemployed people, and to what degree do they support that the state takes responsibility for related provisions? 2. What is the overall relationship between actual and perceived performance on the one hand and preferences for state responsibility on the other, and how consistent is the effect of perceived performance across countries? 3. What is the relative contribution of actual and perceived performance on responsibility preferences, compared to that of other relevant individual and country level characteristics?

Results indicate that performance has predictive power in addition to the more classical indicators of interest and ideology. Overall, the effect is negative in the sense that persons who perceive the standard of living of

16

Svallfors

certain groups as unfavorable are more supportive of government intervention targeting these groups. Nevertheless, the strength of the negative effect varies along with general attitudes toward the target group and across countries. Chapter 3, by Helena Blomberg, Johanna Kallio, Olli Kangas, Christian Kroll, and Mikko Niemelä, focuses on groups with a greater-thanaverage risk of facing social problems (such as poverty and social exclusion) as compared to the rest of the population. It rests on the assumption that various individual circumstances that pose a greater-than-average risk for facing social problems and/or being dependent on welfare state benefits might influence attitudes toward the role of the state in welfare issues in a distinct way. The national institutional and situational settings in the countries are further assumed to affect not only the risk patterns but also the way in which people perceive their future risks and how these perceived risks mediate the effects of various present risks on welfare state attitudes. Through similar mechanisms, it is also believed that the various kinds of risks mentioned have an effect on the evaluations of the performance of the welfare state. Results indicate that the Nordic regime, closely followed by the Continental regime, has succeeded best in creating a “subjective” sense of security against future social risks, which seems concordant with the “factual” performance of welfare states in this respect. Evaluations of welfare state task performance seem also to correspond with factual welfare state performance in terms of poverty and other social problems: High-risk groups in the Nordic countries and in Continental Europe are most satisfied with the performance of the welfare state, while high-risk groups belonging to the Southern European, the Anglo-Saxon, and particularly the Eastern European model have a far more negative view of the task performance of the welfare state. Regarding welfare state attitudes, high-risk groups in Eastern and Southern Europe stress government responsibility to the greatest extent, followed by the high-risk groups in the Nordic regime. High-risk groups in the Anglo-Saxon and, above all, in the Continental model are least in favor of government responsibility. In Chapter 4, Christian Staerklé, Tiina Likki, and Régis Scheidegger focus on how normative beliefs—understood as socially shared values, perceptions, and expectations about social order—shape welfare attitudes. In their theoretical framework, the authors distinguish among four basic



Welfare States and Welfare Attitudes

17

categories of beliefs about social order that are associated with welfare legitimacy: moral order (measured by authoritarianism and social distrust), free market (measured by perceived welfare dependency), social diversity (measured by ethnocentrism), and structural inequality (measured by egalitarianism). The first issue raised in the chapter concerns the structure of these normative beliefs, which is empirically tested. Second, the authors investigate the social contexts that give rise to different normative beliefs by studying how level of education and perceived vulnerability affect normative beliefs and welfare attitudes. A mediation model is proposed to show how normative beliefs function as the mechanisms explaining the effect of social position on welfare attitudes. A third question concerns the nature of the relationship between normative beliefs and welfare attitudes. The authors propose that social psychological processes of differentiation and boundary definition provide the normative foundations of welfare legitimacy. Fourth, the contextual variation of normative mechanisms is examined by looking at how membership in different European countries and welfare regimes shapes the role of normative beliefs as organizing principles of welfare attitudes. The overall results achieved the following: 1. Confirmed a structure of conceptions of social order organized as a function of the four basic categories of normative beliefs. 2. Revealed that shared normative beliefs are rooted in memberships in social groups and collective experiences. 3. Showed that authoritarianism, distrust, and egalitarianism led to greater support for the role of government in guaranteeing social rights, while perceived welfare dependency and ethnocentrism were associated with lower support for government involvement. 4. Showed that the relationships between normative beliefs and welfare attitudes varied across countries. Some of this variation could be explained with country-level characteristics, in particular the level of public social spending and the unemployment rate.

In sum, the aim of Chapter 4 is to propose an account of social psychological processes at work in the construction of welfare attitudes. The results demonstrate that such processes are not universal but rather are moderated by the institutional and normative contexts in which they are enacted. Chapter 5, by Jan Mewes and Steffen Mau, addresses the issue of how people living in European countries view the inclusion of foreigners in the

18

Svallfors

welfare state. Group-specific reactions as well as key factors determining the support for openness or welfare chauvinism are the focus of their investigation. Specifically, they confront the question of whether working-class support for welfare chauvinism is linked either to economic or cultural aspects. It has been suggested that the socioeconomic situation of individuals might play the most decisive role in the development of welfare chauvinism. But other studies in the tradition of cultural sociology point out that this class bias might stem from cultural differences between the lower classes and the materially more privileged. Results discussed by Mewes and Mau point to the ongoing importance of class affiliation in the process of attitude formation. It is found that both material and cultural aspects matter for the explanation of welfare chauvinism and its relation to class. As far as the relative strength of these factors is concerned, authoritarianism is a dominant factor behind more skeptical views about the inclusion of foreigners into the welfare state. Findings further show that conventional measures of class, education, and income are of limited use in understanding material well-being, feelings of security and insecurity, or subjective precariousness. These aspects are important, but they seem only partly to be in line with objective status measures. Hence, for the explanation of welfare chauvinism, one must not only take into consideration the cultural aspects, but the economic aspects, too. In Chapter 6, Stefan Svallfors, Joakim Kulin, and Annette Schnabel argue that age and class are two fundamental principles of social stratification; therefore, it is a sociologically central task to analyze their impact on individuals’ attitudes toward social arrangements. The basic research question in this chapter is, to what extent do risk perceptions, beliefs (about claimant groups and about the sustainability of the welfare state), and values work as micro-level mechanisms that tie class/age categories to attitudes toward government responsibilities? Furthermore, class/age differences are explored in a comparative perspective: To what extent do they differ among different national contexts? Are there factors at the country level that systematically affect the relative size of class and age differences? Another question that arises is, do class and age work as cross-cutting cleavages such that small class differences tend to coincide with large age differences, and vice versa? Findings from this study show that perceptions about individual risk and beliefs about the extent of social problems mediated the class–attitudes



Welfare States and Welfare Attitudes

19

relationship. But most of the original association between class and welfare attitudes persisted, showing that other factors than the ones tested here must be at play. Age differences were not affected at all by the suggested micro-level mechanisms. Results further show a large variation across Europe in the extent to which class and age structure welfare attitudes. Some countries are characterized by considerable class and age variation, and others by more modest or even negligible differences. But these differenceswithin-differences do not form any easily interpretable pattern. The only macro-variable that showed any association with age differences in attitudes was the age structure of the population. Here it transpired that older populations tend to be correlated with smaller age differences in attitudes. Furthermore, there is little support for the notion that age and class work as cross-cutting cleavages, so small class differences in attitudes are found where age differences are large, and vice versa. On average, across different dependent variables, age and class differences vary independently of each other. The last empirical chapter, Chapter 7 by Clem Brooks, puts the European findings into comparative perspective by using a national survey from the United States, in which a large section of the ESS 2008 Welfare Attitudes module was replicated. It shows that when it comes to government responsibilities, American attitudes are quite different from the European mean, in that Americans ask for much less responsibility from government for the welfare of citizens. This chapter also brings to bear a novel source of data and theoretical perspective on opinion formation through the use of embedded survey experiments. The first set of experiments attempt to gauge the impact of target groups on perceptions and attributions concerning a number of welfare state consequences. These experiments incorporate issue frames referring, in turn, to “blacks,” the “middle class,” and “immigrants” as target groups. The second set of experiments concern the impact of the economic crisis on policy-attitude formation. These experiments take advantage of the unique historical context afforded by the global financial and economic crisis. Here six ESS baseline items represent control group conditions, with treatment including further reference to the “impact of the economic crisis on all Americans.” Results suggest several novel findings with reference to the context of the United States. First, negative pressures toward changes in welfare perceptions and attributions relate not primarily to African Americans but to

20

Svallfors

immigrants as a key point of symbolic reference. Second, there is evidence of framing impacts associated with “middle class” cues, but these effects are not generally as large in magnitude. Results regarding economic crisis framings follow a very different pattern, showing few framing effects. This is remarkable given the actual and perceived severity of the ongoing economic crisis in the United States. We know that economic crisis perceptions are a reality in the contemporary United States, but the crisis may not have yet done much to reorient U.S. policy preferences concerning government as the central provider of social services and benefits. Through 2009, at least, Americans acknowledged the reality of a far-reaching economic crisis, but they did not appear to believe it sufficient to reorient their preferences regarding the role of government in social policy. If indeed these components of U.S. policy preferences have resisted egalitarian responses to the crisis, the transatlantic gap in these preferences may be large and potentially robust. In the final chapter, Stefan Svallfors summarizes some of the broader themes that have emerged from the book and discusses them as they relate to previous research in the field. He highlights six broader cross-chapter themes that have emerged: 1. A broadening and specification of the dependent variables in comparative attitude analyses 2. Ethnic and cultural diversity as a new challenge and its implications for welfare attitudes 3. Bringing in new explanatory mechanisms into the analyses 4. The uneven impact of macro-variables in welfare attitudes 5. Extensions of the analyses beyond the northwestern hemisphere and in particular the specific welfare attitudes in Eastern Europe 6. Methodological contributions on how to model the macro–micro interplay

Svallfors also points out tensions and unresolved issues emerging from the current volume. These relate to the role of self-interest in attitude formation, the indeterminate relation between theoretical concepts and empirical indicators, the information problem in cross-country comparisons, and the question of dynamics of welfare attitudes. These are issues that will form an important backdrop to future analyses of welfare attitudes in comparative perspective.



Welfare States and Welfare Attitudes

21

no t e 1.  This is the unweighted average, so it should be read as the mean value for all countries in Europe rather than for all individuals in Europe.

r efer ences Alesina, A., and E. Glaeser. 2004. Fighting Poverty in the U.S. and Europe: A World of Difference. New York: Oxford University Press. Andress, H.-J., and T. Heien. 2001. “Four Worlds of Welfare State Attitudes? A Comparison of Germany, Norway, and the United States.” European Sociological Review 17: 337–56. Andress, H.-J., T. Heien, and D. Hofäcker. 2001. Wozu brauchen wir noch den Sozialstaat? Der deutsche Sozialstaat im Urteil seiner Bürger [Do We Still Need a Welfare State? The German Welfare State Judged by its Citizens]. Wiesbaden, Westdeutscher Verlag. Arriba, A., I. Calzada, and E. Del Pino. 2006. Los ciudadanos y el estado de bienestar en España (1985–2005) [The citizens and the welfare state in Spain (1985–2005)]. Madrid, Centro de Investigaciones Sociológicas. Arts, W., and J. Gelissen. 2001. “Welfare States, Solidarity and Justice Principles: Does the Type Really Matter?” Acta Sociologica 44: 283–99. Banting, K. G., and W. Kymlicka, eds. 2006. Multiculturalism and the Welfare State: Recognition and Redistribution in Contemporary Democracies. Oxford: Oxford University Press. Bean, C., and E. Papadakis. 1998. “A Comparison of Mass Attitudes Towards the Welfare State in Different Institutional Regimes, 1985–1990.” International Journal of Public Opinion Research 10: 211–36. Blomberg-Kroll, H. 1999. Kosta vad det kosta vill? Attitydmönster och attitydförändringar hos befolkning och eliter beträffande välfärdsservicen i nedskärningarnas tid [At any cost? Attitude patterns and attitude changes regarding welfare services among the population and elite groups in times of cuts]. Vasa, Institutionen för socialpolitik, Samhälls- och vårdvetenskapliga fakulteten, Åbo akademi. Boeri, T., G. Hanson, and B. McCormick, eds. 2002. Immigration Policy and the Welfare System. Oxford: Oxford University Press. Bonoli, G. 2000. “Public Attitudes to Social Protection and Political Economy Traditions in Western Europe.” European Societies 2: 431–52. Borre, O., and E. Scarbrough. 1995. The Scope of Government. Oxford: Oxford University Press. Brooks, C., and J. Manza. 2007. Why Welfare States Persist: the Importance of Public Opinion in Democracies. Chicago: University of Chicago Press.

22

Svallfors

Castles, F. G. 1993. Families of Nations: Patterns of Public Policy in Western Democracies. Aldershot, UK, and Brookfield, VT: Dartmouth. Castles, F. G., and D. Mitchell. 1992. “Identifying Welfare State Regimes: the Links Between Politics, Instruments And Outcomes.” Governance 5: 1–26. Cook, F. L., and E. J. Barrett. 1992. Support for the American Welfare State: the Views of Congress and the Public. New York: Columbia University Press. Coughlin, R. M. 1979. “Social Policy and Ideology: Public Opinion in Eight Rich Nations.” Comparative Social Research 2: 3–40. ———. 1980. Ideology, Public Opinion & Welfare Policy. Berkeley, CA: Institute of International Studies. Crozier, M., S. P. Huntington, J. O. Watanuki, and the Trilateral Commission. 1975. The Crisis of Democracy: Report on the Governability of Democracies to the Trilateral Commission. New York: New York University Press. Dunleavy, P. 1980. “The Political Implications of Sectional Cleavages and the Growth of State Employment.” Political Studies 28: 364–83. Edlund, J. 1999a. “Progressive Taxation Farewell? Attitudes to Income Redistribution and Taxation in Sweden, Great Britain and the United States.” In The End of the Welfare State? Responses to State Retrenchment, edited by S. Svallfors and P. Taylor-Gooby, 106–34. London: Routledge. ———. 1999b. “Trust in Government and Welfare Regimes: Attitudes to Redistribution and Financial Cheating in the USA and Norway.” European Journal of Political Research 35: 341–70. Esping-Andersen, G. 1990. The Three Worlds of Welfare Capitalism. Cambridge, UK: Polity Press. Evans, G. 1998. “Britain and Europe: Separate Worlds of Welfare?” Government and Opposition 33: 183–98. Gelissen, J. 2000. “Popular Support for Institutionalised Solidarity; a Comparison Between European Welfare States.” International Journal of Social Welfare 9: 285–300. Gilens, M. 1999. Why Americans Hate Welfare: Race, Media, and the Politics of Antipoverty Policy. Chicago: University of Chicago Press. Habermas, J. 1975. Legitimation Crisis. Boston, MA: Beacon Press. Huber, E., and J. D. Stephens. 2001a. Development and Crisis of the Welfare State: Parties and Policies in Global Markets. Chicago: The University of Chicago Press. ———. 2001b. “Welfare State and Production Regimes in the Era of Retrenchment.” In The New Politics of the Welfare State, edited by P. Pierson, 107– 45. Oxford, UK: Oxford University Press. Jæger, M. M. 2006. “Welfare Regimes and Attitudes Towards Redistribution: The Regime Hypothesis Revisited.” European Sociological Review 22: 157–70. Jenssen, A. T., and W. Martinussen. 1994. Velferdsstaten i våre hjerter [The welfare state in our hearts]. Oslo, Ad notam Gyldendal.



Welfare States and Welfare Attitudes

23

Kaase, M., and K. Newton. 1995. Beliefs in Government. Oxford: Oxford University Press. Kohli, M. 1987. “Retirement and the Moral Economy: An Historical Interpretation of the German Case.” Journal of Aging Studies 1: 125–44. Korpi, W. 1989. “Power, Politics, and State Autonomy in the Development of Social Citizenship–Social Rights During Sickness in 18 Oecd Countries since 1930.” American Sociological Review 54: 309–28. Korpi, W., and J. Palme. 1998. “The Paradox of Redistribution and Strategies of Equality: Welfare State Institutions, Inequality, and Poverty in the Western Countries.” American Sociological Review 63: 661–87. ———. 2003. “New Politics and Class Politics in the Context of Austerity and Globalization: Welfare State Regress in 18 Countries 1975–1995.” American Political Science Review 97: 425–46. Larsen, C. A. 2006. The Institutional Logic of Welfare Attitudes: How Welfare Regimes Influence Public Support. Aldershot, UK: Ashgate Publishers Ltd. ———. 2008. “The Institutional Logic of Welfare Attitudes: How Welfare Regimes Influence Public Support.” Comparative Political Studies 41: 145–68. Lynch, J. 2006. Age in the Welfare State: the Origins of Social Spending on Pensioners, Workers, and Children. Cambridge, UK, and New York: Cambridge University Press. Matheson, G., and M. Wearing. 1999. “Within and Without: Labour Force Status and Political Views in Four Welfare States.” In The End of the Welfare State? Responses to State Retrenchment, edited by S. Svallfors and P. TaylorGooby, 135–60. London: Routledge. Mau, S. 2003. The Moral Economy of Welfare States. Britain and Germany Compared. London: Routledge. Mettler, S., and J. Soss. 2004. “The Consequences of Public Policy for Democratic Citizenship: Bridging Policy Studies and Mass Politics.” Perspectives on Politics 2: 55–73. O’Connor, J. 1974. The Fiscal Crisis of the State. New York: St. Martin’s Press. Offe, C. 1984. Contradictions of the Welfare State. Cambridge, MA: MIT Press. Pierson, P. 1993. “When Effect Becomes Cause: Policy Feedback and Political Change.” World Politics 45: 595–628. ———. 2001. “Coping with Permanent Austerity: Welfare State Restructuring in Affluent Democracies.” In The New Politics of the Welfare State, edited by P. Pierson, 428–31. Oxford: Oxford University Press. Roller, E. 1992. Einstellungen der bürger zum wohlfahrtsstaat der bundesrepublik Deutschland [The citizens’ attitudes to the German welfare state]. Opladen, Westdeutscher Vlg. Saunders, P. 1986. Social Theory and the Urban Question. London: Hutchinson Education. ———. 1990. A Nation of Home Owners. London: Unwin Hyman. Scruggs, L., and J. Allan. 2006. “Welfare-State Decommodification in 18 OECD

24

Svallfors

Countries: a Replication and Revision.” Journal of European Social Policy 16: 55–72. ———. 2008. “Social Stratification and Welfare Regimes for the Twenty-First Century: Revisiting the Three Worlds of Welfare Capitalism.” World Politics 60: 642–64. Staerklé, C., C. Delay, L. Gianettoni, and P. Roux. 2007. Qui a droit à quoi? Représentations et légitimation de l’ordre social. [Who is entitled to what? Representations and legitimation of the social order]. Grenoble: Presses Universitaires de Grenoble. Svallfors, S. 1989. Vem älskar välfärdsstaten? Attityder, organiserade intressen och svensk välfärdspolitik [Who loves the welfare state? Attitudes, organized interests and Swedish welfare policies]. Lund, Arkiv. ———. 1993. “Policy Regimes and Attitudes to Inequality: A Comparison of Three European Nations.” In Scandinavia in a New Europe, edited by T. P. Boje and S. E. Olsson Hort, 87–133. Oslo: Scandinavian University Press. ———. 1996. Välfärdsstatens moraliska ekonomi: välfärdsopinionen i 90-talets Sverige [The moral economy of the welfare state: welfare opinions in 1990s Sweden]. Umeå, Boréa. ———. 1997. “Worlds of Welfare and Attitudes to Redistribution: a Comparison of Eight Western Nations.” European Sociological Review 13: 283–304. ———. 2003. “Welfare Regimes and Welfare Opinions: a Comparison of Eight Western Countries.” Social Indicators Research 64: 495–520. ———, ed. 2007. The Political Sociology of the Welfare State: Institutions, Social Cleavages, and Orientations. Stanford, CA: Stanford University Press. Svallfors, S., and P. Taylor-Gooby, eds. 1999. The End of the Welfare State? Responses to State Retrenchment. London: Routledge. Taylor-Gooby, P. 1982. “Two Cheers for the Welfare State: Public Opinion and Private Welfare.” Journal of Public Policy 2: 319–46. ———. 1983. “Legitimation Deficit, Public Opinion and the Welfare State.” Sociology 17: 165–82. ———. 1985. Public Opinion, Ideology, and State Welfare. London: Routledge & Kegan Paul. Van Oorschot, W. 1998. Dutch Public Opinion on Social Security. Loughborough: Loughborough University, Centre for Research in Social Policy. ———. “Making the Difference in Social Europe: Deservingness Perceptions Among Citizens of European Welfare States.” Journal of European Social Policy 16: 23–42. Van Oorschot, W., and L. Halman. “Blame or Fate, Individual or Social? An International Comparison of Popular Explanations of Poverty.” European Societies 2: 1–28.

chapter two

Welfare Performance and Welfare Support Wim van Oorschot and Bart Meuleman

Is welfare support higher in welfare states producing better social outcomes? And conversely, is support lower in countries with worse social outcomes? Often, one tends to believe that the answer to these questions is positive and that the social legitimacy of welfare state programs gains considerably from high-quality and effective provisions (e.g., Esping-Andersen, 1990). Based on the common empirical finding that the individuals who profit more from welfare benefits and provisions usually are more supportive of the welfare state (Kangas, 1997; Svallfors, 1997), it is tempting to assume that the relationship between outcomes and support is positive indeed. In other words, welfare states with better social outcomes are “rewarded” with high levels of popular support, while less-performing institutional setups are “punished” with lacking legitimacy. In our view, however, this is not so self-evident. It could just as well be that in low-performing welfare states, public opinion on welfare state responsibilities reflects a common feeling that extension is necessary to improve general standards of living. Or it could be, as was actually assumed in the welfare state critical “overload theories” of the 1970s (see, e.g., Van Oorschot, 2002; Kumlin, 2007), that public opinion in well-performing welfare states becomes dominated by a general feeling of being overburdened with taxes necessary to uphold the extensive and generous programs that produce the good outcomes. Such “improvement” or “overburden” reactions would imply a negative instead of a positive effect of welfare performance on welfare support. For additional data, please see the Web Appendix tables and figures by following the link in the listing for Contested Welfare States on our website: http://www.sup.org. 25

26

Van Oorschot and Meuleman

At present, it is impossible to identify the exact relationship between welfare outcomes and support. In this chapter we analyze it in detail from different angles. The first angle regards the question of whether it is the actual performance of welfare states that plays a role or the performance as it is perceived by the citizens. The Thomas theorem states that if people define situations as real, then they are real in their consequences (Thomas and Thomas, 1929). Thus, the distinction between actual and perceived performance may be relevant to the degree that in this particular case, the Thomas theorem applies. Concretely, we will test whether the relationship between actual welfare performance and popular support for the welfare state is mediated by individual perceptions of welfare state performance. A second angle concerns differences in welfare target groups. It is known that welfare support usually is higher and more uniform when it regards opinions on welfare state intervention in general but that people are more opinionated, reflecting more strongly their personal structural and cultural positions, when it comes to concrete programs aimed at specific groups (e.g., Kangas, 1997; Svallfors, 2010). With our data we can analyze separately (perceived) welfare performance and welfare support regarding the target groups of elderly people and unemployed people, and test whether support for welfare intervention in favor of these two groups is explained by different mechanisms. Generally, elderly people are regarded substantially more deserving of welfare distribution than are unemployed people because they score higher on the various “deservingness” criteria that people tend to apply in matters of welfare allocation (Van Oorschot, 2006). A third angle regards the levels of analyses. Since actual welfare performance is a country characteristic and perceived performance and welfare support are characteristics of individuals, we will apply multilevel analysis techniques to assess the relative contribution of each of these variables to the variation in people’s welfare support, as well as the direction of their influence. We also include other country- and individual-level characteristics as control variables. In short, with this chapter we aim to contribute to the field by analyzing empirically the relationships among actual welfare performance, perceived welfare performance, and people’s support for welfare state responsibility in 22 present-day European societies. We address three research questions, considering the target groups of elderly and unemployed people for each:



Welfare Performance and Welfare Support

27

1. How do Europeans in various countries perceive the present performance of their welfare state regarding the standard of living of elderly people and unemployed people, and to what degree do they support that the state takes responsibility for related provisions? 2. What is the overall relationship between actual and perceived performance on the one hand and preferences for state responsibility on the other, and how consistent is the effect of perceived performance across countries? 3. What is the relative contribution of actual and perceived performance on responsibility preferences, compared to that of other relevant individualand country-level characteristics?

In the following sections we review relevant literature on opinions regarding welfare state responsibilities and its relationship to welfare performance. We then specify our analytical model with a number of hypotheses. We present our methods, as well as our findings, and conclude with a brief summary and evaluative discussion. w e l fa r e s tat e p e r f or m a n c e a n d su p p or t : p r e v ious r e se a rc h a n d h y p o t h e se s In previous research, welfare support has most often been operationalized as people’s preferences on what Roller (1995) calls the “range” of the role of government (what tasks concerning what policy areas government should take responsibility for) and the “degree” of this role (how much government should spend on certain social policies). A common starting point in studies of welfare legitimacy is that preferences for government intervention are influenced by three categories of variables: interests, ideological dispositions, and institutional arrangements (Gevers et al., 2000). To a certain degree, this approach has been fruitful. Existing research has shown that welfare support is indeed related to structural characteristics reflecting personal interest in social protection (D’Anjou et al., 1995; Kangas, 1997; Svallfors, 1997; Chong et al., 2001) and certain ideological positions, such as a leftist political stance and the endorsement of egalitarian values support (Gevers et al., 2000; Edlund, 2006). The relationship between welfare support and institutional arrangements (as indicated by welfare regime types) is far less clear because the existing studies come to highly contradictory conclusions (Gelissen, 2000; Andress and Heien, 2001; Arts and Gelissen, 2001; Kumlin, 2007; Meier Jaeger, 2007).

28

Van Oorschot and Meuleman

Yet, in spite of these previous studies, current understanding of preferences for government intervention remains limited. The fact that large portions of the interindividual and between-country variations in welfare support remain unexplained suggests that relevant explanatory factors have been neglected. The actual performances of welfare states, as well as the ways in which individuals perceive these performances, are variables that could play an important role in the genesis of welfare support. Nevertheless, this possibility has hardly received any scholarly attention, and very little is known about how welfare state performance and support might be related. A review of comparative European studies on welfare support, including country-level variables,1 shows that actual welfare performance has not been used as an explanatory factor before. Existing studies have revealed only a little more information about the effects of perceived performance on welfare support than about the effects of actual performance. 2 We know of only four internationally published studies that tackle the relationship between perceived welfare state performance and people’s preferences for welfare provision by the state, two of them national (Edlund, 2006; Calzada and Del Pino, 2008) and two others cross-national (Gevers et al., 2000; Kumlin, 2007). As for the national studies, Calzada and Del Pino (2008) report on the results of a Spanish survey in which people were asked about their perceptions of the efficacy of services in the fields of education, health, pensions, and unemployment benefits. About half to two-thirds of Spanish citizens regard the services as effective and efficient, but the interesting issue is that those who had negative perceptions were more, not less, in favor of increased social spending in the various fields. This means that in Spain, at the individual level, the relationship between perceived welfare performance and welfare support is negative. In a Swedish study, Edlund (2006) found that people’s evaluations of the efficacy of health care, sickness-related income benefits, and pensions were for the most part not significantly related to support for increasing or decreasing spending in various welfare fields. But where correlations were significant, they were negative in some cases (among contract workers, public employees, and women), and positive in other (among the self-employed, private sector employees, and men). Taken together, these two national studies suggest that in some countries, and among some social groups, dissatisfaction with welfare provisions can relate to higher welfare support, while in other cases it can relate to lower welfare support.



Welfare Performance and Welfare Support

29

Kumlin’s (2007) cross-national study points to this kind of variation. On the basis of ESS 2002 data for 16 Western European countries, the study found a positive relationship at the individual level: Satisfied people are more supportive of state intervention, and dissatisfied people are less supportive. However, this relationship differs strongly among countries in degree and direction, and the differences are not related to any welfare state characteristics of the countries represented. Finally, Gevers and colleagues (2000) report insignificant relationships between the perceived adequacy of health care systems and preferences for government intervention in this sector. Research Hypotheses In short, little previous knowledge is available to supply answers to our research questions. Our theoretical point of departure is that possible relationships between actual welfare state performance and welfare support are mediated by perceived welfare state performance. First, we anticipate a relationship between actual and perceived welfare state performance, since the interpretations that individuals make of a situation are likely to be rooted in (yet not determined by) the actually existing situation. Thus, citizen’s perceptions on the welfare outcomes are expected to be more positive in well-performing welfare states than in countries with less favorable social outcomes. This leads to a first research hypothesis: Hypothesis 1: Good actual welfare state performance leads to more positive perceptions of welfare state performance.

In turn, this perceived welfare state performance could affect welfare support. The sign of this relationship, however, is not immediately clear. Possibly, perceptions that welfare systems are delivering good outcomes enhance support for these systems, while perceptions of low performance undermine support because negative experiences may have caused citizens to distrust government intervention and its effectiveness. These reward and punishment mechanisms imply that the relationship between perceived welfare performance and welfare support is a positive one. Yet, from a different perspective, this relationship could be negative as well, caused by improvement or overburden reactions. Possibly, people with negative performance perceptions are especially supportive of state responsibility in the provision of welfare because they see it as a necessary process to improve standards of living for vulnerable groups. Or it might be that those who are ­satisfied

30

Van Oorschot and Meuleman

with the current situation do not feel the need to continue support for government intervention because they feel “overburdened” with the taxes necessary to uphold the extensive programs that produce the (perhaps in their eyes, too) good outcomes. As a result, contradictory hypotheses can be formulated for the relationship between perceived performance and support. 3 Hypothesis 2A: The relationship between perceived welfare performance and welfare support is positive (reward/punishment reactions). Hypothesis 2B: The relationship between perceived welfare performance and welfare support is negative (improvement/overburden reactions).

The scarce previous research on the topic has shown that the relationship between perceived welfare performance can vary over social groups (Edlund, 2006; Calzada and Del Pino, 2008) and contextual settings (Kumlin, 2007). In this study, these possibilities are scrutinized in detail by testing the interaction effects between perceived performance on the one hand and interest and ideological variables on the other. Furthermore, we acknowledge that relationships between perceived performance and support could be different according to the specific welfare programs involved and the characteristics of their target groups. In this contribution, we compare support for policies targeting the elderly and the unemployed. Since the elderly are universally seen as more deserving of welfare support than the unemployed (Van Oorschot, 2006), we expect to see less variation among countries in support for welfare for the elderly. This leads to our third hypothesis: Hypothesis 3: The relationship between perceived performance and welfare support will be weaker when the elderly are concerned compared to the unemployed.

data a n d m e t hods To test the relationships between welfare state performance and welfare support, we use the welfare attitudes module included in the fourth round of the European Social Survey (ESS). Due to missing data on key contextual indicators, six countries—Switzerland, Croatia, Israel, Russia, Turkey, and Ukraine—were dropped from the analysis. The remaining data set consists of 42,422 respondents from 22 countries. The countries (and their respective country codes) are Belgium (BE), Bulgaria (BG), Cyprus (CY), Czech



Welfare Performance and Welfare Support

31

ta bl e 2.1 Question wordings for welfare support and perceived performance items Number

Question wording

welfare support For each of the tasks I read out, please tell me on a score of 0 to 10 how much responsibility you think governments should have . . . D17 . . . to ensure a reasonable standard of living for the old D18 . . . to ensure a reasonable standard of living for the unemployed perceived performance D11 What do you think about the standard of living of the pensioners? D12 What do you think about the standard of living of people who are unemployed?

Answer scale

0: should not be governments’ responsibility at all–10: should be entirely governments’ responsibility 0: extremely bad–10: extremely good

Republic (CZ), Germany (DE), Denmark (DK), Estonia (EE), Spain (ES), Finland (FI), France (FR), Great Britain (GB), Greece (GR), Hungary (HU), Latvia (LV), the Netherlands (NL), Norway (NO), Poland (PL), Portugal (PT), Romania (RO), Sweden (SE), Slovenia (SI) and Slovakia (SK). Welfare support is measured by items referring to the preferred degree of government intervention. Respondents were asked to what extent they think the government should be responsible for the standard of living of specific target groups (the exact questions can be found in Table 2.1). Previous research has identified this role-of-government concept as the most important dimension of welfare state legitimacy (Van Oorschot and Meuleman, 2012). Exactly what criteria should be used to operationalize perceived and actual welfare performance is food for discussion. The answer depends on the criteria that are used to evaluate how well welfare states perform. Some might argue that well-performing welfare states provide generous benefit schemes, while others would say that reduction of inequality, high activity rates, or even high levels of happiness are more crucial criteria to evaluate actual welfare performance. Since the concept of a welfare state implies state responsibility for securing some basic modicum of welfare for its citizens, we choose to use the standard of living of respective target groups as a crucial indicator of welfare performance. Using this point of departure, perceived welfare performance is measured by questions about how respondents estimate the standard of living of the respective target groups

32

Van Oorschot and Meuleman

(i.e., pensioners and the unemployed). The ESS contains items that ask the respondent to rate the standard of living on an 11-point scale, ranging from extremely bad to extremely good (see Table 2.1 for the exact wording). Using a similar argument, we operationalize actual welfare performance by looking at target group–specific poverty rates (i.e., incomes lower than the 40 percent of the median equivalized income in the country) in 2008 as reported on the Eurostat website (http://epp.eurostat.ec.europa.eu). Besides these three key concepts, we include a series of possibly confounding variables (on both the individual and contextual levels) as controls in the analysis. Starting at the individual level, we control for the sociodemographic variables gender, age (in four categories: 15–29, 30–44, 45–64, and over 65), and educational level (none or primary, lower secondary, higher secondary and higher nontertiary, and tertiary). Since previous research has shown that indicators of personal and class interest in welfare regulations are relevant predictors of welfare support, we control for several interest variables. Social class is operationalized by the often used EriksonGoldthorpe-Portocarero (EGP) occupational classification scheme. The EGP classification was collapsed into six categories: unskilled workers, skilled workers, routine nonmanuals, service class II, service class I, and the self-employed. Respondents who never held a paying job in their life, and for which this social class variable cannot be constructed, are considered as a separate category. Furthermore, we control for the main activity of the respondents: working, retired, unemployed, or other (consisting mostly of students and homemakers). In the analysis of support for welfare policies targeting the unemployed, the perceived unemployment risk (how likely to become unemployed in the next 12 months: not at all likely; not likely; likely; very likely; don’t know) is added as well. Besides the interest factor, welfare support has also been shown to depend on ideological dispositions as well as on beliefs about and attitudes toward the functioning of the (welfare) state, its schemes, and its benefit claimants. In our analysis two ideological dimensions are taken into account: the endorsement of the principle of equality and political orientation. The principle of equality involves a sum-scale of two items referring to the acceptability of income differences. Political left-right stance has been categorized in four categories: left, center (reference category), and right, and a missing category to cope with the high nonresponse on this variable.



Welfare Performance and Welfare Support

33

Our hypothesis is that individuals who prefer an equal society and who have a leftist political orientation are more supportive of welfare regulations. We include two variables indicating an overall assessment of the country’s political system in general. Political trust is the average of three items, referring to trust in the parliament, politicians, and political parties, respectively (the exact wording of these items is provided in Web Appendix Table A2.1. Satisfaction with democracy (“How satisfied are you with the way democracy works?”) ranges from 0 (extremely dissatisfied) to 10 (extremely satisfied). Those with high levels of political trust and satisfaction are expected to be more in favor of government intervention. Finally, we also include a number of variables that measure attitudes toward benefit claimants and beliefs on welfare schemes. In our analysis of support for policies targeting the elderly, we include measures of the perceived affordability of pensions in the following four categories: 1. [Country] will not be able to afford the present level of old-age pension. 2. [Country] will be able to afford the present level of old-age pension but not to increase it. 3. [Country] will be able to increase the level of old-age pension. 4. Don’t know.

We also measure the general attitude toward the elderly: How positive or negative do you feel toward people over 70? (0 = extremely negative to 10 = extremely positive). In the analysis of support for policies that favor the unemployed, we control for absence of perceived abuse of unemployment benefits: Most unemployed people do not really try to find a job (1 = agree strongly to 5 = disagree strongly). The number of contextual variables that can be included as controls is limited due to difficulties in obtaining comparable data for as many countries as possible and the relatively small sample size at the country level. Therefore, we only control for the most relevant variables that might confound the relationship between actual welfare performance and perceived performance and welfare support. A first candidate is the wealth of a country (measured as the GDP per capita in PPP). Furthermore, the size of the respective target groups might affect the level of awareness of the living conditions of these groups and consequently also affect the level of support for policies directed at these groups.

34

Van Oorschot and Meuleman

Therefore, we include the percentage aged over 65 in the analysis regarding the pensioners and the long-term unemployment rate in the analysis regarding the unemployed. Income inequality is indicated by the income quintile share ratio. Finally, we also control for the social expenditure on the two target groups as a broad and general social policy indicator. All contextual data were retrieved from the Eurostat Statistics Database (http://epp .eurostat.ec.europa.eu/portal/page/portal/statistics/search_database). The context data matrix can be found in Table A2.2. Statistical Models To test the conceptual model depicted above, we will use multilevel modeling (Hox, 2002). This statistical tool was developed to analyze hierarchical data structures. The ESS data are hierarchically structured, since they contain information on citizens that are nested within countries. Using ­multilevel models in cross-national research, however, also has several limitations. Probably the most important one is that the sample size at the country level is very small, which could lead to bias in the estimates (Meuleman and Billiet, 2009). To minimize this problem as much as possible, we will try to keep the number of parameters at the country level as low as possible and perform additional robustness checks. To get a complete view on the relationships among actual welfare performance, perceived performance, and welfare support, both support and perceived performance must be analyzed as dependent variables. Because the analyses are conducted separately for two different target groups (the unemployed and the elderly), four different dependent variables will be analyzed. Before the actual multilevel models are presented, however, we start with some exploratory analyses. f i n di ng s Exploratory Analyses A first objective of this chapter is to answer the questions “How do Europeans in various countries perceive the present performance of their welfare state regarding the standard of living of elderly people and unemployed people?” and “To what degree do they feel the state should take responsibility for related provisions?” In order to answer these questions, we take



35

Welfare Performance and Welfare Support

10

9.5 LV BG 9

GR PT PL Preceived standard of living—pensioners 1

2

HU 8.5

3

4

RO

Role of government— elderly

a closer look at country-means on the perceived performance and welfare support indicators. Figure 2.1 displays the relationship between the perceived standard of living of the elderly (horizontal axis) and support for government intervention to guarantee a reasonable standard of living of the elderly. At first sight, it becomes clear that support for government intervention in favor of the elderly is uniformly high across Europe. The country-means on the support indicator for the elderly range between 7.4 and 9.4, and are thus clearly located at the upper end of the 0 to 10 scale. The country-means for the perceived standard of living of the elderly are considerably lower and show much more variation among countries. In Eastern (and to a lesser

CY ES EE

GB SI

SE

NO 5

FI

6

DK

7

CZ

SK 8

FR

BE NL

7.5

7

DE

Corr: – 0.56

Figure 2.1  Aggregate relationship between perceived standard of living (x-axis) and role of government (y-axis) for the elderly s o u r c e : European Social Survey 2008. n o t e : The intersection of the axes in this figure represents the cross-country mean of role of government and perceived standard of living.

36

Van Oorschot and Meuleman

extent also in Southern) European countries, the perceived standard of living of pensioners is considerably lower than in Western Europe or Scandinavia. Bulgaria has the lowest perceived standard of living (1.5 on a 0 to 10 scale). At the country level, the relationship between perceived performance and welfare support is strongly negative (corr.: –0.56). On average, in countries where the elderly are perceived as less well off, support for government intervention in favor of the elderly is stronger. This is especially true in Eastern European countries, which have very low levels of perceived performance and have one of the highest levels of welfare support. Given the uniform high levels of support, the empirical evidence is clearly supportive of “improvement” mechanisms at the country level. Using the unemployed instead of the elderly as the target group, several similarities are discovered, as shown in Figure 2.2. Again, when it comes to the unemployed, support for government intervention is high. Country means range between 5.9 (Slovakia) and 9.0 (Latvia) on a 0 to 10 scale. Furthermore, country rankings on welfare support for both target groups bear a striking resemblance: Countries with high welfare support for the elderly are usually also more supportive of the unemployed (correlation between country-means: 0.68). Also, when the unemployed are the target group, the aggregate relationship between perceived performance and welfare support is negative (corr.: –0.38), suggesting an “improvement” reaction to low perceived performance. Yet, in addition to these similarities, there are also some remarkable differences. Although still relatively high, welfare support is substantially lower when the unemployed are considered rather than the elderly, and between-country variation in welfare support is more outspoken when the unemployed are the target group. Regarding the perceived standards of living, the unemployed are seen as less well off than the pensioners in almost all countries (Great Britain and Portugal are the only exceptions). Interestingly, and corroborating our Hypothesis 3, the relationship between perceived performance and support is considerably weaker than the one found for the elderly as the target group. When the unemployed are considered, low levels of perceived standard of living in a country apparently do not translate as easily to high average welfare support. The counterintuitive combination of lower perceived standard of living of the unemployed and lower support for welfare benefits for them is

Welfare Performance and Welfare Support 9 LV 8.5

GR

8 RO EE

BG

PT

7.5

Perceived standard of living—unemployed

1

1.5

2

3 7

2.5 HU

Role of government—unemployed



ES CY FI

NO

SE

3.5 SI

4

4.5

DE CZ 6

5

5.5

6

DK

6.5

PL

37

FR

GB

NL BE

SK 5.5

5

Corr: – 0.38

Figure 2.2  Aggregate relationship between perceived standard of living (x-axis) and role of government (y-axis) for the unemployed s o u r c e : European Social Survey 2008. n o t e : The intersection of the axes in this figure represents the cross-country mean of role of government and perceived standard of living.

indicative of deservingness differences between the two target groups. Pensioners are generally perceived as more deserving of social protection than the unemployed because they tend to score more “positively”4 on popular deservingness criteria as control, identity, gratitude, and reciprocity (Van Oorschot, 2006). Therefore, our finding of stronger support for welfare for elderly people is expected. Apparently, the fact that the unemployed are seen as having a lower standard of living, implying that they score more “positively” on the deservingness criterion of need, does not compensate for their more “negative” scores on the other criteria. Also, the deservingness theory explains why the between-country differences in welfare support are larger for the elderly as a target group than for the unemployed. In all European countries, the high deservingness of elderly people is an

38

Van Oorschot and Meuleman

unquestionable matter in public opinion, leading to little between-country variation, while the popular deservingness of unemployed people is less evident and depends more on variable economic situations (it is lower when unemployment rates are lower) (Bryson, 1997; Fridberg and Ploug, 2000; Hills, 2002), leading to more between-country variation in welfare support. These first exploratory analyses provide important insights into the relationship between perceived welfare performance and welfare support. Yet, these insights are preliminary for several reasons. First, the role of actual welfare performance is not taken into account yet. Second, these aggregate findings cannot be extrapolated to relationships between individual-level variables without the risk of committing ecological fallacy. And third, the inclusion of relevant control variables is crucial to be able to exclude spuriousness of the relationships. In the remainder of this results section, we present the results of several multilevel analyses that meet the above-mentioned requirements. Multilevel Analysis of Welfare Support We start by exploring the antecedents of support for welfare policies in favor of the elderly and the unemployed (Table 2.2). To facilitate comparison, the results for both target groups are discussed together. For each of the two dependent variables, three different models are estimated. In Model 1, direct effects of perceived and actual performance are estimated, controlling for relevant individual and context variables. In Model 2, interactions between perceived support and other individual-level variables are explored. Model 3 focuses on cross-country variation in the effect of perceived performance. Before focusing on the explanatory models, however, we take a look at the variance decomposition of the two support variables. 5 For the support variable referring to the elderly, the country-level variance equals 0.226, while the individual-level variance equals 2.724. Concretely, this means that about 8 percent (i.e., the intraclass correlation) of the total variance can be attributed to the country level. The items measuring welfare support for the unemployed have larger variance components at the individual (4.503) and country (0.603) levels. This is a clear indication that public opinion is more divided on unemployment than on pension policies. For the unemployment item, the percentage of country-level variance is also larger (12 percent). For both variables, differences between individuals in

fixed effects Intercept Age (ref. cat.: 16–29) 30–44 45–64 65+ Gender (ref. cat.: female) Male Education (ref. cat.: Primary) Lower secondary Higher secondary Tertiary Subjective income Social class (ref. cat.: unskilled worker) Skilled worker Routine nonmanual Service class II Service class I Self-employed Never worked Main activity (ref. cat.: working) Retired Unemployed Other Likely to be unemployed (ref. cat.: not at all) Working * Not very likely .05 –.01 .06

–.17 .00 –.03 .02 –.06 –.17 ***

**

* *

.05 –.01 .06

–.17 –.01 –.03 .01 –.06 –.17

.01 –.03 –.12 –.06

.02 –.02 –.12 –.06

.12 .12 .11 –.04

*** **

.12 .13 .11

8.41

***

**

* **

*** **

***

Model 2

–.04

***

8.41

Model 1

elderly

.05 –.01 .06

–.17 .00 –.03 .02 –.06 –.16

.01 –.03 –.12 –.06

–.04

.12 .13 .11

8.40

***

**

* *

*** **

***

Model 3

*

***

–.15

**

***

* * ***

**

***

.09 .11 .02

–.19 –.06 –.08 .01 –.16 –.05

–.06 –.12 –.15 –.09

–.09

.02 .06 .09

7.12

Model 1

–.15

.09 .12 .02

–.19 –.06 –.08 .00 –.16 –.05

–.05 –.12 –.16 –.10

–.09

.02 .05 .09

7.13

***

*

**

***

* * ***

**

***

Model 2

unemployed

**

***

***

* * ***

**

(continued)

–.14

.08 .11 .02

–.20 –.06 –.08 .01 –.17 –.06

–.06 –.12 –.16 –.09

–.09

.02 .07 .10

***

Model 3 7.06

ta bl e 2.2 Results of multilevel analyses for support for government intervention in favor of the elderly (N = 38,478) and the unemployed ( N = 37,963)

Working * Likely Working * Very likely Working * Don’t know Principle of equality Political trust Satisfaction with democracy Political stance (ref. cat.: center) Left Right Missing Affordability of pensions (ref. cat.: present level not affordable) Present level affordable Increase affordable Don’t know Attitude toward elderly Attitude toward unemployed Standard of living of pensioners Standard of living of pensioners *   Attitude toward elderly Standard of living of unemployed Standard of living of unemployed *   Attitude toward unemployed Poverty rate among pensioners Poverty rate among unemployed Long-term unemployment rate Income quintile share ratio Long-term unemployment rate *   Standard of living of unemployed * * *

*** *** *** ***

.06 –.06 .08 .01 .14 .27 .29 –.22

–.01

*** *

.20 –.06 –.02

Model 1

***

.06

–.01

***

** *** ***

* * *

*** *

–.22

.01 .14 .27 .30

.06 –.06 .08

.20 –.06 –.02

Model 2

elderly

.00

–.21

.01 .14 .26 .29

.06 –.07 .08

.20 –.06 –.01

***

*** *** ***

* *

*** *

Model 3

t a b l e 2 . 2 ( continued )

–.03 –.35 .31

–.36

.30

.14 –.05 .23

–.03 .05 –.07 .31 .06 .01

*** **

***

***

***

***

***

Model 1

*** **

*

.05 –.02 –.35 .31

***

***

***

***

***

–.36

.31

.14 –.05 .23

–.03 .05 –.08 .31 .05 .01

Model 2

unemployed

*** *** **

–.14

***

***

***

***

***

.13 –.38 .32

–.36

.30

.14 –.06 .21

–.02 .05 –.09 .31 .06 .02

Model 3

% variance reduction level 1 % variance reduction level 2

random effects Residual variance Var. random intercept Var. random slope (slvpens) Covariance intercept-slope .08 .45

2.50 .12

*** ***

.08 .45

2.50 .12

*** ***

.08 .46

2.50 .12 .01 .02

*** *** ** .09 .60

4.08 .24

*** **

.09 .60

4.08 .24

*** **

.10 .53

4.05 .28 .04 .06

*** ** **

42

Van Oorschot and Meuleman

the same country are far more important than differences between countries. Nevertheless, the variation present at the country level is too large to be ignored, and it confirms our choice for the multilevel approach. Model 1.  Model 1 in Tables 2.2 and 2.3 clearly indicates that perceptions of the standard of living of target groups are significantly related to welfare support. In the case of both target groups, this effect is negative. When the standard of living of a target group is perceived as inadequate, it increases support for government intervention in favor of that group. Lower levels of welfare support are found among individuals having more positive perceptions of the standard of living of target groups. In other words, the improvement and/or overburden mechanisms (Hypothesis 2B) turn out to be more correct than punishment and/or reward reactions (Hypothesis 2A). Since these effects are found controlling for a whole series of more classical predictors for welfare support, perceived welfare performance provides an important addition to the literature. The effects are not only statistically significant, but they are also substantial in size, and perceived standard of living is among the most powerful predictors in the model. A one-standard-deviation increase in perceived standard of living decreases welfare support by 0.22 when the elderly are mentioned as the target group and even by 0.36 in the case of the unemployed. The effects size thus differs markedly between target groups: Welfare support for the unemployed is thus more strongly dependent on individual perceptions of welfare outcomes than welfare support for the unemployed. Remarkably, the exploratory analyses presented above showed that, aggregated at the country level, the relationship between perceived performance and welfare support was strongest when the elderly were the target group. Thus, perceived performance is related to welfare policies in the case of the two target groups, but the concrete mechanisms differ. Support for policies targeting the unemployed are primarily dependent on purely individual perceptions of welfare performance; support for policies targeting the elderly is more strongly affected by a nationally shared image of the standard of living of the elderly. Model 1 also gives us a view of the impact of actual welfare performance (indicated by target group–specific poverty rates) on welfare support. For both target groups, near-zero and insignificant effect parameters are found. This is not surprising, since we are dealing with the direct effect of actual performance here, net of the individually perceived performance.

t a bl e 2 .3 Results of multilevel analyses for perceived standard of living of the elderly (N = 38,478) and the unemployed (N = 37,963) Elderly fixed effects Intercept Age (ref. cat.: 16–29) 30–44 45–64 65+ Gender (ref. cat.: female) Male Education (ref. cat.: primary) Lower secondary Higher secondary Tertiary Subjective income Social class (ref. cat.: unskilled worker) Skilled worker Routine nonmanual Service class II Service class I Self-employed Never worked Main activity (ref. cat.: working) Retired Unemployed Other Likely to be unemployed (ref. cat.: not at all) Working * Not very likely Working * Likely Working * Very likely Working * Don’t know Principle of equality Political trust Satisfaction with democracy Political stance (ref. cat.: center) Left Right Missing Affordability of pensions (ref. cat.: present level   not affordable) Present level affordable Increase affordable Don’t know Attitude toward elderly Attitude toward unemployed Poverty rate among the elderly Poverty rate among the unemployed Social expenditure Long-term unemployment rate random effects Residual variance Var. random intercept % variance reduction level 1 % variance reduction level 2

Unemployed

4.37

***

3.71

***

–.23 –.20 –.10

*** **

–.11 –.28 –.16

* *** **

.02

.01

–.01 –.04 –.01 .28

–.05 .00 –.05 .16

***

–.13 –.04 –.03 –.04 .06 .09

**

–.17 .24 .05

** ***

–.07 .39 .29

** *** ***

–.05 .06 .05 .17 .03 –.05 –.11

*** ***

–.06

–.03 –.02 –.02 .03 .13 .13 –.07 –.23 –.07 .06 .09 –.23 –.14 –.13 .23 .23

***

** * *** *

*** *** *** ***

–.12 .10 –.06

*** *

.19 .06 –.03 –.05 –.24

*** *** ***

.32

*

–.25 .34 –.31

* * **

3.57 .58

*** **

2.79 .25

*** ***

.10 .58

.10 .80

44

Van Oorschot and Meuleman

Besides the key indicators of perceived and actual welfare performance, Model 1 contains several other predictors of welfare support that are used as control variables here. The expectations that personal or class interests drive welfare support are partially confirmed. In line with interest theory, individuals with a higher (subjective) income are less in favor of government intervention. Compared to the lowest educational group (primary education), those who completed tertiary education are significantly less supportive of government intervention. These effects are found for welfare support for both target groups, although the effects are somewhat stronger when the unemployed are the target. Higher support for policies targeting the unemployed is also found among those who are unemployed and those who are actually working but perceive that they are very likely to become unemployed in the near future. Furthermore, the youngest age cohort is significantly less supportive of social policy in favor of the elderly than the other cohorts. Yet, several other findings put interest theory in perspective. We expected to find significantly higher levels of support for pension policies among the retired, but this is not the case. And of all occupational classes, once we control for all other possible factors (including egalitarian values), the skilled workers are found to be least supportive of government intervention, irrespective of the target group mentioned. Overall, the explanatory power of interest variables is limited. Models containing only interest variables (not given here) explain less than 2 percent of the individual-level variation and roughly 10 percent of the differences among countries. The claim that ideological dispositions and attitudinal variables affect welfare support receives ample empirical support. Those who endorse the principle of equality and have a more leftist political stance turn out to be more supportive of government intervention. Political trust is less relevant in understanding levels of welfare support. While those with high levels of trust in the key political institutions are somewhat less supportive of intervention in favor of the elderly, no effect is found in the case of the unemployed. Satisfaction with democracy has no significant impact on support for government intervention. Furthermore, opinions and beliefs about welfare policies and their target groups turn out to be relevant predictors. Negative attitudes toward the elderly in general undermine support for pension policies. Respondents who believe that an increase in the current level of pensions is affordable are clearly more supportive than those who believe that an increase in, or even the current level of, pensions is not affordable.



Welfare Performance and Welfare Support

45

Higher levels of support are found among persons who are uncertain (“I don’t know”) about the affordability question as well. People who are of the opinion that abuse of unemployment benefits is, generally speaking, nonexistent are less willing to support government intervention in favor of the unemployed. Besides individual-level controls, the effects of relevant contextual control variables are analyzed as well. Due to the relatively small number of countries in the analysis, we decided to limit the number of context variables. We started by estimating preliminary models, in which only one context variable was included each time (besides all individual-level variables). Only context controls that turned out to affect welfare support significantly were included in Model 1. In the case of support for government intervention in favor of the elderly, none of the contextual controls turned out to be significant. Support for policies targeting the unemployed turned out to be affected by income inequality and unemployment rates. Higher levels of welfare support are found in countries with heavily skewed income distributions and where long-term unemployment is low. Model 2.  The most important conclusion from Model 2 is that perceived welfare performance is negatively related to welfare support, irrespective of the specific target group mentioned. The effect of perceived performance was thereby assumed to be similar across societal groups and countries. Because previous research (Edlund, 2006; Kumlin, 2007) suggests this might not be the case, we explored possible variations in the effect of perceived performance by including interaction terms with perceived standard of living. Additional models containing interactions between perceived standard of living and the other individual-level variables were specified (results not shown). Two such interaction effects were found to be significant (and are taken up in Table 2.2). Referring to the elderly, perceived performance was found to interact with positive attitudes toward the elderly; in the case of the unemployed, an interaction effect with (absence of) perceived benefit abuse was found. The two interaction effects are in fact very similar, since in both cases it concerns a positive interaction with a variable reflecting positive opinions on the respective target group. For those with negative attitudes toward the elderly, support for government intervention in favor of this group is very much contingent on negative evaluations of the standards of living. Among those with more positive attitudes toward the elderly,

46

Van Oorschot and Meuleman

welfare support is constantly high, irrespective of opinions on the standard of living. Similarly, among individuals opining that the unemployed do not really want to work, negative perceptions of the standard of living are almost a necessary condition for welfare support. Those who perceive less abuse of unemployment benefits are prepared to support welfare intervention independent of the assessment of the standard of living they make. So, generally speaking, we can say that the more positive the feelings toward a target group, the smaller (i.e., closer to zero) the effects of perceived performance. Or, in other words, people who have more positive images of a target group are more unconditionally in favor of supporting them. Model 3.  Finally, we performed additional tests to see whether the effect of perceived standard of living varies across countries and, if so, whether this variation can be explained by country-level variables. The effect of perceived standard of living on welfare support is allowed to vary across countries by including a random slope. For both support items, the variance of this random slope is statistically significant, indicating that cross-country variation in the effect size is present indeed. Figures 2.3 and 2.4 display the country-specific effect parameters (with 95 percent confidence intervals). Taking the elderly as the target group (Figure 2.3), the effects of perceived standard of living on welfare support range between –0.43 (in Portugal) and –0.01 (in Cyprus). Thus, a negative effect is found in all countries, although the parameter is not significantly different from zero in Cyprus and Finland. Looking at Figure 2.4, the cross-country variation in the effect of perceived standard of living of the unemployed on support is more marked. In 19 out of 22 countries, a significantly negative effect is found (ranging between –0.78 in Bulgaria and –0.14 in Finland). In Denmark and Norway, no significant effect is present, while Cyprus is a clear outlier with a strong positive effect of perceived performance on support. To assess whether the detected cross-country variation in the effect size relates to contextual characteristics, we tested whether cross-level interactions between the context variables and the perceived standard of living were present. In the case of support for intervention targeting the elderly, no such cross-level interactions were found. As the country ranking in Figure 2.3 already suggests, no clear pattern is present in the strength of effect of perceived performance. Analyzing the support item referring to the unemployed, however, a significant cross-level interaction effect with long-term



Welfare Performance and Welfare Support

47

0.2 0.1 0 –0.1 –0.2 –0.3 –0.4 –0.5 –0.6 PT RO DE FR HU GB PL SE GR CZ EE NO BE DK NL ES SI SK BG LV FI CY

Figure 2.3  Country-specific effects of perceived standard of living of pensioners on welfare support in favor of the elderly, with 95% confidence intervals s o u r c e : European Social Survey 2008.

unemployment is found. The effect of the perceived standard of living of the unemployed is more negative (thus, stronger) in countries with high long-term unemployment rates. This cross-level interaction remains significant even after removing outlying case Cyprus. Under conditions of high unemployment (i.e., when unemployment systems are under pressure), the perception that the unemployed are needy is thus a stronger condition to support unemployment policies. It could be that under such conditions, the debate about the deservingness of unemployed people intensifies, leading to a stronger polarization, and thus variation, in perceptions of living standards. Multilevel Analyses of Perceived Performance The analyses presented above have shown that perceived welfare performance affects welfare support, although some variation exists in the size of the effect between groups with different opinions on the target groups and between countries. A direct effect of actual welfare performance was not

48

Van Oorschot and Meuleman

0.6 0.4 0.2 0 –0.2 –0.4 –0.6 –0.8 –1 BG HU SK RO FR EE PT GB DE CZ SI PL LV ES GR SE BE NL FI NO DK CY

Figure 2.4  Country-specific effects of perceived standard of living of the unemployed on welfare support in favor of the unemployed, with 95% confidence intervals s o u r c e : European Social Survey 2008.

found. Yet, rather than a direct one, we anticipated an indirect effect (mediated by perceived welfare performance; see Hypothesis 1). To investigate this possibility, additional multilevel analyses are performed with perceived standard of living as a dependent variable and actual welfare performance as an explanatory variable (see Table 2.3). As before, a distinction is made according to target groups, and the necessary individual- and context-level controls are included. Empty models (full results not given) confirm that the perceived standards of living of target groups are strongly context-dependent. Respectively, 26 percent (standard of living of pensioners) and 29 percent (standard of living of unemployed) of the total variation in the perceived standard of living are located at the country level. However, the results show us that this does not necessarily mean that performance perceptions are strongly shaped by actual welfare performance. Controlling for relevant individual and contextual characteristics,6 the poverty rate among the elderly has no significant effect on the perceived standard of living of the elderly. In the case



Welfare Performance and Welfare Support

49

of the unemployed, we do find that poverty among the unemployed has a significant effect on welfare support. And as expected, better actual welfare state performance (indicated by lower poverty rates) lead to a more positive perceived performance (as measured by the perceived standard of living). Putting the pieces of the jigsaw together, we come to the following conclusions: Actual welfare performance affects welfare support indirectly, via perceived welfare performance, when the unemployed are referred to. That is, in countries where unemployed people are poorer, the public perceives their living standards as worse and as a result supports more strongly that the state should provide welfare for the unemployed. With the elderly as the target group, however, no relationship between actual performance and welfare support is found, neither directly nor via perceived welfare performance. This difference in findings could be related to the fact that in all (except two) countries, the actual poverty rate for the group of unemployed is very much higher than the rate for the group of elderly (see Web Appendix Table A2.2). This could mean that there generally is a larger public awareness and debate about poverty among the unemployed, which could make people more sensitive, and at the same time more opinionated, on the issue of their standard of living, leading to a closer connection between actual poverty levels and standard of living perceptions. Since they are not the topic of the chapter, the results of the control variables are discussed very briefly here. To a large extent, the perceived standard of living is affected by the same interest variables as welfare support. Members of the youngest birth cohort (16–29 years) and those with a higher subjective income have a more positive perception of the standard of living of both target groups. Compared to unskilled workers, skilled workers have a more positive image of the living conditions of pensioners, while the self-employed have a more positive view of the standard of living of the unemployed. Compared to the working, retired persons give lower estimates of the standards of living of both targets group. The unemployed and working persons who opine that they are very likely to become unemployed perceive the standard of living of the unemployed as more negative. But, interestingly enough, the unemployed are the group with the most positive perception of the standard of living of the elderly. This could signal a social comparison on the part of the unemployed, who in most of the countries on average are substantially poorer than the elderly. From their point of view the elderly are not that bad off at all.

50

Van Oorschot and Meuleman

Regarding the ideological factors, similar effects are found for both target groups. People who endorse the principle of equality and have low levels of political trust have a less positive image of the standards of living of target groups. Left-oriented persons have a less positive view on the standard of living of the unemployed, while no significant differences are found regarding the standard of living of the elderly. Furthermore, positive opinions on the respective target groups coincide with more positive perceptions of standards of living. Perceived standards of living turn out to be affected by several context variables, but the picture is very different according to the target group mentioned. Perceived living conditions of the elderly appear to be better in more unequal societies. Again, this could be seen as a result of social comparison. In more unequal societies, the position of the elderly, as being a highly deserving group provided with pension income, could on average be (seen to be) better than that of other needy groups. Regarding the unemployed, better perceived living conditions are detected in countries with lower long-term unemployment rates and higher social expenditure. c on c l usion In this chapter, we attempted to shed more light on the relationships between the outcomes that welfare states produce and popular support for welfare state. Despite the rapidly growing number of studies on the antecedents of welfare support, this topic has not been dealt with in detail before. We approached the relationship between welfare performance and welfare support using three different angles. First, we made a distinction between how welfare states actually perform and the individual perceptions that citizens have of this performance. Our theoretical argument is that the effect of actual welfare performance on welfare support is mediated by perceived welfare performance. Second, we acknowledged that the genesis of welfare support can be different, depending on the target group of the policies. And third, we argued that a distinction should be made between mechanisms at the individual level and the country level. Concretely, this chapter provides a multilevel analysis of the relationships among actual welfare performance, perceived welfare performance, and support for policies favoring two different target groups: the elderly and the unemployed. We set out to answer several research questions. Our first goal was to sketch a picture of how Europeans perceive the performance of their wel-



Welfare Performance and Welfare Support

51

fare state regarding the standard of living of elderly people and unemployed people, and to what degree they support that the state takes responsibility for related provisions. Our analyses show that support for government intervention in favor of the elderly is uniformly high across Europe. Policies favoring the unemployed receive considerable support as well, although the levels of support are substantially lower than in the case of the elderly. At the same time, the cross-country variation in welfare support for the unemployed is much more prominent than for the elderly. Europeans are far less positive on the living conditions of the respective target groups, since most country averages fall below the midpoint of the scale. The standards of living of the unemployed are perceived as even less favorable than those of the elderly. Perceived welfare performance is especially low in Central/ Eastern and Southern European countries. The main objective of this chapter, however, is to unveil the relationships between welfare performance and welfare support. First of all, we found a substantial and significant effect of perceived welfare performance on welfare support. This effect is detected controlling for several relevant individual and context variables, meaning that the introduction of perceived welfare performance provides an important addition to the existing knowledge. In the case of both target groups, the relationship between perceived performance and support is a negative one. More positive perceptions of the performance of the welfare state are thus not automatically rewarded with high support, and negative performance perceptions do not necessarily lead to reactions of abandoning the welfare state. Instead, the highest levels of support are found among people who perceive that the welfare state is not able to provide good standards of living for the welfare state. This points in the direction of a popular demand to improve social conditions by increasing the intensity of government intervention. In the theory section, the possibility of overload or overburden reactions was discussed as well. Given the relatively high support for government intervention and the low levels of perceived welfare performance, such mechanisms appear not to apply in the European countries we analyzed. A negative relation between perceived performance and welfare support is found for both target groups. Interestingly enough, however, the individual-level effect is strongest for the unemployed, while the aggregate relationship is stronger in the case of the elderly. Welfare policies that favor the unemployed are primarily dependent on purely individual perceptions

52

Van Oorschot and Meuleman

of welfare performance, while support for policies targeting the elderly is more strongly affected by a nationally shared image of the standard of living of the elderly. At first sight, the negative effect of welfare performance on support for government intervention reported here contradicts the findings in Chapter 3, where it is concluded that positive evaluations of the task performance of the welfare state instigate more positive welfare attitudes. In our opinion, however, the findings of both chapters are not incompatible because very different indicators are used. Blomberg and colleagues (see Chapter 3) operationalize evaluations of task performance with two questions about the perceptions of possible positive social repercussions of the welfare state: “To what extent do you agree or disagree that social benefits and services in [country] prevent widespread poverty?” “To what extent do you agree or disagree that social benefits and services in [country] lead to a more equal society?”

These questions are situated on a more general level and are closer to abstract pro-welfare principles compared to the standard-of-living questions we use. We are of the opinion that both operationalizations lead to complementary insights into the genesis of welfare attitudes: Welfare support is strongest among individuals who believe that the welfare state is a good instrument to reduce poverty but who also believe that the actual standard of living of target groups is unfavorable. Additionally, we tested whether levels of perceived welfare performance are influenced by actual welfare performance (as measured by target group– specific poverty rates). When the unemployed are included as a target group, such an effect is found. Better actual performance (i.e., lower poverty rates among the unemployed) leads to more positive perceptions of the perceived standard of living of the unemployed. Together with the earlier finding that positive performance perceptions decrease welfare support, we can say that actual welfare performance has a small but negative indirect effect on support for government intervention. In the case of the elderly, no such relationship between actual and perceived welfare performance was found. In sum, perceived welfare performance turns out to be a far more decisive factor for explaining welfare support than actual welfare performance. In short, this chapter makes several contributions to the existing literature. What we would like to stress is that in the analysis of welfare support,



Welfare Performance and Welfare Support

53

it is fruitful and even necessary to make distinctions according to the specific groups that are targeted by the welfare state. In our case we could analyze the individual and context level determinants of the degree to which people would want government to take responsibility for the standard of living of the elderly and the unemployed. These two target groups are quite different from each other in terms of popular deservingness (much higher for elderly than for unemployed) (e.g., Van Oorschot, 2006). As we suggested in the interpretations of specific results, this factor may play a role in our further understanding of how different support levels for welfare targeted at different groups are produced. For instance, the lower deservingness of unemployed people may account for the combined fact that people perceive them as having a lower living standard than the elderly, while at the same time people support welfare provision to the unemployed less. We found that in EU countries, the high deservingness of the elderly is undisputed, while the deservingness of the unemployed generates more widely varying viewpoints. This probably is the reason that support for policies targeting the unemployed primarily depends on individual perceptions of welfare performance, while support for policies targeting the elderly is more strongly affected by a nationally shared image of the standard of living of the elderly. In addition, the lower popular deservingness level of the unemployed could mean that people are both more sensitive and more opinionated on the issue of the standard of living of the unemployed, leading to a closer connection between actual poverty levels and standard of living perceptions in the case of the unemployed. This stronger opiniation may also explain why support for welfare policies that target the unemployed shows much more variation among individuals as well as among countries and why this variance is more strongly structured along interest variables and ideologies than in the case of the elderly. How exactly deservingness perceptions play a role in producing differences in welfare support in interaction with other determinants is an important subject for future studies. no t e s 1.  The studies reviewed are Andress and Heien (2001), Meier Jaeger (2007), Meier Jaeger (2009), Svallfors (1997), Svallfors (1999), Svallfors (2003), Paugam (2003), Papadakis and Bean (1993), Linos and West (2003), Deitch (2004), Bowles and Gintis (2000), Bonoli (2000), Blekesaune and Quadagno (2003),

54

Van Oorschot and Meuleman

Kluegel and Miyano (1995), Pfeifer (2009), Forma (1999), Arts and Gelissen (2001), Gelissen (2000), Kumlin (2007). 2.  Certainly, there is a strand of research concerning people’s evaluations of the efficacy of local bureaucracies and their trust in local government (see, e.g., Van de Walle, 2004), as well as concerning perceptions of welfare state performance and their effects on political ideology, behavior, and trust (Huseby, 2000; Kumlin, 2006; Petterson, 2007). However, little is known about the issues raised here. 3.  According to these theoretical arguments, perceived welfare performance has a causal effect on perceived welfare performance. However, causality might also run in the other direction. According to social-psychological theory (such as cognitive dissonance theory), it is possible that individuals adapt their evaluation of welfare state performance to be in line with their principled position on welfare state support. Unfortunately, our cross-sectional research design does not allow us to test for this possibility. 4.  In deservingness theory, a more positive score on a popular deservingness criterion means that the target group at issue is seen as more deserving because of its specific position on the criterion. For instance, the elderly score more positively on the criterion of control than unemployed people because the elderly are blamed less for being old than the unemployed are blamed for being unemployed. 5.  These variance components are derived from so-called empty random intercept models (Hox, 2002). Due to lack of space, complete results of this preliminary step are not given here. 6.  Given the relatively small number of countries in our multilevel models, to limit the number of contextual variables, we started by estimating the effect of all contextual variables separately (with individual-level variables included). Only the context controls that were found to be significant were included in the final models presented here.

r efer ences Andress, H. J., and T. Heien. 2001. “Four Worlds of Welfare State Attitudes? A Comparison of Germany, Norway and the United States.” European Sociological Review 17: 337–56. Arts, W., and J. Gelissen. 2001. “Welfare States, Solidarity and Justice Principles: Does the Type Really Matter?” Acta Sociologica 44: 283–99. Blekesaune, M., and J. Quadagno. 2003. “Public Attitudes Toward Welfare State Policies: a Comparative Analysis of 24 Nations.” European Sociological Review 19: 415–27. Bonoli, G. 2000. “Public Attitudes to Social Protection and Political Economy Traditions in Western Europe.” European Societies 2: 431–52. Bowles, S., and H. Gintis. 2000. “Reciprocity, Self-Interest, and the Welfare State.” Nordic Journal of Political Economy 26: 33–53.



Welfare Performance and Welfare Support

55

Bryson, C. 1997. “Benefit Claimants: Villains or Victims?” In British Social Attitudes, the 14th Report, edited by R. Jowell, J. Curtice, A. Park, L. Brook, K. Thomson, and C. Bryson, 73–88. Aldershot, UK: Ashgate Publishers Ltd. Calzada, I., and E. Del Pino. 2008. “Perceived Efficacy and Citizen’s Attitudes Toward Welfare State Reform.” International Review of Administrative Sciences 74: 555–74. Chong, D., J. Citrin, and P. Conley. 2001. “When Self-Interest Matters.” Political Psychology 22: 541–70. D’anjou, L. J. M., A. J. Steijn, and D. V. Aarsen. 1995. “Social Position, Ideology and Distributive Justice.” Social Justice Research 8: 351–84. Deitch, C. 2004. “Gender and Popular Support for the Welfare State: Cross-National Trends in a Period of Restructuring.” Annual Conference ISA-RC19 “Welfare State Restructuring: Processes and Social Outcomes.” Paris. Edlund, J. 2006. “Trust in the Capability of the Welfare State and General Welfare State Support: Sweden 1997–2002.” Acta Sociologica 49: 395–417. Esping-Andersen, G. 1990. The Three Worlds of Welfare Capitalism. Cambridge: Polity Press. Forma, P. 1999. Interests, Institutions and the Welfare State: Studies on Public Opinion Towards the Welfare State. Helsinki: STAKES. Fridberg, T., and N. Ploug. 2000. “Public Attitudes to Unemployment in Different European Welfare Regimes.” In Welfare Regimes and the Experience of Unemployment in Europe, edited by D. Gallie and S. Paugam, 334–50. Oxford: Oxford University Press. Gelissen, J. 2000. “Popular Support for Institutionalized Solidarity: a Comparison Between European Welfare States.” International Journal of Social Welfare 9: 285–300. Gevers, J., J. Gelissen, W. Arts, and R. Muffels. 2000. “Public Health Care in the Balance: Exploring Popular Support for Health Care Systems in the European Union.” International Journal of Social Welfare 9: 301–21. Hills, J. 2002. “Following or Leading Public Opinion? Social Security Policy and Public Attitudes since 1997.” Fiscal Studies 23: 539–58. Hox, J. 2002. Multilevel Analysis: Techniques and Applications. New Jersey: Lawrence Erlbaum Associates. Huseby, B. 2000. Government Performance and Political Support. Norway: University of Trondheim, Department of Sociology and Political Science. Kangas, O. 1997. “Self-Interest and the Common Good: the Impact of Norms, Selfishness and Context in Social Policy Opinions.” Journal of Socio-Economics 26: 475–94. Kluegel, J., and M. Miyano. 1995. “Justice Beliefs and Support for the Welfare State in Advanced Capitalism.” In Social Justice and Political Change: Public Opinion in Capitalist and Post-Communist States, edited by J. Kluegel, D. Mason, and B. Wegener, 81–108. New York: Walter de Gruyter.

56

Van Oorschot and Meuleman

Kumlin, S. 2006. “Learning from Politics? The Causal Interplay Between Government Performance and Political Ideology.” Journal of Public Policy 26: 89–114. ———. 2007. “Overloaded or Undermined: European Welfare States in the Face of Performance Dissatisfaction.” In The Political Sociology of the Welfare State: Institutions, Cleavages and Orientations, edited by S. Svallfors, 80– 116. Stanford: Stanford University Press. Linos, K., and M. West. 2003. “Self-Interest, Social Beliefs, and Attitudes to Redistribution: Re-addressing the Issue of Cross-National Variation.” European Sociological Review 19: 393–409. Meier Jaeger, M. 2007. “Are the ‘Deserving Needy’ Really Deserving Everywhere? Cross-Cultural Heterogeneity and Popular Support for the Old and Sick in Eight Western Countries.” In Social Justice, Legitimacy and the Welfare State, edited by S. Mau and B. Veghte, 73–96. Aldershot, UK: Ashgate Publishers Ltd. ———. 2009. “United but Divided: Welfare Regimes and the Level and Variance in Public Support for Redistribution.” European Sociological Review, Advance Access published January 12, 2009. Meuleman, B., and J. Billiet. 2009. “A Monte Carlo Sample Size Study: How Many Countries Are Needed for Accurate Multilevel SEM?” Survey Research Methods 3: 45–58. Papadakis, E., and C. Bean. 1993. “Popular Support for the Welfare State: a Comparison Between Institutional Regimes.” Journal of Public Policy 13: 227–54. Paugam, S. 2003. “Social Precarity and Attitudes to Society and the Welfare State.” Paper presented at the EUROConference on “Institutions and Inequalities,” Helsinki, September 20–24, 2003. Petterson, M. 2007. “The Relationship Between Public Service Dissatisfaction and Political Action: Does Institutional Design Matter?” In The Political Sociology of the Welfare State: Institutions, Cleavages and Orientations, edited by S. Svallfors, 149–88. Stanford: Stanford University Press. Pfeifer, M. 2009. “Public Opinion on State Responsibility for Minimum Income Protection: A Comparison of 14 European Countries.” Acta Sociologica 52: 117–34. Roller, E. 1995. “The Welfare State: the Equality Dimension.” In The Scope of Government, edited by O. Borre and E. Scarbrough, 165–97. New York/ Oxford: Oxford University Press. Svallfors, S. 1997. “Worlds of Welfare and Attitudes to Redistribution: a Comparison of Eight Western Nations.” European Sociological Review 13: 283–304. ———. 1999. “Political Trust and Attitudes Towards Redistribution.” European Societies 1: 241–68.



Welfare Performance and Welfare Support

57

———. 2003. “Welfare Regimes and Welfare Opinions.” Social Indicators Research 64: 495–520. ———. 2010. “Public Attitudes.” In The Oxford Handbook of the Welfare State, edited by F. G. Castles, S. Leibfried, J. Lewis, H. Obinger, and C. Pierson, 241–51. Oxford: Oxford University Press. Thomas, W. I., and D. S. Thomas. 1929. The Child in America: Behavior Problems and Programs. 2d ed. New York: Alfred A. Knopf. Van De Walle, S. 2004. “Perceptions of Administrative Performance: the Key to Trust in Government.” Doctoral Thesis. Leuven: Catholic University of Leuven. Van Oorschot, W. 2002. “Individual Motives for Contributing to Welfare Benefits in the Netherlands.” Policy and Politics 30: 31–46. ———. 2006. “Making the Difference in Social Europe: Deservingness Perceptions Among Citizens of European Welfare States.” Journal of European Social Policy 16: 23–42. Van Oorschot, W., and B. Meuleman. 2012. “Welfarism and the Multidimensionality of Welfare State Legitimacy: Evidence from The Netherlands, 2006.” International Journal of Social Welfare 21: 79–93.

chapter three

Attitudes Among High-Risk Groups Helena Blomberg, Johanna Kallio, Olli Kangas, Christian Kroll, and Mikko Niemelä

Popular attitudes toward the welfare state have traditionally been assumed to be dependent on long-term class-related interests, short-term selfinterests, and/or the values and norms that have been internalized by the individuals in the society in question (Taylor-Gooby, 1985: 397; Svallfors, 1995; Svallfors, 2006). Inspired by new institutionalist theories’ assumptions of policy-feedback effects of welfare state institutions, the emphasis has also, to a growing extent, turned toward investigating what effects the welfare state’s institutions have on attitudes. In connection herewith, there has also been a growing interest in the search for a larger variety of, and/ or in achieving a greater “refinement” regarding, factors and mechanisms through which the welfare state, as well as the individual’s own position in relation to the welfare state, might influence the individual’s attitudes (e.g., Mau, 2003; Larsen, 2006). In line with these ambitions, this chapter aims at contributing to the knowledge on attitude differences and attitude formation by focusing on groups with an identifiably greater than average risk of facing social problems (such as poverty and social exclusion) compared to the rest of the population. Our choice to focus on two different “objective” high-risk groups has been guided by some different, although interlinked, perspectives. First, as stated above, it enables us to study from a fairly unexplored perspective the different mechanisms through which individuals who are the most dependent in a particular welfare state evaluate and support that For additional data, please see the Web Appendix tables and figures by following the link in the listing for Contested Welfare States on our website: http://www.sup.org. 58



Attitudes Among High-Risk Groups

59

state’s activities. Hereby, it aims at contributing to the theoretical reasoning regarding the importance of self-interests for welfare state support. Second, in many European countries (as well as in other OECD countries), the number of claimants of various welfare benefits has been rising—for example, because of growing unemployment and structural changes in the economy. Yet, remarkably little cross-national analysis of “high-risk groups” in different European welfare regimes has taken place to date. As a result, we have limited knowledge on whether, compared to the rest of the population, groups that are already at or have a greater than average risk of moving toward the margins of the welfare state have different perceptions about the welfare state’s performance and different attitudes toward its role. Third, in our view, the issue of welfare state legitimacy needs to be viewed not only from the perspective of general public support but from the perceptions and attitudes of the groups most affected by and most dependent on the welfare policies conducted as well. Many scholars have argued that one should not focus solely on dominating welfare policy areas in quantitative terms, such as large social insurance systems, and the support for them among the general public. They believe that focus should also be directed at the legitimacy of policies affecting those at the margins of the welfare state because “it is here the limits—and the contents—of social citizenship are tested, and it is here that any differences in European social policy will be most obstructive” (Leibfried, 2000: 191). Therefore, we believe it is important to closely scrutinize welfare-related perceptions and attitudes among different high-risk groups in different types of welfare states. Since a fundamental task of the welfare state is to protect citizens against social risks, focusing on attitudes and attitude formation among the groups in society facing the greatest social risks could provide valuable information on the legitimacy of European welfare systems. Thus, our research design rests on the assumption that groups that have a greater than average risk of facing social problems (e.g., being sick, being an immigrant) and/or being dependent on welfare state benefits might perceive the state’s role in welfare issues in a distinct way compared to the share of the population that is not exposed to these risks. The national institutional and situational settings in the countries investigated—that is, as reflected through welfare regime type—are further assumed to affect not only the risk patterns but also the way people perceive their future risks. In addition, it is important to explore how these perceived risks mediate the

60

Blomberg, Kallio, Kangas, Kroll, and Niemelä

Institutional and situational setting

High-risk groups • Potential high-risk • Transfer-dependent groups

Perceived future risks

Welfare state attitudes: scope and responsibilities of the welfare state

Evaluation of task performance Figure 3.1  Conceptual model of the study

effects of any present risks on attitudes toward the welfare state. Through similar mechanisms, it is also believed that the various risks mentioned have an effect on evaluations of the welfare state’s performance. Also, these evaluations are expected to influence attitudes toward the state’s role in welfare issues (e.g., Edlund, 2006; Kumlin, 2007), both by themselves and in interaction with perceived future risks (Figure 3.1). h ig h - r i sk g rou p s’ at t i t u de s In studies on the impacts of interests and ideologies on welfare attitudes, indicators of social risk have usually been included as explanatory variables only implicitly. Arguably, most of the traditional factors, such as income, educational background, social class, employment status, age, and gender, can be viewed as reflecting, among other things, differences in social risks. Such studies have shown that generally the unemployed, those in the lowest income groups, members of the working class, and the least educated display the strongest support for the welfare state (e.g., Jaeger, 2006; Svallfors, 2006; Blekesaune, 2007; Pfeifer, 2009). Although such attitudinal patterns partly coincide with higher risks of facing social problems, at present or in the future, these indicators can be assumed to reflect a variety of circumstances besides such risks. It is therefore not always easy to draw conclusions about which aspects of these indicators have the most influence on opinions of the welfare state. Indica-



Attitudes Among High-Risk Groups

61

tors of position within the system of social stratification might be regarded as capturing a “structural” risk of facing certain social problems. However, there are other “levels” of risks that often cut across the above-mentioned categories. Individuals belonging to such groups face a higher risk of experiencing social problems compared to others in an otherwise similar social position. Such groups may be defined in different ways. Factors such as sickness, financial difficulties, immigrant status, and so forth can be assumed to indicate a potentially higher than average risk of facing social problems or being “vulnerable” (see Fridberg and Kangas, 2008) compared to people who lack such characteristics. From a selfinterest perspective, people belonging to such groups can be assumed to support the welfare state because they want to secure themselves against these risks. Thus, people demand insurance against their need for welfare services and benefits (see Varian, 1980; Meltzer and Scott, 1981; Rehm, 2007). For instance, it has been observed that poor health raises the probability of an individual favoring government responsibilities for a decent standard of living (Jaeger, 2006). Also, factors like a bad economy or discontent with one’s personal financial situation have been found to correlate with stronger support for state intervention in economic redistribution (Boeri et al., 2001; Blekesaune, 2007: 397). Another way of determining a potential “high risk” of facing social problems is by identifying those who depend on other types of welfare state benefits. A closely related concept used in international research is “transfer classes,” which refers to those outside the labor force, the unemployed, oldage pensioners, the disabled, and persons with low incomes. It does, however, remain somewhat unclear to what extent transfer classes should be expected to differ in their attitudes toward the welfare state in relation to other “classes.” Some results have shown that most members of transfer classes appear to be more in favor of an extensive welfare state than people in the labor force (Gelissen, 2000). However, other studies have shown that among the transfer classes, only the unemployed prefer a higher level of solidarity and support public health care provision more often than respondents belonging to the labor force. The unemployed are also more often in favor of the need principle in welfare provision (Gevers et al., 2000; Arts and Gelissen, 2001). One reason for these results might be related to the fairly broad definition of “transfer classes,” which usually include both people who might possibly have to depend on the welfare state someday and those who are already doing so. Therefore, we use the concept of “transfer-dependent

62

Blomberg, Kallio, Kangas, Kroll, and Niemelä

groups” instead to clarify that we include in this category only those who are actually clearly dependent on the welfare state. Although there is some overlap with the “potential high-risk groups,” here it is assumed that those (also) who belong to the transfer-dependent groups could provide more information on attitude formation because of their dependence. These groups have specific interests in supporting welfare institutions because they look at social security from a recipient’s point of view. At the same time, personal experiences with services and transfers do not necessary lead to more positive views toward the welfare state. Bad experiences with the system in practice could, depending on the welfare state’s arrangements, also lead to disappointments and thus more negative attitudes toward the welfare state. This is especially true of encounters with selective services and benefits, which tend to be less generous, more stigmatizing, and more controlling compared to more universal services and benefits (see Soss, 1999; Rothstein, 2000; Kumlin and Rothstein, 2005). However, regardless of the kinds of services and benefits involved, encounters with the welfare state might result in more negative views in general and thereby lead to more negative attitudes toward it. For instance, a recent Finnish study (Muuri, 2010) showed that recipients of welfare services and benefits are more likely to regard them as not working properly and their levels as being too low compared to nonrecipients. Such results point to the importance of also including evaluations of welfare state performance in the equation. There is, however, much less comparative research on the perceived consequences of welfare policies than on popular opinions toward the welfare state and its responsibilities (see Van Oorschot, 2010; Wendt et al., 2010). Therefore, not very much is known about the connection between popular evaluations of the welfare state’s performance and attitudes toward the welfare state’s responsibilities; existing research is contradictory. In a study by Kumlin (2007: 110, 111), greater dissatisfaction with welfare state performance tended to be associated with lower levels of support for state intervention. However, such an impact was not found in all the countries studied; the strongest connection was in Nordic and Continental countries. Some earlier national studies have pointed to some correlations between dissatisfaction at a more concrete level and a demand for increased spending and greater opposition to cutbacks in the welfare state (Blomberg, 1999; Edlund, 2006).



Attitudes Among High-Risk Groups

63

Immediate interests (and thus different experiences with the welfare system) are not the only factor that affects popular support for the welfare state and the evaluation of its performance. People’s expectations regarding their future—for example, concerning how they perceive their future risks—might play an important role in their support of the welfare state and possibly also have an impact on the evaluation of its present performance (e.g., Gelissen, 2000; Andress and Heien, 2001). Thus, risk perceptions can be assumed to constitute an intermediating factor in the relationship between risk group membership and welfare state attitudes. Correlations have been found between personal perceptions of a high risk of becoming unemployed and the degree of support for unemployment insurance. This explains, for example, the degree of opposition to cutbacks in social security in Finland (Kallio, 2008: 13) and the willingness to pay for unemployment insurance in France, Germany, Italy, and Spain (Boeri et al., 2001: 21). According to a study by Andress and Heien (2001: 351), people in Germany, Norway, and the United States who felt that their future economic situations were secure were more often opposed to the welfare state. w e l fa r e r e g i m e s , w e l fa r e s tat e p e r f or m a n c e , a n d w e l fa r e at t i t u de s Although the high-risk groups that were investigated differ by definition from the rest of the population in their respective countries, it is likely that these groups are also affected (to a greater or lesser extent) by the same factors that influence the rest of the population and that give rise to country- and/or regime-specific variations in attitudes. Therefore, we present a short review of the findings at the population level regarding the attitudinal dimensions included in our analysis. In several comparisons of popular attitudes in different types of welfare states or welfare regimes, citizens in the Nordic countries have been found to display the strongest support for state intervention and state responsibilities in welfare matters, while support in the Anglo-American welfare states has been the weakest, with the Central European countries placing themselves in between the other two regimes. The differences among regimes (as well as among different socioeconomic groups within the regimes) have been found to be particularly clear when it comes to attitudes

64

Blomberg, Kallio, Kangas, Kroll, and Niemelä

regarding the state’s responsibility to provide a decent standard of living for the unemployed, to offer a job to everyone who wants one, and to reduce income differences between people with high incomes and those with low incomes. Differences have been smaller when it comes to attitudes regarding the state’s responsibilities to offer health care for the sick and to provide a decent standard of living for the elderly. Another common conclusion in previous research has been that the patterns of attitudinal group differences are fairly similar in all developed welfare states as regards the impacts of social class, employment status, and gender (Svallfors, 1997; 2003; 2004; 2006; see also Blekesaune and Quadagno, 2003: 420; Jaeger, 2007: 89). Some studies, however, seem to question this pattern, although they are usually based on partly different selections of countries and on differing operationalizations. Therefore, some analyses have found small attitudinal differences among regimes—that is, toward stronger support for the welfare state in liberal regimes (Gelissen, 2000) or toward higher welfare state support in countries belonging to the Southern and Eastern welfare models (Pfeifer, 2009). Thus, earlier results on general attitudinal patterns in welfare regimes do not necessarily provide a clear basis on which to formulate a hypothesis regarding attitudes among high-risk groups in various countries or regimes. Taking this into account, we have—while aware of the critique that has been presented regarding welfare regime theory in general—chosen to apply a strategy that builds on a study by Fridberg and Kangas (2008: 39, 41). This strategy also satisfies the requirements regarding the sample size of the groups in focus. Thus, our ambition is to analyze (regime-specific) differences in welfare state performance in relation to issues that are central to the groups studied. Fridberg and Kangas analyzed different welfare models along two dimensions that seem to be particularly relevant when investigating high-risk groups: the universality of the system and the generosity of provisions. The Nordic or Scandinavian welfare regime is characterized by a combination of universalism and generosity, whereas the Continental and, to some extent, the Southern European welfare regimes are characterized by fairly generous but less universal benefits. Countries belonging to the Anglo-Saxon regime are, as a rule, neither universal nor generous, possibly with the exception of the United Kingdom, where the welfare system is fairly universal but not generous. In comparison with the above regimes, the Eastern European



Attitudes Among High-Risk Groups

65

welfare regime would appear to be neither universal nor generous. On the basis of earlier European Social Survey (ESS) data, Fridberg and Kangas’s analysis revealed significant variations among the preceding welfare regimes1 as regards to the prevalence of social problems among the groups, which is similar to our category of “potential high-risk groups.”2 In the Nordic regime, it was found that social policy has been comparatively successful in preventing poverty, insecurity, and social isolation of high-risk groups. The situation for high-risk groups seems to be the least advantageous in countries included in the Eastern European regime. Countries belonging to the Continental, Anglo-Saxon, and Southern European regimes place themselves in between these two “poles”; all of them, however, diverge clearly from the situation in the Nordic regime. These results seem to be concordant with earlier findings regarding the Nordic welfare states’ ability to reduce income inequalities and prevent poverty due to their largely universal social policy programs, while other welfare regimes have a stronger emphasis on means testing and private and exclusive occupational programs (e.g., Korpi and Palme, 1998). Therefore, we believe that the amount of respondents who anticipate future risks for encountering social problems will be the lowest in the Nordic welfare regime and the highest in the Eastern European welfare regime. We further assume, also with reference to the character of the respective regimes, that satisfaction with welfare state task performance among highrisk groups would be the highest in the Nordic welfare model and the lowest in the Eastern European model. However, it is difficult to formulate clear assumptions about the support for a (comprehensive) welfare state among high-risk groups in different welfare regimes. Although little research is available to guide assumptions regarding the attitudes of high-risk groups toward the welfare state, it further seems reasonable to assume that the attitudes of different types of risk groups are not equally strongly affected by the institutional design of the welfare model. We assume that the perceptions, evaluations, and attitudes of transferdependent groups are influenced by the institutional design of the welfare model to a greater extent than those of our “potential high-risk” groups that are not (yet) clients of the welfare state. Thus, we assume a clearer variation between welfare regimes concerning the “attitudinal mechanisms” (see Figure 3.1) when transfer-dependent groups are concerned compared to potential high-risk groups.

66

Blomberg, Kallio, Kangas, Kroll, and Niemelä

r e se a rc h qu e s t ion s , va r i a bl e s , a n d m e t hods As we said, our aim is to contribute to the knowledge on attitude differences and attitude formation by focusing on groups with an identifiably greater than average risk of facing social problems (such as poverty and social exclusion) compared to the rest of the population. We focus on attitudes and attitudinal patterns of high-risk groups regarding the responsibility of the government in various welfare-related issues in an attempt to test whether the inclusion of perceptions of future risks and/or the evaluations of welfare state performance could improve the understanding of risk groups’ attitudes. The following questions are addressed in this study: • How do different high-risk groups perceive their future risks in different European welfare models? • How do different high-risk groups evaluate the welfare state’s task performance? • What are the attitudes of different risk groups toward the scope and responsibilities of the welfare state? • Do perceived risks and evaluations affect the attitudes of high-risk groups?

The relevant high-risk groups are empirically constructed on the basis of variables included in the ESS-core module 2008. The “potential highrisk” group variable is constructed on the basis of whether the respondents are hampered by health problems, find it difficult to cope with their present income, and were not born in the country in question, respectively. These indicators were chosen because they reveal, in different ways, a greater than average potential risk of facing more permanent social problems (such as poverty and social exclusion) compared to the rest of the population. The other indicator chosen to measure a greater than average risk of facing more permanent social problems is whether a respondent belongs to a transfer-dependent group; in this study, this is taken to mean those who have reported living in households that rely mainly on social transfers as their main source of income (pensions excluded). 3 Recipients of welfare benefits may also, because of their close encounters with the system, make differing evaluations of the welfare state compared to the rest of the population. Perceived risks, in turn, are measured by the respondents’ perceptions toward the risk of unemployment and economic hardship in the future.



Attitudes Among High-Risk Groups

67

Detailed information about these indicators can be found in Web Appendix Tables A3.1 and A3.2. The variable measuring the evaluation of welfare state task performance is constructed on the basis of two questions: 1. To what extent do you agree or disagree with the statement that social benefits and services prevent widespread poverty? 2. To what extent do you agree or disagree with the statement that social benefits and services lead to a more equal society?4

The dependent variable of welfare attitudes toward governments’ responsibilities is based on the responses to the following statements: 1. It should/should not be governments’ responsibility to ensure a job for everyone who wants one. 2. It should/should not be governments’ responsibility to ensure adequate health care for the sick. 3. It should/should not be governments’ responsibility to ensure a reasonable standard of living for the old. 4. It should/should not be governments’ responsibility to ensure a reasonable standard of living for the unemployed.

Methods applied in the empirical analysis include an examination of frequencies, mean values, and path analysis. Descriptive analyses of frequencies are applied to explore the size of high-risk groups in different countries, as well as in order to examine differences in the evaluation of task performance and in attitudes toward the responsibilities of government within and among the countries. Path analysis is utilized to examine the causal relationships between high-risk groups, perceived risks, evaluations, and attitudes. This method allows us to examine the mediating effects of perceived risks and evaluations of welfare state task performance on the attitudes of the two types of high-risk groups identified compared to the rest of the population and to explore possible differences among the paths in the respective regimes. r e su lt s The overlap in different welfare regimes between various potential and transfer-dependent risk groups and those perceiving the future risks of unemployment and economic hardship as being high, respectively, is shown

68

Blomberg, Kallio, Kangas, Kroll, and Niemelä ta bl e 3.1 Perceived future risk by high-risk group (percentage)

unemployment Sick Economic strain Immigrant Transfer-dependent groups economic hardship Sick Economic strain Immigrant Transfer-dependent groups

All countries

AngloSaxon

Nordic

Continental

Southern

Eastern

17.1 31.8 26.5

15.6 36.4 22.4

11.0 24.3 19.8

13.3 31.2 22.2

18.9 32.2 33.1

20.3 31.9 26.4

44.4

52.6

38.9

40.5

44.6

47.7

45.2 70.8 38.4

39.5 73.5 28.4

14.9 56.7 18.9

22.4 54.0 21.8

51.4 67.3 40.0

63.5 75.9 61.7

61.9

71.0

40.8

49.0

59.8

81.7

in Table 3.1. Among the potential high-risk groups, perceptions of future risks are most common in the Eastern and Southern European countries, followed by the risk groups in the Anglo-Saxon model. Potential high-risk groups in the Nordic countries, rather closely followed by risk groups in Continental Europe, do not in general perceive their risk of future unemployment or economic hardship as being particularly high. However, even in the Nordic countries and Continental Europe, those having economic difficulties at present very often perceive that their future economic situation is going to be difficult. Further, transfer-dependent groups in all welfare regimes perceive their risk of future unemployment and economic hardship as substantial. Nevertheless, differences among the risk groups in different welfare regimes appear on this indicator as well: The figures for transfer-dependent groups in the Nordic countries, again closely followed by figures for transfer-­ dependent groups in Continental Europe, are lower than the figures for Southern Europe and much lower than the figures for the Anglo-Saxon and Eastern European risk groups. It can also be noted that differences among regimes are more pronounced in the case of the perceived risk of future economic hardship than in the case of the perceived risk of future unemployment, regardless of the risk group studied. The data presented in Table 3.2 show that people belonging to various high-risk groups evaluate welfare state task performance differently



69

Attitudes Among High-Risk Groups

ta bl e 3.2 Evaluation of country’s social benefits and services by high-risk group (percentage) All countries

AngloSaxon

Nordic

Continental

prevented widespread poverty: agree or strongly agree Sick 53.0 56.2 63.8 71.2 Economic strain 47.6 52.9 67.2 64.4 Immigrant 55.6 55.6 67.6 70.8 Transfer-dependent groups 57.3 53.4 69.4 67.9 Risk groups’ mean 53.4 54.5 67.0 68.6 lead to a more equal society: agree or strongly agree Sick 45.7 37.3 61.5 58.2 Economic strain 40.3 40.0 59.5 52.8 Immigrant 51.7 50.8 67.4 63.3 Transfer-dependent groups 50.1 40.4 64.0 59.7 Risk groups’ mean 47.0 42.1 63.1 58.5 both (mean value, scale 1–5) Sick 3.25 Economic strain 3.14 Immigrant 3.34 Transfer-dependent groups 3.35 Risk groups’ mean 3.27

Southern

Eastern

55.1 57.8 51.9

40.7 38.4 41.4

50.6 53.9

47.9 42.1

55.0 57.1 51.0

32.4 28.4 35.7

53.0 54.0

36.1 33.2

3.19 3.25 3.38

3.55 3.53 3.64

3.55 3.43 3.61

3.40 3.46 3.30

2.97 2.89 3.00

3.27 3.27

3.62 3.59

3.53 3.53

3.34 3.38

3.08 2.99

in different welfare regimes. Individuals belonging to both types of highrisk groups—potential high-risk groups or transfer-dependent groups— evaluate the performance of the welfare state most positively in the Nordic countries and in Continental Europe. Risk groups belonging to the Southern European, the Anglo-Saxon, and particularly the Eastern European models have a far more pessimistic view of welfare state task performances. Further, percentages among the different risk groups vary as well. For example, respondents with economic difficulties are somewhat less inclined than other risk groups to agree with the statement that social benefits and services in one’s country have led to an equal society. However, risk groups’ mean values for the same questions do not differ substantially from one another. Finally, in every regime, risk groups more often agree with the evaluation that the social benefits and services provided by their country’s welfare system have succeeded in preventing widespread poverty than they do regarding the evaluation that the benefits and services in their country have led to a more equal society.

70

Blomberg, Kallio, Kangas, Kroll, and Niemelä

Table 3.3 shows the attitudes of high-risk groups toward the welfare state’s responsibilities. Consistent with earlier findings (Svallfors, 2003; 2004; 2006), attitudinal differences among regimes (as well as among different risk groups) are smallest concerning the government’s responsibility to ensure health care for the sick and a reasonable standard of living for the old. Differences among regimes (as well as among different types of risk groups) are clearer concerning the government’s responsibility to ensure a job for everyone who wants one and to ensure a reasonable standard of living for the unemployed. High-risk groups in Eastern Europe and Southern Europe are most in favor of welfare state responsibility, followed by risk groups in the Nordic and Anglo-Saxon countries, while high-risk groups in Continental Europe are least in favor of welfare state responsibility. Concerning the questions on the desired tasks of the welfare state, respondents whose economic situation is difficult or who live on social benefits usually have the highest mean scores, while immigrants display the lowest mean scores on these questions. This is true for all the groups included and all the welfare regimes, with the exception of immigrants in the Eastern European regime. Finally, we look at the results from the path analyses. Here, the potential high-risk groups and transfer-dependent groups, respectively, are analyzed separately for every welfare state regime. Perceived risks, welfare state task evaluation, and attitudes toward the responsibilities of the government are treated as continuous variables, while the other variables are dichotomous. In contrast to regression analyses, path analysis can handle the effect of two or more dependent variables. The path coefficients (Beta values) can be used to show direct and indirect effects (paths) in the model. The direct effect between the independent and dependent variable is displayed in Table 3.4 and Figure 3.2, in which significant effects in each welfare regime are shown by arrows following the conceptual model in Figure 3.1. The indirect effects (the path coefficients for each path) are also presented in Table 3.4. All analyses are performed first with the inclusion of the dependent variable and all the independent variables, and then a new model is performed that includes only those effects between variables that were statistically highly significant (greater than .001) in the initial model, with the exception of the Anglo-Saxon model, where effects greater than .05 are included due to the limited size of its data. The results from the final

t a bl e 3.3 Welfare attitudes toward the responsibilities of government by high-risk group, mean values All countries

AngloSaxon

Nordic

Continental

Southern

government should ensure a job for everyone who wants one (1–10) Sick 7.20 6.28 6.37 5.82 7.66 Economic strain 7.73 6.45 6.71 6.46 7.81 Immigrant 7.10 6.10 6.48 5.80 7.52 Transfer-dependent groups 7.27 6.50 6.67 6.38 8.26 Risk groups’ mean 7.33 6.33 6.56 6.12 7.81 government should ensure adequate health care for the sick (1–10) Sick 8.82 8.89 8.82 8.17 8.91 Economic strain 8.85 8.81 8.78 8.27 8.78 Immigrant 8.70 8.29 8.59 8.03 8.94 Transfer-dependent groups 8.65 8.67 9.71 8.34 9.07 Risk groups’ mean 8.76 8.67 8.98 8.20 8.93

Eastern 7.34 7.99 8.17 7.92 7.86 9.05 8.98 9.13 8.62 8.95

government should ensure a reasonable standard of living for the old (1–10) Sick 8.68 8.72 8.55 7.79 8.85 9.02 Economic strain 8.83 8.74 8.64 8.07 8.72 9.02 Immigrant 8.53 8.30 8.37 7.60 8.80 9.16 Transfer-dependent groups 8.49 8.61 8.45 7.94 8.92 8.68 Risk groups’ mean 8.63 8.59 8.50 7.85 8.82 8.97 government should ensure a reasonable standard of living for the unemployed (1–10) Sick 7.25 6.24 7.39 6.28 7.82 Economic strain 7.60 6.35 7.57 6.46 7.94 Immigrant 7.24 6.10 7.13 6.33 7.56 Transfer-dependent groups 7.53 6.63 7.70 6.91 8.11 Risk groups’ mean 7.41 6.33 7.45 6.50 7.86 total mean (0–40) Sick Economic strain Immigrant Transfer-dependent groups Risk groups’ mean

7.48 7.64 7.94 7.90 7.74

31.70 32.80 31.30

29.96 30.20 28.45

30.89 31.52 30.37

27.92 29.06 27.57

32.87 33.04 32.54

33.21 33.41 34.10

31.69 31.87

30.35 29.74

31.48 31.07

29.29 28.46

33.87 33.08

32.91 33.41

n o t e : 0 = should not be the government’s responsibility at all; 10 = should be entirely the government’s responsibility.

Continental, Anglo-Saxon, and Southern European welfare regimes

High-risk groups

Perceived future risks

• Potential high-risk • Transfer-dependent groups

x x

Welfare state attitudes: scope and responsibilities of the welfare state

Evaluation of task performance

x = Not in the Southern European WR

Nordic welfare regime

High-risk groups

Perceived future risks

• Potential high-risk • Transfer-dependent groups

Welfare state attitudes: scope and responsibilities of the welfare state

Evaluation of task performance Eastern European welfare regime

High-risk groups

Perceived future risks

• Potential high-risk • Transfer-dependent groups

Welfare state attitudes: scope and responsibilities of the welfare state

Evaluation of task performance = Potential high-risk groups = Transfer-dependent groups Figure 3.2  The studied paths in different welfare regimes compared to the rest of the population s o u r c e : European Social Survey 2008.

.25 * .13 = .01

.11 * .18 = .02

ns

.10 * ns

.06

.19

.27 * ns

.24 * .09 = .02

.08

.05

.30 * .09 = .03

.25 * .05 = .01

ns

.06

.27 * ns

.35 * .05 = .02

.10

.08

.31 * .06 = .02

.09

2. Group-riskattitude

ns * –.04

–.04 * ns

ns * .02

ns * .02

ns * .05

ns * .05

ns * .11

ns * .11

ns * .12

ns * .12

3. Grouptask-attitude

.11 * –.05 * –.04 = .00

.25 * –.04 * ns

.27 * –.02 * .02 = .00 .10 * –.03 * .02 = .00

.30 * –.07 * .05 = –.00 .24 * –.07 * .05 = .00

.25 * ns * .11

.27 * ns * .11

.31 * –.04 * .12 = –.00 .35 * –.04 * .12 = –.00

4. Group-risktask-attitude

.18

.13

ns

ns

.09

.09

.05

ns

.05

.06

5. Riskattitude

–.05 * –.04 = .00

–.04 * ns

–.03 * .02 = .00

–.02 * .02 = –00

–.07 * .05 = –00

–.07 * .05 = –00

ns * .11

ns * .11

–.04 * .12 = –00

–.04 * .12 = –00

6. Risk-taskattitude

–.04

ns

.02

.02

.05

.05

.11

.11

.12

.12

7. Taskattitude

2 + 4 = .02

1 + 2 = .20

1 + 4 = .06

1 + 4 = .05

1 + 2 + 4 = .10

1 + 2 + 4 = .09

2 = .01

1 = .08

1 + 2 + 4 = .12

1 + 2 + 4 = .11

Total effect from groups to attitude

n o t e : ns = not significant (omitted relationship); column 1 pertains to direct effect, and columns 2, 3, and 4 pertain to indirect effects. Columns 5–7 are included in 2–4.

eastern Potential highrisk group Transfer group

southern Potential highrisk group Transfer group

continental Potential highrisk group Transfer group

nordic Potential highrisk group Transfer group

anglo-saxon Potential highrisk group Transfer group

1. Groupattitude

ta bl e 3. 4 Path-analysis for high-risk groups in different welfare regimes

74

Blomberg, Kallio, Kangas, Kroll, and Niemelä

models are presented in Table 3.4 and Figure 3.2, which show the significant direct effects. The RMSEA (goodness-of-fit) is less than .05 for all models, except for the Anglo-Saxon model, for “transfer-dependent groups,” where RMSEA = 0.1, thus supporting the adequacy of the models. The path analyses for the Continental and Anglo-Saxon welfare regimes resulted in similar (see Figure 3.2 and Table 3.4) paths for the highrisk groups studied. In these welfare state regimes, potential high-risk and transfer-dependent groups are more in favor of the welfare state than other groups, and they also perceive their personal risks of facing future economic difficulties and unemployment as being greater than do other groups. Further, perceived risks function as an intermediating factor between “objective” risk and welfare state attitudes: Those who perceive their economic future as endangered are more in favor of state responsibility than those who do not expect their future situation to be difficult. There is also an indirect effect, as indicated by the path from risk groups via perceived risk and evaluation of welfare state task performance to welfare state support. Analyses of the Southern European regime reveal a rather similar picture compared to the other regimes (see Figure 3.2 and Table 3.4). However, in contrast to the other models, perceived risk is not connected at all to attitudes. Results for the Nordic countries also display a partly similar pattern: Potential high-risk groups have more positive attitudes toward welfare state responsibility than do “nonrisk” groups. However, in the Nordic regime, no direct effect of belonging to a transfer-dependent group on welfare state support can be detected. Instead, perceived risks of future economic and employment problems seem to function as an intermediating factor between transfer dependency and welfare state support: Those respondents who perceive their future situation as gloomier display a somewhat stronger support than those who perceive their future as being brighter. Thus, perceived risks can be seen as an intermediating factor affecting the welfare state attitudes of transfer-dependent groups. It is also interesting to note that in the Nordic regime, there is no causal path from potential high-risk groups via perceived risk to welfare state attitudes and, consequently, no path from risk groups via perceived risk or evaluation of welfare state task performance to welfare state support (see Figure 3.2). The path analyses concerning potential high-risk groups in the Eastern European welfare regime show that potential high-risk groups are clearly



Attitudes Among High-Risk Groups

75

more in favor of state responsibility than other respondents and, further, that they perceive their personal risk of future hardship as being greater than do other groups. Furthermore, perceived risks function as an intermediating factor between “objective” risk and welfare state attitudes. In contrast to the path analyses for the other regimes, a direct negative connection can be seen between high-risk groups and a (negative) evaluation of welfare state task performance. The path analyses for transfer-dependent groups show similarities with the analyses for the Nordic regime in that there is no direct effect between belonging to the transfer-dependent groups and the support for the welfare state (see Figure 3.2 and Table 3.4). The analyses also allow us to detect some other “general” results: Perceived risk per se is in several analyses an equally significant independent variable, as is a more “objective” high-risk status. The analyses also show that evaluations of welfare state performance are not connected with highrisk status (except for the connection between potential high-risk groups and negative evaluations in the Eastern European regime). c on c l usion This chapter analyzes different types of risk groups and their attitudes toward the welfare states’ responsibilities. If we summarize the results of our empirical analyses by focusing on the perceptions of different (“objectively” discerned) high-risk groups, we find that the Nordic regime, closely followed by the Continental regime, has succeeded best in creating a “subjective” sense of security against future social risks, which seems concordant with the “factual” performance of welfare states in this respect. Also, evaluations of welfare state task performance seem to correspond with factual welfare state performance in terms of poverty and other social problems: High-risk groups in the Nordic countries and in Continental Europe are most satisfied with the performance of the welfare state, while high-risk groups belonging to the Southern European, the Anglo-Saxon, and particularly the Eastern European models have a far more negative view of the task performance of the welfare state. Regarding welfare state attitudes, our results show that it is the highrisk groups in Eastern and Southern Europe who stress government responsibility to the greatest extent, followed by the high-risk groups in the Nordic regime. High-risk groups in the Anglo-Saxon and, above all, the

76

Blomberg, Kallio, Kangas, Kroll, and Niemelä

Continental models are the least in favor of government responsibility. On the basis of population data from 14 countries and responses to the questions used in this chapter, Pfeifer (2009) draws the conclusion that “a generous welfare state with high expenditure levels provides comprehensive protection . . . leading to “saturation” in the population. There is no strong demand for further state involvement.” Regarding high-risk groups, such a straightforward interpretation of the interplay between welfare state generosity and attitudes does not necessarily seem to hold true. Since the strongest support for state responsibility can be found somewhat paradoxically in regimes that provide both the weakest and the strongest protection (according to Fridberg and Kangas, 2008), respectively, it seems as if other factors are at play—factors that might even be divergent in different regimes. It should also be remembered that even though we can find differences among risk groups in different welfare regimes, state responsibility for welfare is strongly supported among high-risk groups in all European welfare regimes. We also looked at whether there are differences in the mediating connections between high-risk groups and welfare state support in different welfare regimes. The mediating factors studied were “perceived risks” and “evaluations of welfare state task performance.” A number of conclusions can be drawn in this respect. First, potential high-risk groups (the sick, those experiencing economic strain, and immigrants) are more in favor of welfare state responsibility than “nonrisk groups” in all welfare regimes. The connection is significant in all welfare models, although it is particularly strong in the Eastern European welfare regime. The latter connection does not stand out as too surprising, since risk groups seem, on the basis of earlier research, to be particularly vulnerable in the Eastern European regime (see Fridberg and Kangas, 2008). Also, respondents on social benefits (seen to represent “transferdependent groups”) are more in favor of state responsibility for welfare than non-high-risk groups in some of the welfare regimes studied, but not in the Nordic and Eastern European regimes. One explanation here could be that benefit recipients in the Nordic and Eastern European welfare regimes are more “similar” to the rest of the population in comparison to recipients in the other welfare regimes. In the Eastern European welfare regime, recipients’ feelings about the benefits in question would, then, not differ substantially from other respondents, since both economic and social vulnerability



Attitudes Among High-Risk Groups

77

also are common among those not receiving these benefits. In the Nordic countries, one could also speculate on whether the results, to the extent that attitudes are the product of self-interests, are because the compositions of the transfer-dependent groups in question are not as different from one another as they are in the general population in the Nordic regime but because of the universal and fairly generous character of many of the benefits. It is also interesting to note that respondents who depend on social benefits do not have particularly critical views of welfare state task performance compared to potential high-risk groups or the population at large. Concerning our mediating variables, “perceived risk” seems to function as a factor affecting the attitudes of risk groups in some, but not all, of the welfare regimes studied. It seems clear that perceptions of future risks affect attitudes in the Continental, the Eastern European, and (with some reservations due to sample size) the Anglo-Saxon welfare regimes: Those who expect to encounter social problems in the future support welfare state responsibilities to a greater extent than do those who have more positive perceptions about their future. As we saw, in the Nordic model, perceived risk does not affect the attitudes of potential high-risk groups toward the welfare state, but here, too, we find that “perceived risk” functions as a mediating factor for those on social benefits. Overall, perceived risk per se seems in many respects to be an equally important determinant for attitudes as belonging to a more objectively defined risk group. Finally, concerning our other mediating factor—“evaluations,”—we must conclude that our assumption that risk groups would more often than others evaluate the welfare state’s abilities to fulfill some of its fundamental tasks in a negative way does not seem to hold true in any of the very different welfare regimes studied (with the exception of a rather weak connection regarding potential high-risk groups in the Eastern European model). In other words, those with scarce resources and with personal experiences from welfare services and benefits, often (presumably) selective ones, are not more skeptical toward welfare state performance than others. In sum, our analyses show that it would seem fruitful to include a perspective of risk groups when studying attitudes toward welfare state responsibilities and also to include perceptions of the future in the equation. However, since we concentrated on a less investigated type of independent variable (along with some intermediary variables), it is not surprising that we can explain only a small part of the variance in attitudes. 5 The

78

Blomberg, Kallio, Kangas, Kroll, and Niemelä

­ echanisms related to this type of independent variable seem to follow the m same logic concerning the Continental European, the Anglo-Saxon, and, with some reservation, the Southern European welfare regimes, while it seems partly divergent in the Nordic and Eastern welfare regimes. no t e s 1.  In the study by Fridberg and Kangas (2008), 21 countries were included, while our study involved 26 countries (see Web Appendix Table A3.2). 2.  These groups were called “vulnerable groups,” which included the poorly educated, the sick, the poor, the unemployed, the aged, those who claimed to feel unsafe or felt they had no friends, and immigrants. 3.  Respondents dependent on pensions cannot as a rule be assumed to constitute a “potential high-risk group” as defined here. 4.  Although the redistributional ambitions vary greatly among welfare regimes, we believe also the latter question to be relevant from our perspective, since all welfare systems can be regarded to embrace a certain ambition of leveling inequalities in society. The varying extent and nature of these ambitions could at the same time be expected to be reflected in the attitudes of the groups in focus of this study. 5.  However, separate path analyses (not shown here), including traditional individual-level variables (age, gender, etc.), do not substantially increase the total explained variance. By including “regime” as an explanatory variable in a regression model, the explained variance would have increased substantially. For example, regression analyses (not shown here) show that “regime” explains over 10 percent of the variance concerning “welfare state attitudes.”

r efer ences Andress, H.-J., and T. Heien. 2001. “Four Worlds of Welfare State Attitudes? A Comparison of Germany, Norway, and the United States.” European Sociological Review 17: 337–56. Arts, W., and J. Gelissen. 2001. “Welfare States, Solidarity and Justice Principles: Does the Type Really Matter?” Acta Sociologica 44: 283–99. Blekesaune, M. 2007. “Economic Conditions and Public Attitudes to Welfare Policies.” European Sociological Review 23: 393–403. Blekesaune, M., and J. Quadagno. 2003. “Public Attitudes Toward Welfare State Policies.” European Sociological Review 19: 415–27. Blomberg, H. 1999. “Do Cutbacks Pay Off? Perceived Changes in the Standard of Municipal Services and Attitudes Towards Services Among Citizens and



Attitudes Among High-Risk Groups

79

Municipal Decision Makers in Finland.” International Journal of Social Welfare 8: 206–20. Boeri, T., A. Börsch-Supan, and G. Tabellini. 2001. “Would You Like to Shrink the Welfare State? A Survey of European Citizens.” Economic Policy 16: 7–50. Edlund, J. 2006. “Trust in the Capability of the Welfare State and General Welfare State Support: Sweden 1997–2002.” Acta Sociologica 49: 395–417. Fridberg, T., and O. Kangas. 2008. “The Welfare State, Poverty and Social Exclusion.” In Nordic Social Attitudes in a European Perspective, edited by H. Ervasti, T. Fridberg, M. Hjerm, and K. Ringdal, 22–47. Cheltenhamn, Northampton: Edward Elgar. Gelissen, J. 2000. “Popular Support for Institutionalised Solidarity: a Comparison Between European Welfare States.” International Journal of Social Welfare 9: 285–300. Gevers, J., J. Gelissen, W. Arts, and R. Muffels. 2000. “Public Health Care in the Balance: Exploring Popular Support for Health Care Systems in the European Union.” International Journal of Social Welfare 9: 301–21. Jaeger, M. M. 2006. “What Makes People Support Public Responsibility for Welfare Provision: Self-Interest or Political Ideology? A Longitudinal Approach.” Acta Sociologica 49: 321–38. ———. 2007. “Are the ‘Deserving Needy’ Really Deserving Everywhere? Crosscultural Heterogeneity and Popular Support for the Old and the Sick in Eight Western Countries.” In Social Justice, Legitimacy and the Welfare State, edited by S. Mau and B. Veghte, 73–94. Aldershot, UK: Ashgate Publishers Ltd. Kallio, J. 2008. “Kansalaisten asennoituminen hyvinvointivaltion uudelleenmuotoiluun” [“Citizens’ attitudes towards the reformulation of the welfare state”]. Sosiologia 45: 3–20. Korpi, W., and J. Palme. 1998. “The Paradox of Redistribution and Strategies of Equality: Welfare State Institutions, Inequality, and Poverty in the Western Countries.” American Sociological Review 63: 661–87. Kumlin, S. 2007. “Overload or Undermined? European Welfare States in the Face of Performance Dissatisfaction.” In The Political Sociology of the Welfare State: Institutions, Social Cleavages, and Orientations, edited by S. Svallfors, 80–116. Stanford, CA: Stanford University Press. Kumlin, S., and B. Rothstein. 2005. “Making and Breaking Social Capital: The Impact of Welfare-State Institutions.” Comparative Political Studies 38: 339–65. Larsen, C. A. 2006. The Institutional Logic of Welfare Attitudes: How Welfare Regimes Influence Public Support. Aldershot, UK: Ashgate Publishers Ltd. Leibfried, S. 2000. “Towards a European Welfare State?” In The Welfare State Reader, edited by C. Pierson and F. G. Castles, 190–210. Cambridge, UK: Polity Press.

80

Blomberg, Kallio, Kangas, Kroll, and Niemelä

Mau, S. 2003. The Moral Economy of Welfare States: Britain and Germany Compared. London: Routledge. Meltzer, A. H., and F. R. Scott. 1981. “A Rational Theory of the Size of Government.” The Journal of Political Economy 89: 914–27. Muuri, A. 2010. “The Impact of the Use of the Social Welfare Services or Social Security Benefits on Attitudes to Social Welfare Policies.” International Journal of Social Welfare 19: 182–93. Pfeifer, M. 2009. “Public Opinion on State Responsibility for Minimum Income Protection: a Comparison of 14 European Countries.” Acta Sociologica 52: 117–34. Rehm, P. 2007. “Who Support the Welfare State? Determinants of Preferences Concerning Redistribution.” In Social Justice, Legitimacy and the Welfare State, edited by S. Mau and B. Veghte, 47–72. Aldershot, UK: Ashgate Publishers Ltd. Rothstein, B. 2000. “Trust, Social Dilemmas and Collective Memories.” Journal of Theoretical Politics 12: 477–502. Soss, J. 1999. “Welfare Application Encounters.” Administration & Society 31: 50. Svallfors, S. 1995. “The End of Class Politics? Structural Cleavages and Attitudes to Swedish Welfare Policies.” Acta Sociologica 38: 53–74. ———. 1997. “Worlds of Welfare and Attitudes to Redistribution: a Comparison of Eight Western Nations.” European Sociological Review 13: 283. ———. 2003. “Welfare Regimes and Welfare Opinions: A Comparison of Eight Western Countries.” Social Indicators Research 64: 495–520. ———. 2004. “Class, Attitudes and the Welfare State: Sweden in Comparative Perspective.” Social Policy & Administration 38: 119–38. ———. 2006. The Moral Economy of Class: Class and Attitudes in Comparative Perspective. Stanford, CA: Stanford University Press. Taylor-Gooby, P. 1985. Public Opinion, Ideology and State Welfare. London: Routledge & Kegan Paul. Van Oorschot, W. 2010. “Public Perceptions of the Economic, Moral, Social and Migration Consequences of the Welfare State: an Empirical Analysis of Welfare State Legitimacy.” Journal of European Social Policy 20: 19. Varian, H. 1980. “Redistributive Taxation as Social Insurance.” Journal of Public Economics 14: 49–68. Wendt, C., J. Kohl, M. Mischke, and M. Pfeifer. 2010. “How Do Europeans Perceive Their Healthcare System? Patterns of Satisfaction and Preference for State Involvement in the Field of Healthcare.” European Sociological Review 26: 177.

chapter four

A Normative Approach to Welfare Attitudes Christian Staerklé, Tiina Likki, and Régis Scheidegger

Citizens have many different reasons to support or oppose government responsibility in the domain of social welfare. They may want to restrict social rights for undeserving or deviant beneficiaries, they may want to safeguard their own financial interests, they may want to exclude immigrants and other outgroups from welfare coverage, or they may express solidarity with fellow citizens in precarious life situations. Such reasons are likely to determine the extent to which citizens consider government institutions to be responsible for handling social issues such as inequality, unemployment, and public health. The match between citizens’ expectations and actual government intervention in social issues then shapes the legitimacy of welfare arrangements, thus providing the social foundation of collective responsibility. We use the generic term of normative beliefs to subsume individually held, but socially shared, values, perceptions, and expectations that provide normative support for welfare attitudes, understood as policy preferences in the domain of government responsibility and redistribution of economic resources. Four issues related to normative beliefs are addressed in this chapter. The first issue concerns the structure of the beliefs that are associated with welfare attitudes. We suggest that four basic categories of beliefs account for the normative foundation of welfare intervention, organized as a function of different conceptions of social order. Second, we investigate the social contexts that give rise to these normative beliefs by studying how level of education and perceived vulnerability impact normative beliefs and welfare attitudes. A third question concerns the nature of the relationship between normative beliefs and welfare attitudes. We propose that social 81

82

Staerklé, Likki, and Scheidegger

psychological processes of differentiation and boundary definition provide the normative foundations of welfare legitimacy. And fourth, we investigate contextual variation of these beliefs by examining how membership in different European countries moderates the role of normative beliefs as organizing principles of welfare attitudes. In order to tackle these issues, we develop a normative approach to welfare attitudes that sets out to explain the psychological and cultural processes that account for the relationship among social position, normative beliefs, and policy attitudes. Two empirical strategies are used: First, we propose a mediation model of welfare attitudes with individual social position as the independent variable, welfare attitudes as the outcome variable, and theoretically defined normative beliefs as mediating variables between social position and policy attitudes. Second, we investigate the relationships between normative beliefs and welfare attitudes as a function of national and institutional contexts, first by comparing the impact of normative beliefs as a function of welfare regimes and second through multilevel modeling. c u lt u r a l a p p roac h e s t o w e l fa r e at t i t u de s The question of public legitimacy of social government intervention has attracted much recent scholarly interest (e.g., Andreß and Heien, 2001; Aalberg, 2003; Blekesaune and Quadagno, 2003; Mau, 2003; Svallfors, 2006; Van Oorschot, 2006; Mau and Veghte, 2007; Staerklé et al., 2007). Many accounts of perceived government legitimacy rely on the assumption that self-interest is a key driving force of welfare attitudes (e.g., Hall and Soskice, 2001; Rehm, 2007). Yet, this rationalist view of human motivation whereby citizens support only those policies they benefit from has been challenged by researchers who emphasize the importance of normative, moral, and cultural factors in explaining political attitudes (e.g., Sears and Funk, 1990; Mau, 2004; Van Oorschot, 2006). A cultural approach to welfare attitudes, for example, suggests that members of a cultural community elaborate, share, and perpetuate meanings of the role of government in their everyday lives (see Chamberlayne et al., 1999; Clarke, 2004; Van Oorschot, 2006). The encompassing Nordic welfare states have, for example, been explained with the historically prevailing Lutheran model of the relationship between a caring central authority and citizens who willfully subject



A Normative Approach to Welfare Attitudes

83

themselves to this authority (Soerensen, 1998). Self-interest itself may be a cultural norm rather than an inherently individual motivation (Miller, 1999b). In this view, self-interest is perceived by citizens (and maybe also by large parts of the “elite” such as policy makers and researchers) of the Western world not only as the most “normal” explanation of human behavior (“Everybody is self-interested”) but also as the most appropriate justification (“It’s good to be self-interested”). In addition to studying the normativity of behaviors, a cultural analysis of policy attitudes also focuses on widely shared beliefs about beneficiary groups and cultural patterns of perceived deservingness. In such analyses, the impact of normative images of the poor (Geremek, 1994) and perceptions of causes of poverty are studied as policy attitude determinants (e.g., Van Oorschot, 2000; Sachweh et al., 2007). One widespread cultural belief about the origins of social hardship is welfare dependency, according to which government intervention is actually the problem rather than the solution to the dire consequences of life course misfortunes. In line with neoconservative ideology (e.g., Murray, 1984), overly generous welfare policies are claimed to lead to attitudes of passivity, immorality, selfishness, consumerism, and hedonism. Notwithstanding the fact that research has not been able to empirically uncover such deleterious effects of welfare states, cultural beliefs about welfare dependency remain powerful rhetorical devices to justify welfare retrenchment. This cultural focus on welfare attitudes reflects contemporary value pluralism (e.g., Young, 2007), which also guides our normative approach. In a far-reaching analysis on the cultural values underlying welfare attitudes, Feldman and Zaller (1992) show that political cultures of countries provide people with ideological values that often conflict with one another. In their analysis of policy attitudes in the United States, they describe a central conflict between the cultural values of individualism and humanitarianism, which many authors have cited—for example, in terms of conflicts between capitalism versus democracy (McClosky and Zaller, 1984), the protestant work ethic versus egalitarianism-humanitarianism (Katz and Hass, 1988), and freedom versus equality (Rokeach, 1973). The value pluralism of contemporary political cultures gives rise to widespread attitudinal ambivalence that is therefore not the result of lacking political competence. According to Feldman and Zaller (1992), “Conflict and ambivalence are interpreted not as confusion, inconsistency, or lack of

84

Staerklé, Likki, and Scheidegger

sophistication, but as a problem of reconciling the multiple values, beliefs, and principles simultaneously present in the political culture.”A comparative analysis of welfare attitudes should take into account that contemporary political cultures across different national contexts are characterized by a large array of competing cultural values and models of social organization (Hochschild, 1981). Such value configurations represent normative reference markers that feed and orient lay political thinking (Young, 2007) and are presumably present in any democratic society as “ideological dilemmas” (Billig, 1989). Yet, a cultural approach may be prone to a conceptual pitfall when individuals are seen as unidirectionally “influenced” by cultural beliefs that are supposedly shared to an equal degree by members of a given social group. Instead, it seems more fruitful to consider such cultural beliefs as the product of everyday practice, social experience, and membership in social groups, and thus as organizing principles of social relations and social order (Staerklé, 2009). a nor m at i v e a p p roac h t o w e l fa r e at t i t u de s The normative approach to welfare attitudes we offer in this chapter is based on classic work in European social psychology (Moscovici, 1961/1976; Tajfel, 1978; 1981; Doise, 1986), in particular social representations theory, which was initiated in the 1960s by Serge Moscovici (see Doise et al., 1993; Moscovici, 2000; Augoustinos et al., 2006; Bauer and Gaskell, 2008). At the most general level, social representations are collective definitions of reality, made up by shared normative beliefs that enable communication, consent, and dissent within and among social groups (Elcheroth et al., 2011; Staerklé et al., 2011). Social representations are not “external” to individuals in the sense that they would exert an “influence” on them. Rather, individuals’ thinking is based on representations without which they could not make sense of social realities and communicate with one another about these realities. Social representations therefore organize social relationships by providing normative reference knowledge—expressed as “values,” “beliefs,” or “identities”—that make up the symbolic environment of citizens and their various ingroups. In a social representational account, welfare attitudes are analyzed as expressions of political lay thinking (Staerklé, 2009) that is fed by shared



A Normative Approach to Welfare Attitudes

85

understandings regarding social order and the just distribution of resources in a society (see Mau, 2004). Beliefs concerning the perceived entitlement of potential beneficiary groups are a key component of such political lay thinking. Normative beliefs are thus seen as collectively elaborated and shared knowledge that reflects different models of organizing a society— that is, conceptions of social order. In its most basic sense, the notion of social order refers to the norms, rules, and laws that define our living together as a group, community, or society. Welfare attitudes should therefore be understood in relation to the normative beliefs that underlie social order and make the social bond possible (Van Oorschot and Komter, 1998). But because democratic dissent can always challenge an established way to organize the society, any order may be contested, and its stability can never be taken for granted. Such competing models of social order prescribe different types of relationships among the individual members, their reference groups, and the institutions and authorities that govern them. Some forms of social order, for example, emphasize the collective interests of the group to which individual rights and duties are subordinated, whereas others value individual rights that have priority over community rights. It is our assumption that citizens refer to such competing models of social order when taking up a stance toward social government intervention. Hence, we view normative beliefs such as egalitarianism, authoritarianism, or welfare dependency as reflecting specific conceptions of social order debated in the public sphere, rather than as individual values or even personality dimensions. The differential endorsement of these beliefs reflects the support or the rejection of normative models of social order put forward by political groups, policy makers, and other voices heard in the political debate. Normative beliefs therefore function as organizing principles (Bourdieu, 1984; Doise et al., 1993) of welfare attitudes that crystallize competing definitions of social order. The relative individual importance of these representations of social order as foundations of welfare attitudes varies according to one’s position in the social hierarchy. In our mediation model, we focus on individual variation of conceptions of social order as a function of education and perceived vulnerability (see Lewin-Epstein et al., 2003). The normative beliefs underlying welfare attitudes will be studied on the basis of the model of lay conceptions of social order (Staerklé, 2009). Originally developed to analyze the Swiss Welfare State Survey (Staerklé

86

Staerklé, Likki, and Scheidegger

et al., 2007), this model distinguishes four domains of shared normative beliefs called moral order, free market, social diversity, and structural inequality. These domains refer to four prototypical forms of social order that allow integrating various belief items contained in ESS 4 into a single, theory-based framework of normative beliefs. On the individual level, these conceptions are not mutually exclusive but rather combine in multiple ways. The model assumes that the basic psychological process underlying normative beliefs is differentiation between social categories (Tajfel, 1978). Through differentiation, individuals subjectively define boundaries between social groups (Lamont and Molnar, 2002), in particular the boundaries between positively connoted, deserving social categories and negatively connoted, undeserving categories. Differentiation thereby translates value conflicts at a societal level into a psychological process that sets apart positively evaluated ingroups and negatively evaluated outgroups, organizes perceptions of cleavages in the society, and ultimately determines perceived entitlements of social categories. The psychological process of differentiation is also an essential element of political theories that emphasize the role of social antagonisms in democratic politics (e.g., Mouffe, 1993). These theories suggest that policy decisions are basically strategies to regulate relationships between groups and individuals representing contrasting values. In our view, perceived antagonisms between groups are at the core of normative beliefs. The four conceptions of social order—moral order, free market, social diversity, and structural inequality—represent the outcome of the crossing of two fundamental polarities: The first polarity opposes normative versus categorical forms of differentiation, and the second polarity defines whether differentiation is based on identity (symbolic) versus positional (material) criteria (Table 4.1). Normative differentiation (moral order and free market) establishes boundaries between norm-conforming and norm-violating individuals, while categorical differentiation (social diversity and structural inequality) creates boundaries between groups defined by ascribed group membership (Duckitt, 2001; Kreindler, 2005). In the process of normative differentiation, social cleavages are understood as the result of willful individual actions—for example, in normative beliefs that imply a distinction between lazy and hard-working people, such as welfare dependency. The meaning of these actions is defined with respect to perceived conformity



A Normative Approach to Welfare Attitudes

87

ta bl e 4.1 Model of lay conceptions of social order Social identity

Social position

normative differentiation Principle of categorization Core antagonisms Principle of social regulation Welfare policies

Moral order Morality “Good” and “bad” Conformity, similarity Private support, charity

Normative belief

Authoritarianism, distrust

Free market Productivity “Winners” and “losers” Equity, self-interest Private responsibility, insurance Welfare dependency

categorical differentiation Principle of categorization Core antagonisms Principle of social regulation Welfare policies Normative beliefs

Social diversity Social heterogeneity Majority and minority Intergroup differentiation Majority favoritism vs. group rights Ethnocentrism vs. multiculturalism

Structural inequality Social class, status Dominants and subordinates Inequality management Elite favoritism vs. redistribution Social dominance vs. egalitarianism

with important ingroup values—in this example, the work ethic—and the boundaries are represented as permeable (anyone can work hard if he or she truly wants to). The process of categorical differentiation, in turn, is reflected in normative beliefs that pit against one another the social categories defined through ascribed membership. Boundaries are therefore constructed as impermeable, such as in relationships between ethnic and cultural minorities and majorities (as in the normative belief of ethnocentrism) or between subordinate and dominant groups (as in beliefs of social dominance). These two forms of differentiation processes can either rely on identity-based, nonquantifiable criteria (moral order and social diversity) or on material and quantifiable attributes (free market and structural inequality). This second polarity thus opposes boundary definitions related to social identities and symbolic motives on the one hand and to social position, social hierarchy, and instrumental motives on the other. The four conceptions represent simplified models of social order, and the normative beliefs that reflect these conceptions are deployed and debated in the political arena—for example, by interest groups who seek to define and impose specific meanings of social categories (such as the “lazy unemployed”). By specifying an emblematic antagonism for each type of

88

Staerklé, Likki, and Scheidegger

social order, the model defines four types of perceived entitlements: In the moral order conception, perceived legitimacy of welfare benefits is based on similarity and value conformity, in the free market conception on proportionality and individual contributions, in the social diversity conception on ascribed group membership, and in the structural inequality conception on one’s position in the social hierarchy. The analysis of the relationships between normative beliefs and welfare attitudes is presumed to provide indirect evidence of how perceived entitlements lead to supporting or opposing intervention in the domain of social welfare. Normative beliefs are expected to interact with the institutional environment, which is at the same time the source and the product of shared definitions of reality. While institutional arrangements grow out of preexisting belief and value systems, they also give rise to new beliefs and consolidate existing ones. Analyses of the match between collective levels of support for welfare arrangements and national welfare regimes are based on the assumption that national welfare institutions exert a socializing and norm-stabilizing force that influences public opinion (Andreß and Heien, 2001). Our approach shares the important assumption with such normative institutionalist approaches that the norms and rules of institutions shape the actions of those acting within them (e.g., Rothstein, 1998; Mau, 2004). But in contrast to Mau’s (2004) institutionalist model, which takes national arrangements (weak vs. strong conditionality and comprehensive vs. residual level of welfare provision) as the defining dimensions of normative conceptions of reciprocity, our model of lay conceptions of social order takes a more bottom-up approach based on normative beliefs held by citizens and the psychological processes implied by these beliefs. Yet, our normative approach also assumes a homology between social-institutional and cognitive-psychological regulation. Accordingly, perceived welfare legitimacy (or illegitimacy) should be based on normative beliefs, evidenced with correlations between welfare attitudes and normative beliefs. These relationships are expected to vary according to the normative environments provided by different national and institutional contexts. We empirically analyze such institutional variation by first comparing the role of normative beliefs in predicting welfare attitudes across types of welfare regimes (Esping-Andersen, 1990) and then employing multilevel models to analyze how the relationship is affected by two continuous country-level variables: the country’s proportion of social



A Normative Approach to Welfare Attitudes

89

expenditure and its level of unemployment. Social expenditure was chosen because in strong welfare states the question of defining boundaries between beneficiaries and nonbeneficiaries of social provisions is likely to be a central political question and thus relevant for the analysis of public welfare legitimacy. Level of unemployment, in turn, should affect the extent to which citizens are personally aware of social risk and uncertainty through their own experiences with unemployment and familiarity with unemployed persons around them. In order to provide more detailed rationales as to how the country-level features should affect the legitimizing role of normative beliefs, we now move on to examine separately the four conceptions of social order. f ou r c on c e p t ion s of so c i a l or de r In the conception of moral order, social order is based on the principle of morality and conformity with consensual and established norms and values (as in Etzioni’s (1994) conservative communitarianism). Social order is explained with citizens’ respect for common values such as morality, selfreliance, and discipline: The good citizens are those who represent “our” values, whereas those who disrespect them, with deviant and disorderly behavior, are categorized as “bad” citizens. This form of social order is supported and justified with authoritarian modes of thinking, characterized by intolerance of deviance and submissiveness to authorities (see Duckitt, 1989). Two rationales linking authoritarian beliefs to welfare attitudes seem plausible. First, an authoritarian view of social order may be incompatible with the endorsement of extensive government responsibility in that repression of deviance, rather than redistribution, is seen as the correct answer to social problems. In this view, deviant and nonconforming individuals are not considered worthy of state support, leading to a conditional view of welfare and lower overall levels of welfare support. Some authors have already noted a current tendency in many Western European countries to complement or even substitute social welfare schemes with punitive and repressive policies (see Young, 1999; Garland, 2001). Based on this perceived incompatibility between social and repressive government action, the first hypothesis therefore suggests a negative relationship between authoritarian beliefs and welfare support (“The more authoritarian, the less social government intervention”; H1a). We would expect to find such a

90

Staerklé, Likki, and Scheidegger

relationship particularly in Western European liberal contexts, where high levels of social expenditure have come under political attack from neoliberal movements and where the shift from social to penal regulation of social problems has been most clearly evidenced (Wacquant, 2009). An alternative rationale suggests that endorsing both authoritarian beliefs and extensive government responsibility may not necessarily be mutually exclusive. A positive relationship (“The more authoritarian, the more social government intervention”; H1b) reflects support for a strong government that is expected both to help people in need and to deal effectively with crime and other forms of social insecurity. We therefore expect the relationship between authoritarianism and government responsibility to be more strongly positive in national contexts with high levels of social insecurity, in the form of low social expenditure and high unemployment, compared to contexts with relatively less social insecurity. A second aspect of the moral order conception concerns the extent to which people perceive others as trustworthy group members. Social trust has been shown to be central for coordination and cooperation in groups (Deutsch, 1973; Putnam, 1993). We can therefore expect that social distrust should lead to a lower willingness to support resource redistribution within the group or society (“The more distrust, the less social government intervention”; H2a). An alternative and more counterintuitive hypothesis in light of the research on the positive effects of trust on welfare legitimacy (e.g., Rothstein, 1998), however, suggests a positive relationship (“The more distrust, the more social government intervention”; H2b), interpreted as a response to contexts where people feel they cannot trust others and therefore need the state to compensate for the weakened social ties through extensive government programs. Hence, we expect the relationship between distrust and government responsibility to be positive in contexts of high social insecurity—that is, where social expenditure is low and unemployment is high. The free market conception of social order consists of beliefs about competitive motivations and takes productivity and individual performance as the basis of normative boundary definition. On grounds of economically liberal principles, free market thinking assumes that the basic human motivation is self-interest (Walster et al., 1978). Meritocratic criteria differentiate lazy “losers” from productive “winners,” and individuals are expected to engage in competitive relations with one another. Equity policies (such as



A Normative Approach to Welfare Attitudes

91

individualized insurance regimes), as well as means-tested schemes, are the main policy types in this conception. Alleged welfare dependency—the idea that government support makes people lazy and irresponsible—is therefore a core belief in the free market conception. This free market conception of welfare is evidenced with a negative relationship between welfare dependency and government responsibility (“The lazier the beneficiaries, the less social government intervention”; H3). We expect this free market conception to be widespread across European countries but nevertheless anticipate that alleged welfare dependency plays a more prominent role in welfare attitude construction in countries with extensive welfare states and low unemployment rates. In such contexts welfare abuse is indeed more likely to figure in media and political discourse. Hence, the relationship between welfare dependency and social government intervention should be more strongly negative in these contexts. The conception of social diversity relies upon ascribed group membership and is based on an a priori distinction between social groups—in particular between national, ethnic, and cultural groups. It thereby provides a conceptual framework for integrating beliefs concerning cultural diversity and multiculturalism in the study of welfare attitudes. This conception is more complex than the conceptions of moral order and free market because normative beliefs in contemporary societies may depict group differences as either positive—for example, in multiculturalism and in movements defending rights of particular groups (see Isin and Wood, 1999)—or as negative— for example, in racist and ethnocentric thinking. We expect to find mostly negative relationships between ethnocentrism and government responsibility across European countries (“The more ethnocentric, the less social government intervention”; H4). Such negative relationships are assumed to denote a motivation to limit welfare coverage of immigrant and other minority groups, a process amply illustrated in the U.S. context, where racial prejudice is a major driving force of opposition to welfare (Gilens, 1999). We therefore expect the relationship between ethnocentrism and government responsibility to be more negative in national contexts with high levels of social expenditure. In these relatively wealthy contexts immigration is also more common and more politicized than in less wealthy contexts (Castles and Miller, 2009). In the final conception, normative beliefs are structured by perceived social hierarchies of status and power—that is, by structural inequalities.

92

Staerklé, Likki, and Scheidegger

In this conception, boundaries are drawn between allegedly impermeable subordinate and dominant social categories. These categories are seen as being in a competitive relation of negative interdependence with each other: The demands by subordinate groups directly threaten the well-being of the dominant groups. As in the social diversity conception, normative beliefs promoted by politically active groups and individuals can either portray structural inequalities as legitimate, fair, and “normal,” or as illegitimate and unfair. In order to justify social cleavages, individuals may, for example, endorse the beliefs put forward by groups that point to the moral, social, and intellectual superiority of those in privileged positions (i.e., “legitimizing myths” in social dominance theory; see Sidanius and Pratto, 1999), or they may support egalitarian beliefs that consider class differences as the illegitimate result of social reproduction and inherited privileges (Bourdieu, 1984). Therefore, individuals promoting an egalitarian social order are likely to defend redistributive policies, in particular progressive tax policies and universalist social welfare programs (Rothstein, 1998). Welfare legitimacy following the logic of this conception is evidenced with a positive relationship between egalitarianism and government responsibility (“The more egalitarian, the more social government intervention”; H5), denoting a motivation to expand welfare coverage to subordinate groups. Since the need to address inequalities is strongest in countries with high levels of social insecurity, we expect this relationship to be more positive in countries with relatively low levels of social expenditure and high levels of unemployment. a n a ly t ic mode l An analytic model will guide the treatment of the ESS IV data on 28 European countries. First, we examine the dimensionality and the factorial structure of five normative beliefs defined by the model of lay conceptions of social order: authoritarianism, distrust, welfare dependency, ethnocentrism, and egalitarianism. Then we examine the impact of social position (assessed with level of education and perceived material vulnerability) on the five conceptions of social order (path “a” in Figure 4.1). Multilevel regression analyses then investigate the joint impact of social position (path “b”) and normative beliefs (path “c”) on government responsibility. Next, we present a mediation model that analyzes the indirect effects of social position on



A Normative Approach to Welfare Attitudes

Level 1

93

Normative belief (moral order, free market, social diversity, structural inequality)

a

c

d Social position (education, vulnerability)

Level 2

b

Welfare attitude (government responsibility)

National context (social expenditure, unemployment)

Figure 4.1  Mediation model of welfare attitudes

welfare attitudes, mediated by normative beliefs for all countries together (paths “a” + “c”). In order to examine country-level variation of normative welfare legitimacy, we first look at differences and similarities in normative mechanisms across welfare regimes. Then, in the final step, we test crosslevel interactions to explore the extent to which the relationships between normative beliefs and welfare attitudes are moderated by two country-level variables: social expenditure and level of unemployment (path “d”). data a n d m e a su r e s We use the complete ESS IV dataset (third release) that in May 2010 included 28 countries (total N = 54,988). Variables were constructed in order to test the mediation model presented in Figure 4.1. We first describe the social position variables, then the conceptions of social order, and finally the two dependent variables. Level of education and perceived material vulnerability were used as variables assessing respondents’ position in the social structure. Level of education was chosen because it is a reliable and objective indicator of social position. It is highly correlated with other indicators of social position, such as household income (r = .38), and it has the advantage of having few

94

Staerklé, Likki, and Scheidegger

missing values (contrary to household income). We assume that the level of education determines to a large extent the social position of individuals in terms of income, status, and prestige (Bourdieu, 1984). A single item with seven levels of education was used: not completed primary education (3.2  percent of the sample); primary or first stage of basic schooling (12.4  percent); lower secondary or second stage of basic schooling (19.4 percent); upper secondary (34.6 percent); postsecondary, nontertiary (7.3 percent); first stage of tertiary (21.8 percent); and second stage of tertiary (1.1 percent). The item was treated as an ordinal variable. The second indicator of social position was material vulnerability, which represents a subjective interpretation of one’s social position in terms of the likelihood of being exposed to different material risks. We call this measure “material vulnerability” based on the assumption that situations of need, scarcity, and destitution give rise to feelings of uncertainty, frailty, and doubt. The indicator is an aggregate scale composed of four items of perceived likelihood of life course events with negative material consequences occurring in the next 12 months: becoming unemployed and looking for work, having less time for paid work than desired because of the care given to family members, not having enough money for household necessities, and not receiving health care in case of illness. The average Cronbach’s alpha is .76 and ranges from .55 in Denmark to .85 in Cyprus. Material vulnerability has a moderate negative correlation with level of education (r = –.13), which suggests that people with lower levels of education are more likely to find themselves in situations of material vulnerability. Fifteen items assessing the four conceptions of social order were selected (Table 4.2). The moral order conception was measured with two three-item scales of authoritarianism and distrust; the free market conception was represented by a five-item scale of welfare dependency; the social diversity conception was measured with a three-item ethnocentrism scale; and the structural inequality conception was measured with a single egalitarianism item. A principal components analysis with VARIMAX rotation confirmed the organization of these items into the expected dimensions. In order to formally test the factorial structure of these dimensions, a confirmatory factor analysis on four dimensions was then performed without the single egalitarianism item. A reasonably good model fit confirmed the fourdimensional structure of the items (CFI = .96; RMSEA = .05).



95

A Normative Approach to Welfare Attitudes ta bl e 4.2 Five measures of conceptions of social order, with descriptive statistics for overall sample

Conception Moral order

Indicator (scale) Authoritarianism (1–5)

Distrust (0–10)

Free market

Social diversity

Structural inequality

Welfare dependency (1–5)

Ethnocentrism (0–10)

Egalitarianism (1–5)

Item

M

SD

N

Terrorist suspect should stay in prison until police satisfied Schools should teach children to obey authority People who break the law should get much harsher sentences You can’t be too careful Most people try to take advantage of you People mostly look out for themselves

4.08

.99

53,300

4.06

.94

54,603

3.97

.97

53,736

5.32 4.73

2.58 2.47

54,786 54,360

5.47

2.45

54,626

Social benefits make people lazy Social benefits make people less willing to care for one another Social benefits make people less willing to look after themselves/family Most unemployed people do not really try to find a job Employees often pretend they are sick to stay at home

2.97

1.14

53,374

2.90

1.09

52,778

2.78

1.08

52,773

3.05

1.11

53,770

3.02

1.09

51,484

Immigration is bad for the country’s economy Country’s cultural life is undermined by immigrants Immigrants make the country a worse place to live

5.21

2.49

51,762

4.65

2.61

51,801

5.25

2.36

51,532

For a society to be fair, differences in the standard of living should be small

3.72

.97

53,799

These findings point toward the heuristic value of the model of lay conceptions of social order that organizes normative beliefs into four general, superordinate categories. The structure of normative beliefs is thus defined by differential representations of social order and is composed of two moral order subdimensions (authoritarianism and distrust), one free market dimension (welfare dependency), and one social diversity dimension

96

Staerklé, Likki, and Scheidegger

(ethnocentrism). The single item on egalitarianism is used as a proxy for the structural inequality conception. With respect to the overall means of the measures presented in Table 4.2, it can be observed that distrust, welfare dependency, and ethnocentrism yielded means around the respective midpoints of the scales; that is, there was no clear overall tendency to either refute or support these beliefs. Authoritarianism, however, was supported by a majority of the respondents (mean of 4.04 on a 5-point scale), as was egalitarianism (mean of 3.72 on a 5-point scale). Yet, important differences among countries are hidden behind these overall means. Preliminary analyses revealed that the five conceptions were all positively related to one another, with the exception of a negative relationship between welfare dependency and egalitarianism (r = –.06). The overall correlations ranged from .10 (between egalitarianism and ethnocentrism) to .29 (between ethnocentrism and distrust). Government responsibility, our dependent variable, was measured with four items asking participants to rate the extent to which governments should be responsible for ensuring jobs for everyone, health care for the sick, a better standard of living for the old, and a better standard of living for the unemployed. The global mean of the 4-item summary score shows that there is a strong overall tendency to support government intervention (M = 7.81 on a 10-point scale). nor m at i v e be l i e f s a s a f u n c t ion of e duc at ion l e v e l a n d v u l n e r a bi l i t y Membership in social categories defined, for example, by gender, age, education, or class represents a proxy for contexts in which common experiences are likely to give rise to shared reference knowledge and similar normative beliefs (Bourdieu, 1984). As an example, the socioeconomic position of citizens affects their perceptions of social reality (Wegener, 1987), which leads low-status groups to perceive greater social inequality than do highstatus groups (Aalberg, 2003; Staerklé et al., 2007). More generally, people who are living in unfavorable conditions are likely to develop worldviews in line with their living conditions (Castel, 1995). In the present study, we assumed that similar levels of education give rise to relatively similar life experiences and thus to common normative



A Normative Approach to Welfare Attitudes

97

beliefs. Likewise, the common experience of material vulnerability should lead to specific understandings of social reality. We first carried out a series of multilevel regression analyses predicting each of the five conceptions with social positions. In terms of the mediation model presented in Figure 4.1, this step thus investigates the relationship between the independent variables and the mediator variables (path “a”).1 With respect to age and gender, the findings show that older generations were more likely to endorse authoritarianism, welfare dependency, ethnocentrism, and egalitarianism, but they were somewhat more trusting than younger generations. A similar pattern emerged for gender, where women scored higher than men on the same conceptions as older generations, with the exception of perceived welfare dependency (on which men scored higher) and a much weaker effect on ethnocentrism. Lower levels of education predicted stronger endorsement of all five conceptions. Respondents with lower levels of education were more authoritarian, more distrustful, believed more strongly in the debilitating effects of welfare, and considered immigrant presence as a threat. That is, they were more likely to engage in both normative and categorical differentiation by drawing boundaries between norm-conforming and norm-violating individuals on the one hand, and by establishing stronger boundaries between national and immigrant categories on the other. At the same time, however, they also expressed higher support for egalitarianism, thereby supporting a more inclusive boundary definition in the domain social inequality. Material vulnerability, in turn, also led to distrust, ethnocentrism, and egalitarianism but was unrelated to authoritarianism and negatively related to welfare dependency. Overall, education exerted a stronger pressure on opinions than material vulnerability, this difference being particularly striking for authoritarianism and welfare dependency. Distrust, however, was better predicted by (high) vulnerability than by (low) education level. The strong effects of education level are likely to reflect the fact that education determines both cultural and economic capital (Bourdieu, 1984). People with a higher level of education are higher up in the social hierarchy, are less punitive and more trusting, and are less suspicious and less ethnocentric, but they are also less egalitarian. In short, individuals with high status seem more optimistic and more carefree and less concerned about social issues such as delinquency, welfare abuse, and immigrant presence. The definition of normative boundaries and the perception of social cleavages,

98

Staerklé, Likki, and Scheidegger

in turn, is firmly anchored in low-status positions, presumably because perceiving society as being structured as a function of clearly differentiated groups provides a sense of ontological stability and security that is more likely to be jeopardized in low-status positions (Young, 1999). nor m at i v e l e g i t i m ac y of so c i a l g ov e r n m e n t i n t e rv e n t ion We now turn to the central question of the nature of the relationship between social positions and normative beliefs on the one hand and welfare attitudes on the other. This analysis provides insights into the normative legitimacy of welfare intervention as it reveals the extent to which normative beliefs are associated with the support for or opposition to social government intervention. In terms of our analytic model, we look at the effects of the independent and mediator variables on the dependent variable. In order to control for country-level variance in welfare attitudes, we used multilevel regression analysis to test the effects of social positions and conceptions on welfare attitudes. Similarly to the analyses in the previous section, both the two social position variables and the five conceptions were centered around their country means in order to obtain unbiased and comparable estimates of the relationships among individual-level variables (Raudenbush and Bryk, 2002: 139). In the first step, social position variables of education level and material vulnerability were introduced together with age and gender (path “b”). Regarding age and gender, the findings show that elderly people and women ask for more government intervention than do younger generations and men. Concerning the two social position variables, higher levels of education were related to less support for government responsibility, while material vulnerability predicted stronger endorsement of government intervention. In sum, people in dominant social positions were more likely to reject redistributive government policies, while people in subordinate positions endorsed them, a result in line with prior research on welfare attitudes (e.g., Svallfors, 2006). In a second step, the five conceptions were introduced in addition to the variables included in the first step (path “c,” controlling for path “b”). The strongest effects were found for egalitarianism, welfare dependency, and authoritarianism: Social government intervention was supported with



A Normative Approach to Welfare Attitudes

99

a belief based on the illegitimacy of the categorical differentiation between dominant and subordinate groups. The negative effect of welfare dependency suggests that normative boundary drawing between deserving and undeserving beneficiaries, through the belief in the detrimental effects of government help, is a key driving force of opposition to welfare intervention across European countries. While these two factors are well known in welfare attitude research (e.g., Van Oorschot, 2000), the strongly positive effect of authoritarianism is more intriguing. In line with our second rationale (Hypothesis 1b), this finding indicates that social government intervention is backed up with (rather than challenged by) authoritarian beliefs that call for obedience and punitiveness, and thus for a “strong” government that deals effectively with crime and insecurity. On the level of public attitudes, social government thus goes hand in hand with disciplinary government (see Wacquant, 1999). Similarly, and again contrary to the usual assumption that social trust should support welfare intervention (Rothstein, 1998), distrust in fellow citizens fuels calls for social government intervention but to a lesser degree than authoritarianism. Ethnocentrism, finally, also legitimizes opposition to welfare states (presumably in order to limit social rights for immigrants), but since its effect was weaker than any of the other conceptions, its role as a generalized normative foundation of welfare opposition seems minor. To summarize, the normative legitimacy of European welfare states seems to be grounded in three general belief systems: First, welfare legitimacy is grounded in a principled position of egalitarianism that considers inequalities between dominant and subordinate groups to be a threat to a fair social order. Second, welfare legitimacy is supported with authoritarian beliefs according to which governments should play a more extensive role in safeguarding a social order threatened by immoral and dangerous people, through repressive policies and an emphasis on conformity and obedience. This interpretation is supported by the finding that welfare intervention is also associated with generalized distrust and the feeling that one cannot rely on fellow citizens for help in difficult life situations. Hence, and at odds with the principle that social trust underlies welfare support, people rely on government intervention to compensate for the lack of social support around them. Third, welfare legitimacy is undermined by stereotypical and stigmatizing images that are put forward in the political debate by advocates of a free market organization of social order who aim to restrict

100

Staerklé, Likki, and Scheidegger

social rights for allegedly undeserving citizens. These findings underscore the importance of normative boundary drawing in welfare legitimacy, since beliefs that imply the differentiation of citizens in terms of their conformity with either moral or economic values determine to a large extent perceived welfare legitimacy. Welfare legitimacy is thus supported by the authoritarian credo that more punishment is needed for rule-breakers, while the belief in the ubiquity of free-riding at the expense of hard-working citizens undermines welfare legitimacy. m e di at ion mode l The findings in Table 4.3 also show that the introduction of the conceptions in the model decreased all main effects of the position variables on welfare attitudes, indicating that the impact of these positions on welfare attitudes was at least partially mediated by normative beliefs (see MacKinnon et al., t a bl e 4 .3 Multilevel analysis: Level 1 main effects of social positions and conceptions on welfare attitudes government responsibility m 2 a social positions

m 2 b conceptions

B

SE

B

SE

level 1 Intercept

76.96***

(1.11)

77.17***

(1.11)

Social position Age Female Education Material vulnerability

.63*** 1.36*** –1.20*** 1.34***

(.07) (.14) (.08) (.08)

.28*** .95*** –.84*** 1.01***

(.07) (.13) (.08) (.08)

2.42*** .53*** –3.24*** –.30*** 3.27***

(.08) (.08) (.07) (.08) (.07)

Conceptions Authoritarianism Distrust Welfare dependency Ethnocentrism Egalitarianism level 2 D-2 log likelihood (df)

824.2(4)***

5,092.5(5)***

n o t e : Dependent variable scale ranges from 0 to 100. All predictor variables (except gender) are standardized. Δ-2 log likelihood difference for the M2a model is calculated in comparison to the intercept-only model. Δ-2 log likelihood difference for the M2b model is calculated in comparison to the M2a model. * = p < .05; ** = p < .01; *** = p < .001



A Normative Approach to Welfare Attitudes

101

2007). This section specifically addresses this mediating function of normative beliefs with structural equation models (SEM), using the global sample of 28 countries. We began by testing a hybrid model in which the effects of the two social position variables (education and material vulnerability) on welfare attitudes were mediated by the five conceptions, without including the direct effects from the position variables to the attitude. 2 Figure 4.2 sums up the findings for the model. The model resulted in satisfactory fit indices (CFI = .93; RMSEA = .05).3 In line with our expectations, the analysis reveals that four of the five conceptions (with the exception of ethnocentrism) mediated the effects of both position variables on government responsibility (i.e., the indirect effects were significant). As already shown, lower levels of education were related to a greater motivation to differentiate individuals and groups on both normative and categorical grounds. In the second part of the mediation, from the conceptions to the welfare attitudes, only welfare dependency was negatively linked to welfare attitudes, while all other beliefs led to a greater demand for social government intervention. These mediation analyses thus demonstrate the existence of indirect, mediating effects between social positions and welfare attitudes through

Authoritarianism (R2 =.08) –.25*** –.16***

Education

Distrust (R2 =.12)

–.19***

.10***

Material vulnerability

.28*** –.09***

.17*** .13***

7*

.03

–.13***

–.15***

.2

**

**

Welfare dependency (R2 =.02) Ethnocentrism (R2 =.07)

*

Government responsibility (R2 =.17)

–.23***

.01 *

**

.19

Egalitarianism (R2 =.05)

Figure 4.2  Normative beliefs mediate the link between social position and welfare attitudes s o u r c e : European Social Survey 2008. *** = p < .001

102

Staerklé, Likki, and Scheidegger

normative beliefs, in particular authoritarianism, welfare dependency, and egalitarianism. It should be noted, furthermore, that individuals with a low level of education display a paradoxical pattern: They endorse authoritarian and egalitarian beliefs that legitimize welfare intervention, while at the same time accepting beliefs of welfare dependency that undermine welfare legitimacy. The explanation of this seeming inconsistency is likely to be found in the strategies that low-education groups employ to make sense of their unfavorable social position: While calling for strong government intervention to bring more stability and security to their lives, they also use the work ethic as an identity resource to distance themselves from allegedly value-violating and stigmatized groups such as welfare beneficiaries (Castel, 1995). This leads low-education groups to endorse beliefs of perceived abuse and welfare dependency. This pattern was not found among individuals with high levels of material vulnerability, who perceived less welfare dependency than those with low vulnerability. In other words, material vulnerability is related to beliefs that support welfare legitimacy, not those that undermine it. w e l fa r e r e g i m e s a n d nor m at i v e w e l fa r e l e g i t i m ac y Up to now, we have analyzed welfare legitimacy on the entire sample of 28 countries. There are, however, strong variations in the revealed mechanisms across countries. Normative beliefs concerning social order and attitudes toward welfare states emerge in specific historical and institutional contexts. National citizens are therefore likely to share a common normative heritage that accounts for greater similarities in the beliefs and attitudes of citizens residing in the same country, compared to citizens from different countries. Furthermore, some countries are more similar to one another than others, forming clusters of countries. As a first step toward understanding the country-level variation in normative welfare legitimacy, we look at differences and similarities in normative mechanisms across welfare regimes by using an adapted version from Esping-Andersen’s (1990) wellknown typology. The rationale for this analysis is that welfare regimes, much like the conceptions of social order, reflect different normative models of social order and that some correspondence can be established among the four con-



A Normative Approach to Welfare Attitudes

103

ceptions and types of welfare regimes. While our analysis remains largely exploratory, we nevertheless expect that conceptions of social order should have greater predictive weight of welfare attitudes in matching regimes; that is, dominant normative beliefs should provide stronger justification of welfare legitimacy. The moral order conception is closest to the continental-conservative regime with its principles of subsidiarity and private welfare responsibility. Therefore, authoritarianism and distrust (as indicators of moral order) should be important predictors in this regime. The free market conception clearly matches the liberal Anglo-Saxon regime with low decommodification and market-differentiated provisions that should be supported by welfare dependency. The social diversity conception is not represented in Esping-Andersen’s typology, but it allows investigating the relationship between negative perceptions of cultural diversity (ethnocentrism) and welfare provision, which has been an important focus of recent welfare research (e.g., Banting and Kymlicka, 2006). The structural inequality conception, finally, is closely related to the social democratic Nordic regime, with high decommodification and strong state intervention. We expect egalitarianism to be an important predictor in this context. Eastern and Southern countries are treated as separate types of regimes, notwithstanding important institutional variations within these categories. We do not have specific predictions concerning these two regime types. We conducted a regression analysis that includes the two position variables and the five normative beliefs (as well as controlling for age, gender, and country), separately in five welfare regimes. The countries representing each regime were the same as in Chapter 3, with the only difference being that we excluded Switzerland and the Netherlands from the analysis due to the mixed character of these regimes. The standardized Beta coefficients for the effect of each normative belief on government responsibility by regime are presented in Figure 4.3. The results show that authoritarian beliefs had the greatest positive effect on welfare support in Southern and Eastern regimes, followed by continental-conservative regimes (although their effect remained positive everywhere), thereby partially confirming our hypothesis of a strong relationship between moral order conceptions and welfare attitudes in conservative regimes. Demands for normative conformity and repressive policies, then, seem to go hand in hand with support for social government intervention in all regimes. The effects of distrust are less straightforward, since

104

Staerklé, Likki, and Scheidegger

0.3 .21*

.18*

0.2

.20*

.19* .18*

.18*

.17*

.13* 0.1

.09* .06*

.05*

.03

.02

0 –.02

–.01

–.02

00 –.03+

Egalitarianism

–.08*

–.17*

–0.2 –.20*

Welfare dependency Ethnocentrism

–.06* –0.1

Authoritarianism Distrust

.17* –.21*

–.23*

–0.3 Anglo-Saxon

Nordic

Continental

Southern

Eastern

Figure 4.3  Predicting government responsibility as a function of conceptions of social order across five welfare regimes s o u r c e : European Social Survey 2008. n o t e : Regression coefficients are standardized and controlled for age, gender, and country in each regime cluster. * = p < .001; + = p < .05

distrust played no role in structuring welfare attitudes in Anglo-Saxon, Continental, or Southern regimes, but it had a negative effect on welfare support in Nordic countries and a positive effect in the Eastern regime. Nevertheless, the pattern of variation across regimes was similar to the one observed for authoritarianism. In other words, in the East, perceiving other people as untrustworthy and self-interested leads to higher demand for social state intervention, while in the North, distrust is linked to less support for government intervention. The results suggest, then, that the effect of trust is dependent on context, as formulated in our two alternative hypotheses for moral order. The negative effect of welfare dependency on welfare support was strong across all regimes, indicating that perceived welfare abuse seems to be a universal reason to oppose social government intervention. Similarly, egalitarianism had a universally positive effect on welfare support in all regimes, although its effect was (as expected) strongest in the Nordic and weakest in the Eastern regime. Finally, ethnocentrism was unrelated to



A Normative Approach to Welfare Attitudes

105

welfare support in the Eastern, Continental, and Anglo-Saxon regimes. Ethnocentrism had a negative effect on support for government intervention only in the Nordic and Southern regimes, and even there the effects were small compared to other normative beliefs. a m u lt i l e v e l a p p roac h t o c ou n t ry-l e v e l va r i at ion i n nor m at i v e w e l fa r e l e g i t i m ac y A regime-based approach gives a first idea of how normative mechanisms differ across groups of countries. Such an approach does not, however, take into account differences among countries in the same regime. This can be achieved by multilevel modeling that allows testing whether continuous variables measured on the country level account for the differences observed among countries. We began by analyzing the extent to which the two country-level characteristics—public social security expenditure (as percentage of GDP) and the national unemployment rate—affect welfare attitudes. To test their direct effects on welfare attitudes, we conducted separate analyses for both country-level variables by introducing them into the multilevel models, including all individual level variables presented in Table 4.3. Social expenditure had a negative effect (p < .05) and unemployment rate had a positive effect (p < .05) on welfare support. In other words, in countries with high levels of public social security expenditure, people were less in favor of social government intervention. This is likely to be a compensation effect inasmuch as the desired government intervention depends on the current level of government intervention. In countries where government already plays an important social role, citizens are less likely to ask for even more government responsibility. It should be noted, however, that even in countries where the overall mean for government responsibility was among the lowest, it still remained clearly on the side of general support for government responsibility. In countries with higher unemployment, people were more in favor of government responsibility, suggesting that the experience and awareness of the risk of unemployment lead to stronger support for government responsibility. In the next step, we were interested in the extent to which the effect of normative beliefs in explaining welfare attitudes differs across countries. In order to calculate cross-national variation in the relationships among the five conceptions and government responsibility, we ran separate models, each incorporating the random slope of one conception (together with all

106

Staerklé, Likki, and Scheidegger

the individual-level main effects presented in Table 4.3). The results of the random slope tests show that the relationship between each of the conceptions and welfare attitudes varied significantly across countries—that is, normative beliefs did not have the same effect on welfare attitudes across the 28 countries. The slope variance was greatest for authoritarianism, while the slope variances of the other four conceptions were smaller in magnitude. This finding suggests that welfare legitimacy based on a desire for a strong and repressive government varies greatly across countries. The conception with the second largest slope variance was ethnocentrism, indicating that countries also differ in the extent to which embracing or rejecting social diversity organizes attitudes toward government responsibility. In the final step, we investigated the extent to which the two countrylevel variables (social expenditure and unemployment rate) explain the variations in explanatory power of the five normative beliefs. In order to do so, we introduced cross-level interaction terms into the models described above for testing slope variations. We tested one model for each conception country-level variable combination (i.e., one cross-level interaction at a time) and included in each model the main effect of the country-level variable composing the interaction. Table 4.4 summarizes the moderating effects of welfare expenditure and unemployment variables on the belief-attitude relationships. The results show that, with the exception of welfare dependency, the effects of all normative beliefs on support for government responsibility ta bl e 4.4 Multilevel analysis: Cross-level interactions between conceptions and level 2 indicators on welfare attitudes government responsibility Level 2 indicator

Level 1 conception

B

SE

Social expenditure*

Authoritarianism Distrust Welfare dependency Ethnocentrism Egalitarianism

–.94** –.66** –.11 –.70** –.53*

(.33) (.22) (.21) (.23) (.21)

Unemployment rate*

Authoritarianism Distrust Welfare dependency Ethnocentrism Egalitarianism

.75* .60* .40* .60* .46*

(.34) (.23) (.19) (.24) (.22)

n o t e : The cross-level interaction models are compared to their respective random-slope models. Underlined interaction effects are depicted in Figures 4.4a–c. * = p < .05; ** = p < .01; *** = p < .001

(a) Government responsibility as a function of authoritarianism and social expenditure 84 82

78 76

**

3*

80 B

=

6 3.

Low Social Expenditure (–1 SD)

**

High Social Expenditure (+1 SD)

.75*

B=1

74 72 Low Authoritarianism (–1 SD) High Authoritarianism (+1 SD)

(b) Government responsibility as a function of ethnocentrism and social expenditure 84 82 80

B = .24 Low Social Expenditure (–1 SD)

78 76

B=–

High Social Expenditure (+1 SD) 1.17*

**

74 72 Low Ethnocentrism (–1 SD)

High Ethnocentrism (+1 SD)

(c) Government responsibility as a function of distrust and unemployment rate 84 82 80

5***

B = .9

78 Low Unemployment (–1 SD) 76

B = –.24

High Unemployment (+1 SD)

74 72 Low Distrust (–1 SD)

High Distrust (+1 SD)

Figure 4.4  Cross-level interactions on government responsibility s o u r c e : European Social Survey 2008.

108

Staerklé, Likki, and Scheidegger

were moderated by the level of social expenditure. The two strongest effects were found for authoritarianism (Figure 4.4a) and ethnocentrism (Figure 4.4b). As predicted, the positive effect of authoritarianism on support for government responsibility was stronger in countries with lower levels of welfare spending (e.g., Estonia, Latvia, Romania, Russian Federation, Turkey), compared to countries with higher levels of welfare spending (e.g., Belgium, Denmark, Finland, France, Germany, Sweden), even though the slopes were positively significant for both contexts. In other words, in the (mostly ex-communist) contexts characterized by a weak social state and high levels of social insecurity, authoritarianism was a stronger basis of welfare legitimacy than in (mostly) Western European national contexts. Egalitarianism showed the same pattern as authoritarianism, with a stronger positive relationship between egalitarianism and government responsibility in countries with low welfare expenditure. Similarly, distrust was positively related to welfare support in countries with low welfare expenditure and unrelated to welfare support in countries with high welfare expenditure. At odds with our expectations, the impact of welfare dependency did not vary across countries, confirming the results of the regime analysis that also suggested that the normative differentiation between deserving and undeserving individuals represents a fairly uniform basis for rejecting welfare intervention across European countries. Ethnocentrism, in turn, did not predict government responsibility in countries with lower social expenditure, but it had, as predicted, a negative effect on support for government responsibility in countries with higher levels of social expenditure. It thus appears that in national contexts where the state takes an important role in redistributing resources among citizens, a negative view of social diversity is linked to less support for government involvement. This is an important finding because it suggests that the importance of prejudice and ethnocentrism as delegitimizing forces of welfare states is contingent upon national contexts. It is interesting to note that although higher (mean) levels of ethnocentrism are generally found in Eastern and Central European countries, such exclusionary attitudes actually play a more powerful political role in defining welfare attitudes in Western European countries with highly developed welfare states. Unemployment rate moderated the effect of all conceptions on support for government responsibility. Authoritarianism and egalitarianism were both more strongly related to government responsibility in countries



A Normative Approach to Welfare Attitudes

109

with higher levels of unemployment (e.g., Spain, Slovakia, Turkey) than in countries with less unemployment (e.g., Denmark, the Netherlands, Norway, Switzerland). Welfare dependency had a stronger negative effect on government responsibility in countries with lower levels of unemployment, suggesting that the prevalence of unemployment attenuates the negative influence of perceived welfare dependency on welfare support. The delegitimizing role of welfare dependency is thus particularly common in relatively well-off contexts. Concerning distrust (Figure 4.4c), perceiving people as self-interested and untrustworthy was related to greater support for government responsibility in countries with high levels of unemployment, while no such link existed in countries with lower levels on unemployment. Additional analyses show that in high-unemployment contexts, overall levels of social trust are lower than in countries with less unemployment. These higher levels of social distrust in countries marked by precariousness and vulnerability seem to lead to a stronger reliance on institutional, government-based support also at the individual level. Finally, ethnocentrism was related to less welfare support in countries with low unemployment and unrelated to welfare attitudes in countries with higher unemployment. Again, attitudes toward foreigners play a more important role in determining welfare attitudes in relatively more privileged contexts characterized by low unemployment, compared to the less privileged contexts of high unemployment. c on c l usion In this chapter, we proposed a normative approach to welfare attitudes, based on the general idea that welfare attitudes are shaped by widespread normative beliefs that are anchored in citizens’ everyday experiences. We showed how five normative conceptions of social order—authoritarianism, distrust, welfare dependency, ethnocentrism, and egalitarianism—were called upon to either support or undermine the legitimacy of government responsibility in the domain of social welfare and the protection of social rights. These shared beliefs reflect the psychological process of differentiation through which individuals define subjective boundaries between social categories: between good and bad people, between winners and losers, between immigrants and nationals, and between dominant and subordinate groups.

110

Staerklé, Likki, and Scheidegger

In a first step, factor analyses confirmed the expected fourfold structure of conceptions of social order: moral order (measured by authoritarianism and distrust), free market (measured by welfare dependency), social diversity (measured by ethnocentrism), and structural inequality (measured by egalitarianism). The structure proposed by the model of lay conceptions of social order is thus relevant and applicable to comparative international data. Second, we investigated how level of education and perceived material vulnerability gave rise to the endorsement of normative models of social order. Individuals with lower levels of education systematically expressed greater endorsement of beliefs based on normative boundary drawing and were thus more prone to perceive social cleavages. Respondents with lower levels of education were more authoritarian, expressed higher levels of social distrust, endorsed more strongly beliefs about the undeservingness of beneficiaries, and held more negative attitudes toward immigrants. At the same time, they were also more egalitarian. Material vulnerability, in turn, was less uniformly related to normative beliefs. A sense of risk and vulnerability gave rise to higher levels of social distrust, ethnocentrism, and egalitarianism, but it was unrelated to authoritarianism. In contrast to the low-education group, materially vulnerable people perceived less welfare dependency than materially secure people. Therefore, it appears that individuals with material difficulties are more aware of the difficulties and life course risks faced by fellow citizens, may themselves live on welfare benefits, or may expect to require welfare benefits someday. This awareness leads them to reject a label of welfare abusers for themselves or anyone else dependent on benefits. Our third question concerned the nature of the relationship among social position, normative beliefs, and welfare attitudes. The overall pattern that arose was clear: In line with prior research, members of low-status groups—people with low levels of education and a strong sense of material vulnerability, females, and older generations—were more supportive of social government responsibility. With respect to the normative beliefs, authoritarianism, distrust, and egalitarianism led to greater support for government responsibility, while perceived welfare dependency and ethnocentrism were associated with less government involvement. The central normative foundations of welfare legitimacy were authoritarianism, welfare dependency, and egalitarianism. These findings suggest that perceptions of a dangerous and immoral society where deviance and crime are



A Normative Approach to Welfare Attitudes

111

perceived as major social problems fuels support for social government intervention, while stigmatizing images of undeserving citizens undermine welfare legitimacy in virtually all European countries. Egalitarianism, in turn, justifies welfare intervention with the principle that excessive inequalities between dominant and subordinate groups are morally unacceptable. Mediation analyses confirmed that the impact of low social positions on the support for welfare intervention was to a considerable extent due to the endorsement of these normative beliefs. Interestingly, the normative differentiation between good and bad people (authoritarianism and distrust in the moral order conception) and between deserving winners and undeserving losers (welfare dependency in the free market conception) produced opposite effects for welfare legitimacy: Social government intervention was seen as an appropriate solution for a society marred by lacking value conformity, suggesting that in many instances, people call for both social and repressive government functions in order to deal with life course difficulties. When endorsing the belief that free riders take advantage of the work of productive citizens, however, less rather than more government is the solution. Normative differentiation thus plays a key role, both in supporting and undermining welfare legitimacy. On a more general level, our findings suggest that collective definitions of the boundaries of moral communities are a key factor in analyzing welfare legitimacy. Any implementation of justice principles raises the issue of the definition of the boundaries of moral communities—that is, the groups whose members are potential beneficiaries of resource allocation (Opotow, 1990; Miller, 1999a). Because they reflect strategies of boundary definition, perceived entitlements and stereotypical images of beneficiary groups become the normative basis of welfare legitimacy. Based on the model of lay conceptions of social order, we argue that the relationships between normative beliefs and support for government responsibility may illustrate different logics of defining the moral boundaries of welfare recipients. The link between distrust and government responsibility, for example, may indicate that perceived erosion of moral communities fuels calls for institutional regulation and protection; the strong association between distrust and material vulnerability further underscores that precarious living conditions increase perceptions of “broken societies.” Egalitarianism, in turn, gives rise to an inclusive perception of society based on extending welfare coverage to those in disadvantaged living conditions. The negative ­relationship

112

Staerklé, Likki, and Scheidegger

between welfare dependency and government responsibility points to a desire to exclude those who are seen as abusing the welfare system and profiting from others’ work. Ethnocentrism was related, although only in certain contexts, to lower support for general government involvement in ensuring social rights. This can be seen as a strategy of exclusion where immigrants are differentiated from the national ingroup and left outside the network of national solidarity. Such an interpretation of the results in terms of moral inclusion and exclusion is, however, limited by our operationalization of the conceptions of social order. While the items provided by the ESS data set certainly tap into normative beliefs and allow testing our model, they do not explicitly ask about the inclusion or exclusion of specific groups in the sphere of welfare. Our interpretation in terms of moral boundaries is therefore based on the social psychological mechanism assumed to orient the way people reason about welfare. Further research is needed that more explicitly measures the notion of moral communities and allows developing and specifying this aspect of differentiation and boundary definition. The fourth and final aim of the chapter was to map and explain the contextual, country-level variation of these normative beliefs as organizing principles of welfare attitudes. In a regime comparison approach, welfare dependency and egalitarianism emerged as consistent predictors of government intervention across welfare regime types. Authoritarianism played the greatest role in structuring attitudes in Eastern, Southern and continental-conservative regimes, while the exclusion of immigrants was mainly of importance in the Nordic regime (see Staerklé et al., 2010). Finally, distrust had opposite effects on welfare attitudes in the Eastern compared to the Nordic regime. Multilevel analyses showed that the effect of each of the five normative beliefs on welfare attitudes varied significantly across countries. In other words, normative beliefs did not have the same weight as predictors of social government intervention across different national contexts. In countries with relatively low levels of welfare spending, as well as in countries with higher levels of unemployment, the positive effects of authoritarianism, distrust, and egalitarianism on welfare support were stronger than in countries with higher welfare spending and less unemployment. This finding suggests that in these predominantly ex-communist, Eastern, Central, and Southern European countries, welfare legitimacy is more strongly grounded in a general conception of a strong and authoritarian state that is expected



A Normative Approach to Welfare Attitudes

113

to regulate relations between citizens as well as in a principled belief concerning the unfairness of excessive social inequalities. The positive effect of distrust on government responsibility is striking, since research on welfare legitimacy tends to assume the contrary relationship and to consider social trust an important foundation of welfare systems (Putnam, 1993; Rothstein, 1998). In contrast, we view social distrust as a reflection of precarious life situations, characterized by a pessimistic, suspicious, and victimizing outlook on life (Castel, 1995). From this perspective, institutionalized welfare provisions and social government responsibility may represent a stable and reliable source of support that becomes a way to cope with existential insecurity and vulnerability. The patterns were less clear for ethnocentrism and welfare dependency, but significant cross-level interactions allow us to conclude that the negative effects of ethnocentrism on welfare support were stronger in countries with higher levels of welfare spending and lower levels of unemployment. In other words, negative attitudes toward immigrants are a key factor in the delegitimizing of welfare states, mainly in Western European countries. The negative effect of welfare dependency on welfare support, too, was stronger in contexts characterized by lower levels of unemployment. Overall, however, the role of welfare dependency in delegitimizing welfare intervention seems to be more uniform across European countries than the role of the other normative beliefs. Social expenditure and level of unemployment thus turned out to be important moderators of the social processes underlying welfare legitimacy. In general, these findings then point toward the importance of social conditions in countries in understanding the normative foundations of the social legitimacy of welfare state arrangements. An important limit of our contextual analysis regards the choice of country-level variables. We chose to focus on social expenditure because we wanted to show how boundary definition strategies were affected by the level of welfare provision in a country (which they clearly did, as our results have shown). Unemployment, in turn, was chosen, since it should affect personal awareness of social risk and uncertainty, and thus also moderate strategies of normative boundary definition. Future research could also incorporate country-level measures of the dominant normative context or ideological climate, measured, for example, by aggregating individual-level attitude data (Sarrasin et al., in press). People are likely to be aware of the dominant norms

114

Staerklé, Likki, and Scheidegger

prevalent in their societies and disseminated through media and public discourse. Such country-level norms may prove to exert a powerful contextual influence on people’s positioning toward welfare attitudes. It is a matter of debate whether the country-level variables used in this study (and in other chapters) truly capture something unique about the context and allow identifying characteristics of the national context that in themselves explain cross-national differences in belief-attitude relationships. A different interpretation of our cross-level interactions is that rather than levels of social expenditure and unemployment explaining the differences, such variation would be related to more general country profiles or regimes. For example, we can question whether the more strongly positive effect of authoritarianism on welfare support in countries with lower levels of welfare expenditure is truly related to social spending or rather has something to do with the fact that most countries with low social spending in our sample are former communist or Southern countries with a history of authoritarian regimes. Our chapter gives no final answer to whether the multilevel approach is fundamentally superior to the regime approach. The two approaches might be best conceived of as complementary rather than competing, in particular since the multilevel approach allows specifying some relations that would remain unnoticed on a regime level. In sum, the approach advocated in this chapter highlights the importance of motivations other than self-interest in explaining citizen attitudes toward government and social rights. Sharing beliefs that give meaning to everyday experiences, differentiating oneself from others, drawing boundaries between one’s group and other groups, as well as participating in the construction of social order through the endorsement of normative beliefs are such motivations explaining citizen attitudes. Our basic goal was to propose an account of social psychological processes at work in the construction of welfare attitudes. Our results demonstrated that such processes are not universal but rather are moderated by the institutional and normative contexts in which they are enacted. The normative beliefs held by individuals embedded in social groups thus become key elements for understanding the psychology of welfare attitudes. no t e s 1.  Since we were mainly interested in the impact of education and vulnerability in countries, these two variables were group-centered within countries



A Normative Approach to Welfare Attitudes

115

(Enders and Tofighi, 2007). By removing between-country variation, this procedure produces unbiased estimates of the relationships between the individual-level variables as well as of slope variances. 2.  The model is a hybrid model because four conceptions and the measure of government responsibility were included as latent variables. All other variables were included as observed variables. 3.  The models include one proposed modification (a covariation between two of the welfare dependency items).

r efer ences Aalberg, T. 2003. Achieving Justice. Comparative Public Opinions on Income Distribution. Leiden: Brill. Andreß, H.-J., and T. Heien. 2001. “Four Worlds of Welfare Attitudes? A Comparison of Germany, Norway, and the United States.” European Sociological Review 17: 337–56. Augoustinos, M., I. Walker, and N. Donaghue. 2006. Social Cognition: An Integrated Introduction. London: Sage. Banting, K., and W. Kymlicka, eds. 2006. Multiculturalism and the Welfare State. Recognition and Redistribution in Contemporary Democracies. Oxford: Oxford University Press. Bauer, M. W., and G. Gaskell. 2008. “Social Representations Theory: A Progressive Research Programme for Social Psychology.” Journal for the Theory of Social Behaviour 39: 335–53. Billig, M. 1989. Ideological Dilemmas: A Social Psychology of Everyday Dilemmas. London: Sage. Blekesaune, M., and J. Quadagno. 2003. Public Attitudes Toward Welfare State Policies: A Comparative Analysis of 24 Nations.” European Sociological Review 19: 415–27. Bourdieu, P. 1984. Distinction: A Social Critique of the Judgement of Taste. Cambridge, MA: Harvard University Press. Castel, R. 1995. Les métamorphoses de la question sociale: une chronique du salariat. Paris: Fayard. Castles, S., and M. J. Miller. 2009. The Age of Migration. International Population Movements in the Modern World. Guilford: Guilford Press. Chamberlayne, P., A. Cooper, R. Freeman, and M. Rustin. 1999. Welfare and Culture in Europe. Towards a New Paradigm in Social Policy. London: Jessica Kingsley. Clarke, J. 2004. Changing Welfare, Changing States: New Directions in Social Policy. London: Sage. Deutsch, M. 1973. The Resolution of Conflict. New Haven, CT: Yale University Press. Doise, W. 1986. Levels of Explanation in Social Psychology. Cambridge, UK, and New York: Cambridge University Press.

116

Staerklé, Likki, and Scheidegger

Doise, W., A. Clémence, and F. Lorenzi-Cioldi. 1993. The Quantitative Analysis of Social Representations. Hemel Hempstead: Harvester Wheatsheaf. Duckitt, J. 1989. “Authoritarianism and Group Identification: A New View of an Old Construct.” Political Psychology 10: 63–84. ———. 2001. “A Dual Process Cognitive-Motivational Theory of Ideology and Prejudice.” In Advances in Experimental Social Psychology, edited by M. P. Zanna, 41–113. San Diego, CA: Academic Press. Elcheroth, G., W. Doise, and S. Reicher. 2011. “On the Knowledge of Politics and the Politics of Knowledge: How a Social Representations Approach Helps Us Rethink the Subject of Political Psychology.” Political Psychology 32: 729–58. Enders, C. K., and D. Tofighi. 2007. “Centering Predictor Variables in CrossSectional Multilevel Models: A New Look at an Old Issue.” Psychological Methods 12: 121–38. Esping-Andersen, G. 1990. The Three Worlds of Welfare Capitalism Cambridge, UK: Polity Press. Etzioni, A. 1994. The Spirit of Community. New York: Simon and Schuster. Feldman, S., and J. Zaller. 1992. “The Political Culture of Ambivalence: Ideological Responses to the Welfare State.” American Journal of Political Science 36: 268–307. Garland, D. 2001. The Culture of Control: Crime and Order in Contemporary Society. Oxford: Oxford University Press. Geremek, B. 1994. Poverty: A History. Oxford: Blackwell. Gilens, M. 1999. Why Americans Hate Welfare. Race, Media and the Politics of Antipoverty Policy. Chicago: University of Chicago Press. Hall, P. A., and D. Soskice, eds. 2001. Varieties of Capitalism. The Institutional Foundations of Comparative Advantage. Oxford: Oxford University Press. Hochschild, J. L. 1981. What’s Fair? American Beliefs About Distributive Justice. Cambridge, MA: Harvard University Press. Isin, E. F., and P. K. Wood. 1999. Citizenship and Identity. London: Sage. Katz, I., and R. G. Hass. 1988. “Racial Ambivalence and American Value Conflict: Correlational and Priming Studies of Dual Cognitive Structures.” Journal of Personality and Social Psychology 55: 893–905. Kreindler, S. A. 2005. “A Dual Group Processes Model of Individual Differences in Prejudice.” Personality and Social Psychology Review 9: 90–107. Lamont, M., and V. Molnar. 2002. “The Study of Boundaries in the Social Sciences.” Annual Review of Sociology 28: 167–95. Lewin-Epstein, N., A. Kaplan, and A. Levanon. 2003. “Distributive Justice and Attitudes Toward the Welfare State.” Social Justice Research 16: 1–27. Mackinnon, D. P., A. J. Fairchild, and M. S. Fritz. 2007. “Mediation Analysis.” Annual Review of Psychology 58: 593–614. Mau, S. 2003. The Moral Economy of Welfare States. Britain and Germany Compared. London: Routledge.



A Normative Approach to Welfare Attitudes

117

———. 2004. “Welfare Regimes and the Norms of Social Exchange.” Current Sociology 52: 53–74. Mau, S., and B. Veghte, eds. 2007. Social Justice, Legitimacy and the Welfare State. Hampshire, UK: Ashgate Publishers Ltd. McClosky, H., and J. Zaller. 1984. The American Ethos. Public Attitudes Toward Capitalism and Democracy. Cambridge, MA: Harvard University Press. Miller, D. 1999a. Principles of Social Justice. Cambridge, MA: Harvard University Press. Miller, D. T. 1999b. “The Norm of Self-Interest.” American Psychologist 54: 1053–60. Moscovici, S. 1961/1976. La psychanalyse, son image et son public. Paris: Presses Universitaires de France. ———. 2000. Social Representations: Explorations in Social Psychology. Cambridge, UK: Polity Press. Mouffe, C. 1993. The Return of the Political. London: Verso. Murray, C. 1984. Losing Ground. New York: Basic Books. Opotow, S. 1990. “Moral Exclusion and Injustice: An Introduction.” Journal of Social Issues 49: 1–20. Putnam, R. D. 1993. Making Democracy Work: Civic Traditions in Modern Italy. Princeton, NJ: Princeton University Press. Raudenbush, S. W., and A. S. Bryk. 2002. Hierarchical Linear Models: Applications and Data Analysis Methods. Thousand Oaks, CA: Sage. Rehm, P. 2007. “Who Supports the Welfare State? Determinants of Preferences Concerning Redistribution.” In Social Justice, Legitimacy and the Welfare State, edited by S. Mau and B. Veghte, 47–72. Hampshire, UK: Ashgate Publishers Ltd. Rokeach, M. 1973. The Nature of Human Values. New York: The Free Press. Rothstein, B. 1998. Just Institutions Matter: The Moral and Political Logic of the Universal Welfare State. Cambridge, UK, and New York: Cambridge University Press. Sachweh, P., C. G. Ullrich, and B. Christoph. 2007. “The Moral Economy of Poverty: On the Conditionality of Public Support for Social Assistance Schemes.” In Social Justice, Legitimacy and the Welfare State, edited by S. Mau and B. Veghte, 123–42. Hampshire, UK: Ashgate Publishers Ltd. Sarrasin, O., E. G. T. Green, N. Fasel, O. Christ, C. Staerklé, and A. Clémence. 2012. In press. “Opposition to Anti-racism Laws Across Swiss Municipalities: A Multilevel Analysis.” Political Psychology. Sears, D. O., and C. L. Funk. 1990. “The Limited Effect of Economic SelfInterest on Political Attitudes of the Mass Public.” Journal of Behavioural Economics 19: 247–71. Sidanius, J., and F. Pratto. 1999. Social Dominance: An Intergroup Theory of Social Hierarchy and Oppression. Cambridge, UK, and New York: Cambridge University Press.

118

Staerklé, Likki, and Scheidegger

Soerensen, A. 1998. “On Kings, Pietism and Rent-Seeking in Scandinavian Welfare States.” Acta Sociologica 41: 363–75. Staerklé, C. 2009. “Policy Attitudes, Ideological Values and Social Representations.” Social and Personality Psychology Compass 3: 1096–1112. Staerklé, C., A. Clémence, and D. Spini. 2011. “Social Representations: A Normative and Dynamic Intergroup Approach.” Political Psychology 32:759–68. Staerklé, C., C. Delay, L. Gianettoni, and P. Roux. 2007. Qui a droit à quoi? Représentations et légitimation de l’ordre social [Who is entitled to what? Representations and legitimation of social order]. Grenoble: Presses Universitaires de Grenoble. Staerklé, C., J. Sidanius, E. G. T. Green, and L. E. Molina. 2010. “Ethnic Minority-Majority Asymmetry in National Attitudes Around the World: A Multilevel Analysis.” Political Psychology 31: 491–519. Svallfors, S. 2006. The Moral Economy of Class. Stanford: Stanford University Press. Tajfel, H. 1978. Differentiation Between Social Groups. Cambridge, UK, and New York: Cambridge University Press. ———. 1981. Human Groups and Social Categories: Studies in Social Psychology. Cambridge, UK, and New York: Cambridge University Press. Van Oorschot, W. 2000. “Who Should Get What, and Why? On Deservingness Criteria and the Conditionality of Solidarity Among the Public.” Policy & Politics 28: 33–48. ———. 2006. Culture and Social Policy: A Developing Field of Study.” International Journal of Social Welfare 16: 129–39. Van Oorschot, W., and A. Komter. 1998. “What Is It that Ties . . . ? Theoretical Perspectives on Social Bond.” Sociale Wetenschappen 3: 5–24. Wacquant, L. 1999. Les prisons de la misère. Paris: Éditions Raisons d’Agir. ———. 2009. Punishing the Poor: The Neoliberal Government of Social Insecurity. Durham: Duke University Press. Walster, E., G. W. Walster, and E. Berscheid. 1978. Equity: Theory and Research. Boston: Allyn and Bacon. Wegener, B. 1987. “The Illusion of Distributive Justice.” European Sociological Review 3: 1–13. Young, J. 1999. The Exclusive Society: Social Exclusion, Crime and Difference in Late Modernity. London: Sage. ———. 2007. The Vertigo of Late Modernity. London: Sage.

chapter five

Unraveling Working-Class Welfare Chauvinism Jan Mewes and Steffen Mau1

Mass migration is one of the central phenomena of our time. Europe, in particular, has experienced a dramatic growth in immigration over the last decades, and most European countries today have a sizable immigrant population (Castles and Miller, 2003; Hooghe et al., 2008). Immigration does not only mean that foreigners gain access to another territory and its labor market, but it has larger repercussions for the way welfare states are run and legitimized as well. Though not all immigrant groups have the same rights or entitlements to social benefits, most Western welfare states have incorporated foreigners into their welfare systems (Banting and Kymlicka, 2006). This is particularly pronounced for those groups that have been granted permanent residency (Guiraudon, 2002). Indeed, today it is not as much citizenship that establishes the right to social entitlements but residency (Joppke, 1998; Mau, 2007). However, such an inclusionary practice does not take place without raising problems, since it requires a more encompassing concept of solidarity. Moreover, immigrants tend to be proportionally more reliant on state welfare, and consequently the public might increasingly perceive them as a group that receives more social benefits (see Boeri et al., 2002). This can result in conflict, because as soon as foreigners take up permanent residence in a given territory, it is in the public interest to include them in the welfare system in order to minimize problems arising from ethnic segregation and marginalizaFor additional data, please see the Web Appendix tables and figures by following the link in the listing for Contested Welfare States on our website: http://www.sup.org.

119

120

Mewes and Mau

tion. At the same time, it is clear that the inclusion of immigrants or groups who are not considered as “belonging” could undermine the legitimacy of a social security system based on solidarity with one’s own community (Banting and Kymlicka, 2006; Van Oorschot 2006, see also Chapter 7). This chapter addresses the issue of how residents of European countries and Israel view the inclusion of foreigners. In particular, we are interested in class-specific reactions as well as in key country-level factors related to opposition to inclusion, which we call “welfare chauvinism.” Welfare chauvinism refers to those views that support, according to Kitschelt (1997: 22), a “system of social protection [only] for those who belong to the ethnically defined community and who have contributed to it.” More specifically, our contribution addresses the question of why the lower classes are prone to welfare chauvinism and whether this relationship is linked either to economic or to cultural aspects. In part of the literature, it is suggested that the socioeconomic situation of individuals plays the most decisive role in the development of welfare chauvinism (e.g., Scheve and Slaughter, 2001; Malchow-Møller et al., 2008). Other studies point out that this workingclass affinity toward welfare chauvinism might be related to cultural differences between the lower classes and the materially more privileged strata (Houtman, 2003; Achterberg and Houtman, 2009; Van der Waal et al., 2010). According to the former explanation, it is either the real or the perceived competition over scarce and valued resources that drives the desire to exclude immigrants. Hence, individuals in precarious socioeconomic situations are expected to be more prone to welfare chauvinism. The sociocultural approach, in contrast, emphasizes that it is the higher salience of authoritarian dispositions in the working class that eventually triggers welfare chauvinism. Thus, it would be the authoritarian values and attitudes among working-class members in particular (see Lipset, 1959; Van der Waal et al., 2010) that make them prone to adopt negative sentiments vis-à-vis the inclusion of foreigners. We will start by discussing these two major theoretical approaches, explaining the link between class and welfare chauvinism. In our analysis we try to tap both the economic and the cultural aspects of class, assuming that both dimensions matter (see Kriesi, 2008; Kriesi et al., 2008). However, we assert that highlighting the effects of authoritarianism, education, and income on welfare chauvinism might not be enough. Rather, we suggest additionally employing a measure of perceived material risk, since the effect of competition should be most likely moderated through individuals’



Unraveling Working-Class Welfare Chauvinism

121

perceptions of their current status and their subjectively perceived vulnerability (see also Chapter 3). In other words, we suggest that objective class positions are only partly conducive to measuring people’s feelings of status insecurity and risk proneness. Several working hypotheses for our empirical analysis will be derived from this discussion. Empirically, we will, like the other authors in this book, refer to data from the fourth round of the European Social Survey (ESS 2008). Because we are interested in the characteristics at both the country and the individual levels, we will run logistic multilevel models. In order to gain a more easily interpretable picture of how strong the covariates of interest impact on support for welfare chauvinism, we will also present a range of predicted probabilities derived from our multilevel models. Finally, our concluding section will briefly summarize our results and discuss possible policy implications regarding the emerging power of populist right-wing movements in Europe. c l a s s a n d w e l fa r e c h au v i n i sm In much of the literature on the welfare state, the working class appears as a strong supporter of egalitarianism and a redistributive welfare state. One of the most prominent welfare state theories, the power resource theory as proposed by Walter Korpi (1983), sees the working class as the main force behind universalist and egalitarian welfare states. At its core, it asserts that working-class power achieved through organization by labor unions or leftist political parties produces more egalitarian distributional outcomes (Stephens, 1979; Korpi, 1983). The theory suggests a rather undifferentiated working class with strong preferences for a comprehensive and universal welfare state and views it as being the “natural ally” of leftist parties. Indeed, several studies have revealed that it is especially the members of the lower social classes who support redistributive welfare policies (e.g., Svallfors, 1997; Svallfors, 2006). However, while this theory seems to explain welfare state support in the context of a relatively self-contained welfare state, it runs into trouble when it comes to support for the inclusion of foreigners. The question to be raised here is whether, in the face of immigration, the egalitarianism ascribed to the working class proves to be severely limited. In other words, do the universalism and egalitarianism associated with the working class fail when it comes to the willingness to support the inclusion of foreigners? Most of the empirical studies in the field provide supportive

122

Mewes and Mau

evidence of this: They demonstrate that it is particularly members of the working class who are prone to draw group-specific boundaries when it comes to the idea of economic egalitarianism—that is, they tend to limit their solidarity to their own ethnic (or national) group (e.g., Kitschelt, 1997; Raijman et al., 2003; Gorodzeisky and Semyonov, 2009; Green, 2009). This resentment vis-à-vis the inclusion of migrants in welfare benefits schemes and the granting of equal social rights has been labeled as “welfare chauvinism” (Kitschelt, 1997). Our chapter focuses on explaining this class-specific character of welfare chauvinism and draws on two different theoretical traditions. According to the different strands of group conflict theory (e.g., Blumer, 1958; Sherif, 1966; Blalock, 1967; Olzak, 1992; Bobo, 1999), it is the relatively weak economic position of the working class that determines the derogation of immigrants. On the other hand, a recent study by Van der Waal and colleagues (2010) suggests that welfare chauvinism might have cultural and value-related, rather than economic, roots. Socioeconomic Approaches to Welfare Chauvinism Why do members from the lower strata of society tend to support the exclusion of immigrants more often than those from the upper classes? A socioeconomic explanation is first of all provided by theories based upon a self-interest framework. According to the split labor market theory (Bonacich, 1972), a class bias as regards support for welfare chauvinism might result from differentials in the price of labor for the same occupation. At the core of this, it is suggested that, under conditions of labor cost differentials between native workers and a foreign labor force, nonimmigrants might be substituted by low-wage immigrants. In turn, nonimmigrant employees might react with hostility toward the competing immigrants (see Olzak, 1992: 23). In a European context, it has been suggested that “migrants are usually in direct economic competition with low-wage earners and manual laborers, as most immigrants are employed in manual labor” (Quillian, 1995: 591). Thus, it may be expected that members from the lower social strata are more likely to perceive the inflow of immigrants as an economic threat than do higher-status groups (Ceobanu and Escandell, 2010). However, the split labor market theory has been criticized for failing to explain variation in intergroup hostility across different countries and over time (Quillian, 1995). It has also been labeled as being relatively



Unraveling Working-Class Welfare Chauvinism

123

“mechanistic and deterministic” (Bernstein 1996: 246), since it tends to reduce racial and ethnic relations to economic conflicts alone. In the same way as the split labor market theory, the realistic conflict theory (Sherif, 1966; Blalock, 1967; LeVine and Campbell, 1972; Bobo and Hutchings, 1996; Olzak, 1992) assumes that intergroup conflicts might result from actual or perceived competition over scarce and valued social resources such as jobs, power, and status (see also Stephan and Stephan, 2000). Yet, according to Bobo and Hutchings (1996: 955), contrary to the former explanation, current contributions to the realistic conflict theory emphasize that “feelings of competition and hostility emerge from historically and collectively developed judgments about the positions in the social order that ingroup members should rightfully occupy relative to members of an outgroup” (italics added). According to Bobo (1999), ingroup membership itself is not very important in explaining sentiments against an outgroup. Instead, he emphasizes that the “sense of group position” (Blumer, 1958) becomes important only when members of the dominant group perceive their proprietary claims as being endangered by members from groups who are regarded as subordinate. In the context of our research question, access to welfare benefit schemes and rights might be framed as such a proprietary claim by “indigenous” citizens from European countries. From a classoriented research perspective, it is important to emphasize that intergroup conflicts of this kind should especially occur when individuals from both immigrant and nonimmigrant groups hold similar social positions—for example, “work in similar niches on the labor market” (Coenders et al., 2009). Realistic conflict theory focuses strongly on the social context of attitude formation, making it highly applicable for explaining attitudinal differences among countries (Quillian, 1995: 148). In line with our previous arguments, it can be expected that people from countries with relatively weak economic performance (as measured by their GDP and related growth indicators) turn out to be especially prone to welfare chauvinism. Indeed, empirical evidence has indicated that individuals who live in less affluent countries are more inclined to support welfare chauvinism (Green, 2009), demonstrate a lower level of solidarity with immigrants (Van Oorschot, 2007), and tend to be more prejudiced against foreigners (Quillian, 1995) than on average. In addition, there is also evidence that an increase in unemployment gives rise to anti-immigrant attitudes among Europeans (Meuleman et al., 2009).

124

Mewes and Mau

Authoritarianism and Welfare Chauvinism: A Sociocultural Approach In contrast to this, a recent study by Van der Waal and colleagues (2010) emphasizes that welfare chauvinism may have cultural rather than socioeconomic roots. The cultural-sociological explanation of working-class affinity to welfare chauvinism is inspired by literature explaining changing voting behavior that—in addition to taking into account the “old politics” dimension (left-right dimension)—focuses on the authoritarian-libertarian dimension of “new politics” (Kitschelt, 1994; Kitschelt, 1997; Knutsen and Kumlin, 2005; Brooks, 2006). 2 In this context the authoritarian-libertarian distinction can help to explain why the working class is often not egalitarian (or universalist) at all when it comes to the inclusion of foreigners. It is said that the higher educated are more prone to adopt libertarian values, whereas the less educated are more in favor of an authoritarian ideological stance. This suggestion is backed by the empirically well-established correlation between education and authoritarianism (Adorno et al., 1950; Altemeyer, 1988; Houtman, 2003; Hello et al., 2006). Following, we discuss how authoritarianism is supposed to trigger welfare chauvinism on the one hand and how education is related to authoritarianism at the other hand. We start by providing a short overview of the concept of authoritarianism. Research into authoritarianism underlines that authoritarians hold prejudiced and hostile attitudes toward a variety of different outgroups. The seminal work of Adorno and colleagues (1950) tried to link multifold phenomena such as anti-Semitism, racism, and ethnocentrism to a single underlying psychological dimension: the authoritarian personality. Though the concept attracted many social psychologists, it faced several criticisms on both theoretical and empirical grounds (Altemeyer, 1981; Altemeyer, 1988). One of the fallacies of Adorno and his colleagues was that they never really defined authoritarianism.3 John Duckitt (1989) redefined the concept of authoritarianism in order to overcome several difficulties in the original concept. He especially comes to terms with the “problem of reductionism” characteristic of earlier approaches: While authoritarianism research started with the aim of explaining collective social behavior (Stellmacher and Petzel, 2005), according to Duckitt (1989: 68), most often the construct “was reductionistically forced into a purely individualistic mold by being conceptualized as a personality trait dimension.” In order to tackle



Unraveling Working-Class Welfare Chauvinism

125

this problem, Duckitt conceptualized authoritarianism as a group phenomenon. In Duckitt’s model, authoritarianism is regarded as the individual “perception of what the appropriate relationship between the group and the individual should be” (Duckitt, 1989: 72). Conceptually, authoritarianism is treated as one dimension with two extremes: At one extreme would be the belief that the purely personal needs, inclinations, and values should be subordinated as completely as possible to the cohesion of the group and its requirements. At the other extreme would be the belief that the requirements of group cohesion should be subordinated as completely as possible to the autonomy and self-regulation of the individual member. These two extreme positions could be labeled authoritarianism and libertarianism. . . .

Hence, in conceptualizing authoritarianism as a group phenomenon (see Stellmacher and Petzel, 2005), the construct becomes highly interesting for sociological research. Still, from our discussion it is not clear how authoritarianism might relate to welfare chauvinism and what, after all, class membership might have to do with it. In this regard, it is useful to recall that Lipset’s (1959) seminal study on “working-class authoritarianism” was, on empirical grounds, criticized for overlooking that it seems to be education rather than class per se that explains differences in authoritarianism (Houtman, 2003; Svallfors, 2006: 125). Indeed, by now it is well established that education has a direct and negative effect on authoritarianism (Altemeyer, 1988; Hello et al., 2006; Stubager, 2008). It can be thus suggested that education might be a suitable predictor for authoritarianism (see Van der Waal et al., 2010). In the same manner as theorized in the socioeconomic explanation for welfare chauvinism, perceived threat by outgroups is pivotal to understanding how authoritarian dispositions might translate into manifest outgroup hostility (in the context of our research question: support for welfare chauvinism). By and large, it can be assumed that threats “activate authoritarian behaviors in persons who already possess authoritarian dispositions” (Stellmacher and Petzel, 2005: 248). Only persons with authoritarian dispositions are expected to show a negative reaction toward other groups in situations that are subjectively perceived as threatening. Recent experimental studies have shown that the situational activation of authoritarian dispositions does seem to be a strong indicator of prejudice against ethnic outgroups (Cohrs and Asbrock, 2009; Duckitt and Sibley, 2010). ­According

126

Mewes and Mau

to Van der Waal and colleagues (2010: 360), the cultural explanation of welfare chauvinism might empirically even outweigh the socioeconomic one completely: They suggest that “a weak economic position has no impact on welfare chauvinism whatsoever, while a weak cultural position does, and quite strongly so.” In sum, comparing the two alternative explanations for the support of welfare chauvinism, the causal chain of the cultural explanation for welfare chauvinism might, on the face of it, seem to be very similar to the socioeconomic one. Nevertheless, the mechanism that makes working-class members relatively prone to support the exclusion of immigrants is fundamentally different according to the two approaches. What both of the alternative models have in common is that they depict the willingness to support welfare chauvinism as being largely dependent on the element of the subjective perception of threat caused by ethnic outgroups (see Kunovich, 2004: 26; Sides and Citrin, 2007: 478). Yet, according to the socioeconomic explanation, it is primarily the comparatively low status in the societal hierarchy of members of the working class that makes them susceptible to perceiving a threat from immigrants, whereas the cultural explanation for welfare chauvinism highlights the prevalence of authoritarian dispositions as being of key importance for the formation of anti-immigrant attitudes. h y p o t h e se s From our theoretical discussion of the relationship between class and welfare chauvinism, we derive several working hypotheses in the following. Because our analysis refers to the influence of factors at both the national and individual levels, two kinds of hypotheses are derived. First, we develop hypotheses at the individual level, since our main purpose is to test the rival hypotheses derived from the socioeconomic and the cultural-sociological approaches to welfare chauvinism. In a second step, we present additional hypotheses at the country level. Here, our research interest lies in gauging important contextual factors that might mediate or moderate welfare chauvinism. Individual-Level Hypotheses First of all, our empirical analysis is based on the premise of a negative relationship between class membership and support for welfare chauvinism. In consequence, our first working hypothesis reads as follows:



Unraveling Working-Class Welfare Chauvinism

127

Members from upper social classes are less prone to welfare chauvinism than members from lower social classes (H1).

We expect a class-based hierarchy with regard to welfare chauvinism, with the members of the highest-class position displaying less support for welfare chauvinism and the members on the lowest end of the class scale as being the most attracted to the idea of setting barriers to the inclusion of foreigners. Assuming that we are able to identify such a class-based pattern in regard to the support of welfare chauvinism, we turn to the cultural-­sociological theory of working-class authoritarianism (see Van der Waal et al., 2010). In a first step, we omit class membership from our analysis and substitute it by two related dimensions: education and socioeconomic status. In line with Van der Waal and colleagues, we assume that education is an indicator of authoritarianism. In monitoring respondents’ socioeconomic status, we also control for respondents’ perceptions of their own material risk and vulnerability (Mau et al., 2012; see also Chapter 3). Thus, our approach emphasizes that material well-being is not an objectively measurable dimension only but that it is driven to a large degree by circumstances different from the labor market position an individual holds or the income that he or she earns. In this regard, our aim should be to ensure that the two dimensions of education and socioeconomic status significantly impact on the probability of support for welfare chauvinism so it can be warranted that class is indeed correlated with these two parameters. Therefore, we hypothesize the following: The higher the educational attainment, the higher the income, and the lower the level of perceived material risk, the lower the proneness to welfare chauvinism (H 2).

Finally, we test the hypothesis that it is authoritarianism specific to the working class, rather than its members’ comparably weak economic status, that drives their welfare chauvinism. Therefore, in addition to the aforementioned covariates, in the next step, we control for individual authoritarian dispositions. Our primary aim is to test whether the effect of socioeconomic status is outweighed by the cultural one, as in the (noncomparative) study by Van der Waal and colleagues. Our hypothesis accordingly reads as follows:

128

Mewes and Mau Controlling for education and authoritarianism, the effect of income and perceived material risk on the probability of supporting welfare chauvinism should disappear (H3).

Country-Level Hypotheses In the second step of our empirical analysis, we test for the effects of a number of country-level characteristics. Referring to the diverse strands of realistic conflict theory, we begin by taking into account both economic and immigration measures, assuming that the way individuals evaluate the overall economic conditions of the country they are living in impacts on their readiness to support the inclusion of foreign newcomers. With respect to economic factors, we first introduce “classic” indicators such as the national unemployment rates, the respective development during the last five years (2003–2008), the GDP per capita, and its development during the years from 2003 to 2008. In this regard, it might be expected that both the current level of unemployment as well as the unemployment growth rate (2003–2008) would reveal a positive impact on welfare chauvinism, whereas GDP per capita and the respective growth rate (2003–2008) are assumed to negatively impact on welfare chauvinism. Also, to our knowledge, we will be the first to control for the national sectoral composition of the labor market. More precisely, we focus on the share of the workforce employed in industry and agriculture, assuming that it is especially those two sectors that attract most of the low-skilled immigrants coming from outside the OECD to find a job in Europe (see Ceobanu and Escandell, 2010). We assume that job competition between the unskilled or low-skilled members of the native population and immigrants might especially occur in the context of low-income employment in the industrial and agricultural sectors. As a final measure for countries’ economic conditions, we employ the aggregate country-means of the perceived material risk scale, thereby being able to take into account an overall measure for the subjective economic climate in the countries of our sample. In addition, we also control for the level of immigration, assuming that the inflow of foreigners (measured in absolute numbers) might, in line with realistic conflict theory, increase the willingness to support exclusionary policies with regard to immigrants (Quillian, 1995).4 In sum, we derive the following hypothesis from realistic conflict theory:



Unraveling Working-Class Welfare Chauvinism

129

People who live in countries that are relatively economically prosperous, that exhibit high shares of the labor force working in the service sector, and that are inhabited by a comparably low number of foreigners are less prone to support welfare chauvinism than individuals living in countries with comparably precarious economic conditions, high shares of the labor force working in the industrial and agricultural sectors, and high numbers of foreigners  (H4).

Whereas there are several potential hypotheses that focus on the role of the country level in the formation of public opinion on immigration that may be derived from realistic conflict theory, the knowledge about sociocultural macro-level factors that potentially trigger welfare chauvinism is limited. Thus, we concentrate on the aggregate means of authoritarianism at the country level only. These country-specific means of our authoritarianism scale are labeled “collective authoritarianism” in the following. Overall, we assume high levels of collective authoritarianism to eventually affect even those who are relatively libertarian, making them more prone to support welfare chauvinism compared to those living in countries with lower levels of collective authoritarianism. Therefore, we hypothesize the following: Individuals who live in countries with relative high levels of collective authoritarianism are more supportive of welfare chauvinism than individuals who live in countries with relatively low levels of collective authoritarianism (H 5).

data a n d m e t hod The main objective of our chapter is to scrutinize the relationship between class and welfare chauvinism in a cross-national perspective. Like all the chapters in this book, we use data from the fourth round of the European Social Survey, fielded in late 2008 to early 2009. This survey offers information about respondents from 28 countries. However, due to the listwise deletion of missing data that was necessary for our data analysis, and also because of a lack of valid macroindicators, we had to reduce our sample to 25 countries, containing responses from some 34,947 individuals. 5 The included countries are (in alphabetical order) Belgium (n = 1,524), Croatia (n = 628), the Czech Republic (n = 1309), Denmark (n = 1263), Estonia (n = 1249), Finland (n = 1969), France (n = 1780), Germany (n = 2149),

130

Mewes and Mau

Greece (n = 1160), Hungary (n = 1047), Israel (n = 1381), Latvia (n = 1471), the Netherlands (n = 1513), Norway (n = 1459), Poland (n = 1211), Portugal (n = 816), Romania (n = 1126), the Russian Federation (n = 1702), Slovenia (n = 900), Spain (n = 1527), Sweden (n = 1613), Switzerland (n = 1322), Turkey (n = 1738), Ukraine (n = 1143), and the United Kingdom (n = 1947). In the following section, we first introduce the response variable. We continue by discussing the covariates of interest. This section concludes with a discussion of the appropriate multilevel techniques that were used to estimate our statistical models. Response Variable To measure individual support for welfare chauvinism, we used item D38 of the ESS 2008. The wording of the respective question is: “Thinking of people coming to live in [country] from other countries, when do you think they should obtain the same rights to social benefits and services as citizens already living here?” The respective response categories are: 1. “Immediately on arrival.” 2. “After living in [country] for a year, whether or not they have worked.” 3. “Only after they have worked and paid taxes for at least a year.” 4. “Once they have become a [country] citizen.” 5. “They should never get the same rights.”

Our indicator is quite novel because it does not ask, as most of the other surveys do, whether foreigners should be granted equal rights at a very general level. Rather, it differentiates between different types of statuses that largely mirror legal categories. Whereas, as Gorodzeisky (2011: 100) says, most studies in the field “have not distinguished conceptually between public support for exclusion and other forms of hostility toward foreigners and have unfortunately used the terms of prejudice, discriminatory attitudes, hostility, antiforeigner sentiment, and exclusionary views interchangeably,” the specific wording of the ESS4 item assesses Europeans’ support for welfare chauvinism quite well. In our statistical multinomial multilevel models, the first category (“immediately”) serves as the reference category. It is important to note that we regard the last category (“never”) as being the most valid indicator for welfare chauvinism. Even though the penultimate category (“once they



Unraveling Working-Class Welfare Chauvinism

131

have become a citizen”) reflects welfare chauvinism to a certain degree, the very process of naturalization differs markedly regarding the necessary requirements across countries (see, e.g., Janoski, 2010, with respect to varieties of naturalization practices in the OECD). For this reason, we concentrate on the estimates for the category “they should never get the same rights” only. A word of caution with respect to the concept of “immigrant” across the different countries in our sample: Since the ESS4 does not allow measuring welfare chauvinism on a multi-item scale, we cannot say whether people actually understand the same thing when they declare their readiness to grant immigrants equal rights and access to welfare benefits (see, e.g., Meuleman et al. (2009) on measurement equivalence across countries). In addition, to cite the rather extreme case of Israel, the term “immigrant” does not distinguish between incoming members from privileged majority groups, such as Jewish immigrants coming to Israel, and those belonging to disadvantaged minorities, such as noncitizen Palestinians working in Israel (see Raijman, 2010: 89). Similarly, our response variable does not allow taking into account historical particularities: For example, until the late 1990s, Germany clearly distinguished between ethnic Germans (Aussiedler) and other groups of immigrants—a distinction with quite different consequences regarding immigrants’ entitlement to rights and their access to welfare benefits. Certainly, our analysis might be affected by these and other intercultural biases regarding the concepts of immigrants and immigration. Yet, since we are interested in the general relationship between class and welfare chauvinism—and against the background of a lack of richer comparative data—our analyses should still be regarded as an important contribution to the question of how different contextual settings might affect class-specific views on foreign “outsiders.” Covariates Because we conduct multilevel regression analyses, we are able to introduce both variables at the country and individual levels in our statistical models. This subsection starts with a discussion of the covariates at the individual level, followed by a presentation of the country-level variables of interest. Basically, our conceptual model rests on the premise that class affiliation is linked to welfare chauvinism. Hence, we employ the widely used

132

Mewes and Mau

EGP class schema to measure the respondents’ class affiliations in the first step of our multivariate analysis (see also Chapter 6). Erikson and Goldthorpe (1992: 37) explain, “The aim of [this] class schema is to differentiate positions within labor markets and production units or, more specifically, one could say, to differentiate such positions in terms of the employment relations that they entail.” Respondents’ class categories were derived from the ISCO-88 occupation categories available in the ESS 2008. In order to “translate” the respective categories into the EGP scale, we referred to a slightly modified version of the Ganzeboom/Trento SPSS program (Leiulfsrud et al., 2002) that was developed for use in the ESS (our modifications were only due to a change in the ESS variable set since the release of the first round of the ESS in 2002). Here we use a six-class EGP schema that consists of six categories: service class I (EGP I); service class II (EGP II); routine nonmanuals (EGP III); small self-employed with and without employees (EGP IV); skilled manual workers and lower-grade technicians (EGP V and VI); and semi- and nonskilled agricultural and manual workers (EGP VII, used as the reference category in the multivariate analysis). Respondents for whom the EGP schema was not applicable (students, homemakers, unemployed, retirees, etc.) were categorized in the category “other.” For reasons of clarity, results for this last category are not shown in the following discussion of the empirical results. In order to measure education, we use the number of years the respondents have spent in educational systems (years of education). In order to measure the effect of income, which is one of the two indicators for the material well-being of individuals, we refer to the monthly household’s net income divided through the square root of the number of household members. In the ESS 2008, the income variable is already harmonized according to the country mean, so the ten categories are the ten income deciles of the respective country. As a further measure of the respondents’ socioeconomic situations, we include the level of perceived material risk (Mau et al., 2012) in our set of independent variables. To tap this subjective risk, we use three items from the ESS 2008: D47, D49, and D50. The exact wording of these items was as follows: Please tell me how likely it is that during the next 12 months . . . 1. you will be unemployed and looking for work for at least four consecutive weeks.



Unraveling Working-Class Welfare Chauvinism

133

2. there will be some periods when you don’t have enough money to cover your household necessities. 3. you will not receive the health care you really need if you become ill.

The response categories ranged from 1 (“agree strongly”) to 5 (“disagree strongly”). The reliability of this perceived material risk scale is satisfactory (Cronbach’s alpha: .72). As a means of gauging authoritarianism, we refer to the three items “Schools must teach children to obey authority”; “People who break the law should be given much harder sentences than they are these days”; and “If a man is suspected of planning a terrorist attack in [country], the police should have the power to keep him in prison until they are satisfied he was not involved” (we use the same authoritarianism scale as in Chapter 4). The possible answers ranged from 1 (“agree strongly”) to 5 (“disagree strongly”). From these three items, we computed an authoritarianism scale, with higher scores indicating stronger tendencies toward authoritarianism. In this way, our authoritarianism scale fits well in the conceptual framework of Duckitt’s (1989: 71f.) three dimensions of authoritarianism: conventionalism, authoritarian submission, and authoritarian aggression. Due to data restrictions, we can here concentrate on the third element of authoritarianism only (see Chapter 4). Our scale pertains to the following questions in Duckitt’s theoretical framework: 1. How severe should be or are punishments and condemnation for nonconformity to group norms and rules? 2. How strictly are or should be such punishments and condemnation for nonconformity enforced?

Although the reliability of this scale is not fully satisfying (Cronbach’s alpha is .55), the use of the authoritarianism scale seems to be justified given that it contains only three items (Cortina, 1993). Additionally, we added a relatively small number of individual-level control variables that are supposed to either foster or weaken welfare chauvinism. First, we assume that the older the people, the lower the probability of supporting the inclusion of immigrants (Coenders and Scheepers, 2003). Therefore, we control for respondents’ ages. Second, we monitor the size of the community in which the respondents are living, hypothesizing that cities and bigger towns are, in contrast to villages and rural areas, more cosmopolitan places, offering the possibility for various intercultural ­encounters

134

Mewes and Mau

(Sassen, 1991) that in turn might positively impact on people’s attitudes toward the culturally different (Mau et al., 2008). Third, we control for the gender of the respondents, though few results point to a significant gender effect in regard to support for welfare chauvinism. Finally, we contend that our results could be severely biased by respondents with a migratory background. Therefore, we employ a dummy checking whether the respondents have either only foreign citizenship, if they were born outside the respective country in question, or whether any parent was born abroad. If any of the given possibilities applies, the respective respondent will be indicated as having a migratory background. In terms of estimating the effect of economic macro-level indicators, we focus attention on the possible impacts of unemployment (and its change over time), economic output, and economic development over the past five years. According to the realistic conflict theory, welfare chauvinism should be especially marked in those countries that score rather poorly on all those economic indicators. Additionally, we analyze how the sectoral composition of the country’s labor force affects support for anti-immigrant attitudes. Also, based upon realistic conflict theory, we assume that respondents from countries in which comparably large shares of the population work in the industrial and agricultural sectors are more supportive of welfare chauvinism than respondents living in countries in which larger shares of the economically active population work in the service sector. In order to quantify the country-specific economic strength and its development over time, we employ two measures: First, we monitor for GDP per capita (Purchasing Power Parity, US$) in the year of the interview (2008). The values were divided by 1000 to obtain interpretable results. Additionally, we control for economic growth between 2003 and 2008 (we decided to look at a five-year time window in regard to potential countrylevel effects over time, see Kunovich 2004: 32). Here, we introduce the absolute change of the GDP per capita (as measured in PPP, US$) between 2003 and 2008. As a means of registering the possible effects of unemployment, we control for the share of unemployed at the time of the interview (2008), as well as for the respective change over time (referring to the period 2003–2008). In the latter case, we conducted linear regressions on the unemployment rates over the period 2003–2008. The respective regression slopes are used as macro-level determinants in our multilevel models. In line with realistic conflict theory, we also assume that objections to the



Unraveling Working-Class Welfare Chauvinism

135

equal treatment of immigrants and natives should be most marked in those European countries whose labor markets are easily entered by unskilled migrants. To measure this aspect, we introduce the country-specific share of the labor force working in the agricultural and industrial sectors as an additional covariate. Similarly, we employ the national share of industrial and agricultural output that amount to the country’s overall GDP. Data for the economic characteristics of the 25 countries included in our statistical sample were taken from Eurostat, the OECD, and IndexMundi (www .indexmundi.com). Because attitudes toward immigrants should not be investigated without also taking into account the role of immigration in the countries in question, we introduce the absolute numbers of foreigners living in each of the analyzed countries as an additional macro-level determinant in our analyses. With two exceptions, our immigration data stem from Eurostat. Due to a lack of appropriate Eurostat data in two cases, we decided to use NationMaster (www.nationmaster.com) to include information on Bulgaria and the Ukraine. Because of the general problem in obtaining valid and rich country-specific information on immigration for all the countries in our sample, we could not monitor additional, more sophisticated measures, such as the share of foreign-born citizens or the structure of immigration (e.g., refugees vs. labor immigrants; OECD migrants vs. non-OECD migrants). Finally, we also employ two macro-level indicators that were obtained by calculating the aggregate means of two individual-level variables: authoritarianism and perceived material risk. In this way, we are able to measure how collective authoritarianism and collectively perceived risk impact on the willingness to support immigrants’ exclusion among the Europeans in our sample. Method Because the response variable is nominal (see discussion above), we applied unordered multinomial logistic multilevel analyses. All of the multilevel models were calculated with MLWIN 2.20, using second-order penalized quasi-likelihood estimation (PQL), which is regarded as superior compared to first-order maximum quasi-likelihood estimation (MQL), given that it produces more reliable level 2 estimates and more precise standard errors (Hox, 2010: 119–21). Also, a sandwich estimator was used to obtain robust

136

Mewes and Mau

standard errors, which are regarded as lowering the influence of statistical outliers, and to be more reliable in cases in which the data violate the assumption of normality (Maas and Hox, 2004). While we used the reference category “immediately after arrival” as the reference category of our multinomial multilevel models throughout the chapter, in the footnotes we also briefly discuss which kinds of results are yielded when changing the models’ reference category to “after a year whether or not have worked.” As will be discussed, the findings from the two alternative statistical approaches are very similar, allowing us to conclude that our models are relatively robust and stable. r e su lt s Let us now turn to our empirical analyses. We start by having a first look at the country-specific response patterns regarding the question as to when immigrants should be granted access to welfare benefits and equal rights. Table 5.1 summarizes the distribution of responses in the countries of our sample. As a first step, we consider which countries score especially high with regard to the support of welfare chauvinism. Descriptive Findings According to our findings, it is especially the Eastern European countries in which a considerable share of respondents opts for the possibility of setting high barriers in respect to the inclusion of immigrants. In Hungary, Latvia, the Czech Republic, and the Russian Federation, more than every tenth respondent exhibits welfare chauvinism. In Turkey, the figure is even 19 percent of the respondents who express the wish for the exclusion of immigrants. Other Eastern European countries in our sample score higher on welfare chauvinism than on the European average, too (Ukraine: 9 percent; Romania: 8 percent). Notable exceptions from this pattern are only Poland (2 percent) and Estonia (3 percent). Thus, our results are clearly at odds with those from Coenders and colleagues (2009), who claimed there would be no empirical evidence for the supposition that anti-immigrant sentiments would be stronger in (former) EU candidate countries than in the European average (which amounts to 7 percent in our sample). A large share of interviewees who favor the exclusion of migrants from equal treatment and access to welfare benefit schemes is also found in Greece



Unraveling Working-Class Welfare Chauvinism

137

ta bl e 5.1 Support for the inclusion of immigrants: Country-specific response patterns (percentage of respondents within countries) when should immigrants obtain rights to social benefits/ services?

Country Belgium Croatia Czech Republic Denmark Estonia Finland France Germany Greece Hungary Israel Latvia Netherlands Norway Poland Portugal Romania Russian Fed. Slovenia Spain Sweden Switzerland Turkey Ukraine United Kingdom

Immediately on arrival 6.1 7.8 2.4 13.2 4.6 5.0 12.2 8.4 9.9 1.4 22.4 4.5 7.8 13.5 5.2 11.6 6.7 4.8 4.2 11.4 19.8 12.5 10.0 8.2 5.6

After a year, After worked whether or and paid not have taxes at least worked a year 10.4 4.8 5.5 15.1 5.2 12.1 10.7 11.8 4.2 3.1 18.7 3.2 9.7 12.7 7.5 13.4 7.1 5.0 5.1 9.2 16.2 12.9 10.2 6.6 5.4

48.4 30.1 35.1 32.1 40.7 37.0 47.0 43.4 38.3 29.8 16.7 37.7 35.2 34.5 38.4 55.4 27.4 31.7 33.3 52.1 31.1 56.1 18.9 28.2 48.6

Once they have become citizens

They should never get the same rights (welfare chauvinism)

29.0 57.3 42.1 37.5 46.4 43.1 25.1 29.9 32.8 51.9 35.0 40.2 44.5 37.4 46.7 15.7 50.4 46.2 50.7 21.7 32.1 15.7 41.9 47.9 31.5

6.0 0.0 14.9 2.1 3.2 2.7 5.0 6.5 14.8 13.8 7.2 14.4 2.8 1.9 2.1 3.9 8.3 12.3 6.7 5.6 0.7 2.9 19.0 9.2 8.9

s o u r c e : European Social Survey 2008; ni = 34,947; nj = 25. n o t e s : The question was “Thinking of people coming to live in [country] from other countries, when do you think they should obtain the same rights to social benefits and services as citizens already living here?” The respective response categories were 1—Immediately on arrival (unconditional inclusion); 2—After living in [country] for a year, whether or not they have worked; 3—Only after they have worked and paid taxes for at least a year; 4—Once they have become a [country] citizen; and 5—They should never get the same rights (welfare chauvinism).

(14 percent). The same cluster of countries that is supportive of welfare chauvinism in the form of complete exclusion scores high with regard to the category “inclusion based on citizenship.” Given the fact that naturalization in most of these countries is far from easy to achieve, we may conclude that the respondents of the Eastern parts of Europe plus those from Greece show a comparatively strong tendency toward welfare chauvinism.

138

Mewes and Mau

In contrast, Sweden and Israel display the highest level of support for the unconditional inclusion of immigrants. Concerning Sweden, we observe the strongest tendency toward lowering barriers for the inclusion of immigrants. While the share of respondents that opt for “never” is very small (less than 1 percent), about every fifth Swede in our sample calls for the immediate inclusion of foreigners in terms of access to social rights and services. In the same vein, Israel turns out to be highly supportive of welfare universalism (which means here that egalitarian measure should not be directed exclusively to the members of one’s own ethnic community). Yet, the Israeli sample turns out to be more biased than the Swedish one: Whereas every fourth Israeli opts for unconditional inclusion (which is the highest value of all countries in that category), 7 percent of Israeli respondents support welfare chauvinism. Here, public opinion seems to be relatively polarized. This observation is in line with the specific and ambivalent situation of Israeli immigration described above. Interestingly, respondents from Spain and Portugal also turn out to be—compared to most of the other European countries—extremely supportive when it comes to granting foreigners access to welfare benefits and equal social rights. At least in the Spanish case, this result seems to be a little bit surprising, since Spain has experienced a large amount of clandestine immigration during the last years. In general, we find that people living in the Scandinavian countries plus those living in Spain and Portugal tend to be more in favor of granting immigrants access to social rights and benefits than the European average, whereas the majority of Eastern European countries, as well as Greece, generally exhibit a relatively high level of welfare chauvinism. In any case, a large share of Europeans opt at least for the conditional inclusion of foreigners; that is, though they basically agree with granting foreigners access to social benefits and services, immigrants should fulfill certain conditions in order to obtain equal status (living and working in the respective country for at least one year, or just living in the country for a year). In contrast, only a minority of Europeans supports an immediate inclusion of people from foreign countries. This result is in line with other findings where people expressed reservations toward an unconditional inclusion of immigrant groups into the welfare state (Sefton, 2005). Taxpaying and employment seem to be crucial as regards the conditions for inclusion into the welfare system. Only a minority would deny access to the



Unraveling Working-Class Welfare Chauvinism

139

benefit system as long as migrants pay their “dues.” In fact, it seems that the willingness to redistribute in favor of others depends on whether they perceive them as playing fair—that is, “as showing willingness to reciprocate when it is their turn to make a contribution” (Miller, 2006: 334). Individual-Level Determinants of Welfare Chauvinism Let us now turn to the results of our multilevel analyses (for a descriptive overview in regard to level 1 covariates, see Web Appendix Table A5.1). For reasons of clarity, we only present results for the response category of interest (welfare chauvinism).6 In Table 5.2, Model I, we test the hypothesis that members of lower classes, compared to those from the higher classes, are more prone to support welfare chauvinism. Indeed, our data substantiate our hypothesis of a relationship between class affiliation and attitudes toward the inclusion of immigrants (H1). Compared to the reference group—that is, semi- and nonskilled workers in industry and agriculture— the service classes (classes I and II) and the routine nonmanual workers (class III) are significantly less in favor of an exclusionist ideological outlook to people immigrating into their countries. Based upon the observation of class-based differences with regard to the desired treatment of “newcomers” to Europe, we continue by testing the effects of the variables “income,” “perceived material risk,” and “education.” These explanatory factors turn out to be relatively strongly correlated with the class dimension. Indeed, a t-test (not shown) reveals that members of the highest EGP classes score the highest with regard to income as well as to years spent in the education system. On the other hand, it is the members of the working classes who score highest on perceived material risk. By excluding the EGP class measure and substituting it through the related factors of income, material risk, and education, it can be shown that all of these class subdimensions have a significant impact on the probability of support for welfare chauvinism (Table 5.2, Model II). Our second hypothesis (H 2), according to which the likelihood of support for welfare chauvinism declines with income, material well-being, and education, is thus corroborated by our data. In the next step of our analysis, we examine whether part of the variance may be explained by the level of authoritarianism (Table 5.2, Model III). Indeed, our results show that people exhibiting strong authoritarian dispositions have significantly higher odds of supporting welfare ­chauvinism

ta bl e 5.2 Determinants of welfare chauvinism (multinomial logistic multilevel models) model i Coefficient class (egp)a I II III IV V, VI Income

–1.537 (0.139) –1.237 (0.132) –0.721 (0.087) –0.227 (0.084) 0.032 (0.072)

model ii

Significance

Migratory background (1 = yes)

*

Community size Constant Variance (country-level)

0.005 (0.006) 0.000 (0.000) 0.097 (0.065) –1.574 (0.233) –0.099 (0.031)

Significance

*

Authoritarianism

Female (1 = yes)

Coefficient

*

Education

Age (squared)

Significance

*

Perceived material risk

Age

Coefficient

model iii

* *

–0.036 (0.014)

*

–0.026 (0.014)

0.191 (0.028) –0.123 (0.006)

*

0.168 (0.028) –0.090 (0.006) 0.871 (0.033) 0.003 (0.006) 0.000 (0.000) –0.115 (0.040)

*

0.014 (0.006) 0.000 (0.000) –0.089 (0.040)

*

–1.494 (0.081) –0.055 (0.017)

*

*

*

–1.505 (0.081) –0.027 (0.017)

0.542 (0.141)

–0.018 (0.123)

–0.037 (0.116)

0.414 (0.120)

0.351 (0.102)

0.309 (0.091)

* * *

*

*

s o u r c e : European Social Survey 2008; ni = 34,947; nj = 25. n o t e s : Reference category is “Immediately on arrival” (unconditional inclusion). aThe reference category in the class scheme is EGP class VII. For reasons of clarity, only results concerning the response category “They should never get the same rights” are shown. Unstandardized coefficients are stated, and standard errors appear in parentheses. Significant level * ≤ 0.05. Empty model: within-country variance π 2 /3 (the standard value in regard to multilevel logistic regression). Empty model: between-country variance 0.601 (0.173).



Unraveling Working-Class Welfare Chauvinism

141

than individuals who are more libertarian (as measured by our authoritarianism scale). According to the sociocultural approach to welfare chauvinism, the effect of income should diminish or even disappear (Van der Waal et al., 2010) when controlling for authoritarianism. Our analysis does indeed provide evidence for this. In fact, the effect of income not only diminishes, but it even becomes insignificant. In contrast, the significant positive impact of perceived material risk (which was employed to grasp how individuals evaluate their socioeconomic conditions subjectively) remains robust when controlling for authoritarianism. Therefore, our results are partly at odds with those from Van der Waal and colleagues (2010: 360), who claimed that in the Dutch case, “a weak economic position has no impact on welfare universalism whatsoever, while a weak cultural position does, and quite strongly so” (italics added). The proposition that the socioeconomic dimension does not play any role once we control for authoritarianism (H3) cannot be confirmed by our data. In sum, the first part of our analysis, which focuses on the level of individuals, reveals that both theoretical approaches to welfare ­chauvinism—the cultural one (Van der Waal et al., 2010) and the socioeconomic one—prove to be useful empirically. Country-Level Determinants of Welfare Chauvinism Taking into consideration that the introduction of our individual-level covariates also leads to a decrease of variance between countries (see Table 5.2 notes on statistical information about the empty model), we observe that some part of the between-country variance is due to compositional effects only. However, this decrease in variance compared to the “empty model” is only modest. Thus, there must be specific features at the country level that might explain part of the variance in support for the exclusion of immigrants. For this reason we calculated additional multinomial multilevel models containing detailed information about the countries in our sample. In this context, we used Model III in Table 5.2 as the baseline model and introduced the nine relevant macro-indicators derived from our previous discussion separately. We start by looking at the effects of the country-level indicators that were derived from our discussion of the realistic conflict theory. According to the theory, we would assume that high levels of unemployment might trigger perceptions of economic uncertainty, thereby encouraging the de-

142

Mewes and Mau

sire for excluding migrants from welfare benefit schemes. Indeed, our data corroborate this hypothesis: The higher the country level of unemployment, the more the respondents are likely to opt for welfare chauvinism as a reaction to the question of how to include immigrating “newcomers” (Table 5.3, Model IV). However, introducing a longitudinal indicator of unemployment (indicating an increase or a decline in national unemployment rates between 2003 and 2008), we observe a converse effect (Table  5.3, Model V): Surprisingly, it is people from countries with falling unemployment rates who tend to be supportive of welfare chauvinism. While our results might, at first glance, suggest that decreasing unemployment might in the long run make people even more supportive of welfare chauvinism, we would like to raise a word of caution: Due to the model’s marginal contribution to decreasing between-country variance and the rather high probability of error (p < .10), and also because of the regional clustering of countries with relatively strong drops in unemployment rates (it is predominantly Central and Eastern European countries that exhibit large decreases in unemployment between 2003 and 2008), the implications of Model V appear to be negligible. Now that we have discussed the impact of unemployment on welfare chauvinism, how does the overall economic performance of a country affect attitudes toward the inclusion of foreigners? To explore this issue we introduced two different measurements: GDP per capita in 2008 (in US$1000) and the change in GDP per capita (in US$1000) between 2003 and 2008. In line with our expectations, this covariate proves to be significant (Table 5.3, Model VI). According to the model, citizens in economically weaker countries are significantly more inclined to support welfare chauvinism than those from wealthier countries. Moreover, the decrease in betweencountry variance is relatively high in this model, compared to most of the other models that reveal significant impacts of macro-level indicators on our response variable. The economic performance of a country, therefore, as measured by GDP per capita seems to play a major role in the process of attitude formation (see Semyonov et al., 2006: 444). Our findings are in line with those from Sides and Citrin (2007: 496), who reported, on the basis of a comparative study of 12 European countries, that “countries with a higher GDP per capita were less hostile to immigration.” In addition, our results support the findings of Kunovich (2004), who in a sample

Unemployment rate 2008 (%) Unemployment rate (2003–2008, regression slope) GDP per capita in US$1000, 2008 GDP per capita in US$1000, growth (2003–2008, absolute) Share of foreigners in the population 2008 (%) Share of labor force in industry and agriculture (%) Share of labor force in industry and agriculture * Share of foreigners (interaction, main effects held constant) Collective material risk (country average of perceived material risk) Collective authoritarianism (country average of authoritarianism)

Model III (Table 5.2) + . . .

1.118 1.234

0.062 –0.437 –0.026 –0.060 0.033 0.017 0.005

Coeff.

0.213 0.511

0.028 0.240 0.010 0.047 0.020 0.012 0.002

S.E.

* *

*

* † *

Sig.

0.282 0.292

0.291 0.307 0.288 0.305 0.315 0.274 0.279

Country-level variance

–0.027 –0.017

–0.018 –0.002 –0.021 –0.004 +0.006 –0.035 –0.030

∆ decrease in variance compared to the baseline model

s o u r c e : European Social Survey 2008; ni = 34,947; nj = 25. n o t e s : Entries are results from nine separate multilevel models; the baseline model is Model III, Table 5.2. Thus, every model included all the individual-level covariates reported in Model III. * Sig. at p ≤ 0.05; † sig. at p < 0.10

XI XII

IV V VI VII VIII IX X

Model

t a bl e 5.3 Macro-level determinants of supporting welfare chauvinism

144

Mewes and Mau

of 17 European ­countries showed that individual determinants of prejudice were outweighed by contextual factors under poor economic conditions. He found, moreover, that the decoupling of individual-level determinants and prejudice seemed to be particularly pronounced in Eastern Europe. Accordingly, the Eastern European tendency to support welfare chauvinism in our findings might be interpreted as a consequence of this region’s poor economic strength rather than of the individual characteristics of the citizens. Nevertheless, further research has to address the question in more detail of which factors account for Eastern European welfare chauvinism. In contrast to the aforementioned results, the absolute growth of GDP does not seem to play any role in the formation of attitudes in regard to immigrants (Table 5.3, Model VII): Respondents from countries that grew fast in terms of the absolute purchasing power parity (measured in US$1000) between 2003 and 2008 do not tend to be less supportive of welfare chauvinism than those from nations that grew less quickly.7 We also observe that the share of immigrants in a country has no effect on support for welfare chauvinism (Table 5.3, Model VIII). From the point of view of realistic conflict theory, this result means that the sheer presence and number of foreigners do not necessarily cause welfare chauvinism. This result is in line with those from a comparative European study across 20 countries (Hjerm, 2007), in which it was demonstrated that neither the actual nor the perceived size of the foreign(born) population had an effect on the formation of anti-immigrant attitudes.8 In addition, we introduced the percentage of the labor force working in industry and agriculture as another explanatory variable in our model. We do not observe a significant impact of this covariate (Table 5.3, Model IX) in the first instance. However, this situation clearly changes when simultaneously employing the share of foreigners and the share of the labor force employed in industry and agriculture (Table 5.3, Model X). Now we observe that the two positive main effects become significant (not shown). Moreover, we also witness a significant interaction between the two covariates, indicating that it is especially respondents from countries with relatively high numbers of immigrants and with a comparably large share of the workforce employed in industry and agriculture that are particularly prone to adopt welfare chauvinist attitudes. Given the remarkable decrease in between-country variance brought about in this model, our results ­suggest



Unraveling Working-Class Welfare Chauvinism

145

employing the sectoral composition of the labor force in future studies that deal with the contextual forces of anti-immigrant attitudes. Next, we employed the country-level average from our perceived material risk scale as a further explanatory variable in our model (Table 5.3, Model XI). Interestingly, we found a significant and—given the reduction in variance between countries—quite strong effect. One could suggest that economic uncertainty might become “epidemic” in the sense that even people with low levels of risk perception might start to perceive immigrants as an economic threat due to the general social climate. Furthermore, this result could also be interpreted as an “identity” effect: Though people do not actually suffer from precarious economic conditions themselves, they could, nevertheless, develop perceptions of solidarity with their fellow citizens who are materially more vulnerable. Finally, we controlled for collective authoritarianism, assuming that countries in which comparably large shares of the population exhibit authoritarian dispositions might foster a more authoritarian climate in general, eventually leading to an overall increase in the readiness to support welfare chauvinism—even among those with more libertarian ideological outlooks. Indeed, this covariate reveals a positive effect on welfare chauvinism (Table 5.3, Model XII). Admittedly, our knowledge about the mechanisms that account for this effect of collective authoritarianism is rather limited. In sum, our results for the contextual factors corroborate our hypothesis derived from realistic conflict theory, according to which individuals, under conditions of comparatively precarious material conditions, might become more prone to develop anti-outgroup attitudes (H4). Furthermore, our data also support our hypothesis (H 5) concerning the role of collective authoritarianism: High aggregate scores of authoritarianism make individuals more prone to adopt welfare chauvinist attitudes than under conditions of collective libertarianism, despite their individual dispositions on the authoritarianism-libertarianism dimension.9 Predicted Probabilities Now that we have presented the results from our multinomial multilevel models, it still remains unclear how strong the key covariates of interest impact on the probability of opting for welfare chauvinism. Therefore, we predicted probabilities for a series of ideal-type cases. As a baseline, one should

146

Mewes and Mau

consider that, according to our multilevel baseline model (Table 5.2, Model III), the average probability of opting for welfare chauvinism would be p = .06 (or 6 percent), given that we controlled for the average respondent of an average country scoring on average on each of the employed covariates. Tables 5.4 and 5.5 present the results from several prediction procedures we undertook with the help of the “customized predictions” feature in MLwiN 2.20. The “min” column depicts the probability of support for welfare chauvinism, given that an individual has the minimum observed value at the covariate of interest and that he or she scores on average for all other covariates. In contrast, the “max” column indicates the probability of favoring the exclusion of immigrants from access to social benefits and rights, given that an individual has the maximum observed value on the explanatory variable at stake and that he or she scores on average with regard to all other explanatory variables in our model. The same logic applies to the predictions at the macro level. In Table 5.5, we combined some of our key covariates to demonstrate their maximal impact on the support of welfare chauvinism, as predicted by our multinomial multilevel model. Nevertheless, whereas the examples for the predicted probabilities might appear to be somewhat hypothetical, it has to be noted that the examples for the minimum and maximum values were estimated on the basis of data that were actually observed in our data set. Given the scenarios in Tables 5.4 and 5.5, we now know more about how strongly the key covariates of interest impact on the probability of support for welfare chauvinism. For example, changing class position from EGP class I (service class I) to class VII (nonskilled manual workers) increases the probability of support for welfare chauvinism, all other things being equal, from 4 percent to 11 percent. Because we are interested in disentangling the relationship between class and welfare chauvinism, we estimated a set of customized predictions in regard to the four class-related covariates of interest. While we clearly see (Table 5.4) that education seems to be the strongest individual-level determinant of welfare chauvinism (with a shift from p = .01 in the case of the lowest educational level to p = .12 in the case of the highest one), we cannot say with certainty that this bias stems primarily from differences in the prevalence of authoritarian dispositions, given that the related probability differential amounts to only 7 percentage points. In the case of perceived material risk, the difference between the observed maximum value and the

ta bl e 5. 4 Predicted mean probabilities to support welfare chauvinism (EGP)†

Class Education Income Authoritarianism Perceived material risk Unemployment rate GDP per capita Collective material risk

Min

Max

.11 .12 .08 .02 .05 .05 .09 .04

.04 .01 .05 .09 .08 .08 .04 .11

s o u r c e : European Social Survey 2008; ni = 34,947; nj = 25. n o t e s : Predicted probabilities based on Model III, Table 5.3 (except † class, which is based on Model I, Table 5.2, not including income, education, authoritarianism, and risk; Class (min) = class VII; Class (max) = class I. For all those variables not appearing in the rows of this table, we used the respective mean. For example, given an individual who has the minimum observed value on the education variable and who scores an average (in the case of continuous variables, on the group-centered mean) on all the other explanatory variables (gender, age, migratory background, community size, income, and perceived material risk), we predict a probability of p = 0.12 (12% “chance”) to support welfare chauvinism (example for the interpretation of row 2).

t a b l e 5 .5 Predicted mean probabilities for a range of combinations with regard to key covariates Education (min.) and authoritarianism (max.) Education (max.) and authoritarianism (min.)

.16 .01

Education (min.); authoritarianism (max.) and perceived risk (max.) Education (max.); authoritarianism (min.) and perceived risk (min.)

.22 .00

Education (min.); authoritarianism (max.); perceived risk (max.) and GDP (min.) Education (min.); authoritarianism (max.); perceived risk (max.) and GDP (max.) Education (max.); authoritarianism (min.); perceived risk (min.) and GDP (min.) Education (max.); authoritarianism (min.); perceived risk (min.) and GDP (max.)

.28 .15 .00 .00

s o u r c e : European Social Survey 2008; ni = 34,947; nj = 25. n o t e s : Predicted probabilities in regard to the first four cases based on Model III, Table 5.2. Predicted probabilities for the last four cases based on Model VIII, Table 5.4. For all those variables not appearing in the rows of this table, we used the respective mean. For example, given an individual who has the minimum observed value on the education variable and the maximum observed value on the authoritarianism scale, and who scores average on all the other explanatory variables (gender, age, migratory background, community size, income, and perceived material risk), we predict a probability of p = 0.16 (16% “chance”) to support welfare chauvinism (example for the interpretation of row 1).

148

Mewes and Mau

observed minimum value is just .03 in terms of a change in the probability of support for welfare chauvinism. Thus, Van der Waal et al. (2010) might be partially right in calling for a cultural-sociological turn as regards the analysis of welfare attitudes. Yet, when the national context of attitude formation is taken into consideration as well, the effect of socioeconomic context becomes striking. This is not so much because of the primary effects of GDP, unemployment, and risk at the country-level, but rather because the interplay of individual-level and country-specific characteristics and the nonlinear nature of logistic models point to the importance of engaging contextual factors. Table 5.5 (examples 3 and 4) depicts a scenario in which two ideal types are contrasted: Whereas the individual-level characteristics are the same in these two cases (with an otherwise average respondent scoring highest on authoritarianism and perceived risk and lowest on education), we only altered the value of GDP per capita, employing both the highest and the lowest GDP value of our sample. In this context, the probability of support for welfare chauvinism dramatically increases from 15 percent (in the case of a country with the highest GDP in our sample) to 28 percent (when employing the lowest GDP value of the sample). In contrast (see Table 5.5, examples 5 and 6), the manipulation of GDP does not alter the situation significantly when looking at an ideal-type respondent with the highest level of education and the lowest levels of perceived risk and authoritarianism (p = .00 in both of the examples). In sum, our results are a telling example for the empirical appropriateness of realistic conflict theories in comparative analyses (Quillian, 1995; Kunovich, 2004). c on c l usion In this chapter, we addressed the question of why working-class members, when compared to members from higher social strata, are more likely to adopt welfare chauvinist attitudes. Welfare chauvinism may be understood as the preference to grant entitlements to rights and access to welfare state benefits only to the members of one’s own ethnic community (see Kitschelt, 1997). By and large, theories of realistic or perceived group conflict (e.g., Sherif, 1966; Blalock, 1967; Olzak, 1992; Quillian, 1995) predict that antiimmigrant sentiments in low-skilled and less educated groups refer to a relatively weak socioeconomic position in the (realistic or perceived) com-



Unraveling Working-Class Welfare Chauvinism

149

petition between “native” blue-collar workers and immigrants coming to Europe (Ceobanu and Escandell, 2010). Resting on this, it can be suspected that native working-class members are especially prone to subjective feelings of ethnic competition over scarce and valued social resources (Scheepers et al., 2002). Other studies suggest that it is less (perceived) competition but rather authoritarian dispositions in the working class that underpin its welfare chauvinism (Van der Waal et al., 2010). We tried to disentangle the roots of the class-specific adoption of welfare chauvinist attitudes from a cross-national perspective. Taking into account income and the level of perceived material risk as indicators for the individuals’ socioeconomic situation on the one hand and authoritarianism on the other hand, it could be shown that neither of the two explanations are wrong: In line with the results by Van der Waal and colleagues (2010), the effect of income disappeared by controlling for authoritarian dispositions. At the same time, however, our results also revealed the robust and strong effect of perceived material risk on welfare chauvinism. Thus, both of the dimensions, the material one as well as the cultural one, ultimately seem to matter for the explanation of welfare chauvinism and its relation to class. Moreover, our analysis sheds some light on the contextual factors that moderate the interplay between class position and support for welfare chauvinism. With respect to economic factors, economic output (GDP) especially appears to be a strong moderator for the support of welfare chauvinism among the less privileged social strata. For example, the estimated average probability of support for welfare chauvinism nearly doubled among the high-risk group of the precarious, authoritarian, low-skilled working-class members when controlling for the highest observed GDP value instead of for the lowest one. In contrast, economic conditions do not seem to account at all for changing the likelihood of support for welfare chauvinism among members of the “low-risk” group: According to our model, those with relatively high levels of education, low levels of subjective material risk, and antiauthoritarian dispositions are hardly at risk of becoming welfare chauvinists, regardless of the economic conditions they live in. Thus, our results emphasize once again the importance of carefully treating results from logistic regressions (Mood, 2010). In sum, there seems to be no reason to disregard economic factors, especially at the country level, when examining welfare chauvinism and

150

Mewes and Mau

related phenomena. Based on our findings, however, we have clear indications that conventional measures of class, education, and income (at the individual level) are of limited use in understanding the role of risk, security, and precariousness. Hence, for the explanation of welfare chauvinism, one must not only take into consideration “objective” cultural and economic conditions, but also individual perceptions and expectations. From a broader policy-oriented perspective, our results may also be of some relevance regarding the question of how to develop measures to deal with populist right-wing parties in European countries. Empirical studies have shown that “workers and the old middle class [self-employed craftsmen and small shop owners, the authors] are . . . overrepresented among new radical right voters” (Rydgren, 2007: 249). Clearly, anti-immigrant policies are at the very core of the programs of right-wing populist parties. Whereas our findings do not allow us to say why certain groups are actually more prone to vote for these parties than others, our results give some hints as to why working-class members might be attracted by the antiimmigrant rhetoric of the radical right. In terms of the question of how to possibly confront the working-class tendency to support chauvinism politically, our results appear to be quite ambivalent. Since authoritarian dispositions are generally considered to be learned and socialized during childhood and adolescence, the cultural explanation for welfare chauvinism provides little reason to hope for short-term solutions. Certainly, in the long term, an ongoing expansion of national educational systems might be beneficial to the weakening of authoritarian values and thus for the spreading of attitudes of resentment. But in the short run, any policy interventions focused on reducing support for welfare chauvinism should, according to our findings, concentrate on fighting unemployment and on providing favorable conditions for economic prosperity. These purely economic means might substantially dampen chauvinist attitudes among unskilled workers—but only to the extent to which this frequently disadvantaged group is able to profit from overall improving socioeconomic conditions. no t e s 1.  This research was financed by the German Research Foundation [grant number MA 3282/3-1] within the EUROCORES program HumVIB of the European Science Foundation. We are particular obliged to Staffan Kumlin for his valuable comments and suggestions on the WAE book conference at the Rock-



Unraveling Working-Class Welfare Chauvinism

151

efeller Centre in Bellagio, Italy (November 2010). Our thanks also go to our colleagues of the WAE HumVIB network—and especially to Stefan Svallfors—for their useful comments on previous versions of this study. 2.  A similar view has been established in the field of political psychology: Whereas there was a long-standing tradition of portraying ideological attitudes as being structured along a “single (liberal) left to right (conservative) dimension” (Duckitt and Sibley, 2009: 98), the extensively documented lack of empirical support for that relationship eventually led to the view that related values and attitudes must be approached as being organized around two relatively orthogonal dimensions: As Duckitt (2001: 46) explains, “One dimension has been labeled authoritarianism, social conservatism, or traditionalism, at one pole, versus openness, autonomy, liberalism, or personal freedom at the other pole. The second dimension has been labeled economic conservatism, power, or belief in hierarchy or inequality at its one pole versus egalitarianism, humanitarianism, social welfare, or concern at its other pole.” 3.  Similarly, Lipset in his famous work on “working-class authoritarianism,” lacked a proper definition of the concept (Svallfors, 2006: 124). 4.  Referring to theories of realistic conflict, as well as on Alesina and Glaeser’s (2004) challenging “diversity hypothesis,” which maintains that ethnic diversity negatively impacts on redistribution, a growing body of literature has started to investigate the role of the size and composition of immigrant and foreigner populations in a cross-national perspective (e.g., Semyonov et al., 2006; Hjerm, 2007; Schneider, 2008; Mau and Burkhardt, 2009; Staerklé et al., 2010; Hjerm and Nagayoshi, 2011). Given that it was our aim to maximize the number of countries in our sample and due to the lack of appropriate official statistics in respect to the investigated countries, we only took into consideration the absolute number of foreigners (in percent of the total population) in the country. 5.  Most of the missing cases are omitted due to the lack of information on household income. The missing response rate concerning household income ranged from 4.4 percent (in the case of Norway) to 58.1 percent (in the case of Portugal). Yet, because income is a pivotal variable in our multilevel models, we were forced to work with the comparably small data sample. We decided against using the possibility of multiple imputation, since we considered exactly those variables as being important for conducting the estimations that already are covariates of key importance in our models (such as class and education). Moreover, we rejected the option of using household income as a categorical variable and to treat the “missings” as an additional category because we wanted to take into consideration the important role of household size. 6.  Tables with results concerning the three other response categories (“after a year, whether or not have worked”; “after having worked and paid taxes at least a year”; “once they have become a citizen”), along with information regarding the country-level covariances between the response categories, can be obtained from the authors.

152

Mewes and Mau

7.  Indeed, all the countries contained in our sample witnessed an increase in GDP per capita between 2003 and 2008. Thus, according to the results from our model, we cannot say what happens to countries whose GDP either remains static or even declines (see Semyonov et al., 2006: 444). 8.  Yet, these kinds of conclusions might not provide the full picture, taking into account that we were, due to a lack of data, not able to measure the share of the foreign-born population. In the same vein, we lack data on the composition of the foreign population (Hjerm and Nagayoshi, 2011). For example, the results of a previous study by Mau and Burkhardt (2009) indicate that it is the share of foreign-born citizens from non-OECD countries that might trigger exclusionary attitudes toward immigrants. In much the same way, the studies by Gorodzeisky and Semyonov (2009: 415) and by Schneider (2008: 60) suggest that the relative size of the immigrant population with non-Western/non-European background promotes chauvinist and hostile attitudes toward immigrants. 9.  We additionally conducted multilevel analyses in which we used a different reference category in order to test the stability of our findings. Using the category “after a year, whether or not have worked” as the new reference category, it was revealed that the results with respect to the category of interest (“they should never get the same rights”) were, by and large, very similar to those derived from the original multilevel analyses: With respect to the direction and the significance of the results at the individual level, the effects of class, education, and perceived material risk did not differ in the alternative models. Yet, we observed differential results concerning the impact of income: While the negative effect of income disappeared when holding constant authoritarian dispositions in the original model (Table 5.2, Model III), it remains robust in the alternative model. At the country level, we observed two deviations from the original findings: First, there was no effect of unemployment in the redefined model, neither with regard to the cross-sectional unemployment data nor in terms of the related development over time (2003–2008). As a second deviation from the original results, we witnessed a significant and positive main effect concerning the share of the workforce being employed in industry and agriculture. In this way, the latter finding emphasizes even more the important role of the labor force’s sectoral composition in testing hypotheses derived from realistic conflict theory. Regarding the effects of GDP, collective authoritarianism, and collective risk, the alternative models do not notably differ from the original models. In conclusion, the results discussed in this chapter seem to be fairly robust.

r efer ences Achterberg, P., and D. Houtman. 2009. “Ideologically Illogical? Why Do the Lower-Educated Dutch Display So Little Value Coherence?” Social Forces 87: 1649–70.



Unraveling Working-Class Welfare Chauvinism

153

Adorno, T., E. Frenkel-Brunswik, D. Levinson, and N. Sanford. 1950. The Authoritarian Personality. New York: Harper. Alesina, A. and E. L. Glaeser. 2004. Fighting Poverty in the US and Europe: A World of Difference. Oxford: Oxford University Press. Altemeyer, B. 1981. Right-Wing Authoritarianism. Winnipeg, Canada: University of Mannitoba Press. ———. 1988. Enemies of Freedom: Understanding Right-Wing Authoritarianism. San Francisco: Jossey-Bass. Banting, K. G., and W. Kymlicka, eds. 2006. Multiculturalism and the Welfare State: Recognition and Redistribution in Contemporary Democracies. Oxford: Oxford University Press. Bernstein, D. 1996. “Expanding the Split Labor Market Theory: Between and Within Sectors of the Split Labor Market of Mandatory Palestine.” Comparative Studies in Society and History 38: 243–76. Blalock, H. M. 1967. Toward a Theory of Minority Group Relations. New York: Wiley. Blumer, H. 1958. “Race Prejudice as a Sense of Group Position.” The Pacific Sociological Review 1: 3–7. Bobo, L. D. 1999. “Prejudice as Group Position: Microfoundations of a Sociological Approach to Racism and Race Relations.” Journal of Social Issues 3: 445–72. Bobo, L., and V. L. Hutchings. 1996. “Perceptions of Racial Group Competition: Extending Blumer’s Theory of Group Position to a Multiracial Social Context.” American Sociological Review 61: 951–72. Boeri, T., G. Hanson, and B. McCormick, eds. 2002. Immigration Policy and the Welfare System. Oxford: Oxford University Press. Bonacich, E. 1972. “A Theory of Ethnic Antagonism: The Split Labor Market.” American Sociological Review 37: 547–59. Brooks, C. 2006. Voters, Satisficing, and Policymaking: Recent Directions in the Study of Electoral Politics. Annual Review of Sociology 32: 191–211. Castles, S., and M. Miller. 2003. The Age of Migrations: International Population Movement in the Modern World. London: MacMillan. Ceobanu, A. M., and X. Escandell. 2010. “Comparative Analyses of Public Attitudes Toward Immigrants and Immigration Using Multinational Survey Data: A Review of Theories and Research.” Annual Review of Sociology 36: 309–28. Coenders, M., M. Lubbers, and P. Scheepers. 2009. “Opposition to Civil Rights for Legal Migrants in Central and Eastern Europe: Cross-National Comparisons and Explanations.” East European Politics and Societies 23: 146–64. Coenders, M., and P. Scheepers. 2003. “The Effect of Education on Nationalism and Ethnic Exclusionism: An International Comparison.” Political Psychology 24: 313–43.

154

Mewes and Mau

Cohrs, J. C., and F. Asbrock. 2009. “Right-Wing Authoritarianism, Social Dominance Orientation and Prejudice Against Threatening and Competitive Ethnic Groups.” European Journal of Social Psychology 39: 270–89. Cortina, J. M. 1993. “What Is Coefficient Alpha? An Examination of Theory and Applications.” Journal of Applied Psychology 78: 98–104. Duckitt, J. 1989. “Authoritarianism and Group Identification: A New View of an Old Construct.” Political Psychology 10: 63–84. ———. 2001. A Dual-Process Cognitive-Motivational Theory of Ideology and Prejudice. In Advances in Experimental Social Psychology, edited by M. P. Zanna, 41–113. San Diego: Academic Press. Duckitt, J., and C. G. Sibley. 2009. “A Dual-Process Motivational Model of Ideology, Politics, and Prejudice.” Psychological Inquiry 20: 98–109. ———. 2010. “Right-Wing Authoritarianism and Social Dominance Orientation Differentially Moderate Intergroup Effects on Prejudice.” European Journal of Personality 24: 583–601. Erikson, R., and J. H. Goldthorpe. 1992. The Constant Flux: A Study of Class Mobility in Industrial Societies. Oxford: Clarendon Press. Gorodzeisky, A. 2011. “Who Are the Europeans that Europeans Prefer? Economic Conditions and Exclusionary Views Toward European Immigrants.” International Journal of Comparative Sociology 52: 100–13. Gorodzeisky, A., and M. Semyonov. 2009. “Terms of Exclusion: Public Views Towards Admission and Allocation of Rights to Immigrants in European Countries.” Ethnic and Racial Studies 32: 401–23. Green, E. G. T. 2009. “Who Can Enter? A Multilevel Analysis on Public Support for Immigration Criteria Across 20 European Countries.” Group Processes & Intergroup Relations 12: 41–60. Guiraudon, V. 2002. “Including Foreigners in National Welfare States: Institutional Venues and Rules of the Game.” In Restructuring the Welfare State: Political Institutions and Policy Change, edited by B. Rothstein and S. Steinmo, 129–56. New York: Palgrave. Hello, E., P. Scheepers, and P. Sleegers. 2006. “Why the More Educated Are Less Inclined to Keep Ethnic Distance: An Empirical Test of Four Explanations.” Ethnic and Racial Studies 29: 959–85. Hjerm, M. 2007. “Do Numbers Really Count? Group Threat Theory Revisited.” Journal of Ethnic and Migration Studies 33: 1253–75. Hjerm, M., and K. Nagayoshi. 2011. “The Composition of the Minority Population as a Threat: Can Real Economic and Cultural Threats Explain Xenophobia?” International Sociology 26: 815–43. Hooghe, M., A. Trappers, B. Meuleman, and T. Reeskens. 2008. “Migration to European Countries: A Structural Explanation of Patterns, 1980–2004.” International Migration Review 42: 476–504. Houtman, D. 2003. “Lipset and ‘Working-Class’ Authoritarianism.” American Sociologist 34: 86–105.



Unraveling Working-Class Welfare Chauvinism

155

Hox, J. J. 2010. Multilevel Analysis: Techniques and Applications. 2d ed. New York: Routledge. Janoski, T. 2010. The Ironies of Citizenship: Naturalization and Integration in Industrialized Countries. Cambridge, UK, and New York: Cambridge University Press. Joppke, C. 1998. “Immigration Challenges the Nation State.” In Challenge to the Nation State: Immigration in Western Europe and the United States, edited by C. Joppke, 5–46. Oxford: Oxford University Press. Kitschelt, H. 1994. The Transformation of European Social Democracy. Cambridge, UK, and New York: Cambridge University Press. ———. The Radical Right in Western Europe. Ann Arbor: University of Michigan Press. Knutsen, O., and S. Kumlin. 2005. “Value Orientations and Party Choice.” In The European Voter, edited by J. Thomassen, 125–66. Oxford: Oxford University Press. Korpi, W. 1983. The Democratic Class Struggle. London: Routledge & Kegan Paul. Kriesi, H. 2008. “Rejoinder to Liesbet Hooghe and Gary Marks, ‘A Postfunctional Theory of European Integration: From Permissive Consensus to Constraining Dissensus.’” British Journal of Political Sciences 39: 221–24. Kriesi, H., E. Grande, R. Lachat, M. Dolezal, S. Bornschier, and T. Frey. 2008. West European Politics in the Age of Globalization. Cambridge, UK, and New York: Cambridge University Press. Kunovich, R. M. 2004. “Social Structural Position and Prejudice: an Exploration of Cross-National Differences in Regression Slopes.” Social Science Research 33: 20–44. Leiulfsrud, H., I. Bison, and H. Jensberg. 2002. Social Class in Europe: European Social Survey 2002/3. Oslo/Trento: NTNU Social Research. LeVine, R. A., and D. T. Campbell. 1972. Ethnocentrism: Theories of Conflict, Ethnic Attitudes, and Group Behaviour. New York: Wiley. Lipset, S. 1959. “Democracy and Working-Class Authoritarianism.” American Sociological Review 24: 482–501. Maas, C. J. M., and J. J. Hox. 2004. “Robustness Issues in Multilevel Regression Analysis.” Statistica Neerlandica 58: 127–37. Malchow-Møller, N., J. R. Munch, S. Schroll, and J. R. Skaksen. 2008. “Attitudes Towards Immigration: Perceived Consequences and Economic Self-Interest.” Economic Letters 100: 254–57. Mau, S. 2007. “Mitgliedschaftsräume, wohlfahrtsstaatliche Solidarität und Migration.” In Moderne (staats)Bürgerschaft, edited by J. Mackert and H.-P. Müller, 215–34. Wiesbaden: VS. Mau, S., and C. Burkhardt. 2009. “Migration and Welfare State Solidarity in Western Europe.” Journal of European Social Policy 19: 213–29. Mau, S., J. Mewes, and A. Zimmermann. 2008. “Cosmopolitan Attitudes Through Transnational Social Practices?” Global Networks 8: 1–24.

156

Mewes and Mau

Mau, S., J. Mewes and N. M. Schöneck. 2012. “What Determines Subjective Socio-Economic Insecurity? Class and Context in Comparative Perspective.” Socio-Economic Review, advance access, published March 22, 2012. Meuleman, B., E. Davidov, and J. Billiet. 2009. “Changing Attitudes Toward Immigration in Europe, 2002–2007: A Dynamic Group Conflict Theory Approach.” Social Science Research 38: 352–65. Miller 2006. “Multiculturalism and the Welfare State: Theoretical Reflections.” In Multiculturalism and the Welfare State: Recognition and Redistribution in Contemporary Democracies, edited by K. G. Banting and W. Kymlicka, 323–38. Oxford: Oxford University Press. Mood, C. 2010. “Logistic Regression: Why We Cannot Do What We Think We Can Do and What We Can Do About It.” European Sociological Review 26: 67–82. Olzak, S. 1992. The Dynamics of Ethnic Competition and Conflict. Stanford: Stanford University Press. Quillian, L. 1995. “Prejudice as a Response to Perceived Group Threat: Population Composition and Anti-immigrant and Racial Prejudice in Europe.” American Sociological Review 4: 816–60. Raijman, R. 2010. “Citizenship Status, Ethno-National Origin and Entitlements to Rights: Majority Attitudes Towards Minorities and Immigrants in Israel.” Journal of Ethnic and Migration Studies 36: 87–106. Raijman, R., M. Semyonov, and P. Schmidt. 2003. “Do Foreigners Deserve Rights? Determinants of Public Views Towards Foreigners in Germany and Israel.” European Sociological Review 19: 379–92. Rydgren, J. 2007. “The Sociology of the Radical Right.” Annual Review of Sociology 33: 241–62. Sassen, S. 1991. The Global City. Princeton: Princeton University Press. Scheepers, P., M. Gijsberts, and M. Coenders. 2002. “Ethnic Exclusionism in European Countries: Public Opposition to Civil Rights for Legal Migrants as a Response to Perceived Ethnic Threat.” European Sociological Review 18: 17–34. Scheve, K. F., and M. J. Slaughter. 2001. “Labor Market Competition and Individual Preferences over Immigration Policy.” The Review of Economics and Statistics 83: 133–45. Schneider, S. 2008. “Anti-Immigrant Attitudes in Europe: Outgroup Size and Perceived Ethnic Threat.” European Sociological Review 24: 53–67. Sefton, T. 2005. “Give and Take: Public Attitudes to Redistribution.” In British Social Attitudes: The 22nd Report, edited by A. Park, J. Curtice, K. Thomson, C. Bromley, M. Philipps, and M. Johnson. London: Sage. Semyonov, M., R. Raijman, and A. Gorodzeisky. 2006. “The Rise of AntiForeigner Sentiment in European Societies, 1988–2000.” American Sociological Review 71: 426–49.



Unraveling Working-Class Welfare Chauvinism

157

Sherif, M. 1966. Group Conflict and Cooperation: Their Social Psychology. London: Routledge & Kegan Paul. Sides, J., and J. Citrin. 2007. “European Opinion About Immigration: The Role of Identities, Interests and Information.” British Journal of Political Sciences 37: 477–504. Staerklé, C., J. Sidanius, E. G. T. Green, and L. E. Molina. 2010. “Ethnic Minority-Majority Asymmetry in National Attitudes Around the World: A Multilevel Analysis.” Political Psychology 31: 491–519. Stellmacher, J., and T. Petzel. 2005. “Authoritarianism as a Group Phenomenon.” Political Psychology 26: 245–74. Stephan, W. S., and C. W. Stephan. 2000. “An Integrated Theory of Prejudice.” In Reducing Prejudice and Discrimination, edited by S. Oskamp, 23–46. New Jersey: Erlbaum. Stephens, J. D. 1979. The Transition from Socialism to Capitalism. Urbana: University of Illinois Press. Stubager, R. 2008. “Education Effects on Authoritarian–Libertarian Values: a Question of Socialization.” British Journal of Sociology 59: 327–50. Svallfors, S. 1997. “Worlds of Welfare and Attitudes Toward Redistribution: A Comparison of Eight Western Nations.” European Sociological Review 13: 283–304. ———. 2006. The Moral Economy of Class: Class and Attitudes in Comparative Perspective. Stanford: Stanford University Press. Van Der Waal, J., P. Achterberg, D. Houtman, W. De Koster, and K. Manevska. 2010. “Some Are More Equal than Others: Economic Egalitarianism and Welfare Chauvinism in the Netherlands.” Journal of European Social Policy 20: 350–63. Van Oorschot, W. 2006. “Making the Difference in Social Europe: Deservingness Perceptions Among Citizens of European Welfare States.” Journal of European Social Policy, 16: 23–42. ———. 2007. “Solidarity Towards Immigrants in European Welfare States.” International Journal of Social Welfare 17: 3–14.

chapter six

Age, Class, and Attitudes Toward Government Responsibilities1 Stefan Svallfors, Joakim Kulin, and Annette Schnabel

The problem we tackle in this chapter concerns the mechanisms that tie the categorical location of people—where they are positioned in the social structure of their societies—to their welfare attitudes. By now, researchers know a great deal about categorical differences in attitudes toward the welfare state and how they differ among contexts. Such research has shown, for example, that workers are generally more supportive of the welfare state than the salariat and the self-employed (Svallfors, 1997; 2003; 2004; 2006; Linos and West, 2003); that women are more supportive of extensive welfare policies than men (Svallfors, 1997; Bean and Papadakis, 1998; Andress and Heien, 2001; Blekesaune and Quadagno, 2003; Linos and West, 2003); that generational differences are often quite small (Goerres, 2009; Edlund and Svallfors, 2011); and that age still matters because attitudes toward the welfare state vary along the life cycle (Svallfors, 2008; Busemeyer et al., 2009). Comparative research has also shown that although patterns of class differences in welfare attitudes are basically similar across national contexts, their magnitude varies quite substantially, reflecting political circumstances and historical traditions (Svallfors, 2006; Kumlin and Svallfors, 2007). Attitude differences among age groups also vary substantially among country contexts, although in a less patterned way (Busemeyer et al., 2009). However, it is still the case that questions of why certain categorical locations tend to be tied to specific welfare attitudes and why the magFor additional data, please see the Web Appendix tables and figures by following the link in the listing for Contested Welfare States on our website: http://www.sup.org. 158



Attitudes Toward Government Responsibilities

159

nitude of categorical differences in attitudes varies across countries have received scant attention. In this chapter, we therefore take a stab at finding mechanisms, both at the micro- and at the macro-(country-) levels, that may explain class and age differences in attitudes toward government responsibilities. Age and class are two fundamental principles of social stratification, so it is a sociologically central task to analyze their impact on individuals’ attitudes toward social arrangements. The welfare state is often depicted as revolving around class conflicts, since it redistributes risks and resources among differently endowed groups in the population. But the welfare state may also be conceived of as a generational contract, in which the working population agree to support the young and the elderly through a system of transfers, services, contributions, and taxes. In return, they expect support in their own old age and for their own dependent offspring. In this implicit intergenerational contract, there is always a possibility of age conflicts arising. Such issues have come increasingly to the fore because of the changing age structure of European countries. A combination of increasing longevity and declining birth rates has given rise to severe demographic challenges for all European welfare states in trying to accommodate the problems posed by aging populations. Here, both life cycle and generational factors are likely to affect welfare attitudes in varying degrees across European countries. We ask to what extent risk perceptions, beliefs (about claimant groups and about the sustainability of the welfare state), and values work as micro-level mechanisms that tie class/age categories to attitudes toward government responsibilities. By “mechanism” we here understand factors that provide links between causes and outcomes and that are found at a “deeper” level than the initially observed outcome. That is, in this case we ask to what extent processes at a deeper cognitive level, such as object-transcendent subjective states, may explain class and age patterns in attitudes toward specific welfare policies. Much previous research in the field is characterized by an implicit or explicit assumption that actors’ self-interest is the mechanism that ties structural location to attitudes. Categorical differences in attitudes tend, often without much discussion, to be attributed to the self-interest of actors who are differently placed in the matrix of risks and resources. A quite explicit assumption of self-interest is found in the political-economical approach, where demands for insurance and redistribution are seen as driven

160

Svallfors, Kulin, and Schnabel

exclusively by the individuals’ relative incomes and their risk exposure in the labor market (Iversen and Soskice, 2001; Cusack et al., 2006). 2 The assumption that self-interest is the driving force behind the formation of welfare attitudes sits uneasily with an emerging body of research, stemming from such different quarters as survey research, experimental economics, and evolutionary biology. Here, empirical findings suggest that, in fact, self-interest exerts a quite limited influence on human orientations and actions. The basic orientation of actors seems rather to be reciprocal, driven by considerations about justice, and modified by their basic human values (Mau, 2003; 2004; Gintis et al., 2004; Kumlin, 2004; Fehr and Fischbacher, 2005; Fong et al., 2005). As expressed by the experimental economists Ernst Fehr and Urs Fischbacher (2005: 151), “During the previous decades, experimental psychologists and economists have gathered overwhelming evidence that systematically refutes the self-interest hypothesis.” So a substantial part of research about the structuring of welfare attitudes seems implicitly or explicitly to be driven by notions that in other regions of the social sciences are discarded as fundamentally refuted. Hence, there are good reasons to probe more deeply into the possible micro-level mechanisms at work. Furthermore, we explore class/age differences in a comparative perspective: To what extent do they differ among different national contexts? We ask whether the structural and institutional contexts in which actors are located affect the magnitudes of class/age differences, as suggested (for class) by previous research (Kumlin and Svallfors, 2007). Are there factors at the country level that systematically affect the relative size of class and age differences? We also ask if class and age work as cross-cutting cleavages such that small class differences tend to coincide with large age differences, and vice versa. Class and age may be seen as competing cleavages, over time or crossnationally. Precisely for this reason, Kohli (2006) argues that polities that are strongly structured by class are less likely to develop strong age-based conflicts. We ask if that claim can be corroborated in our data. The findings we present in this chapter refute many of the claims by previous research, including our own. We show that although the suggested micro-mechanisms explain some of the class–attitude association, most of this association remains even after taking into account several strong candidates for plausible micro-level mechanisms. The mechanisms explain absolutely no part of the age–attitude association. Furthermore, we dem-



Attitudes Toward Government Responsibilities

161

onstrate that only a part of the astonishing differences across countries in the strength of the class/age–attitude association can be explained by the suggested macro-level mechanisms. We also show that, in fact, neither the class–age trade-off hypothesis nor the emerging age conflict scenario receives much support from our analysis. This chapter is structured as follows: We start by presenting the overall framework for our analysis, including a discussion of what possible mechanisms and contextual effects we discern. Next, we introduce the empirical indicators and the statistical approach of our analyses. Three empirical sections, in turn, address the overall age/class differences in different countries, explore the micro-level mechanisms that tie structural location to welfare attitudes, and investigate the macro-(country-) level factors that explain variation in categorical differences. The penultimate section provides additional analyses of the development over time of the class and age differences for a few selected country cases. The concluding section summarizes the main results and discusses implications of the findings. c at e g or i e s , m e c h a n i sm s , a n d c on t e x t s Categorical inequality is structured along the lines of defined borders between social categories (Tilly, 1998; Massey, 2007). Examples of such categories are different classes, different ethnic groups, men and women, and young and old. Orientations may vary among groups because such groups may be more or less endowed with crucial resources and more or less exposed to risks, as well as differently placed in networks of interaction and communication. In turn, these factors give rise to different interests, as well as different beliefs about reality and different values about what is proper, just, and acceptable. The question we raise in this chapter is, to what extent do these mechanisms explain differences among age and class categories in their welfare attitudes? The concept of “mechanism” is contested, and several competing notions exist (Hedström and Swedberg, 1996; Mayntz, 2003; Opp, 2005). Here, we understand mechanisms as factors or processes that generate links between causes and outcomes and that occur at a “deeper” level than the cause and the outcome itself (Hedström and Swedberg, 1996). We argue that subjective orientations at a more fundamental level, such as interests, beliefs, and values, may serve to explain orientations at a more situational level, such as attitudes toward specific social and political objects.

162

Svallfors, Kulin, and Schnabel

According to an interest-based perspective, the risk that people are exposed to through their position in the labor market is a fundamental cause behind their attitudes toward government intervention. As used by Iversen and collaborators, security is a basic requirement of the welfare state, and the demand for security is determined by the level of risk in the labor market (Iversen and Soskice, 2001; Cusack et al., 2006). In this chapter, we tackle this issue by asking to what extent people’s perceived risks for unemployment and sustenance problems mediate age and class differences in welfare attitudes. We do this in the conviction that in order to affect attitudes, risks have to be perceived as such by individuals. But we also add present or past unemployment as an indicator of “realized risk” in the labor market. A second set of factors relates to the beliefs that actors hold about the actual state of affairs. We understand beliefs as cognitive understandings of the real conditions in which actors are located. 3 We enter indicators of two kinds of beliefs that we think may be pertinent to welfare state issues. The first is a belief about the extent of the social problems that welfare policies are supposed to address. How many unemployed, sick, and needy do people think there are in their countries? And do such beliefs explain some of the differences in welfare attitudes between classes and age groups? The second is a belief about welfare state sustainability. Do people think that welfare state commitments can be financed and sustained in the long run? Given aging populations and postindustrial challenges, concerns about welfare state sustainability are well justified. But to what extent do such beliefs account for class or age differences in welfare attitudes? A third set of factors includes the values people hold. Values should here be understood as a set of desirable abstract and motivational goals that transcend specific objects or situations and that function as guiding principles in a person’s life (Rokeach, 1973; Schwartz, 1992; 1994). This is in contrast to attitudes, which are evaluations that refer to specific objects and situations (Rokeach, 1968: 550; Eagly and Chaiken, 1998; see Kulin, 2011).4 However, here we risk encountering endogeneity problems—that is, the danger that what we are trying to explain—variation among social categories in attitudes—is so closely and almost by definition linked to values as explanatory mechanisms that explanations risk becoming empty and self-evident. For example, showing that egalitarian values are linked to demands for redistribution seems self-evident and void of any real explanatory content.



Attitudes Toward Government Responsibilities

163

We handle this problem by not using social values related to the ordering of social and political relations as our value indicators. Instead, we use indicators of basic human values, as developed by Schwartz (1994; 2001) and included in the European Social Survey. Schwartz identifies ten basic human values, which in turn can be summarized as four higher‑order value types along two orthogonal dimensions: self-transcendence versus selfenhancement, and openness to change versus conservation. These four value types are included in our analyses to test to what extent they are associated with welfare attitudes and whether they explain parts of the age/class differences in such attitudes. These basic value orientations are deep-seated personal dispositions that do not carry the endogeneity problems of socially and politically constructed values in explaining attitudinal variations across categories. If values actually work to translate class positions into specific attitudes, one could then in turn ask whether this is an effect of different values being induced in different classes or of values being devices for self-selection into specific occupations (see Brooks and Svallfors, 2010). There are good a priori grounds to expect that risk perceptions, beliefs about the size of social problems and welfare state affordability, and values may explain parts of the associations between class/age location and welfare attitudes. Classes and age groups are differently exposed to labor market risks and problems of sustenance; they may therefore differ also in their risk perceptions and in their view of social problems. It is also possible that younger age groups may have a different and more skeptical view of the future affordability of the welfare state than older groups and that this may affect their welfare attitudes. Since values relate to deep-seated dispositions about what is good and acceptable, we could expect them to be structured by fundamental social divisions such as age and class and, if so, perhaps work to mediate the class/age differences in welfare attitudes. Our analytical strategy is therefore to test to what extent we can “empty” the original association between categorical location and attitudes. That is, to what extent is the original association between age/class position and welfare attitudes modified once indicators of possible mechanisms are introduced in the models? m ac ro -l e v e l fac t or s We also ask to what extent country contexts influence the relationship between categorical location and welfare attitudes. Such an influence could

164

Svallfors, Kulin, and Schnabel

affect the patterns or magnitude of categorical differences in attitudes: In some kinds of contexts, are the links between class/age location and welfare attitudes stronger than in others? We explore a number of features of the macro-context in the course of our analyses. The first concerns some basic structural features of the countries in our analyses—namely, the level of actual inequality and the demographic situation of the countries in question. The question we raise here is whether more inequality tends to create larger or smaller class/age differences in welfare attitudes. A priori, one could expect categorical differences in welfare attitudes to be smaller in countries where institutions such as wage-setting systems or welfare states distribute goods and burdens in a more egalitarian way. Put crudely, those with a strong market position have less to lose, and those with a weak market position have less to gain from further redistribution in more egalitarian systems compared to more inegalitarian ones. Hence, we would expect age/class differences in welfare attitudes to be smaller under such circumstances. But in a slightly counterintuitive way, previous research shows that class and income differences in redistributive attitudes actually tend to be larger where inequality is smaller (Kumlin and Svallfors, 2007; Rueda and Pontusson, 2010). Two slightly different interpretations pertain to these findings. One interpretation argues that low levels of inequality are typically intertwined with encompassing welfare and labor market policies. Such policies constitute focal points for citizens’ political thinking; they affect the extent to which citizens think about inequality and redistribution as political responsibilities in the first place. Such institutions tend to have a self-reinforcing effect: The more redistribution that is achieved through public policy, the more unprivileged citizens and their political representatives think about remaining inequalities (Kumlin and Svallfors, 2007: 39). A related but slightly different interpretation is that in groups with stronger market positions, highly redistributive institutions tend to promote resistance against further redistribution (Kumlin and Svallfors, 2007: 39; Rueda and Pontusson, 2010). We also ask if a more problematic demographic situation, in terms of the dependency ratio and share of elderly among the population, tends to create larger age differences in attitudes concerning government responsibilities. One could easily envisage that as the groups of elderly grow in relation to the working population, and problems of welfare sustainability then



Attitudes Toward Government Responsibilities

165

become more acute, tensions between age groups would grow in relation to welfare policy priorities. The second aspect of the macro-context concerns the institutional setup of the welfare state. Here, we include different measures of the size of the welfare state, as well as the age profile of social spending. We ask whether larger and more generous welfare states tend to create larger or smaller age/ class differences in welfare attitudes. On the one hand, we could expect extensive welfare policies to mitigate distributive conflicts among groups in the population. On the other hand, we could also—as already discussed in relation to inequality –envisage that redistributive policies and government intervention in general make people more aware of distributive issues, thus sharpening categorical differences. We also ask whether welfare states that are relatively more geared toward providing for the elderly tend to increase age-based attitudinal differences (see Lynch, 2006). It has been argued that it is likely that welfare states that are more geared toward spending on the elderly will exacerbate age conflicts over welfare spending, since this will leave young and working-age people with less resources from the public purse (Busemeyer et al., 2009: 208). The last aspect we cover is the political articulation of distributive issues. Why should we expect political articulation to matter in forging the nexus between categorical location and attitudes? Here we should note that tying one’s categorical location to welfare attitudes is a fairly complex mental exercise. It involves perceiving one’s own position and linking that position to the possibilities and desirability of redistribution and protection through public policies. Organized interests such as political parties and trade unions help people make this mental connection, and in doing so, they may help to forge a stronger link between categorical location and welfare attitudes. The empirical question we raise is whether the articulation of socioeconomic (distributive) issues in the political manifestos of political parties tends to create larger class/age differences in attitudes. This question is based on the assumption that what is indexed in party manifestos is a reflection of the issues that are actually debated in the political arena (Budge, 2001: Ch. 3). Even if party manifestos in themselves are read by few people, they can be assumed to reflect a wider political debate, which in turn is hypothesized to affect group differences in attitudes. Previous

166

Svallfors, Kulin, and Schnabel

research ­indicates that a larger amount of such socioeconomic articulation does indeed go together with larger class differences in redistributive attitudes (Kumlin and Svallfors, 2007). We test whether this result holds for a larger set of countries and a different set of dependent variables and whether age differences in attitudes are also affected. We also test whether higher levels of class organization, as measured by trade union density, tend to create larger class differences in attitudes. i n dic at or s a n d a n a ly t ic a l s t r at e g i e s As with all chapters in this volume, our analyses build on data from the welfare attitudes module of the European Social Survey 2008.5 In order to capture both “class-relevant” and “age-relevant” issues, we use four separate dependent variables. The first is the widely used item that asks respondents whether they agree that “the government should take measures to reduce differences in income levels.” Answers are given on a 5-point Likert scale, which we transformed into a 0–10 range to make it roughly comparable to our other dependent variables. Our three other dependent variables are constructed from a set of items where respondents are asked their views about what the responsibilities of governments should or should not be. Answers are given on an 11-point scale, where 0 means it should not be governments’ responsibility at all, and 10 means it should be entirely governments’ responsibility. Items include government’ responsibilities to do the following: 1. Ensure a job for everyone who wants one 2. Ensure adequate health care for the sick 3. Ensure a reasonable standard of living for the old 4. Ensure a reasonable standard of living for the unemployed 5. Ensure sufficient child care services for working parents 6. Provide paid leave from work for people who temporarily have to care for sick family members

From this item battery, we construct three indices to tap attitudes toward specific aspects of government intervention. These three subdimensions concern, in turn, health care and pensions, labor market–related measures, and family support intervention. These subdimensions are differently related to class and the life cycle, so it makes sense to treat them separately rather than summarizing them into an overall measure.6 We



Attitudes Toward Government Responsibilities

167

would expect class differences to be more prominent than age differences when it comes to support for redistribution and labor market policies, since they cover especially class-pertinent issues. The reverse could be expected for health care/support for the old and for family support policies. These policies are clearly related to different life cycle stages and less so to class. Turning to possible mechanisms, we first enter a set of empirical indicators of risk perceptions. These items ask respondents how likely they think it is that they will encounter welfare problems themselves or, more specifically, that during the next 12 months, they will experience the following: 1. They will be unemployed and looking for work for at least four consecutive weeks. 2. There will be some periods when they don’t have enough money to cover their household necessities. 3. They will not receive the health care they really need if they become ill.

Answers are given on a 4-point scale, from not at all likely to very likely. These indicators are entered as separate variables in the analysis. We also include four dummy variables to capture present and past unemployment experiences.7 Beliefs about welfare needs are measured by asking respondents to judge the size of various claimant groups and the needy: Of every 100 people of working age in [country], how many would you say: 1. Are unemployed and looking for work? 2. Are long-term sick or disabled? 3. Do not have enough money for basic necessities?

Answers are given on an 11-point scale, with evenly spaced percentage intervals (from “0–4” to “more than 50”). These indicators are then entered as continuous variables.8 Beliefs about welfare state sustainability are measured by asking about the assessment of future resource availability for basic welfare provisions such as pensions and health care: There is some debate nowadays about the cost of [providing public health care/pensions] in [country]. Thinking about 10 years from now, which of these statements comes closest to your own opinion? In ten years’ time: 1. [Country] will not be able to afford the present level of [public health care/old-age pension].

168

Svallfors, Kulin, and Schnabel 2. [Country] will be able to afford the present level of [public health care/ old-age pension] but not to increase it. 3. [Country] will be able to afford to increase the level of [public health care/old-age pension].

These two indicators are entered in the analysis as two couples of dummy variables. As indicators for values, the ESS basic human values scale items—a reduced version of the Portrait Values Questionnaire developed by Schwartz (1994; 2001)—are used. Each value is measured by two or three items as indicators, which are verbal portraits of persons with different goals and aspirations. Respondents are then asked how similar a described person is to them with regard to a particular value. For instance, as an indicator of the value universalism: She/he thinks that every person in the world should be treated equally and that everyone should have equal opportunities in life. The items indicating different values all range from 1 to 6, where 1 = “not like me at all”; 2 = “not like me”; 3 = “a little like me”; 4 = “somewhat like me”; 5 = “like me”; and 6 = “very much like me.” A full description of items for the different value dimensions is found in Appendix 1 at the end of this chapter. We estimate factor scores as indicators of the four higher-order values types: self-enhancement, self-transcendence, conservation, and openness to change. Since factors represent the relative importance of observed variables associated with a particular factor, a person’s score on that factor can be estimated based on the scores for each of the observed variables. Computing factor scores enables control for measurement error by excluding any variation in the observed variables that is not attributable to the latent construct. Earlier studies utilizing the ESS values instrument and consequent data have shown that factor loadings for the observed variables seldom exceed 0.4– 0.6, which makes a strong case for constructing factor scores rather than additive indices. The factor scores for each of our constructs are centered on their respective means, and the standard deviation is set to one.9 The coding of our two independent categorical variables also requires some consideration. Rather than taking “age” as a linear chronological concept, we created a number of life-course-defined age categories as starting points for comparison (see Mayer and Schoepflin, 1989). We use five age categories in the analysis: 16–25, 26–39, 40–50, 51–64, and 65+. The rationale is that the first group consists of young people, typically in their



Attitudes Toward Government Responsibilities

169

establishment phases regarding labor market participation and family life, and typically with larger problems of unemployment and low income; the second is the phase of actual or potential parenthood; the third consists of career peaks and the transition of children into adulthood; the fourth consists of “empty nests” and transition into postwork conditions; and the last is the phase of retirement and increasing frailty. Any age categorization is to some extent arbitrary as to the number of categories and their exact cutoff points. Hence, we test the robustness of our results by moving cutoff points three years backward and forward. These robustness checks show that results are generally quite stable regardless of the exact cutoff point.10 The other key independent variable in the analyses—class—is indicated by the well-known Erikson-Goldthorpe class schema. This perspective views classes as constituted by employment relations (Erikson and Goldthorpe, 1992; Goldthorpe, 2000). Classes are aggregations of positions in labor markets and production units and can be operationalized through a combination of occupation title and employment status. This class schema has been the subject of numerous evaluations focusing on both its construct validity and its criterion validity. In general, such evaluations tend to confirm the validity of the class schema on both counts. That is, classes thus defined tend to differ in terms of important social outcomes, as well as in terms of the employment relations they are supposed to reflect. We collapse the Erikson-Goldthorpe schema into six classes: unskilled workers, skilled workers, routine nonmanuals, service class II, service class I, and the self-employed.11 As a summary measure of the strength of the association between class/ age categories and attitudes, we use the Κ index (Hout et al., 1995; Brooks and Svallfors, 2010). Since we want a measure that indicates the attitudinal dispersion between class/age categories, it makes sense to simply use the standard measure of dispersion: the standard deviation. Hence, the Κ index is the standard deviation of the B-coefficients in the regression models (with the reference category set to 0), and the Κ values can be compared across variables and across countries. Furthermore, reductions in the Κ index show how the original association changes once various indicators of mechanisms are introduced into the models (Brooks and Svallfors, 2010).12 By way of illustration, Table 6.1 shows the regression estimates for Finland regarding attitudes toward income redistribution policies and the

170

Svallfors, Kulin, and Schnabel

ta bl e 6.1 Class, age, gender, and attitudes toward income redistribution for Finland, 2008 B Constant Service class I Service class II Routine nonmanual Skilled workers Self-employed   (unskilled workers = 0) Age 26−39 Age 40−50 Age 51−64 Age 65+ (Age 16−25 = 0) Gender (men = 0) K class K age

S.E.

t

Sig.

6.162

0.225

27.336

0.000

−1.701 −0.399 −0.127 0.063

0.173 0.157 0.165 0.187

−9.831 −2.542 −0.767 0.338

0.000 0.011 0.443 0.736

−0.771 0.998 1.225 1.712 1.512

0.178 0.185 0.188 0.177 0.183

−4.333 5.395 6.520 9.693 8.241

0.000 0.000 0.000 0.000 0.000

0.314

0.11

2.843

0.005

0.60 0.61

resulting Ks for class and age. As shown, there are substantial and statistically significant class and age differences in that the service class and the self-employed (as expected) are less in favor of redistribution than workers, while the middle-aged and older are (though less than expected) more in favor of policies for income redistribution than the young. These class and age differences in attitudes are then translated into comparatively large Ks for both class and age (see Finland’s position in Figure 6.1).13 Finally, a large set of country-level variables have been compiled from different databases and added to the data set. Since we only present results for a small subset of these indicators, we provide their exact construction and sources adjacent to the analyses and in more detail in Appendix 2 at the end of this chapter. ag e a n d c l a s s di f f e r e n c e s ac ros s c ou n t r i e s How do the relative class and age differences in attitudes toward welfare policies vary across countries? Figure 6.1 and Web Appendix Figures A6.1– A6.3 display the size of the Κ index for class (the vertical axes) and age (the horizontal axes) for all countries and dependent variables. (The Κ values are displayed in Table 6.2, Model I.)



Attitudes Toward Government Responsibilities

171

Standard deviations of coefficients (K) .8

CZ

.7 .6

CH DE

.5

K class

PL NL

GB

FR BE SI

SK

FI UA

BG SE NO EE

.4

DK

CY

RU

HU LV HR ES PT GR RO

.3 .2 .1 0 0

.1

.2

.3

.4

.5

.6

.7

.8

K age Figure 6.1  Attitudes toward income redistribution policies s o u r c e : European Social Survey 2008.

A number of observations can be made about the figures. One is that both class and age Κs are clearly larger for the first index (Figure 6.1) than for the others (Web Appendix Figures A6.1–A6.3). This is perhaps not surprising, since the issue of redistribution is posed to pit different groups against one another to a larger extent than is the case for the other indices. Attitudes toward labor market policies also have a clear class profile, although the sizes of the coefficients vary substantially across countries. That attitudes toward support for the elderly and health, and family support have less clear class grounding is to be expected; these are not policies with a clear class profile. Here we should instead note the fairly modest age differences.

172

Svallfors, Kulin, and Schnabel

The differences are confirmed when we compare the age and class differences across different policies. Class differences regarding redistribution are clearly larger than those found among age categories (average Κ for class is .46 compared to .29 for age); they are also clearly larger for labor market policies (Κ = .30 vs. .16). Strikingly, class differences are even slightly larger than age differences when it comes to the clearly life-cycle-related policies (K = .20 and .19 vs. .16 and .18, respectively). Given the clear age profile of both these policy sets, this finding is an indication that age does not constitute such an important dividing line in welfare policy issues. It is clearly not the case that age is emerging as a major fault line in these issues. ta bl e 6.2 Attitudes toward welfare policies in 26 European countries Panel A: Class income redistribution policies K class Belgium Bulgaria Croatia Cyprus Czech Rep. Denmark Estonia Finland France Germany Great Britain Greece Hungary Latvia Netherlands Norway Poland Portugal Romania Russia Slovakia Slovenia Spain Sweden Switzerland Ukraine Average

labor market policies

health and elderly support policies

family support policies

I

II

I

II

I

II

I

II

.52 .55 .26 .35 .80 .38 .43 .61 .56 .54 .48 .26 .33 .30 .60 .43 .64 .23 .24 .37 .64 .48 .26 .43 .58 .56 .46

.45 .41 .30 .45 .63 .32 .39 .52 .39 .32 .24 .27 .28 .29 .45 .39 .58 .17 .17 .34 .53 .44 .14 .43 .52 .57 .38 (83%)

.23 .65 .28 .37 .36 .18 .44 .29 .30 .42 .22 .18 .27 .25 .17 .23 .58 .39 .20 .22 .43 .29 .16 .23 .22 .30 .30

.20 .62 .24 .36 .23 .19 .31 .24 .24 .26 .18 .22 .22 .13 .17 .19 .43 .24 .25 .22 .38 .19 .14 .21 .19 .23 .25 (83%)

.09 .20 .33 .12 .17 .10 .16 .11 .17 .27 .12 .07 .35 .09 .12 .13 .27 .19 .21 .10 .18 .12 .10 .03 .16 .19 .20

.09 .12 .25 .19 .22 .11 .11 .12 .13 .15 .11 .10 .22 .10 .11 .11 .21 .19 .25 .07 .20 .12 .15 .03 .18 .14 .15 (75%)

.16 .26 .33 .22 .08 .10 .17 .20 .17 .15 .26 .10 .25 .17 .15 .29 .33 .23 .15 .09 .35 .10 .13 .18 .11 .26 .19

.17 .14 .22 .22 .07 .11 .13 .19 .09 .12 .12 .12 .23 .10 .10 .25 .28 .10 .33 .07 .22 .08 .07 .18 .12 .23 .15 (79%)

(continued)



Attitudes Toward Government Responsibilities

173

t a b l e 6 . 2 (continued) Panel B: Age income redistribution policies K age Belgium Bulgaria Croatia Cyprus Czech Rep. Denmark Estonia Finland France Germany Great Britain Greece Hungary Latvia Netherlands Norway Poland Portugal Romania Russia Slovakia Slovenia Spain Sweden Switzerland Ukraine Average

labor market policies

health and elderly support policies

family support policies

I

II

I

II

I

II

I

II

.28 .35 .15 .08 .39 .30 .42 .60 .24 .13 .08 .19 .17 .20 .43 .40 .42 .06 .15 .48 .56 .17 .23 .39 .12 .62 .29

.30 .38 .13 .26 .46 .33 .45 .60 .31 .14 .18 .24 .11 .30 .50 .43 .47 .11 .27 .18 .20 .18 .31 .50 .14 .60 .31 (107%)

.10 .19 .05 .17 .15 .16 .25 .22 .17 .06 .13 .10 .07 .21 .25 .20 .31 .18 .24 .29 .32 .23 .20 .19 .30 .16 .16

.13 .21 .16 .24 .23 .16 .21 .23 .17 .04 .17 .11 .10 .27 .29 .22 .31 .28 .18 .20 .27 .26 .22 .23 .24 .20 .21 (131%)

.15 .11 .11 .15 .18 .21 .16 .10 .11 .10 .15 .07 .16 .14 .10 .17 .08 .17 .25 .18 .15 .10 .17 .19 .17 .19 .16

.13 .11 .12 .17 .25 .18 .12 .08 .14 .12 .13 .10 .20 .17 .09 .17 .12 .16 .06 .11 .19 .17 .19 .18 .12 .15 .14 (88%)

.20 .09 .09 .10 .10 .41 .17 .20 .12 .18 .21 .11 .15 .12 .16 .18 .17 .14 .18 .31 .13 .13 .12 .23 .33 .23 .18

.21 .07 .13 .10 .24 .41 .11 .16 .15 .17 .20 .16 .16 .20 .17 .20 .11 .20 .09 .17 .18 .12 .15 .21 .27 .22 .17 (94%)

At the same time, however, we should note the astonishing variation across country contexts in the class–attitude and age–attitude associations. A country like Poland is characterized by consistently large class differences, and with some exceptions this also applies, for example, to Bulgaria and Slovakia. No easily discernible pattern of country differences emerges. In contrast to several of the other chapters in this book, we find no clear regime patterning of class and age differences in attitudes. Neither can previous research that found particularly large class differences in northwestern Europe be corroborated (Kumlin and Svallfors, 2007). The class differences in Denmark are strikingly small across the board, while the other Nordic countries occupy a middling position (with the exception of attitudes

174

Svallfors, Kulin, and Schnabel

toward redistribution in Finland). None of the other northwestern countries display particularly large class differences; they are just as large on average in central-continental countries such as Switzerland or Germany. Looking at the Mediterranean countries, class differences are very small in Spain and Greece, in line with previous findings, but are substantially larger in Portugal for several of the indicators. Overall, we find a quite mixed and varied pattern. It should be noted that no real support is found for the notion that age and class are competing cleavages. In the case of attitudes toward redistribution, it is rather the case that large class and large age differences appear in tandem (R = .53; see Figure 6.1). It is only in the case of attitudes toward family support policies that we find a (weak) tendency that small class differences go with large age differences, and vice versa (R = −.31). For the two remaining policy sets, no clear pattern regarding the age–class trade-off is found (R = .17 and −.15; see Web Appendix Figures A6.1–A6.3). m ic ro -l e v e l m e c h a n i sm s What micro-level factors can explain the association between age/class location and welfare attitudes? As described in the theory and data sections, we enter indicators for a set of risk factors and perceptions, a set of cognitive beliefs about welfare needs and sustainability, and a set of basic human values. To what extent do these factors work as mechanisms in translating age/class location to welfare attitudes? Perhaps the most striking finding to come out of these exercises is that several of the proposed factors contribute absolutely nothing to an explanation of the class or age differences in welfare attitudes (see Web Appendix Tables A6.1–A6.4).14 None of the value dimensions matter at all in these respects; this goes also for the beliefs about welfare sustainability indicators. We want to emphasize that this does not mean that these factors are without their own importance and effects when it comes to structuring welfare attitudes. On several counts, they do have clear associations, in the expected direction, with welfare attitudes. However, they do not work as mechanisms in tying categorical location to welfare attitudes.15 But risk perceptions and beliefs about welfare needs mediate a substantial part of the class–attitudes associations. In Table 6.2, we display findings for class and age for all countries and for each of the dependent



Attitudes Toward Government Responsibilities

175

variables, first (Model I) without indicators of mechanisms, and then with indicators of individual risk perceptions and beliefs about the size of needy groups (Model II). As shown, differences in risk perceptions and beliefs about welfare clearly explain some of the association between class categories and welfare attitudes. About 20 percent of the original association between class categories and attitudes can be explained by these factors. But most of the variation between class categories is still there after entering these factors. Substantial variations exist among countries and variables in how much the original association changes when indicators of mechanisms are entered. In Britain, a considerable part of the association is thus explained (50 percent in the case of attitudes toward redistribution). In Sweden, almost none of the original class–attitude association is explainable. However, no clear country pattern is discernible in how much of the categorical variation is explained. The mechanism variables explain virtually nothing of the association between age categories and attitudes. On average, it is even the case that the associations between age categories and welfare attitudes become stronger once the proposed mechanisms are entered. Whatever explains age differences in welfare attitudes is not to be found among the factors we have analyzed here. m ac ro -l e v e l fac t or s So far, we have seen that age and class differences in welfare attitudes vary substantially across countries. We have also seen that a sizeable fraction of the association between class categories and welfare attitudes, and no part of the age–attitude association, can be explained by the risk, beliefs, or values indicators at our disposal. But are there factors at the macro-level that systematically affect the relative size of class/age differences across countries? The scattered and partly surprising patterns displayed suggest that this may be a difficult exercise. As described in previous sections, we ask whether more egalitarian societies are characterized by smaller or larger class/age differences; whether a more problematic demographic situation regarding aging populations tends to exacerbate age divisions; whether larger and more age-biased welfare states tend to create larger or smaller attitudinal differences; and, finally,

176

Svallfors, Kulin, and Schnabel

whether the extent to which socioeconomic (distributive) issues are politically articulated affects age/class differences in attitudes. Analytically, we proceed by taking the Κs for age and class, and plot and regress them onto a number of different indicators of the macro-properties of the different countries. We display a subset of all these scatterplots and report regression findings in the text. Other results are available from the authors. Again, what is most striking is that many of the proposed macro-­factors have no relation at all to the size of class/age differences. For example, age differences in attitudes are not affected at all by the age structure of social expenditure, unlike previous expectations in this regard (Busemeyer et al., 2009: 208).16 Nor does the size or generosity of the welfare state affect the size of age or class differences in attitudes. Regarding inequality, we find, as displayed in Figure 6.2, some support for previous findings that small income inequality is connected with larger class differences in redistributive attitudes (B = −1.35, p = .02)17 (Kumlin and Svallfors, 2007; Rueda and Pontusson, 2010).18 However, no inequality effects on age differences were detected, nor is there any clear pattern regarding class differences for the other indices. Moving to the measures of political articulation, we find that the total amount of political argument and information about socioeconomic redistributive issues, as measured in the program content of political parties, has no relation at all to class or age differences in attitudes. Nor does the trade union density have any clear correlation with the size of attitudinal class differences. This finding is in contrast to some previous research, where class differences in attitudes were clearly larger in countries where a larger share of the political program content dealt with these issues and where trade union density was higher (Kumlin and Svallfors, 2007). By way of illustration, Web Appendix Figure A6.4 shows the virtual zero correlation between political articulation of socioeconomic issues and attitudes toward income redistribution. The polarization between the main right-wing and the main left-wing party displays a couple of unexpected associations with class differences in attitudes. As shown in Web Appendix Figure A6.5, the degree of polarization in socioeconomic issues has a negative association with attitudinal class differences, even if the relationship is quite weak (B = −.006; p = .09). That is, attitudinal class differences are on average smaller where arguments in distributive issues are more polarized between left and right.



Attitudes Toward Government Responsibilities

177

Standard deviations of coefficients (K) CZ

.8 .7

SK NL FI

.6

DE

K class

.5

SI NO SE DK

.4

PL CH FR BE

EE RU

HU

.3

GR

RO

.2

GB

ES

.1 0 .2

.25

.3

.35

.4

.45

Gini coefficient LIS 2000 Figure 6.2  Income inequality and attitudes toward income redistribution policies s o u r c e : European Social Survey 2008.

The association is not strong, but it runs counter to what has been found in previous research, where stronger polarization was connected to larger class differences in attitudes (Kumlin and Svallfors, 2007). A similar but even weaker pattern applies to class differences regarding health care and support for the elderly (B = −.003; p = .11).19 Regarding age differences in attitudes, the single factor that seems to be of importance is the age structure of the population. Here, the interesting finding is that older populations typically display smaller age differences in attitudes. Findings for attitudes toward labor market policies and the old-age dependency ratio are displayed in Figure 6.3 (B = −.013; p = .002). Similar results pertain to the other indices, except the Family Support Index, and regardless of whether the old-age dependency ratio for 1990 or the share of population over 65 is used as the alternative macro-measure. At the very least, this finding suggests that aging populations are not followed by increasing age conflicts around the welfare state. In sum, the findings in this section negate many of the assumptions of previous research. It should be emphasized that this is not an effect of including also the countries of Eastern Europe. Even if one were to exclude

178

Svallfors, Kulin, and Schnabel Standard deviations of coefficients (K) SK

.3

PL

CH

RU RO

.2

K age

CY

UA CZ

EE

NL SI

FI LV ES BG PT FR DK

NO

SE

GB BE

.1 HU

0

15

20

GR

HR

25

DE

30

Old-age dependency ratio 2010 Figure 6.3  Old-age dependency ratio and attitudes toward labor market policies s o u r c e : European Social Survey 2008.

the former communist countries, the same set of “nonfindings” would apply. We will return in the concluding section to the question of how this should be interpreted, but before that, we will delve into the development over time of class and age differences in four selected countries. c l a s s a n d ag e di f f e r e n c e s i n s c a n di n av i a a n d e a s t e r n e u rop e In relation to previous research, one of the striking findings in the previous sections is that class differences in attitudes are fairly modest in the Scandinavian countries, while they are fairly substantial in some of the former communist countries of Eastern Europe. How should this be interpreted? Is it an indication of transformations of the class–attitude link in contemporary Europe, of a temporary “crisis effect” or simply of patterns that have received scant attention in previous research? Fully addressing these questions lies beyond the scope of the present chapter, but before concluding, we ask how the link between class/age location and welfare attitudes has changed recently in four selected countries. The only ESS welfare attitude



Attitudes Toward Government Responsibilities

179

indicator that is part of the core module, and therefore repeated biannually, is the question about income redistribution. In this section, we analyze how age and class differences have changed for this indicator in four selected countries: Denmark, Sweden, Poland, and Slovakia. We choose these four countries based on the following considerations. Denmark and Sweden have the most egalitarian income distributions and represent the most encompassing and generous welfare states in the world. In previous research, they have also been characterized by comparatively large class differences in attitudes toward public policies and redistribution (Svallfors, 2006; Kumlin and Svallfors, 2007). These findings have not been replicated in this chapter. Poland and Slovakia represent the emerging or rudimentary welfare states of Eastern Europe, whose communist legacy is often expected to blur class patterns of political articulation and attitudes. But in the analyses presented in this chapter, they still display comparatively large class and age differences in attitudes. Results are presented in Table 6.3. For Denmark and Sweden, we find that class differences in redistributive attitudes decrease over time, and especially so between the last two observation points. Class differences are substantially smaller in both countries in 2008 than they were in 2002. Support for redistributive policies increases in all classes in the two Scandinavian countries between 2006 and 2008, but it increases more among the most skeptical—that is, among members of service class I. Poland and Slovakia instead display persistent and on balance even growing class ­differences

t a bl e 6 .3 Class and age differences in attitudes toward income redistribution for Denmark, Sweden, Poland, and Slovakia, 2002−2008 (standard deviations of coefficients (K)) 2002

2004

2006

2008

k class Denmark Sweden Poland Slovakia

.57 .64 .46 —

.46 .57 .55 .50

.55 .59 .46 .75

.38 .43 .64 .64

k age Denmark Sweden Poland Slovakia

.17 .26 .52 —

.17 .33 .61 .38

.27 .28 .51 .32

.30 .39 .42 .56

180

Svallfors, Kulin, and Schnabel

in attitudes 2002–2004 to 2008. In 2008, class differences in redistributive attitudes are clearly larger in the two ex-communist countries than they are in the two Scandinavian countries. In 2002, class differences in Poland were smaller than in the two Scandinavian countries. The increased class difference in Poland and Slovakia is mostly an effect of decreasing support for redistribution among the self-employed and in service class I. Age differences grow steadily over time in the two Scandinavian countries. In 2008, they are almost on par with class differences, while they were much smaller than class differences in 2002. In Poland, age differences decrease between 2004 and 2008, while they jump considerably in Slovakia from 2006 to 2008. In all four countries, it is the case that the older age groups are more in favor of redistributive measures than the younger ones. So we find that it seems indeed to be the case that the surprisingly small class differences in the Scandinavian countries are a quite recent phenomenon, as are the growing age divisions in these two countries. The comparatively large class differences in some Eastern European countries, on the other hand, were there already by the first ESS round, although they have so far not been noted by analysts. Since then, they have grown even larger. In the concluding section, we will return to the question about how to interpret these and other findings. c on c l usion The main empirical conclusions that emanate from this chapter are mostly of a negative kind. In this respect, they have some interesting implications for previous research, including our own (Svallfors, 2006; Kumlin and Svallfors, 2007), and provide some important challenges for future research. To begin, the results show that there is indeed astonishing variation across Europe in the extent to which class and age affect welfare attitudes. Some countries are characterized by considerable class and age variation, and others by more modest or even negligible differences. But these ­differences-within-differences do not form any easily interpretable pattern. No clear regime patterning emerges, in contrast to some other chapters in this volume. Previous research has often pointed to class differences in attitudes as particularly large in the Nordic or northwestern countries (Wright, 1985; 1997; Svallfors, 2006; Kumlin and Svallfors, 2007), but no such pattern can be detected here. Instead, class differences in welfare attitudes are



Attitudes Toward Government Responsibilities

181

particularly large in some of the Eastern European countries, but on the whole, no obvious regional pattern can be found. The findings that relative age/class differences are nonpatterned across countries are basically confirmed by the attempts to find macro-variables that are systematically associated with the size of the attitude differences among categories. These exercises were on the whole negative, in that few of the macro-variables were clearly associated with the size of class differences. The only macro-variable that showed any association with age differences in attitudes was the age structure of the population. Here it transpired, contrary to expectations, that older populations tend to be correlated with smaller age differences in attitudes. Especially surprising was the fact that no clear relationship was found between the political articulation of distributive issues and class/age differences in attitudes. Theoretical arguments and previous findings suggested that such a relationship should emerge (Kumlin and Svallfors, 2007). It is unclear whether this should be interpreted as a true nonrelationship or as an indication that the use of Political Manifesto data to indicate the political debate in countries is problematic. The time lag between the last observation in the Manifesto data and our survey data is problematically large. As we saw in the focused Scandinavia versus Eastern Europe comparison, important attitudinal shifts have taken place since 2002. Furthermore, it is open to debate whether the contents of manifestos truly represent the wider political debate in all countries. It is also possible that there are counteracting mechanisms that we have not been able to find and test that cancel out the impact of political articulation on categorical differences in welfare attitudes. More research is clearly needed in order to tease out these relationships. Furthermore, there is little support for the notion that age and class work as cross-cutting cleavages so that small class differences in attitudes are found where age differences are large, and vice versa. On average, across different dependent variables, age and class differences vary independently of each other. An important difference between the age and the class patterns is that the latter are fairly consistent across variables and countries. With almost no exceptions, we find workers to be more supportive of extensive welfare policies than the self-employed and the higher salariat. Age patterns, on the other hand, are far less consistent, and in the few cases where we find

182

Svallfors, Kulin, and Schnabel

c­ onsiderable age differences, they point in different directions. For example, the old are consistently less in favor of public welfare responsibilities than the young in Switzerland, while the young in Denmark are considerably less supportive of family policies than the middle-aged (see the positions of Switzerland and Denmark in Web Appendix Figure A6.3). National idiosyncrasies rather than systematic factors seem to drive age differences in welfare attitudes, where such are found. To us, this would suggest that age is less likely to emerge as a future important social cleavage when it comes to welfare policies. Of course, perceptions regarding the necessity and impact of different public policies may still differ among age groups, giving rise to age differences in attitudes toward spending for specific public policies (Svallfors, 2008; Busemeyer et al., 2009). But as we have seen in this chapter, such differences do not translate into sharply different and systematic views about government responsibilities. Furthermore, we have seen that older populations tend to be associated with smaller age differences in attitudes. As Europe’s populations become gradually older, there are absolutely no indications that this is in fact translated into sharper age conflicts about government responsibilities for various welfare state interventions. Age groups are tied together through families and networks, often with considerable intergenerational transfers taking place within extended families (Kohli, 1999; 2006; 2008). Furthermore, social rights tied to specific stages in the life course make recipients appear as “deserving” of benefits, something that goes especially for the elderly (Van Oorschot and Halman, 2000; Svallfors, 2008). As humans, we also expect “automatic age mobility,” which makes this cleavage different from, for example, class and gender or ethnicity. We would therefore argue that these facts in combination make the emergence of “age politics” a less likely outcome. Overall, the findings in this chapter constitute important challenges for future research. Are previous findings, which indicated a clear macropatterning of class differences in redistributive attitudes, in need of ­fundamental revision, or are the present findings dependent on data collected at a peculiar point in time? For the latter, it would speak that data were collected in the middle of a severe economic crisis, affecting countries differently. As we saw in the case of Denmark and Sweden, it is clearly the case that patterns of age/class differences changed substantially from 2006



Attitudes Toward Government Responsibilities

183

to 2008. On the other hand, one would have to be entirely open to the possibility that there simply are no macro-factors, at least of the kind we have tested here, that systematically affect relative differences in the magnitude of class/age differences across countries. A related interpretation is that we are witnessing some important shifts over time. Previous research was to a large extent dependent on data from the 1990s and at best early 2000s. The focused comparison of developments over time in Denmark and Sweden on the one hand and Poland and Slovakia on the other showed that class differences across countries may be in a stage of transformation. Class differences clearly decline in the Scandinavian countries, while they grow in Eastern Europe. In combination, these trends wipe out previously discernible patterns. Do we see the decline of class as an important social cleavage in the mature and generous welfare states of the North and its rise in the patchy and much less generous welfare states of the former communist countries of Eastern Europe? Yet, it is also possible that we are witnessing some temporary crisis effects that will subside once a more normal state of affairs is reentered. The search for micro-level mechanisms that link class/age location with welfare attitudes showed that perceptions about individual risk and beliefs about the extent of social problems mediated the class-attitudes relationship. But most of the original association between class and welfare attitudes persisted, showing that factors other than those tested here must be at play. Age differences were not affected at all by the suggested mechanisms. It should be emphasized that this result cannot be blamed on the items intended to measure the proposed mechanisms. They work in the expected direction, so people who perceive more risk, believe that social problems are large, are less concerned about future welfare sustainability, and have more self-transcendent and conform values all support more extensive government welfare responsibilities than others. But this only to a limited extent affects the links between class/age location and welfare attitudes. These explorative exercises of finding micro-mechanisms that link age/ class location with welfare attitudes suggest that such mechanisms should largely be sought elsewhere. What could these other possible micro-level mechanisms be? First, we should note that there is no indicator of immediate short-term self-interest in the data set. Our indicators of self-interest relate rather to perceptions about risks in the near future. It is possible that perceptions about gains and losses in the here and now could mediate some

184

Svallfors, Kulin, and Schnabel

of the class/age differences in welfare attitudes. Furthermore, we have no indicators of broader orientations toward the future, where considerations about alternative futures, and their implications for one’s own welfare and of society at large, can be expected to play a role in forming present-day attitudes. Finally, the networks in which people are embedded, that provide them with a sense of what their interests are and who their friends and foes are, could also play a role in explaining categorical differences in welfare attitudes. We leave these considerations for future explorations into the explanatory mechanisms of welfare attitudes and their differences across groups. a p p e n di x 1: h u m a n va l u e s di m e n sion s a n d t h e i r r e sp e c t i v e i t e m s Main question: Now I will briefly describe some people. Please listen to each description and tell me how much each person is or is not like you. Self-enhancement value dimension items POWER #1: It is important to her/him to be rich. S/he wants to have a lot of money and expensive things. POWER #2: It is important to her/him to get respect from others. S/he wants people to do what s/he says. ACHIEVEMENT #1: It is important to her/him to show her/his abilities. S/he wants people to admire what s/he does. ACHIEVEMENT #2: It is important to her/him to be successful and that people recognize her/his achievements.

Self-transcendence value dimension items UNIVERSALISM #1: S/he thinks it is important that every person in the world should be treated equally. S/he believes everyone should have equal opportunities in life. UNIVERSALISM #2: It is important to her/him to listen to people who are different from her/him. Even when s/he disagrees with them, s/he still wants to understand them. UNIVERSALISM #3: S/he strongly believes that people should care for nature. Looking after the environment is important to her/him. BENEVOLENCE #1: It is very important to her/him to help the people around her/him. S/he wants to care for their well-being. BENEVOLENCE #2: It is important to her/him to be loyal to her/his friends. S/he wants to devote herself/himself to people close to her/him.



Attitudes Toward Government Responsibilities

185

Openness to change value dimension items SELF-DIRECTION #1 Thinking up new ideas and being creative are important to her/him. S/he likes to do things in her/his own original way. SELF-DIRECTION #2 It is important to her/him to make her/his own decisions about what s/he does. S/he likes to be free and not depend on others. STIMULATION #1 S/he likes surprises and is always looking for new things to do. S/he thinks it is important to do lots of different things in life. STIMULATION #2 S/he looks for adventures and likes to take risks. S/he wants to have an exciting life. HEDONISM #1 Having a good time is important to her/him. S/he likes to “spoil” her/himself. HEDONISM #2 S/he seeks every chance s/he can to have fun. It is important to her/him to do things that give her/him pleasure.

Conservation value dimension items TRADITION #1 It is important to her/him to be humble and modest. S/he tries not to draw attention to herself/himself. TRADITION #2 Tradition is important to her/him. S/he tries to follow the customs that have been handed down by her/his religion or her/his family. SECURITY #1 It is important to her/him to live in secure surroundings. S/he avoids anything that might endanger her/his safety. SECURITY #2 It is important to her/him that the government ensure her/ his safety against all threats. S/he wants the state to be strong so it can defend its citizens. CONFORMITY #1 S/he believes that people should do what they’re told. S/ he thinks people should follow rules at all times, even when no one is watching. CONFORMITY #2 It is important to her/him always to behave properly. S/he wants to avoid doing anything people would say is wrong. VALUES AND CATEGORIES 1 Very much like me 2 Like me 3 Somewhat like me 4 A little like me 5 Not like me 6 Not like me at all 7 Refusal 8 Don’t know 9 No answer

Respondents answering 7–9 on any of the questions have been removed from the analysis.

186

Svallfors, Kulin, and Schnabel

a p p e n di x 2 : m ac ro -va r i a bl e s Variable

Source

Definition

Old-age dependency ratio (1990 and 2010) Change in old-age dependency ratio 1990–2010 Population over 65 (2005) (%) Gini coefficient disposable income Social expenditure total (2002) Elderly/Nonelderly spending ratio (ENSR) (2005)

UNDP Human Development Report UNDP Human Development Report Eurostat Luxembourg Income Study website OECD Social Expenditure Social Data Base Eurostat

Population 65+/Population 16–64

Trade union density

(Franzese, 1998; Samanni et al., 2008)

Social spending as % of GDP Average spending for old-age in relation to social spending for the nonelderly Union membership as percentage of labor force

Variables from the Comparative Manifestos Data Set Indicators of political articulation have been sought in the Comparative Manifestos data set (Klingemann et al., 2006). This data set indexes the program contents in election manifestos for a number of Western countries for all elections from 1945 and for Eastern and Central European countries from 1990. The data set is based on a classification of the smallest significant units (“quasi-sentences”) of each election manifesto on a number of themes (56 categories distributed among 7 policy areas) (Klingemann et al., 2006). The two measures that we display findings for are as follows: • The share of program content among all political parties that deals with socioeconomic (that is, redistributive) issues • The degree of polarization between the dominant left-wing party and the dominant right-wing party in the socioeconomic/redistributive dimension

These data are averages for the time period 1980–2003 to create a sufficiently large number of observations for all countries (choosing the time period 1960–2003 yields indistinguishable results from those presented). The dimensions in this analysis have been constructed by merging multiple categories as follows: I. Socioeconomic dimension “left-wing themes”: per403 (“Market Regulation”) + per404 (“Economic Planning”) + per405 (“Corporatism”) + per412 (“Controlled Economy”) + per413 (“Nationalization”) + per503 (“Social Justice”) + per504 (“Welfare State Expansion”) + per701 (“Labor Groups: Positive”)



Attitudes Toward Government Responsibilities

187

II. Socioeconomic dimension “right-wing themes”: per401 (“Free Enterprise”) + per402 (“Incentives”) + per414 (“Economic Orthodoxy”) + per505 (“Welfare State Limitation”) + per702 (“Labor Groups: Negative”) + per704 (“Middle Class and Professional Groups”)

For a detailed account of the contents of each category, see Klingemann et al. (2006: App I–II). By adding or subtracting these theme groups, we arrive at the two measures that we use and report in the analyses: • The share of program content in each country that deals with socioeconomic issues (that is, I + II) • The difference between the dominant left-wing party and the dominant right-wing party in the balance between left-wing and right-wing themes in the socioeconomic dimension (that is, I – II for each party)

no t e s 1.  The research has been financed by the Swedish Council for Social and Working Life Research (FAS). Previous versions of the paper were presented at the Conference of the Council for European Studies, April 15–17, 2010, Montreal; at the ISA World Congress, July 11–17, 2010, Gothenburg; and at the Rockefeller Centre, Bellagio, November 24–25, 2010. Juan Fernandez provided data and advice. Thanks to Michael Shalev, Staffan Kumlin, Edeltraud Roller, John Huber, Marcel Lubbers, Juan Fernandez, Achim Goerres, and several WAE participants for thoughtful comments. 2.  A short, highly formalized paper void of empirical substance has achieved an almost paradigmatic status in the American political science literature in this field (Melzer and Richard, 1981). Even those who are highly critical of their perspective, in which a self-interested demand for redistribution is the prime driver of redistributive policies, feel compelled to cite it. 3.  The concept “beliefs” is sometimes used as an “umbrella concept” that covers all the different aspects of subjective orientations. We use it here in a much more specific sense. 4.  Sometimes in the literature, “values” are used interchangeably with what we here call “attitudes.” We try to make a clear distinction between attitudes and values. 5.  The chapter builds on release 3.0 of data from the European Social Survey 2008, containing data from 26 countries: Belgium (BE), Bulgaria (BG), Croatia (HR), Cyprus (CY), Czech Rep. (CZ), Denmark (DK), Estonia (EE), Finland (FI), France (FR), Germany (DE), Great Britain (GB), Greece (GR), Hungary (HU), Latvia (LV), Netherlands (NL), Norway (NO), Poland (PL), Portugal (PT), Romania (RO), Russia (RU), Slovakia (SK), Slovenia (SI), Spain (ES), Sweden

188

Svallfors, Kulin, and Schnabel

(SE), Switzerland (CH), and Ukraine (UA). Israel and Turkey have been left out from the data set, since they are not European countries. Furthermore, they lack many of the macro-indicators we use in our analyses. The Turkish attitudinal patterns are also highly idiosyncratic, which suggests that the questions in the survey may have quite different meanings in the Turkish context compared to the European one. 6.  The factor loadings for each of the three dimensions—labor market policies, health care/pension policies, and family support policies—range from .87 to .93, with the underlying construct explaining 75 to 87 percent of the variation in the observed variables. Cronbach’s alpha reliability measures are .67, .85, and .74, respectively (for all countries combined). 7.  The three different indicators of past unemployment include “ever unemployed for more than three months,” “ever unemployed more than a year,” and “ever unemployed within last five years.” 8.  We have tried several other specifications of these variables, such as using the midpoints of the categories or grouping answer categories and entering them as a set of dummy variables. Results remain very stable. (SPSS output can be provided by authors.) 9.  A potential problem with computing raw factor scores involves the question of measurement invariance—that is, equivalence of meaning in the constructs across countries. Computing factor scores does not allow for constraining the factor loadings to be set equal across countries, so measurement equivalence cannot be established. Recent studies using more sophisticated methods for assessing equivalence show that the assumption of measurement invariance across countries, with only a few modifications, seems to hold for previous rounds of the ESS using the same measurement instrument for basic human values (e.g., Davidov et al., 2008). This means that the instrument seems to measure approximately the same latent construct across countries, which makes a reasonable, although not optimal, case for utilizing factor scores. 10.  A peculiar exception is found for the Health Care and Support for Elderly index in Bulgaria, where moving the cutoffs three years backward produces medium-sized and statistically significant age differences, in contrast to the small age differences displayed in Table 6.2. 11.  We used a six-class version of the recoding schema constructed by Harry Ganzeboom (http://home.fsw.vu.nl/hbg.ganzeboom/isko88/index.htm), which has been used extensively in comparative attitude research (see, for example, Svallfors, 2006). The coding is based on the occupation and employment status of the respondent. Pensioners, unemployed, and people outside the labor market have been classified according to their last occupation (if any). Recoding algorithms (in SPSS code) may be obtained from the authors on request. We checked the robustness of our results by comparing them to the ones produced by using the European Socioeconomic classification, which is another operationalization of the Erikson-Goldthorpe class schema (some slight peculiarities with the findings from the ESeC schema make us prefer the Ganzeboom schema). We find that although



Attitudes Toward Government Responsibilities

189

a substantial number of respondents fall outside the main diagonal when we cross-tabulate the two schemas, the strength and patterns of association between class and our dependent variables are strikingly similar across the schemas. The relative sizes of class differences among countries are not affected at all. 12.  All these advantages of the Κ index are bought at the expense of (a) not getting indications of the precision of the Κ values (such as standard errors) and (b) not seeing the actual class/age patterns displayed in the tables. We still think the advantages of the measure outweigh the drawbacks. 13.  So the K index for class is the standard deviation of the values −1.701, −.399, −.127, .063, −.771, and 0, and the K index for age is the standard deviation of the values .998, 1.225, 1.712, 1.512, and 0. Changing the reference category does not change the value of the K index. 14.  In order not to further clutter a chapter that already contains so much tabular information, we do not report findings regarding the association between various indicators of mechanisms and attitudes. Remember that our focus here is set on how they mediate the class/age position and welfare attitudes. All output from the full models is available from the authors on request. 15.  This also means that values do not work as self-selection devices in allocating people to class positions (see further Brooks and Svallfors, 2010). 16.  We use the Elderly/Nonelderly Spending Ratio (ENSR), as proposed by Lynch (2006:30–1), to measure the age bias of public spending. 17.  Reliable inequality measures do not exist for all countries. We rely on the measures of disposable income from the Luxembourg Income Study (LIS), which is the only income inequality data set where data are truly comparable across countries. Accordingly, some of the countries are not included in Figure 6.2 because they are not included in the LIS database. Alternative measures of inequality from the LIS database, such as the 90/10 decile ratio, yield results that are consistent with those displayed in Figure 6.2. 18.  Although the focus of Rueda and Pontusson is set on income strata rather than class, the logic and findings are quite compatible with the findings for class in this regard. 19.  Data for Russia and Slovakia are not included in these analyses, since their political systems were semiauthoritarian for at least parts of the Manifesto Data period. Hence, it makes little sense to talk about dominant right-wing and dominant left-wing parties.

r efer ences Andress, H. J., and T. Heien. 2001. “Four Worlds of Welfare State Attitudes? A Comparison of Germany, Norway, and the United States.” European Sociological Review 17: 337–56. Bean, C., and E. Papadakis. 1998. “A Comparison of Mass Attitudes Towards the Welfare State in Different Institutional Regimes, 1985–1990.” International Journal of Public Opinion Research 10: 211–36.

190

Svallfors, Kulin, and Schnabel

Blekesaune, M., and J. Quadagno. 2003. “Public Attitudes Toward Welfare State Policies: A Comparative Analysis of 24 Nations.” European Sociological Review 19: 415–27. Brooks, C., and S. Svallfors. 2010. “Why Does Class Matter? Policy Attitudes, Mechanisms, and the Case of the Nordic Countries.” Research in Social Stratification and Mobility 28, 199–213. Budge, I. 2001. Mapping Policy Preferences: Estimates for Parties, Electors, and Governments, 1945–1998. Oxford and New York: Oxford University Press. Busemeyer, M. R., A. Goerres, and S. Weschle. 2009. “Attitudes Towards Redistributive Spending in an Era of Demographic Ageing: the Rival Pressures from Age and Income in 14 OECD Countries.” Journal of European Social Policy 19: 195–212. Cusack, T., T. Iversen, and P. Rehm. 2006. “Risks at Work: The Demand and Supply Sides of Government Redistribution.” Oxford Review of Economic Policy 22: 365–89. Davidov, E., P. Schmidt, and S. Schwartz. 2008. “Bringing Values Back In: the Adequacy of the European Social Survey to Measure Values in 20 Countries.” Public Opinion Quarterly 72: 420–45. Eagly, A. H., and S. Chaiken. 1998. “Attitude Structure and Function.” In The Handbook of Social Psychology, 4th ed., edited by D. T. Gilbert, 269–322. New York: McGraw-Hill. Edlund, J., and S. Svallfors. 2011. “Cohort, Class and Attitudes to Redistribution in Two Liberal Welfare States: Britain and the United States, 1996–2006.” In Generational Politics and Policies: Comparative Studies of Ageing Postindustrial Democracies, edited by A. Goerres and P. Vanhuysse, 206–24. London: Routledge. Erikson, R., and J. H. Goldthorpe. 1992. The Constant Flux: A Study of Class Mobility in Industrial Societies. Oxford: Clarendon Press. Fehr, E., and U. Fischbacher. 2005. “The Economics of Strong Reciprocity.” In Moral Sentiments and Material Interests. The Foundations of Cooperation in Economic Life, edited by H. Gintis, S. Bowles, R. Boyd, and E. Fehr, 151–91. Cambridge, MA: The MIT Press. Fong, C., S. Bowles, and H. Gintis. 2005. “Reciprocity and the Welfare State.” In Moral Sentiments and Material Interests. The Foundations of Cooperation in Economic Life, edited by H. Gintis, S. Bowles, R. Boyd, and E. Fehr, 277–302. Cambridge, MA: The MIT Press. Franzese, R. J. 1998. Participation, Inequality, and Transfer Database: http:// www-personal.umich.edu/~franzese/Publications.html. Gintis, H., S. Bowles, R. Boyd, and E. Fehr, eds. 2004. Moral Sentiments and Material Interests: the Foundations of Cooperation in Economic Life. Cambridge, MA: MIT Press. Goerres, A. 2009. “Einstellungen zu umverteilender Politik. Babyboomer in Grossbritannien und Westdeutschland im Vergleich.” Zeitschrift für Politikwissenschaft 19: 205–35.



Attitudes Toward Government Responsibilities

191

Goldthorpe, J. H. 2000. Social Class and the Differentiation of Employment Contracts. In On Sociology: Numbers, Narratives and the Integration of Research and Theory, edited by J. Goldthorpe, 206–29. Oxford: Clarendon Press. Hedström, P., and R. Swedberg. 1996. “Social Mechanisms.” Acta Sociologica 39: 281–308. Hout, M., C. Brooks, and J. Manza. 1995. “The Democratic Class Struggle in the United States, 1948–1992.” American Sociological Review 60: 805–28. Iversen, T., and D. Soskice. 2001. “An Asset Theory of Social Policy Preferences.” American Political Science Review 95: 875–93. Klingemann, H.-D., A. Volkens, J. Bara, I. Budge, and M. Mcdonald. 2006. Mapping Policy Preferences II: Estimates for Parties, Electors, and Governments in Eastern Europe, European Union, and OECD 1990–2003. New York: Oxford University Press. Kohli, M. 1999. “Private and Public Transfers Between Generations.” European Societies 1: 81–104. ———. 2006. “Aging and Justice.” In Handbook of Aging and the Social Sciences, 6th ed., edited by R. Binstock and L. George, 456–78. San Diego, CA: Elsevier. ———. 2008. “Generational Equity. Concepts and Attitudes.” In Pension Reform in Europe: Politics, Policies and Outcomes, edited by C. Arza and M. Kohli, 197–214. London: Routledge. Kulin, J. 2011. Values and Welfare State Attitudes. The Interplay Between Human Values, Attitudes and Redistributive Attitudes Across National Contexts. PhD thesis. Department of Sociology, Umeå University. Kumlin, S. 2004. The Personal and the Political: How Personal Welfare State Experiences Affect Political Trust and Ideology. New York: Palgrave Macmillan. Kumlin, S., and S. Svallfors. 2007. “Social Stratification and Political Articulation: Why Attitudinal Class Differences Vary Across Countries.” In Social Justice, Legitimacy and the Welfare State, edited by S. Mau and B. Veghte, 19–46. Aldershot, UK: Ashgate Publishers Ltd. Linos, K., and M. West. 2003. “Self-Interest, Social Beliefs, and Attitudes to Redistribution. Re-addressing the Issue of Cross-National Variation.” European Sociological Review 19: 393–409. Lynch, J. 2006. Age in the Welfare State: the Origins of Social Spending on Pensioners, Workers, and Children. Cambridge, UK, and New York: Cambridge University Press. Massey, D. S. 2007. Categorically Unequal: the American Stratification System. New York: Russell Sage Foundation. Mau, S. 2003. The Moral Economy of Welfare States. Britain and Germany Compared. London: Routledge. ———. 2004. “Welfare Regimes and the Norms of Social Exchange.” Current Sociology 52, 53–74.

192

Svallfors, Kulin, and Schnabel

Mayer, K. U., and U. Schoepflin. 1989. “The State and the Life Course.” Annual Review of Sociology 15: 187–209. Mayntz, R. 2003. “Mechanisms in the Analysis of Social Macro-Phenomena.” Philosophy of Social Science 34: 237–59. Melzer, A. H., and S. F. Richard. 1981. “A Rational Theory of the Size of Government.” Journal of Political Economy 89: 914–27. Opp, K.-D. 2005. “Explanations by Mechanisms in the Social Science. Problems, Advantages and Alternatives.” Mind & Society 4: 163–78. Rokeach, M. 1968. “The Role of Values in Public Opinion Research.” Public Opinion Quarterly 32: 547–59. ———. 1973. The Nature of Human Values. New York: Free Press. Rueda, D., and J. Pontusson. 2010. “Individual Preferences for Redistribution in Western Europe: Self-Interest, Political Articulation, Altruism and Identity.” Working paper, February 2010. Samanni, M., J. Teorell, S. Kumlin, and B. Rothstein. 2008. The Quality of Government Social Policy Dataset, Version November 4, 2008. Gothenburg: University of Gothenburg: The Quality of Government Institute. Schwartz, S. H. 1992. “Universals in the Content and Structure of Values— Theoretical Advances and Empirical Tests in 20 Countries.” Advances in Experimental Social Psychology 25: 1–65. ———. 1994. “Are There Universal Aspects in the Structure and Contents of Human Values?” Journal of Social Issues 50: 19–45. ———. 2001. “Extending the Cross-cultural Validity of the Theory of Basic Human Values with a Different Method of Measurement.” Journal of CrossCultural Psychology 32: 519–42. Svallfors, S. 1997. “Worlds of Welfare and Attitudes to Redistribution: A Comparison of Eight Western Nations.” European Sociological Review 13: 283–304. ———. 2003. “Welfare Regimes and Welfare Opinions: a Comparison of Eight Western Countries.” Social Indicators Research 64: 495–520. ———. 2004. “Class, Attitudes and the Welfare State: Sweden in Comparative Perspective.” Social Policy & Administration 38: 119–38. ———. 2006. The Moral Economy of Class: Class and Attitudes in Comparative Perspective. Stanford, CA: Stanford University Press. ———. 2008. “The Generational Contract in Sweden: Age-Specific Attitudes to Age-Related Policies.” Policy and Politics 36: 381–96. Tilly, C. 1998. Durable Inequality. Berkeley: University of California Press. Van Oorschot, W. and L. Halman., 2000. “Blame or Fate, Individual or Social? An International Comparison of Popular Explanations of Poverty.” European Societies 2: 1–28. Wright, E. O. 1985. Classes. London: Verso. ———. 1997. Class Counts: Comparative Studies in Class Analysis. Cambridge, UK, and New York: Cambridge University Press.

chapter seven

Framing Theory, Welfare Attitudes, and the United States Case1 Clem Brooks

How and why do welfare attitudes change? What conditions tend to exert the greatest pressures toward transformations of opinion? These are fundamental challenges for welfare state scholarship. Indeed, they are of substantial consequence if such forces as demographic shifts in population composition or economic crises have the capacity to remake mass attitudes toward social policy, potentially undermining (or instead buttressing) the legitimacy of welfare states around the globe. To gain new purchase on questions about change in welfare attitudes, I bring to bear a novel source of data and theoretical perspective on opinion formation. The data come from The 2009 Survey of American Policy Attitudes (SAPA). A distinguishing feature of the SAPA data collection is the use of embedded survey experiments. Here, key expectations and the underlying research design derive from scholarship on what has come to be known as “framing theory” (Chong and Druckman, 2007b). Why framing theory? A potential contribution for comparative welfare state scholarship is to highlight processes that induce pressures toward policy-attitude change among individuals. In doing so, framing theory enables researchers to take productive steps toward linking individual- and macrolevel processes, identifying when political conflicts and other environmental changes may generate pressures that ultimately culminate in aggregate opinion shifts. A quite different scenario is when characteristics of the institutional For additional data, please see the Web Appendix tables and figures by following the link in the listing for Contested Welfare States on our website: http://www.sup.org.

193

194

Brooks

environment dispose individuals to instead give identical weight to the same policy considerations over time, facilitating inertia in aggregate opinion. As discussed in greater detail, this is a second point of applicability that framing theory has with respect to comparative welfare state research. With these ideas in mind, the first section of this chapter takes up the challenge of understanding change processes. Here, I use two recent strains of welfare state scholarship by way of example, considering first institutionalist work on policy feedback and path dependency, and moving next to extensions that focus on mass opinion and policy responsiveness. Together, these literatures suggest benefits of a fuller engagement with framing theory’s points of relevance and associated hypotheses. The chapter’s second section discusses the logic of SAPA’s framing experiments. Using these new data, I focus on two candidate sources of pressure toward opinion change in the contemporary United States. There is evidence that the recent economic crisis may have exerted surprisingly modest influence on policy-attitude formation. By the same token, perceptions relating to immigrants as a symbolic target group may be of greater consequence. The chapter provides further perspective on the U.S. results with crossnational comparisons using Round 4, European Social Survey (ESS) data. These comparisons suggest that American welfare attitudes are distinguished in cross-national terms by notably low levels of public support for government responsibility as regards social provision. This result is important for questions concerning change and convergence pressures on welfare attitudes in specific, national settings. In conclusion, I summarize key implications from this chapter’s application of framing theory and survey experiments. t h e qu e s t ion of c h a ng e Institutionalist Approaches How do mature welfare states change? What conditions exert pressure toward change? Questions about welfare state change were central to the classic scholarship on social policy development. In applications of modernization theory, for instance, demographic changes in population composition were seen as powerful levers of welfare state emergence (e.g., Cutright, 1965; Wilensky and Lebeaux, 1965). And in the early classics of power resources scholarship, the ebb and flow of partisan control over national



Framing Theory, Welfare Attitudes, and the U.S.

195

government, itself shaped by historical and country-specific alignments of class forces, were identified as central mechanisms of change and development (e.g., Korpi, 1978; Esping-Andersen, 1990). During the next two decades, a good deal of scholarship evolved into a more ambivalent posture toward questions about change, sometimes as a matter of focus and other times as a matter of adherence to theoretical models arguing for the “locking-in” of institutional development. One condition facilitating this shift in scholarly orientation was accumulating evidence for the maintenance and, at times, growth of social policy provisions (Huber et al., 1993; Pierson, 1994; Iversen and Cusack, 2000). That welfare states were often robust in the face of economic globalization and budgetary constraints in many national contexts propelled institutionally minded scholars to develop new theoretical models of policy persistence in the 1990s and the early years of the new millennium. A watershed in this literature was the publication of Pierson’s (1996; see also Skocpol, 1992) influential account of the new politics of the welfare state. Focusing on feedback effects of policy development, Pierson identified an asymmetry in incentives on the part of government officials with respect to credit-claiming initiatives for welfare maintenance (or expansion) versus blame-avoidance attempts associated with the prospect of retrenchment. So vividly did Pearson’s model appear to capture the internal dynamics of welfare state persistence that a significant portion of institutionalist scholarship was soon reoriented toward the locking-in of welfare state institutions. If, as economic historians had demonstrated (David, 1985), such initially arbitrary and subsequently inefficient arrangements as the QWERTY keyboard had come, over time, to displace alternatives and develop their own legitimacy, why couldn’t the logic of path-dependent development apply to welfare state institutions? Institutionalist models of policy feedback and path-dependency by no means rule out either transformations of or more modest alterations in the historical trajectory of welfare state policies (e.g., Pierson, 2001; Rothstein and Steinmo, 2002). Still, the focus on persistence or monotonic development is strongly anticipated in this scholarly tradition. When it comes to questions about change, an underlying challenge thus regards the conditions under which policies (or policy change) lead to negative consequences that undermine welfare state legitimacy rather than positive, institution-enhancing effects. Put another way, what prevents welfare states from at times gen-

196

Brooks

erating (or becoming the object of) negative reactions on the part of voters, thus departing from path-dependent expectations of monotonic growth? These questions pose constructive challenges for much institutional thinking about welfare states. Some of the most intriguing and seemingly paradoxical evidence on this point has come from the 1996 case of U.S. welfare reform (Soss and Schram, 2007; Dyck and Hussey, 2008). There, policy change did not bring its legislative authors (and the Democratic Party, more generally) the expected degree of electoral credit among citizens, and Americans attitudes toward means-tested “welfare” programs appeared to have changed little, suggesting an anomaly for institutional theory. Instances of this sort underscore the challenge of understanding change dynamics, including with reference to questions about when and why welfare attitudes are influenced by the trajectory of policy. Framing theory, as outlined following, provides a perspective from which to consider some relevant mechanisms and associated hypotheses. Policy Responsiveness The challenge of understanding change dynamics in welfare attitudes can be further appreciated by considering a second, related body of scholarship. This work has sought to extend institutionalist scholarship by incorporating factors relating to mass opinion into research on the policy process. Starting with insights from empirical democratic theory, this literature (e.g., Page and Shapiro, 1983; Burstein, 1998; Erikson et al., 2002) argues that the organization of political institutions within democracies gives officials an incentive to comply with the aggregate preferences of national publics, providing, of course, that key conditions are met. 2 Policy responsiveness happens if governments or specific officials comply indirectly through the outcome of elections. In this scenario, politicians learn about citizen preferences through who wins and who loses in actual electoral contests. In contrast, direct policy responsiveness occurs when parties or coalitions seek to avoid losses prior to elections by implementing policy decisions that are popular with national publics. This type of linkage is less well-known to generalist scholars of politics, yet it has been particularly central to recent empirical scholarship on democratic responsiveness (Stimson, 2004; Wlezien, 2004; Canes-Wrone, 2006). But when applied to the study of comparative welfare states, policy responsiveness scholarship often reveals some of the same ambivalence as institutionalist models with respect to questions about change. Mature



Framing Theory, Welfare Attitudes, and the U.S.

197

welfare states, particularly those characterized by universalistic program design and high levels of spending effort, may be viewed as relatively stable because of high levels of program popularity (Brooks and Manza, 2007; see also Rothstein, 1998). Similarly, linkages between central cleavages such as those based on class and social policy institutions create lasting constituencies and “moral economies” of welfare that may be slow to change (Svallfors, 2006). Exemplifying again the expectation of (positive) policy feedback, welfare state development can create a “policy ratchet” in which conservative parties accept new forms of social provision because, as Huber and Stephens (2001) state, “those reforms were popular with the mass public.” As in the earlier case of institutionalist scholarship, policy responsiveness and related strains of research do not logically preclude transformations in welfare attitudes and institutions. But an underlying expectation in much of this work is that of stability or relatively glacial change; some recent evidence for convergence in country-level welfare opinions thus represents an important paradox (Edlund and Svallfors, 2011). Grasping the conditions under which welfare attitudes change more rapidly or move away from historical patterns of policy support is critical. The time is ripe to perhaps reconsider the locking-in scenario with respect to both welfare policies and policy attitudes. f r a m i ng t h e ory How does framing theory bear on the issues at hand? Framing scholarship argues for the causal relevance of cognitive cues and issue descriptions for understanding the dynamics of the survey response and, in turn, the measurement of policy attitudes. Framing is the process by which these cues and descriptions vary in experimental or naturalistic settings. Systematic research (incorporating control groups) dates back to the classic questionwording experiments of early survey research (for review, see Sniderman and Grob, 1996), where early experiments yielded unexpectedly large impacts in the wake of modest changes in item wording. Subsequent scholarship has provided a firmer theoretical basis for understanding more precisely why (and when) such may shape measured attitudes in the first place. Central to the maturation of framing scholarship has been the incorporation of established psychological theories of cognition (Chong and Druckman, 2007b; see also Gilovich et al., 2002), including, with respect to the

198

Brooks

heuristics and biases tradition of scholarship. Here, individuals’ choices and evaluations are properly understood as subject to conditions of low-information, uncertainty, and ambivalence in beliefs and other predispositions. As summarized in the work of Zaller (1992), the attitude measurement process should be seen as intrinsically probabilistic, with individuals responding to contextual stimuli by sampling from considerations that are accessible in specific contexts. By the same token, because some predispositions such as left/right ideology are more firmly anchored in personality characteristics or habitual behavior, the measurement of attitudes can lead to predictable responses to survey questions (Jost et al., 2009). This is where framing enters into the picture. When frames matter, they tend to influence attitude formation through several distinct processes. The simplest scenario is by making available new information, as when the results of scientific researchers or other experts diffuse through a population. Providing that individuals accept and process such new stimuli, information-related framing has the potential to rapidly reorient policy attitudes. But these limiting conditions are key. If individuals instead engage in what cognitive psychologists term “motivated” reasoning (Kunda, 1990; Ditto and Lopez, 1992; Jost et al., 2004), they may be disposed to selectively accept new stimuli, filtering out those items perceived as inconsistent with prior beliefs. Laboratory studies suggest that the communication of negative self-information, and perhaps bad news more generally, can stimulate resistance, in significant part because of cognitive dissonance (see Festinger, 1957). Under these conditions, framing and alternative cognitive cues may have little to no impact on policy-attitude formation. This brings us to a third scenario, where new cues or issue descriptions mobilize preexisting beliefs on the part of individuals. Rather than communicating new information about the external environment, this framingrelated activity is best understood as priming, where biases to which individuals already subscribe are made accessible (Kinder, 1998). In an illustrative example of this process, Sniderman and colleagues (2004) found that when questions about public fears of violence are conjoined with additional reference to “ethnic minorities,” support for tolerance declines, and attitudes simultaneously become more interconnected. In this case, framing matters because it prompts individuals to develop new connections between preexisting beliefs and the issue at hand. This raises the likelihood that



Framing Theory, Welfare Attitudes, and the U.S.

199

endorsed. 3

new policy positions may be It is not only information that can represent a significant input into policy-attitude formation. Implications for Welfare State Scholarship There are strong points of relevance between the psychological processes identified by framing theory and the study of comparative welfare states. In general, if the macro-environment in which social policy attitudes are formed shifts so as to propel individuals to process new information or weigh alternative considerations, this may alter public responses to policies. If these environmental changes persist, this raises the likelihood of aggregate change in policy opinion. With reference to welfare states, this shift could occur if, for instance, right-wing parties and other market-oriented organizations disseminate new ideas or information that are received by individuals and incorporated into their policy reasoning. A similar outcome could occur if the ideas that are endorsed by right-parties (and other relevant organizations) are held constant, but an asymmetric shift in media access or other means of disseminating policy ideas consistently favors welfare state critics. Of course, for such environmental changes to matter over the long term for aggregate opinion formation, alternative frames sponsored by competing parties and organizations must continue to be at a comparative disadvantage. Since partisan conflict is a virtual certainty, ideological competition has been seen as potentially neutralizing competing frames (Chong and Druckman, 2007a). Under these conditions, aggregate opinion may be left largely intact. This brings us to a second point about framing theory. Specifically, linkages among macro-environmental change, frame communication, and reception by individuals are always subject to numerous contingencies. If, for instance, individuals are inattentive to new ideas, subsequent opinion impacts will tend to be minimal. Of further relevance is the motivated reasoning scenario, where individuals are willfully inattentive to new information/ideas because their predispositions filter which environmental cues are processed (Jost et al., 2009). To probe this scenario empirically, I consider a battery of experiments making reference to the recent economic crisis. This is by itself a historically prominent cue. It may have considerable relevance to welfare attitudeformation, prompting individuals to rethink and demand more benefits and services from government. Indeed, that expectation is central to the

200

Brooks

­ thermostatic” model of opinion change, where fluctuations in the business “ cycle are seen as continually remaking mass opinion (Wlezien, 1995). But opinion responsiveness to economic change (or information about such change in the form of framing) is hardly a given. As outlined earlier, an alternative scenario is that welfare state beliefs operate as such established filters as ideology, making individuals reluctant to relinquish prior policy attitudes. It is well worth emphasizing that the preceding scenario is in line with macro-level scholarship arguing for inertia in policy institutions or welfare opinions. Framing theory also suggests the utility of directly probing the responses of individuals to competing frames. This is useful in mimicking the real-life environment of political competition. Contemporary political environments are those in which campaigning and media activities frequently involve the use of poll-driven results in an effort to sway enough citizens in one direction or another. These communication strategies often take the form of framing, where politicians and other communicative actors seek to define issues, often in a subtle or coded fashion, so as to maximize their advantage (Jacobs and Shapiro, 2000; Druckman and Jacobs, 2006). As an experiment into this aspect of environment/attitude linkage, I use a series of experiments into policy-relevant target groups, where the underlying beneficiaries of social policies are manipulated. These target group cues may, if potent for policy-attitude formation, stimulate preexisting biases involving feelings of sympathy (or antipathy) toward the groups in question. Here, I consider immigrants, the middle-class, the working class, and “all Americans” as suitable targets in the United States context. As before, motivated reasoning may act to neutralize or otherwise make irrelevant treatment effects in which reference is made to groups. It is through considering the empirical results that we can begin to see precisely how such processes may influence welfare attitude-formation. a n e m p i r ic a l a p p l ic at ion usi ng t h e 2 0 0 9 su rv e y of a m e r ic a n p ol ic y at t i t u de s data How might frames that call attention to different groups in U.S. society affect welfare attributions and the level of support for public social provisions? And in the context of a worldwide financial and economic crisis, how might cuing these considerations influence welfare attitudes? These



Framing Theory, Welfare Attitudes, and the U.S.

201

questions carry forward the application of framing theory that I pursue in this chapter. Survey Experiments and Research Design The results I present are based on data from a recent national survey of adult Americans: The 2009 Survey of American Policy Attitudes (SAPA). This survey was carried out by the Center for Survey Research at Indiana University, one of the premier academic survey centers in the United States. A large portion of SAPA was devoted to fielding for the first time in the United States several batteries of items drawn from Round 4 of the European Social Surveys (ESS). Interviews were conducted using computerassisted telephone interviewing (CATI) methods; items were pretested on April 7, 2009, with production beginning May 15 and ending on August 4. Average interview length was 29.8 minutes, and the total number of cases was 1559. Additional details, including survey protocols, interviewer training, and sample characteristics, are summarized in the Appendix.4 A pair of SAPA’s design features merit discussion. First is the generous use of embedded survey experiments, where respondents are randomly assigned to experimental versions of specific items. Because of the use of a randomized design, differences in measured opinion across item version can be attributed to the influence of frames, as indexed by question wording differences. Statistical comparisons across item versions are thus informative. In contrast to conventional, nonexperimental research, subsequent results are less vulnerable to selection bias and other sources of causal uncertainty. A second distinguishing feature of SAPA is the use of baseline policyattitude items drawn from the ESS. The use of established survey instrumentation enhances reliability and the possibility of comparisons across surveys/countries. In all survey experiments, the baseline (unchanged) wording of ESS items is the control group condition, with subsequent treatment-effect versions representing alternative issue frames. Target Group Experiments Using a battery of five ESS items, SAPA’s first set of experiments seek to gauge the impact of target groups on perceptions and attributions concerning a number of welfare state consequences. Classic survey-based scholarship on target groups posits prejudice as frequently influencing policy evaluations (e.g., Schuman et al., 1997; Sears et al., 2000), as long as these

202

Brooks

predispositions are stimulated by framing or other contextual cues. For the current application, a rich vein of welfare state theorizing provides further grounds for viewing target groups as relevant because of underlying associations with beliefs about the deservingness of specific groups (Oorschot, 2008; see also Steensland, 2006). Do target groups indeed matter for individuals’ reasoning about welfare? How much do target group frames shift opinion? Based on past laboratory studies and scholarship on the availability heuristic (Iyengar, 1990; Chong and Druckman, 2007a), we may expect target groups to matter for attitude formation. But absent randomization in the assignment of treatment effects, causal inference from observational data is intrinsically uncertain. The target group experiments provide a means of reducing causal uncertainty, while bringing to bear framing theory tools for welfare state scholarship. Incorporating established scholarship on the racial politics of the U.S. welfare state (Quadagno, 1994; Gilens, 1999), my experiments use issue frames referring, in turn, to “blacks” and the “middle class.” An additional issue frame identifies “immigrants” as a target group. This provides a means of using scholarship on the political relevance of immigration (Nannestad, 2007; see also Alesina and Glaeser, 2004) and its symbolic dimensions to the study of welfare opinions. If, as expected, target groups matter in cuing latent beliefs about group deservingness, we can expect positive results from the framing experiments. If so, which group cues represent the largest treatment effects is informative. That tells us how much immigrant and working/middle-class statuses may by themselves operate as potent mechanisms behind opinion formation on social welfare issues in the United States. The Economic Crisis Experiments The second set of experiments concerns the impact of the economic crisis on policy-attitude formation. These experiments take advantage of the unique historical context afforded by 2009 global financial crisis. Here, six ESS baseline items again represent control group conditions, with a first treatment including further reference to the “impact of the economic crisis on all Americans.” This inclusive and even nationalist-tinged framing is deliberate. It tracks a long-standing literature on the importance of “sociotropic” policies for understanding economic policy attitudes. If reference to



Framing Theory, Welfare Attitudes, and the U.S.

203

this sentiment-rich idea of the American public enhances policy support, that attests to the latent power of this issue frame. It would also mean that we can expect macro-economic change to exert pressure toward opinion change beyond the historical context of the survey itself. The crisis experiments also incorporate a second dimension of framing-related variation. Here, the second experimental condition applies the crisis-related wording from the first experiment, only now substitutes “middle-class Americans” or “working-class Americans” for “all Americans.” The middle- and working-class frames derive from past scholarship (e.g., Jackman and Jackman, 1985; Lipset, 1996), arguing for the ideological power of middle-class identity in the United States and for the accompanying irrelevance or even negative connotation of working-class identity. This should return differential effects on opinion formation across middle- versus working-class treatments. As in the group deservingness experiments, randomized assignment makes it possible to test whether (and which) frames shape opinions. a n a ly se s a n d r e su lt s How Do Target Groups Matter? Let’s begin with the framing experiments into target group effects. These experiments are embedded in the five items that probe respondents’ perceptions of the consequences of social benefits and services provision. Table 7.1 summarizes results, reporting item means across the baseline (control group) and three treatment conditions. F-test statistics for analysis of variance are presented in the table’s final column. For all items in Table 7.1, higher scores indicate more negative views.5 Starting with the economic strain item from the first row, .44 is the average item response for the baseline item replicated from the ESS. By itself, this indicates that in 2009 Americans tacked slightly toward the position that social benefits and services are not a strain on the economy. What happens when the initial, baseline item wording is conjoined with additional reference to key target groups? The first experimental condition yields an item mean of .62. This higher score indicates that the “immigrants” frame facilitates new receptivity to the position that welfare provision has negative consequences for the economy. But the same is not true for either the “blacks” or “middle class” framing experiments: With mean

204

Brooks t a b l e 7. 1 Experiments with target groups: Social benefits and services

[. . .] place too great a strain on the economy. [. . .] make people lazy. [. . .] do not prevent widespread poverty. [. . .] encourage people from other countries to come and live here. [. . .] cost businesses too much.

Baseline

Including for immigrants

Including for blacks

Including for middle class

F-test

.44 .47

.62 .50

.42 .44

.44 .41

24.91* 4.62*

.49

.50

.55

.50

3.11

.70

.74

.61

.67

12.47*

.54

.55

.45

.49

7.52*

* p < .01.

scores of .42 and .44, respectively, these experimental conditions are statistically indistinguishable from the baseline (no-frame) condition.6 Combined with information provided by the F-statistic, we can infer that it is the idea of immigrants as a target group that has a significant potential to reorient beliefs about interrelationships between welfare provisions and the U.S. economy. When Americans are cued to think about immigrants, their welfare attitudes become strikingly more negative. The second and third experiments in Table 7.1 use baseline ESS items that refer, in turn, to the ideas that benefits and service provision “make people lazy” and do not “reduce poverty.” These experiments yield different results. In the “lazy” experiment, it is the contrast between the “middle class” frame versus the control and “immigrants” frames that shows significant differences. Specifically, reference to middle-class recipients leads to more positive welfare attributions. In the poverty experiment, overall group differences are much smaller, and the p-value for overall group differences is .03. If we look more closely at pairwise contrasts using post hoc tests, the only candidate for group differences relates to the contrast between blacks as a target group versus the baseline item.7 What of the experiment using the item concerning immigration as a perceived (and implicitly undesirable) consequence of welfare states? There is clear evidence of framing effects. For instance, both the “black” and “middle class” frames have the effect of reducing negative attitudes relative to the “immigration” frame.



Framing Theory, Welfare Attitudes, and the U.S.

205

The final experiment into evaluations of welfare costs yields generally similar results. In contrast to baseline and “immigrants” conditions, the “black” frame again has the effect of reducing negative attributions concerning costs. The nonsignificant difference across baseline and “immigrants” conditions is an informative result. As discussed further in the chapter’s conclusion, this result indicates that U.S. survey respondents may already consider issues of immigration when evaluating welfare cost issues. How Does the Economic Crisis Matter? I now consider results for the second set of framing experiments. Here, the primary focus is on the impact of the economic crisis on opinion formation. An additional focus is on the crises’ intersection with frames emphasizing middle- versus working-class Americans as potent points of reference. The results are presented in Table 7.2, and, as before, I present item means across control and experimental conditions. To take into account the presence of a pair of intersecting treatment effects, I use a nested ANOVA structure. There are now two F-test results: one pertaining to the crisis-related experimental cues and one for the class-related treatments. The results of these experiments are easy to summarize. With one possible exception, none of the framing experiments returns significant effects on measured opinions. The single exception is the “child care” item, where the F-statistic for the crisis tests yields a p-value of approximately .02. At face value, there is some possible evidence that the crisis cues shape policy attitudes when it comes to child care service provision (though without the class-related treatments also shaping opinions). But the size of this apparent effect is at best modest. There is no evidence for any parallel process at work in the other experiments. What happens if we momentarily refocus to look at the control group versus “all Americans” contrasts? As anticipated by five of the six relevant F-statistics, measured differences in opinion appear relatively trivial. That is true as well with respect to child care provision. Given that we might expect the severity of the crisis in the United States in 2009 to have elicited a greater public demand for new forms of government service provision, this is contrary to expectations. In this context, it is also worth emphasizing the objective and perceived severity of the economic crisis during this time among Americans.8

* p < .01

. . . to ensure a job for everyone . . . [so the impact of the current economic crisis on [. . .] can be reduced] . . . to ensure adequate health care . . . [so the impact of the current economic crisis on [. . .] can be reduced] . . . to ensure a reasonable standard of living for the old . . . [so the impact of the current economic crisis on [. . .] can be reduced] . . . to ensure a reasonable standard of living for the unemployed . . . [so the impact of the current economic crisis on [. . .] can be reduced] . . . to ensure sufficient child care services for working parents . . . [so the impact of the current economic crisis on [. . .] can be reduced] . . . to provide paid leave for people who have to care for sick family members . . . [so the impact of the current economic crisis on [. . .] can be reduced] .59 .39 .35 .48 .46 .44

.36 .33 .50 .48 .45

All Americans

.64

Baseline

.47

.45

.50

.34

.41

.53

Middle-class Americans

t a b l e 7. 2 Experiments with crisis/class frames: [Not] government’s responsibility

.48

.45

.48

.35

.39

.56

Working-class Americans

.12

85.94

2.68

3.48

5.60

4.63

Fcrisis

1.90

.03

.40

.40

.63

3.06

Fclass



Framing Theory, Welfare Attitudes, and the U.S.

207

It’s appropriate to consider what this all means. As in any experiment, there is no guarantee against internal validity limitations. But that said, the experimental cues relating to the crisis (and also to “all/middle-class/ working-class” Americans) are neither opaque nor distant from contemporary media discourse. A different angle may thus be in order. Simply put, the pattern of crisis-related framings having little effect is consistent with the motivated reasoning scenario in which individuals are disposed to selectively process new stimuli, at least with reference to dissonance-inducing elements. If so, the economic crisis in the United States in 2009 was compatible with a degree of inertia with regard to dimensions of U.S. policy attitudes toward government responsibility. This is discussed in further detail in the conclusion to this chapter. Cross-National Perspective What additional purchase does a comparison across countries enable? Using the baseline (unframed) ESS items, I consider differences in aggregate opinion across the United States and 28 countries surveyed in Round 4 of the European Social Surveys. By virtue of its fully randomized design, the SAPA data render impractical comparisons based on scaling items and extracting factor scores; my focus is restricted to single items. Figures 7.1 and 7.2 display country-level item means, with dot plots showing how countries rank for the five items relating to perceptions of the consequences of social benefits and services provision. Keeping in mind that lower scores indicate more positive perceptions, Russia and several other East European countries tend to cluster at the more pro-welfare pole of the distribution. By contrast, public perceptions of welfare state consequences are considerably more negative in a number of Western European polities. This is particularly so with respect to the United Kingdom and France. Turning to the U.S. case, public perceptions tend toward the middle of country rankings with respect to the strain and cost items. The U.S. public’s attributions concerning poverty reduction and immigration are more negative in orientation. For the final item concerning social welfare as a stimulus to “make people lazy,” American attitudes appear relatively positive, showing less negative perceptions than in most Western European countries in the ESS. But recall that the earlier framing experiments showed a significant degree of elasticity in U.S. policy attitudes on most of these issues. As a

Make people lazy

Strain on economy Hungary Slovakia United Kingdom France Romania Portugal Czech Rep. Spain Croatia Germany Belgium Slovenia Switzerland Greece United States Sweden Poland Israel Cyprus Netherlands Denmark Latvia Turkey Norway Bulgaria Finland Estonia Ukraine Russia 0.00

0.63

0.57 0.57 0.54 0.54 0.53 0.52 0.51 0.51 0.50 0.49 0.48 0.48 0.47 0.46 0.46 0.45 0.45 0.45 0.45 0.44 0.44 0.41 0.41 0.41

0.10

0.20 0.30

0.41 0.35 0.40 0.50

0.60

0.70

0.65

United Kingdom Hungary Slovenia Czech Rep. France Poland Slovakia Norway Portugal Belgium Netherlands Romania Israel Turkey Germany Spain Croatia Bulgaria Sweden Finland Switzerland United States Denmark Cyprus Estonia Latvia Greece Ukraine Russia

0.60 0.59

0.58 0.58 0.57 0.56 0.55 0.55 0.55 0.53 0.53 0.52 0.52 0.52 0.51 0.51 0.51 0.50 0.49 0.49 0.48 0.47 0.45 0.44 0.40 0.39 0.36 0.35 0.34

0.00

0.80

0.10

0.20

0.30

0.33 0.40

0.50

0.60

0.70 0.80

Negative welfare perceptions

Negative welfare perceptions

Do not prevent poverty Hungary Latvia Bulgaria Russia United States Ukraine Slovakia Poland Croatia Slovenia Spain Romania Isreal Estonia Turkey Portugal United Kingdom Czech Rep. Greece Finland Norway Germany France Sweden Switzerland Belgium Denmark Netherlands Cyprus 0.00

0.60 0.54 0.54 0.52 0.51 0.50 0.46 0.46 0.44 0.44 0.44 0.43 0.42 0.42 0.41 0.40 0.40 0.38 0.37 0.36 0.36 0.35 0.34 0.34 0.34 0.33 0.33 0.32

0.10

0.20

0.30

0.30 0.40

0.50

0.60

0.70 0.80

Negative welfare perceptions

Figure 7.1  Negative perceptions of strain, laziness, and poverty s o u r c e s : Survey of American Policy Attitudes 2009; European Social Survey 2008.



209

Framing Theory, Welfare Attitudes, and the U.S. Encourage migration

Germany United Kingdom France Norway Switzerland Denmark Spain United States Finland Cyprus Belgium Sweden Greece Netherlands Israel Czech Rep. Slovenia Portugal Croatia Turkey Romania Poland Slovakia Hungary Russia Estonia Latvia Ukraine Bulgaria 0.00

Costs too much 0.75 0.73 0.71 0.70 0.69 0.68 0.68 0.68 0.66 0.65 0.62 0.62 0.60 0.59 0.56 0.56 0.51 0.51 0.50 0.50

0.47 0.42 0.40 0.39 0.37 0.37 0.32 0.31

0.10

0.20

0.25 0.30 0.40

0.50

0.60

0.70

0.80

Hungary France Belgium United Kingdom Spain Croatia Portugal Germany Netherlands Turkey Israel Romania Poland Slovakia Czech Rep. Sweden United States Finland Norway Slovenia Latvia Cyprus Switzerland Estonia Greece Ukraine Denmark Bulgaria Russia 0.00

0.72 0.63 0.62 0.58 0.57 0.57 0.56 0.55 0.54 0.54 0.54 0.53 0.53 0.53 0.52 0.52 0.51 0.51 0.51 0.50 0.50 0.49 0.49 0.49 0.47 0.46 0.45 0.44

0.10

Negative welfare perceptions

0.20

0.30

0.38 0.40 0.50

0.60

0.70

0.80

Negative welfare perceptions

Figure 7.2  Negative perceptions of migration and costs s o u r c e s : Survey of American Policy Attitudes 2009; European Social Survey 2008.

result, priming immigrants as a symbolic target mobilizes more negative attitudes for the strain item. This leads to a pattern of aggregate opinion that would now place the United States as the second most negative country on this issue (behind only Hungary; see Figure 7.1). A second, counterfactual simulation, this time using the immigration item and the “blacks” frame, experimentally resituates U.S. opinion more firmly at the center of the ESS distribution of countries. Results based on the second set of government responsibility items again provide an instructive contrast. As summarized in Figures 7.3 and 7.4, the United States emerges in dramatic fashion as consistently the most antistate of countries with respect to policy preferences. The magnitude of these average item scores merits comment. For four of these six items, the gap between the U.S. score and the next closest score is between two and four times larger than the parallel gap between the second and third most antigovernment countries. Using the SAPA 2009 data, the U.S. public appears unusually low in support for government responsibility for welfare

Health care

Jobs United States Switzerland Netherlands Denmark France United Kingdom Norway Sweden Czech Rep. Belgium Germany Slovakia Cyprus Estonia Finland Slovenia Poland Portugal Romania Turkey Bulgaria Spain Israel Croatia Russia Hungary Greece Latvia Ukraine 0.00

United States Switzerland Romania France Belgium Turkey Netherlands Czech Rep. Slovakia Germany Slovenia Estonia Sweden United Kingdom Finland Cyprus Croatia Portugal Denmark Poland Spain Norway Hungary Bulgaria Russia Israel Ukraine Greece Latvia

0.59 0.52 0.47 0.45 0.41 0.41 0.40 0.40 0.38 0.38 0.37 0.35 0.35 0.34 0.32 0.30 0.28 0.27 0.27 0.26 0.26 0.23 0.22 0.21 0.21 0.19 0.18 0.15 0 .12 0.10 0.20 0.30

0.40

0.50

0.60

0.70

0.00

0.80

0.39 0.23 0.21 0.20 0.20 0.18 0.18 0.16 0.16 0.16 0.14 0.13 0.13 0.13 0.13 0.12 0.12 0.12 0.11 0.11 0.10 0.10 0.10 0.09 0.09 0.09 0.07 0.08 0.07 0.10 0.20

0.30

0.40

0.50

0.60

0.70

Negative welfare perceptions

Negative welfare perceptions

Standard of living for the old United States Switzerland Germany Netherlands Belgium France Romania Czech Rep. Turkey Denmark Slovenia Slovakia Finland Sweden United Kingdom Poland Norway Estonia Croatia Hungary Spain Portugal Greece Cyprus Israel Bulgaria Russia Latvia Ukraine 0.00

0.34 0.28 0.24 0.23 0.21 0.21 0.20 0.18 0.18 0.17 0.17 0.16 0.16 0.15 0.15 0.13 0.13 0.13 0.13 0.12 0.12 0.11 0.11 0.11 0.10 0.08 0.08 0.07 0.06 0.10 0.20

0.30

0.40 0.50

0.60

0.70 0.80

Negative welfare perceptions

Figure 7.3   Not government’s responsibility: Jobs, health care, and the old s o u r c e s : Survey of American Policy Attitudes 2009; European Social Survey 2008.

0.80

Standard of living for the unemployed United States Slovakia United Kingdom Belgium France Switzerland Czech Rep. Netherlands Poland Germany Denmark Slovenia Hungary Russia Bulgaria Estonia Portugal Norway Sweden Finland Israel Cyprus Romania Spain Croatia Turkey Ukraine Greece Latvia 0.00

Child care services United States Netherlands Switzerland United Kingdom Slovakia France Poland Belgium Czech Rep. Turkey Romania Croatia Sweden Norway Germany Denmark Russia Slovenia Finland Estonia Portugal Spain Ukraine Hungary Israel Bulgaria Cyprus Latvia Greece

0.49 0.42 0.40 0.39 0.39 0.37 0.37 0.37 0.36 0.35 0.34 0.33 0.32 0.29 0.28 0.27 0.27 0.27 0.26 0.25 0.25 0.24 0.22 0.23 0.22 0.21 0.17 0.16 0.15 0.10 0.20 0.30

0.40

0.50

0.60

0.70

0.00

0.80

Negative welfare perceptions

0.47 0.37 0.35 0.31 0.29 0.29 0.27 0.27 0.26 0.26 0.23 0.22 0.21 0.20 0.20 0.19 0.19 0.18 0.18 0.18 0.17 0.17 0.17 0.16 0.16 0.15 0.15 0.12 0.11 0.10 0.20

0.30

0.40

0.50

0.60

0.70

0.80

Negative welfare perceptions

Paid family leave United States Switzerland Netherlands United Kingdom France Belgium Slovakia Romania Germany Finland Czech Rep. Sweden Poland Slovenia Estonia Croatia Israel Greece Portugal Norway Denmark Turkey Spain Hungary Cyprus Russia Bulgaria Ukraine Latvia 0.00

0.46 0.39 0.33 0.28 0.28 0.27 0.27 0.26 0.26 0.25 0.25 0.21 0.23 0.20 0.20 0.20 0.20 0.19 0.18 0.18 0.18 0.18 0.18 0.17 0.17 0.17 0.15 0.15

0.10

0.14 0.20 0.30

0.40

0.50

0.60

0.70 0.80

Negative welfare perceptions

Figure 7.4  Not government’s responsibility: Unemployed, child care, and family leave s o u r c e s : Survey of American Policy Attitudes 2009; European Social Survey 2008.

212

Brooks

in comparison to a broad sample of European polities (and also Israel and Turkey). The size of this transatlantic gap in preferences is striking in the contemporary historical era, where convergence pressures across nations continue to prompt popular and scholarly debate. The Magnitude of Framing Impacts So far, analysis of the SAPA data tells us about the statistical significance of framing effects and the patterning of aggregate U.S. attitudes in crossnational perspective. But the evidence of framing effects for the welfare perceptions items calls out for a bit more attention. Now it is useful to consider the magnitude of impacts. Questions of magnitude are common in (nonexperimental) regression analysis. In the context of experimental results, it is likewise good to know not just that framing effects exist but also just how large (or small) they are in substantive terms. We can get a sense of this by comparing with respect to more established influences on policy attitudes. To this end, I incorporate the survey experiments into a regression model of each policy-attitude item. In this setup, survey-embedded treatment effects are covariates in the regression; these operate as main effects alongside a standard roster of more established factors behind policy attitude formation.9 Prior to analysis, it is premature to rule out of the possibility of interaction effects between issue frame and nonexperimental covariates. I consider several such candidates for interaction with respect to framing: economic evaluations, education, race, and ideology. In subsequent tests, there is evidence of significant interactions in 2 out of 20 cases, and these are discussed in the course of the results presented below. What can we learn about the magnitude of framing influence? Let’s use the established factor of economic performance evaluations as a benchmark. As displayed in Web Appendix Figure A7.1, results are range-standardized, so we can directly compare the respective impacts across the framing and economic evaluation covariates.10 The graphical displays reveal a thematic result: All else being equal, framing exerts a larger impact on welfare attributions concerning strain, laziness, and immigration in comparison to economic evaluations. When we consider welfare perceptions of strain, laziness, and immigration, framing effects are not just statistically significant; they are sizable in their own right. Just as shifts in evaluations of how the economy is doing influence social welfare perceptions, changes in the framing of welfare



Framing Theory, Welfare Attitudes, and the U.S.

213

beneficiaries also matter. In the strain results, for instance, we can again see how changing the framing of welfare beneficiaries from blacks to immigrants leads to dramatically more negative perceptions of social welfare consequences. Recalling the range-standardization of these estimates, the framing effect is considerably larger than the equivalent effect of a shift in economic perceptions. In Web Appendix Figure A7.2, we can consider parallel estimates for the two domains of welfare beliefs characterized by interactions involving framing. In the case of poverty attributions, framing effects are heterogeneous with respect to education. This means that just how much framing matters depends on the education level of a survey respondent. Looking, for instance, at respondents possessing the lowest level of education, framing matters considerably: Experimental cuing of “blacks” as a target group enhances negative perceptions of welfare state effects on poverty. But when we look at individuals at the highest level of education, the impact of framing is much smaller and switches direction; this means that cuing of the “blacks” target group frame now reduces negative perceptions of welfare state effects on poverty. In either case, however, the absolute value of framing effects is larger than the parallel estimates for economic evaluations. In general, then, framing may matter more than economic evaluations when we consider individuals’ perceptions of whether the welfare state reduces poverty. Looking at Web Appendix Figure A7.2’s second set of results, the interactions at hand involves framing and economic evaluations. At the lowest level of economic evaluations, framing impacts are substantial, but framing effects are quite negligible at the highest level of economic evaluations. Framing matters more in this domain when respondents view economic performance in positive terms (as measured by the response category, “better than a year ago”). Staying with the interaction effect results for welfare costs, economic evaluations have nontrivial impacts on attitude formation. But these vary in direction, depending on whether we are considering the “middle” class frame or not.11 This is an interesting result, and it indicates that reasoning about the cost dimensions of welfare may elicit the greatest overall parity in the impacts of the two factors under consideration. But as before, the main thematic result is that for the welfare attribution items as a whole, framing has quite sizeable effects, operating alongside more established factors shaping policy attitudes.

214

Brooks

c on c l usion In what ways can framing theory contribute to comparative welfare state scholarship? Scholarship on framing sheds light on some candidate sources of pressure toward opinion change. Frames and framing processes are intriguing because they represent elements of social and historical contexts that link political organizations and other sources of policy-related communication to individual-level processes of policy-attitude formation (Druckman and Jacobs, 2006; see also Zaller, 1992). It should be emphasized, however, that these linkages are rife with contingency. This is not only a result of inattention or information-processing limits on the part of individuals and audiences. Further, if opinions are sufficiently well grounded in emotion-laden beliefs and other biases so as to dispose individuals to resist processing new environmental stimuli, this can lead to opinion inertia. Determining whether inertia scenarios or framinginduced malleability is applicable requires sustained empirical research. A benefit of the experimental survey design is to yield evidence that is far less vulnerable to selection scenarios in comparison to the standard social survey. Starting with the target group experiments, this chapter’s results suggest three novel findings with reference to the context of the United States. First, negative pressures toward changes in welfare perceptions and attributions relate not only to African Americans but also to immigrants as a key point of symbolic reference.12 Second, there is evidence of framing impacts associated with “middle class” cues, but these effects are not generally as large in magnitude. And four of these five experiments suggest that Americans may already be thinking about immigration when reflecting on welfare state consequences. This is evident in cases where the baseline item and “immigrants” condition return statistically indistinguishable responses. Together, these results most clearly exemplify the operation of processes anticipated in the growing literature on framing effects (e.g., Gamson and Modigliani, 1987; Kinder, 1998; Chong and Druckman, 2007b). Given the experiments’ design target group cues, the results suggest that frames may operate through the mobilization of preexisting biases. When framing matters for social policy-related attitudes in the U.S. context, it may tend to do so by getting individuals to rethink welfare issues using ideas to which they already are likely to subscribe.



Framing Theory, Welfare Attitudes, and the U.S.

215

What about the second set of experiments with preferences concerning government responsibility for different arenas of social provision? Here, the results follow a very different pattern, showing virtually no framing effects. This is notable, even remarkable, given the actual and perceived severity of the economic crisis and recession in the United States. If crisis frames represented abstract or hypothetical ideas, ones difficult to envision, we might reasonably blame instrumentation for a lack of evidence concerning framing differences. But the case of the economic crisis instead suggests that the explanation lies elsewhere. Indeed, we know that economic crisis perceptions, as measured by the National Election Studies sociotropic performance item (and replicated in the SAPA data), are a reality in the contemporary United States. The preceding results suggest, then, an intriguing, provisional conclusion: Through 2009, the crisis may not have done much to reorient U.S. policy preferences concerning government as the central provider of social services and benefits. One contender for the explanatory mechanism at work is motivated reasoning. Certainly Americans acknowledged the reality of a far-reaching economic crisis. By the same token, however, they did not appear willing to acknowledge (or incorporate) it in such a way as to reorient preferences regarding the role of government in social policy (see also Brooks and Manza, 2011). In this context, the cross-national comparisons provide perspective. Recall that these comparisons show the United States as consistently ranking as the most antiwelfare nation when it comes to public preferences for government responsibility. If indeed these components of U.S. policy preferences have resisted egalitarian responses to the crisis, the transatlantic gap in these preferences may be large and potentially robust. Notably, however, the same is not true of the welfare perceptions measured by the other ESS items is notable, and this calls out for further research. A key payoff of framing scholarship is, then, to provide evidence of opinion malleability versus inertia scenarios. One time period’s worth of survey data and experiments is, of course, insufficient to reach anything like definitive conclusions on these issues. But they raise fruitful new questions for research that have the promise to enhance scholarly understanding of processes of change relevant to both welfare states and social policy attitudes. The current results, alongside the larger challenge of understanding change versus stability dynamics in opinion formation, make framing

216

Brooks

theory an attractive addition to the toolkit of comparative welfare state scholarship. a p p e n di x : at t i t u de s

t h e 2 0 0 9 s u rv e y o f a m e r i c a n p o l i c y ( s a pa )

The Survey of American Policy Attitudes was developed in cooperation with the Center for Survey Research (CSR) at Indiana University. Data were collected using computer-assisted telephone interviewing methods (CATI), and numbers were randomly generated using the Genesys list-assisted method. This method allows for unpublished numbers and new listings to be sampled. After selecting a random sample of telephone numbers, numbers were matched to a database of business and nonworking numbers, and all matches were subsequently purged from the sample. The sample was nationwide, and at each residential number a respondent from all household members age 18 or older was selected. The data collection staff included 11 supervisors and 54 interviewers. All interviewers received at least 20 hours of training in interviewing techniques before production interviewing. Interviewers were instructed to read questions and response categories at a pace slower than conversation and to use neutral probes and feedback phrases. Both audio and visual monitoring were regularly conducted by the telephone survey supervisors, using the CSR facilities. Monitoring was conducted randomly, with each interviewer being monitored at least once during each 3‑hour shift. All cases with confirmed valid telephone numbers were called up to 15 times, unless the respondent refused or there was insufficient time before the end of the study. Cases with unknown validity (persistent no answers or answering devices) were called a minimum of 8 times, with calls made during the morning, afternoon, evening, and weekend. Interviewers attempted to convert each “refusal” at least twice. When possible, a conversion attempt was made at the first instance of refusal and a second attempt was usually made after a few days. Final Disposition Summary Web Appendix Table A7.1 classifies every case according to its final disposition. These dispositions are based on the guidelines for Final Disposition Codes for RDD Surveys established by the American Association for Public



Framing Theory, Welfare Attitudes, and the U.S.

217

Opinion Research (AAPOR) Standard Definitions for Final Dispositions of Case Codes, 2004. Using AAPOR’s RR3 (Response Rate 3 formula), we calculated the response rate as 0.1527 (or 15.3 percent). This response rate is in line with ongoing declines in participation found in U.S. telephone surveys. Research on the impact of these trends on data quality suggests that declining response rates are not by themselves tantamount to response bias or that probability sampling should necessarily be abandoned in favor of such alternatives as quota sampling (Groves, 2006). A relevant result of experiments into effects on data quality (Keeter et al., 2006) is that low rates are likely a source of bias only insofar as the probability of survey participation is correlated with one or more measured variables of interest. In SAPA, the majority of nonresponses (unknown eligibility or not interviewed) were a product of conditions such as no answers on the line and initial refusals to participate before the start of the survey itself (or any item). no t e s 1.  Direct correspondence to Clem Brooks, Department of Sociology, 1020 E. Kirkwood Ave., Indiana University, Bloomington, IN, 47405-7103, USA; e-mail: [email protected]. This research was supported by grants from the National Science Foundation (SES·0830917) and the Russell Sage Foundation (RSF-83-08-04). 2.  Three such conditions are that policy preferences are real and “accessible” in the minds of citizens, characterized by nonzero levels of salience, and can be mapped onto the distribution of political parties and organizations in a given context (see Aldrich et al. 2006; Brooks 2006; Soroka and Wlezien, 2010). The latter are not always measured in empirical scholarship, and evidence of an association between opinion and policy (in the face of relevance controls) is often seen as providing prima facie evidence. 3.  Framing is thus distinct from persuasion, where explicitly persuasive communication frequently provides additional, confounding information to audiences regarding a speaker’s or source’s strategic intentions. 4.  As discussed there, the response rate to SAPA is in line with well-known trends involving declining U.S. survey participation. Notwithstanding the rising cost of surveys that this generates, there is little evidence that declining response rates by themselves have eroded survey data quality (see Keeter et al., 2006). 5.  Following convention, the 1–5 response categories are recoded to a 0–1 range for all items. 6.  Using, for instance, Tukey’s HSD and Tamhane’s T2 tests, we obtain pvalues for the baseline versus “blacks” condition of .87 and .97, respectively; for the baseline versus “middle class” conditions, p-values are >.99.

218

Brooks

7.  This pairwise contrast leads to p-values of approximately .02 for both HSD and T2 tests. 8.  In 2004, 45 percent of respondents in the National Election Survey viewed the national economy as “worse than a year ago.” In the 2008 NES survey, the corresponding figure was 91 percent, and in the 2009 SAPA survey, it is 72 percent. 9.  As summarized in Web Appendix Tables A7.2 and A7.3, covariates for established factors include economic performance evaluations, age, education, gender, race, class, religious participation, and ideological identification. Inclusion of the ideology covariate represents a methodologically conservative decision, though dropping it from the model generally yields quite comparable coefficient estimates. 10.  For this standardization, I take the difference between the smallest and largest coefficients for the framing covariates; calculations for economic evaluation are the predicted differences in effects vis à vis the top versus bottom response categories. 11.  Note that this interaction between the framing and economic evaluation covariates applies to only a single issue frame. As displayed in Web Appendix Figure A7.1, estimates for economic evaluations are thus independent of framing in three of the four experimental item-assignment conditions. 12.  These results fit well with research on ethnic cleavages in European polities (Alesina and Glaesar, 2004; Nannestad, 2007) suggesting that the widely studied U.S. race/welfare link (Gilens, 1999; Quadagno, 1994) will likely benefit from closer engagement with additional consideration of ethnicity and immigrant status.

r efer ences Aldrich, J., C. Gelpi, P. Feaver, R. Reifler, and K. Sharp. 2006. “Foreign Policy and the Electoral Connection.” Annual Review of Political Science 9: 477–502. Alesina, A., and E. Glaeser. 2004. Fighting Poverty in the US and Europe: A World of Difference. New York: Oxford University Press. Brooks, C. 2006. “Voters, Satisficing, and Public Policy: New Directions in the Study of Electoral Politics.” Annual Review of Sociology 32: 191–211. Brooks, C., and J. Manza. 2007. Why Welfare States Persist: the Importance of Public Opinion in Democracies. Chicago: University of Chicago Press. ———. 2011. Unpublished manuscript. How Did the Crisis Matter? Bloomington: Indiana University. Burstein, P. 1998. “Bringing the Public Back In: Should Sociologists Consider the Impact of Public Opinion on Public Policy?” Social Forces 77: 27–62. Canes-Wrone, B. 2006. Who Leads Whom?: Presidents, Policy, and the Public. Chicago: University of Chicago Press.



Framing Theory, Welfare Attitudes, and the U.S.

219

Chong, D., and J. Druckman. 2007a. “Framing Public Opinion in Competitive Democracies.” American Political Science Review 101: 637–55. ———. 2007b. “Framing Theory.” Annual Review of Political Science 10: 103–26. Cutright, P. 1965. “Political Structure, Economic Development, and National Social Security Programs.” American Journal of Sociology 70: 537–50. David, P. 1985. “Clio and the Economics of QWERTY.” American Economic Review 75: 332–37. Ditto, P., and D. Lopez. 1992. “Motivated Skepticism: Use of Differential Decision Criteria for Preferred and Nonpreferred Conclusions.” Journal of Personality and Social Psychology 63: 568–84. Druckman, J., and L. Jacobs. 2006. “Lumpers and Splitters: The Public Opinion Information that Politicians Collect and Use.” Public Opinion Quarterly 70: 453–76. Dyck, J., and L. Hussey. 2008. “The End of Welfare As We Know It? Durable Attitudes in Changing Information Environment.” Public Opinion Quarterly 72: 589–618. Edlund, J., and S. Svallfors. 2011. “Cohort, Class, and Attitudes to Redistribution in Two Liberal Welfare States: Britain and the US, 1996–2006.” In Generational Politics and Policies: Comparative Studies of Ageing Post-Industrial Democracies, edited by G. Achim and V. Pieter. New York: Routledge. Erikson, R., M. Mackuen, and J. Stimson. 2002. The Macro Polity. Cambridge, UK, and New York: Cambridge University Press. Esping-Andersen, G. 1990. The Three Worlds of Welfare Capitalism. Cambridge, UK: Polity Press. Festinger, L. 1957. A Theory of Cognitive Dissonance. Stanford: Stanford University Press. Gamson, W., and A. Modigliani. 1987. “The Changing Culture of Affirmative Action.” Research in Political Sociology 3: 137–63. Gilens, M. 1999. Why Americans Hate Welfare: Race, Media, and the Politics of Antipoverty Policy. Chicago: University of Chicago Press. Gilovich, T., D. Griffin, and D. Kahneman. 2002. Heuristics and Biases: The Psychology of Intuitive Judgement. Cambridge, UK, and New York: Cambridge University Press. Groves, R. 2006. “Nonresponse Rates and Nonresponse Bias in Household Surveys.” Public Opinion Quarterly 70: 646–75. Huber, E., C. Ragin, and J. D. Stephens. 1993. “Social Democracy, Christian Democracy, Constitutional Structure, and the Welfare State.” American Journal of Sociology 99: 711–49. Huber, E., and J. D. Stephens. 2001. Development and Crisis of the Welfare State: Parties and Policies in Global Markets. Chicago: University of Chicago Press. Iversen, T., and T. Cusack. 2000. “The Causes of Welfare State Expansion: Deindustrialization or Globalization?” World Politics 52: 313–49.

220

Brooks

Iyengar, S. 1990. “The Accessibility Bias in Politics: Television News and Public Opinion.” International Journal of Public Opinion Research 2: 1–15. Jackman, M., and R. Jackman. 1985. Class Awareness in the United States. Berkeley: University of California Press. Jacobs, L., and R. Shapiro. 2000. Politicians Don’t Pander: Political Manipulation and the Loss of Democratic Responsiveness. Chicago: University of Chicago Press. Jost, J., M. Banaji, and B. Nosek. 2004. “A Decade of System Justification Theory: Accumulated Evidence of Conscious and Unconscious Bolstering of the Status Quo.” Political Psychology 25: 881–919. Jost, J., C. Federico, and J. Napier. 2009. “Political Ideology: Its Structure, Functions, and Elective Affinities.” Annual Review of Psychology 60: 307–37. Keeter, S., C. Kennedy, M. Dimock, J. Best, and P. Craighill. 2006. “Gauging the Impact of Growing Nonresponse on Estimates from a National RDD Telephone Survey.” Public Opinion Quarterly 70: 759–79. Kinder, D. 1998. “Communication and Opinion.” Annual Review of Political Science 1: 167–97. Korpi, W. 1978. “Social Democracy in Welfare Capitalism-Structural Erosion, Welfare Backlash, and Incorporation?” Acta Sociologica, Supplement, 97–111. Kunda, Z. 1990. “The Case for Motivated Reasoning.” Psychological Bulletin 108: 480–98. Lipset, S. M. 1996. American Exceptionalism: A Double-Edged Sword. New York and London: Norton. Nannestad, P. 2007. “Immigration and Welfare States: A Survey of 15 Years of Research.” European Journal of Political Economy 23: 512–32. Oorschot, W. V. 2008. “Who Should Get What, And Why?” In Welfare States: Construction, Reconstruction, edited by S. Leibfried and S. Mau, 353–68. London: Edward Elgar. Page, B., and R. Shapiro. 1983. “Effects of Public Opinion on Policy.” American Political Science Review 77: 175–90. Pierson, P. 1994. Dismantling the Welfare State? Reagan, Thatcher, and the Politics of Retrenchment. Cambridge, UK, and New York: Cambridge University Press. ———. 1996. “The New Politics of the Welfare State.” World Politics 48: 143–79. ———. 2001. “Coping with Permanent Austerity: Welfare State Restructuring in Affluent Democracies.” In The New Politics of the Welfare State, edited by P. Pierson, 428–31. Oxford: Oxford University Press. Quadagno, J. 1994. The Color of Welfare. How Racism Undermined the War on Poverty. New York: Oxford University Press. Rothstein, B. 1998. Just Institutions Matter: The Moral and Political Logic of the Universal Welfare State. Cambridge, UK, and New York: Cambridge University Press.



Framing Theory, Welfare Attitudes, and the U.S.

221

Rothstein, B., and S. Steinmo. 2002. “Restructuring Politics: Institutional Analysis and the Challenges of Modern Welfare States.” In Restructuring the Welfare State: Political Institutions and Policy Change, edited by B. Rothstein and S. Steinmo, 1–19. New York: Palgrave Macmillan. Schuman, H., C. Steeh, L. Bobo, and M. Krysan. 1997. Racial Attitudes in America: Trends and Interpretations. Cambridge, MA: Harvard University Press. Sears, D. O., J. Sidanius, and L. Bobo. 2000. Racialized Politics: the Debate About Racism in America. Chicago and London: University of Chicago Press. Skocpol, T. 1992. Protecting Soldiers and Mothers: The Political Origins of Social Policy in the United States. Cambridge, MA, and London: Belknap Press of Harvard University Press. Sniderman, P., and D. Grob. 1996. “Innovations in Experimental Design in Attitude Surveys.” Annual Review of Sociology 22: 377–99. Sniderman, P., L. Hagendoorn, and M. Prior. 2004. “Predisposing Factors and Situational Triggers: Exclusionary Reactions to Immigrant Minorities.” American Political Science Review 98: 35–49. Soroka, S., and C. Wlezien. 2010. Degrees of Democracy: Politics, Public Opinion, and Policy. Cambridge, UK, and New York: Cambridge University Press. Soss, J., and S. Schram. 2007. “A Public Transformed? Welfare Reform as Policy Feedback.” American Political Science Review 101: 111–27. Steensland, B. 2006. “Cultural Categories and the American Welfare State: The Case of Guaranteed Income Policy.” American Journal of Sociology 111: 1273–1326. Stimson, J. A. 2004. Tides of Consent: How Public Opinion Shapes American Politics. Cambridge, UK, and New York: Cambridge University Press. Svallfors, S. 2006. The Moral Economy of Class: Class and Attitudes in Comparative Perspective. Stanford: Stanford University Press. Wilensky, H., and C. Lebeaux. 1965. Industrial Society and Social Welfare: The Impact of Industrialization on the Supply and Organization of Social Welfare Services in the United States. New York: Free Press. Wlezien, C. 1995. “The Public as Thermostat: Dynamics of Preferences for Spending.” American Journal of Political Science 39: 981–1000. ———. 2004. “Patterns of Representation: Dynamics of Public Preferences and Policy.” Journal of Politics 66: 1–24. Zaller, J. 1992. The Nature and Origins of Mass Opinion. Cambridge, UK, and New York: Cambridge University Press.

chapter eight

Welfare Attitudes in Context Stefan Svallfors

The conclusion of this book provides a timely opportunity to highlight a number of broader themes and findings that have emerged through the preceding chapters. As in any work of this scope, it will also be of value to revisit some of the critical tensions and unresolved issues that may well form the basis for future analyses. The findings in each chapter in various ways confirm, question, and extend previous findings in the field. Among the confirmations we find that the welfare state is indeed quite popular among broad sections of the population. The citizens of Europe ask for a wide-ranging public responsibility for their welfare, in line with previous findings in this regard (see, for example, Svallfors, 1997; 2003; Bonoli, 2000; Andreß and Heien, 2001; Aalberg, 2003; Brooks and Manza, 2007; Mau and Veghte, 2007; summaries in Kumlin, 2007; Svallfors, 2010). Even though the level of support varies among groups and across countries, nowhere do we find a widespread resistance against state intervention and redistribution. The contrast to attitudes in the United States, brought out in Chapter 7, is instructive in this regard. There, the support for government intervention is weaker than in any European country. Recent findings, suggesting a transatlantic convergence in welfare attitudes (Edlund, 2009; Edlund and Svallfors, 2011), clearly do not indicate that the United States is presently at some European average level when it comes to support for the welfare state. Findings also confirm that support for extensive welfare policies is stronger in the Nordic countries than in continental Europe and liberal Britain (see Svallfors, 1997; 2003; 2004; Andreß and Heien, 2001; Blekesaune and Quadagno, 2003; Jæger, 2006; Brooks and Manza, 2007). But 222



Welfare Attitudes in Context

223

they also show that support for a wide-ranging public responsibility is even stronger in Southern and Eastern Europe. I will return later to the question of the extension of the analyses beyond the traditional “three worlds of welfare.” The analyses have also confirmed the continuing, although varying, impact of “class” and “class-related” factors (such as income and education and various risk-related factors) on the structuring of welfare attitudes (i.e., Cusack et al., 2006; Edlund and Svallfors, 2011; Kumlin and Svallfors, 2007; Svallfors, 2004; 2006). As shown in several of the chapters, class differences in attitudes toward the welfare state are substantial in many countries, and patterns are fairly consistent across countries. Still, the magnitude of these differences varies substantially among countries, although in a more complex pattern than previous research suggested (see Chapter 6 and Kumlin and Svallfors, 2007). The analyses also confirmed that support for the welfare state has a similar—but far from identical—value base across Europe. Although Chapter 4 showed that there are some variations in how different normative bases link to welfare attitudes, there are still basic similarities in how values such as egalitarianism and universalism translate into support for welfare policies. On a more general note, one should be clear that the commonalities across Europe are just as important as the differences. The social sciences are sometimes obsessed with differences and change, to the expense of noting continuity and similarities (Lieberson, 1985; Erikson and Goldthorpe, 1992; Goldthorpe and Marshall, 1992). But the strong support for an extensive welfare state, the similarities in cleavage structures, and the similar value bases for the welfare state across Europe should be noted before we proceed to take important differences into account. t h e m at ic f i n di ng s However, our main emphases in this volume have been on relations rather than on levels of attitudes. In this respect, our analyses have included a number of additions to the current state of the art. Specific findings were summarized in Chapter 1 and elaborated in each empirical chapter, so there is no need to repeat those here. But a number of broader substantive and methodological themes have emerged and should be highlighted before closing.

224

Svallfors

First, our analyses have considerably broadened and specified the dependent variable(s) in the analysis of welfare attitudes. Previous research has to a large extent focused on what might be termed general welfare state support, regardless of what method was chosen to measure it (as discussed in Chapter 1). Sometimes this question about the preferred range of government responsibilities has been complemented with other questions about views on the intensity of government involvement, most often measured by studying how much tax money people say they want spent on various government activities (Roller, 1995). Only a scattered handful of studies extend their purview of investigation outside these foci. In our analyses here, we go beyond a focus on only welfare attitudes in this restricted sense; we also, crucially, take into account welfare state evaluations, welfare chauvinism, and age- and class-specific policy areas. This widening of focus brings some interesting insights. In Chapter 2, it is found that although support for state intervention is strong in Europe, satisfaction with how well the welfare state performs its tasks is much lower and highly varied across Europe. Chapter 3 notes that groups particularly exposed to risks are clearly less satisfied than others with the way the welfare state works in different respects. These findings point to the simple fact that if we want to measure the legitimacy of the welfare state, it is hardly enough to focus on principled support for state intervention. We must also ask to what extent existing policies are seen as adequate to the task of ameliorating adverse conditions. Here it seems the European welfare states, and in particular those of Eastern and Southern Europe, perform much less well than their citizens would ask for. If there are legitimacy problems in the European welfare states, they rather refer to these kinds of considerations and evaluations than to any principled resistance against state intervention. A second theme where our analyses have contributed to existing research relates to ethnic and cultural diversity as a new challenge to nationally organized welfare policies. In the wake of Alesina and Glaeser’s (2004) influential study, national and comparative studies have analyzed to what extent (real or perceived) ethnic diversity affects welfare attitudes (see the discussion in Chapter 5 and in Banting and Kymlicka (2006). Results in this regard are inconclusive. Some studies find little support for the notion that ethnic diversity undermines support for the welfare state (Taylor-Gooby, 2005; Banting et al., 2006; Finseraas, 2008), but others do (Eger, 2010). At the very least, results suggest that findings from the United States on the



Welfare Attitudes in Context

225

concentration of African Americans cannot be extrapolated to Europe’s immigrant populations without taking the very different histories and contexts of the two continents into account (Castles and Schierup, 2010). But the analyses in this book add several important observations in this regard. One concerns the conditional, and highly varied, support for including immigrants into the welfare state. As shown in Chapter 5, the support for including migrants into the welfare state under the same conditions as the autochthonous population varies dramatically across Europe. But nowhere do we find a majority prepared to grant newcomers access to benefits and services immediately on arrival. On the other hand, pure welfare chauvinism as expressed in the view that immigrants should never receive the same kind of access as those born in the receiving country is clearly a small minority position. Instead, mainstream views about the inclusion of migrants seem to rest on a kind of reciprocal thinking: Once “they” have contributed, they can be granted rights. Hence, these views fit well into the “moral economy” framework in which this book is embedded. Chapter 4 adds some interesting observations in this regard, too. Here it is shown that while ethnocentrism decreases support for welfare state intervention, authoritarianism actually increases it. So skepticism about ethnic diversity and authoritarian dispositions, which are sometimes considered parts of the same attitudinal syndrome (Lipset, 1959), have in fact the opposite effects as value bases for welfare policies. It should be pointed out that in the case of ethnocentrism, the effects are rather weak, and in the case of authoritarianism, the effects are varied across countries, but it is nevertheless interesting to note the opposite effects. Different kinds of support for diversity seem to have quite different implications for welfare state support. In policy terms, the conditional support for including newcomers would seem to point to the importance of access to the labor market. If visible minorities are included in the labor market, they contribute to financing the welfare state and are less easily depicted as “welfare scroungers” by antiimmigrant parties. In a Europe with persistent high unemployment, such labor market inclusion is a major challenge to which current policies and politics unfortunately seem ill-adapted. In regard to existing research, our findings imply that the focus on ethnic fragmentation per se may be slightly misplaced. It is rather the policy response and degree of difficulties in the labor market that matter for

226

Svallfors

a­ ttitudes toward the inclusion of newcomers in the welfare state. The hard core of outright welfare chauvinists is a fairly small minority in most places in Europe, while the majority is prepared to grant access to newcomers provided that they are contributing. A third theme to which this book has added input concerns the explanatory mechanisms behind welfare attitudes and between categorical locations and such attitudes. As pointed out in Chapter 6, much research in the field has been based on an explicit or implicit assumption that selfinterest is the mechanism that links social location with attitudes. When significant associations are found between categorical location and welfare attitudes, such associations have often been explained by the (imputed) selfinterests of actors located in such categories. This is unfortunate, since this assumption runs counter to much of what is now argued in widely different fields—namely, that self-interest has a rather limited role in the formation of attitudes, beliefs, values, and actions. So a broader conception of explanatory mechanisms, and new empirical indicators of such mechanisms, is clearly needed, including values, beliefs, and norms of reciprocity. The analyses in this book have addressed these issues in several ways. First, it is important to be clear that the analyses have in no way discarded the notion of the importance of self-interest altogether. Several of the chapters focus on risk as a key concept in explaining patterns of attitudes. Chapter 3 argues that groups that are more exposed to risks, and more dependent on the welfare state than others, are likely to support welfare state intervention out of purely self-regarding motives. To reduce their own risk of being exposed to income losses due to labor market difficulties or welfare state cutbacks, these groups support a far-ranging public safety net. Chapter 5 also argues that workers’ motives for excluding immigrants from welfare benefits are partially grounded in their materially based fear of low-wage competition from the newcomers. Again, this is a clear case of how material interests are translated into welfare attitudes. Also in Chapter 6, it transpired that risk exposure is one important link between class position and welfare attitudes. A substantial part of the class–attitude link could be explained by the sustenance risks that people in different class positions were exposed to. However, self-interest is in no way close to the full story behind the patterning of welfare attitudes. To begin, even risk exposure has a quite complex causality when it comes to affecting welfare attitudes. As shown in



Welfare Attitudes in Context

227

Chapter 4, it does increase egalitarianism, which is a link toward asking for a more extensive welfare commitment from the state. But it also increases ethnocentrism, which is a path toward weaker support for extensive welfare policies. So both egalitarian and exclusionary attitudes are promoted by risk exposure. More broadly, it is clearly the case that beliefs, values, and notions of deservingness all affect attitudes toward redistribution and welfare policies. Their impact may vary across groups and among countries, but it is nevertheless clear that attitudes toward the welfare state entail notions of social order and legitimacy in the widest sense (compare the arguments and findings in Chapter 4). It seems that future research would have much to gain by moving beyond implicit assumptions of self-interest as the sole driver of welfare attitudes. However, here, research on the matter encounters a difficult problem. We run the risk of entering endogeneity problems—that is, the risk that what we are trying to explain (variations in welfare attitudes) is so closely and almost by definition linked to some of the proposed explanatory mechanisms that explanations risk becoming empty and self-evident. All of the chapters in this book have tried to steer clear of this conundrum, but to what extent we have succeeded remains an open question. This seems to be an important unresolved issue for this whole field of research. A fourth theme to which our analyses have contributed concerns the impact of macro-variables on micro-level processes and outcomes. Here, we should first note the perhaps self-evident finding that the share of variation in welfare attitudes that takes place at the country level is a very minor fraction of the total variation. Even if this is not at all surprising in itself, it serves as a reminder that any notion of coherent national cultures that would uniformly mold the welfare attitudes of their citizens is widely overblown. Instead, a variety of sources for attitudinal variation at the individual level take place within and across country contexts. What has transpired most clearly in the analyses we have provided is that the relation between indicators at the macro-(country-) level and welfare attitudes is a complex and highly contingent one. Any notion that certain structural macro-variables could in general be linked to specific levels of or associations among welfare attitudes is a pipe dream. Instead, the impact of macro-variables is highly diverse, depending on the attitudinal outcomes in question as well as the proposed mechanisms.

228

Svallfors

In this respect, it is important to have in mind that nonfindings may be just as pertinent as findings. For example, we show in Chapter 6 that it is not the case that older populations display more age conflicts than younger populations. If anything, the reverse is true. A higher demographic burden is thus not at all translated into sharper distributional age conflicts or substantial age differences in policy preferences. This is contrary to reasonable expectations of emerging distributional conflicts in aging societies. Explaining it entails realizing that simple self-interest-based models not only fail to capture the pertinent mechanisms at the micro-level, but they also stand at odds with findings regarding the macro–micro links. Instead, socially constructivist modes of approaching the link between macro-variables and micro-outcomes look more promising. In the case of age conflicts, one has to realize that the elderly are considered a highly deserving group for support (Van Oorschot, 2000; Svallfors, 2008), and this means that a demographic burden is not likely to be translated into sharp age conflicts unless the elderly are delegitimized as a recipient group. And this is not likely to occur, given current political and policy setups in Europe. On a more general note, we would contend that it is highly possible that research will not find any consistent and persistent associations between macro-variables and attitudinal outcomes. These associations may simply be too historically specific and too dependent on the full context in which they take place to display any lasting patterns to be analyzed. But research is only at the start here, and future analyses will show to what extent these misgivings may be warranted. A further contribution is our extending of the analyses beyond the traditional “three worlds of welfare capitalism” (Esping-Andersen, 1990). As pointed out in the introduction, comparing welfare attitudes in x worlds of welfare has turned into a veritable cottage industry, but perhaps now with some degree of diminishing returns. In this volume, all chapters instead take a European-wide perspective and include also the countries of Southern and Eastern Europe. In addition to adding specific interesting findings, this approach also highlights some general shortcomings and blind spots of existing research. In particular, the characteristics of the former communist countries of East-Central Europe are worth noting. One should, of course, avoid speaking of these countries as if they were all of the same kind. As Fenger (2007: 16) notes, they are “both mutually differentiated and collectively different



Welfare Attitudes in Context

229

from the Western countries” in their welfare characteristics and outcomes. This certainly goes also for their welfare attitudes. These countries seem to combine a highly combustible combination of attitudes to and evaluations of the welfare state. High demand for state intervention and redistribution (Chapters 1, 2, and 7) is combined with fairly low social trust (Chapter 4), very low satisfaction with performance (Chapters 2 and 3), high perceived risk (Chapter 3), comparatively large and growing class differences in attitudes (Chapter 6), and more widespread authoritarian dispositions and welfare chauvinism than elsewhere (Chapters 4 and 5). In combination, such a mindset seems highly conducive to the kind of ruptures and eruptions many countries in Eastern Europe have experienced in the recent past. Findings such as these evoke the somber thought that so much research effort has been used to analyze a part of Europe and the world where problems of sustenance and legitimacy are in fact the smallest. The three worlds of welfare capitalism represent a privileged corner of the globe, one where democracy and prosperity—in spite of all their shortcomings and gaps— reign supreme. Just by extending comparative analyses to the Southern and Eastern parts of Europe, this book has made a small step toward amending this state of affairs. The picture that emerges is a bleaker and more problematic one than those provided from the three worlds of comparatively favored nations. As the supply of comparative high-quality data improves, it is to be hoped that many more analyses will put countries outside the three worlds of welfare capitalism in comparative perspective. The survey module on which our current analyses build is for the moment the best available for pan-European analyses of welfare attitudes. Perhaps could it in the future be extended to encompass middle-income countries outside Europe. The last broader theme that will be highlighted is a methodological one. Here, we should first emphasize the development of the analytical model itself and its operationalization. In comparison with previous research, this is by far the most comprehensive attempt to tackle welfare attitudes in comparative perspective. There are national surveys that take as broad a perspective as the one we have applied in our European Social Survey module, but they are, of course, not comparable among countries. Existing comparative welfare attitude surveys tend to be much more general and circumscribed than the one we have used in our book. In particular, nowhere do we find an attempt to measure the underlying value bases for welfare

230

Svallfors

attitudes or the possible mechanisms that tie together categorical locations and attitudes. This book has only touched upon many of the themes in the Welfare Attitudes ESS module, and we expect many more analyses to follow down the line. Furthermore, we would expect later research in the field to take this module as its own starting point for extensions and criticisms. The analyses in this book display an interesting variety concerning how to use country units in methodological terms. Some chapters use them for constructing analytical categories that frame micro-level associations (Chapters 3 and 4), others use them as observational units in order to interpret and explain micro-level associations (Chapter 6), and still others use them as level-2 units in various forms of hierarchical modeling (Chapters 2, 4, and 5). This is indicative of the fact that there is no “one best way” to use country units in comparative research. The overblown expectations sometimes laid on multilevel models have been gradually replaced with a more cautious, low-keyed pluralistic approach to the use of countries as explanatory units in comparative research (see Ragin, 1987; Shalev, 2007). The multilevel methods that are now increasingly established as a sort of standard for the field were developed with quite different nested data structures in mind—for example, where a large number of schools or other organizations are sampled, and then individuals are sampled within these higher-level units. In cross-national comparative survey research, what we have are higher-order units (i.e., countries) that are relatively few and chosen on strategic or simple availability grounds. We therefore tend to push this kind of method beyond what it was originally intended to do. There are certainly statistical ways to take the particular data structure into account (see, for instance, the example set in Chapter 5). However, it is important to keep in mind that the problems we encounter in this regard in comparative research are at their core information problems rather than statistical problems per se. So a certain degree of modesty and continuing methodological pluralism is highly warranted. Yet another methodological contribution has been the use of survey-embedded experiments in order to analyze framing effects on welfare attitudes (Chapter 7). The use of survey-embedded experiments allows the malleability of welfare attitudes to be analyzed, by unpacking elements of the framing context in which they are formed. The findings from these analyses (see the summary in Chapter 1) suggest some pertinent framing effects but also some important noneffects. In any realistic application of framing theory,



Welfare Attitudes in Context

231

we would tend to expect just such a mix of results, and the results clearly point in the American context to the near-certain importance of conflicts surrounding immigration to several domains of the welfare state. For the future, it would seem worthwhile to extend these framing experiments to a comparative format. As is well known, the American context differs from the European by having political parties that are less moored to popular movements and social cleavages (Lipset, 1996: 39–46), by being less institutionally dense when it comes to welfare policies (Svallfors, 2007), and by having a less attentive and informed public in political matters (Granberg and Holmberg, 1988; Milner, 2002). An interesting question is whether this will make European publics more or less susceptible than the American one to political articulation and framing in the formation of their welfare attitudes. Will it make them more susceptible because they pay more attention to the wider political debate or less susceptible because their attitudes are more strongly rooted in institutional experiences and political affiliations? This leads to the wider issue of how our analyses could and should be extended in the future, and what tensions, lacunas, and fissures our own analyses display. In closing this book, a few words need to be said about these uncomfortable matters. t e n sion s a n d n e w c h a l l e ng e s Current research on welfare attitudes branches out in two slightly different directions. One is a strict political-economy approach, where welfare attitudes are seen as basically reflecting the calculated self-interest of individuals (for example, Iversen and Soskice, 2001; Cusack et al., 2006; Kitschelt and Rehm, 2006; Rehm, 2007). The other main strand is a political-sociological approach, where welfare attitudes are seen as reflecting not only self-interest but also broader considerations about social justice, social rights, and reciprocity (for example, Mau, 2003; Brooks and Manza, 2007; Svallfors, 2007). Although our book is set firmly in the latter camp, there are clearly tensions regarding the importance ascribed to self-interests in the formation of welfare attitudes. While the analyses in Chapter 3 build on assumptions about self-interest as a key mechanism in shaping welfare attitudes, Chapters 4 and 6 more or less explicitly state that self-interest is only one, and possibly a minor, factor behind the formation of welfare attitudes. While

232

Svallfors

presumably none of these authors would deny the existence of different mechanisms behind welfare attitudes, or that self-interest is one of these, the analytical emphasis is nevertheless quite different across the chapters. This highlights that the issue about the role of calculation and self-interest in the formation of welfare attitudes is hardly settled. We expect future analyses to tackle these issues head-on, possibly with better indicators of both self-interest and other mechanisms than the ones we have used here. A second tension in our volume relates to the indeterminate relationship between theoretical concepts and empirical indicators. Question items that were designed to measure some specific theoretical concept or idea may well be used by other researchers to indicate other concepts and ideas. As displayed in the various chapters, different authors come up with different solutions on how best to capture the different concepts they apply. Furthermore, one person’s indicators of mechanisms can be the next person’s dependent variables. This is unavoidable, but it can certainly create less cumulation and more confusion than what is ideal. Yet another tension relates to the micro–macro linkages, and most specifically on how to deal with regime typologies. As pointed out, simply comparing the welfare attitudes in “x worlds of welfare capitalism” looks like an exercise with diminishing returns. However, we are still using the regime typology to frame our inquiries, as presented in some of the chapters. Here, the regime typology is used mainly to test whether mechanisms and underlying values are organized by welfare regime. Other chapters do not take the same route but instead try to quantify pertinent aspects of the macro-environment, as discussed in the previous section. What these exercises are testimony to is that there are different ways to try and grapple with the information problem at the country level. There are simply too few cases and too many possible variables of interest to ever make analyses of the macro–micro linkages truly robust. Carefully guided by theory, and cautiously proceeding, as the analyses in this book without exception are, we nevertheless walk on thin ice when it comes to the macro– micro problem in comparative attitude research. The last tension that should be brought out in our analyses is a crossdisciplinary one that pits the social-psychological perspective of Chapter 4 against the other chapters. The social-psychological perspective entails a “bottom-up” variety of welfare attitude analysis, where attitudes are seen as grounded in everyday experiences and their interpretations. This is in



Welfare Attitudes in Context

233

contrast to the more “top-down” political-sociological perspective of many of the other chapters, where macro-features and the actions of elites are seen as fundamentally important. It seems that a synthesis of both these perspectives and their suggested mechanisms would be a fruitful way forward. Apart from productive tensions behind our current analyses, some notable elements appear to be more or less missing from this book. One concerns the policy responsiveness to attitudes, or put in different form, the micro-macro aggregation of attitudes to organized political action and policy outcomes. Our analyses have rather dealt with the macro-micro problem—that is, how institutions and other contextual factors affect attitudinal patterns among the population. How such attitudes in turn affect policy outcomes is a complicated issue in its own right, and one to which considerable research effort has been devoted (for example, Burstein, 1998; 2003; Soroka and Wlezien, 2004; 2005; Brooks and Manza, 2006a; 2006c; 2007). In general, there is agreement that there is some degree of congruence between attitudes and policy outcomes, but how much and under what conditions attitudes affect policy remain a bone of contention (Manza and Cook, 2002; Brooks and Manza, 2006a; 2006b; Myles, 2006). Difficult methodological problems beset this line of research, including how to deal with counterfactual analyses (“What would policy development have been if the attitudes had been different?”). Exploring and explaining the attitudes-policy nexus remain important challenges in the field. A second key remaining issue concerns the dynamics of welfare attitudes. Since we build on a pioneering comparative survey, this is, of course, unavoidable, but some reflections about the issue must be made. For example, it may be asked to what extent the findings in this book depend on data being collected at a particular point in time, in the midst of the most severe economic recession in the world economy since the 1930s. There is, of course, no way to firmly decide this empirically, since the data on which we build were collected at this time and none other. But a few observations can be made. One is that we have, in general, been less concerned in this volume with levels of support for the welfare state and much more with patterns of associations between groups, among variables, and across countries. It is perhaps less probable that such associations and patterns would have been fundamentally affected by the economic crisis, even though the changing class patterns found in Chapter 6 may in part be grounded in the economic crisis.

234

Svallfors

A second point to be noted is that the one single indicator of welfare attitudes for which we have a time series from the ESS—the one asking whether government should reduce income differences—shows little consistent change in country means between 2006 and 2008. Only a few of these changes are statistically significant, they go in different directions, and they are not larger than the ones registered between 2004 and 2006. So there is no support here for arguing that the economic crisis should have made people in general more prone to ask for government redistribution (figures from the ESS data files). A third observation stems from the framing experiments in Chapter 7. They showed that articulation of the economic crisis in the framing of survey questions did little to affect Americans’ attitudes toward welfare state intervention. Clem Brooks suggests that this shows the economic crisis so far has had little effect on Americans’ thinking about the welfare state. There is, nonetheless, the alternative interpretation that respondents had already factored in the economic crisis when they answered the unframed ESS baseline question. Furthermore, there is no way to tell how far these findings from the United States can be extrapolated to Europe. But they do seem to suggest that although the economic crisis may have affected welfare attitudes to some extent, it has by no means turned them in any dramatic fashion. These considerations invite the much broader and final question about the dynamics of attitudes in comparative perspective. In fact, of all the outstanding issues we have alluded to in this final chapter, this seems like the one most pressing for answers. There have certainly been analyses of changes in attitudes toward state intervention, using data from the International Social Survey Program. In a nutshell, these data show a considerable transatlantic convergence in attitudes toward state intervention among the rich Western countries from the mid-1990s to the mid-2000s. Behind this convergence lies, on the one hand, decreasing support for employment policies among the publics of Western Europe and, on the other, rising concerns about income inequality in the United States (Edlund, 2009; Edlund and Svallfors, 2011). But the ISSP data are set at such a general level that it is hard to decide exactly what could explain these trends and whether they apply also to other domains and aspects than principled support for state intervention. As shown in Chapter 7, it is still clearly the case that American welfare attitudes are at a considerable distance from the European average.



Welfare Attitudes in Context

235

A broader and more explanatory approach is highly warranted in order to get a better handle on the dynamics of attitudes. In this respect, it seems researchers in the field have often been unduly constrained by a mechanical application of “independent” and “dependent” variables. Research would perhaps be better served by moving toward evolutionary stories, in which the dynamics of attitudes are seen as coevolving with institutional and political change, subject to mutual and recurring feedback loops. The challenge is to do this without becoming speculative and empirically unfounded. But dynamics are also relevant to take into account at an individual level. Currently, there are no longitudinal data sets that allow us to analyze the development of welfare attitudes along the life cycle, or even across an extended stretch of time. Comparative longitudinal data sets simply do not exist, and even existing national panels that to any extent include welfare attitudes tend to be either too short term or too sparse to be of much use. Any move to address this shortage would be most useful for the field. Consequently and furthermore, we know little about the relative importance of the past (that is, biography), the present (that is, the current position), and the future (that is, anticipation) in forming attitudes. In particular, we know very little about how anticipated futures affect presentday attitudes. We all live in the shadow of the future, and the attitudes we hold are to some extent a function of what we believe about alternative futures and their implications for our personal lives and society at large. So longitudinal data are clearly called for. But a repetition of the European Social Survey module we used in this book would also be highly valuable in analyzing the dynamics of attitudes. Even if this were a repeated cross section and not a longitudinal study, it would allow several of the outstanding issues we have mentioned to be addressed. Consider this a call to ourselves and other researchers to make it happen! We have reached the end of our journey. But in research, ending one journey always means embarking on a new one. We hope future travelers along these paths will take advantage of the maps and signposts we have provided, wherever they may be destined to arrive in their own explorations.

r efer ences Aalberg, T. 2003. Achieving Justice. Comparative Public Opinions on Income Distribution. Leiden: Brill.

236

Svallfors

Alesina, A., and E. Glaeser. 2004. Fighting Poverty in the US and Europe: A World of Difference. New York: Oxford University Press. Andreß, H. J., and T. Heien. 2001. “Four Worlds of Welfare State Attitudes? A Comparison of Germany, Norway and the United States.” European Sociological Review 17: 337–56. Banting, K. G., R. Johnston, W. Kymlicka, and S. Soroka, S. 2006. “Do Multiculturalism Policies Undermine the Welfare State? An Empirical Analysis.” In Multiculturalism and the Welfare State: Recognition and Redistribution in Contemporary Democracies, edited by K. G. Banting and W. Kymlicka, 49–91. Oxford: Oxford University Press. Banting, K. G., and W. Kymlicka, eds. 2006. Multiculturalism and the Welfare State: Recognition and Redistribution in Contemporary Democracies. Oxford: Oxford University Press. Blekesaune, M., and J. Quadagno. 2003. “Public Attitudes Toward Welfare State Policies: A Comparative Analysis of 24 Nations.” European Sociological Review 19: 415–27. Bonoli, G. 2000. “Public Attitudes to Social Protection and Political Economy Traditions in Western Europe.” European Societies 2: 431–52. Brooks, C., and J. Manza. 2006a. “Social Policy Responsiveness in Developed Democracies.” American Sociological Review 71: 474–94. ———. 2006b. “Theory and Methods for Comparative Opinion/Social Policy Research: Reply to Myles.” American Sociological Review 71: 499–502. ———. 2006c. “Why Do Welfare States Persist?” Journal of Politics 68: 816–27. ———. 2007. Why Welfare States Persist: the Importance of Public Opinion in Democracies. Chicago: University of Chicago Press. Burstein, P. 1998. “Bringing the Public Back In: Should Sociologists Consider the Impact of Public Opinion on Public Policy?” Social Forces 77: 27–62. ———. 2003. “The Impact of Public Opinion on Public Policy: A Review and an Agenda.” Political Research Quarterly 56: 29–40. Castles, S., and C.-U. Schierup. 2010. “Migration and Ethnic Minorities.” In The Oxford Handbook of the Welfare State, edited by F. G. Castles, S. Leibfried, J. Lewis, H. Obinger, and C. Pierson, 278–91. Oxford: Oxford University Press. Cusack, T., T. Iversen, and P. Rehm. 2006. “Risks at Work: The Demand and Supply Sides of Government Redistribution.” Oxford Review of Economic Policy 22: 365–89. Edlund, J. 2009. “Attitudes Towards State Organized Welfare in 22 Societies: A Question of Convergence?” In The International Social Survey Programme 1984–2009: Charting the Globe, edited by M. Haller, R. Jowell, and T. Smith, 125–46. London: Routledge. Edlund, J., and S. Svallfors. 2011. “Cohort, Class and Attitudes to Redistribution in Two Liberal Welfare States: Britain and the United States, 1996–2006.” In Generational Politics and Policies: Comparative Studies of Ageing Post-



Welfare Attitudes in Context

237

industrial Democracies, edited by A. Goerres and P. Vanhuysse, 206–24. London: Routledge. Eger, M. A. 2010. “Even in Sweden: The Effect of Immigration on Support for Welfare State Spending.” European Sociological Review 26: 203–17. Erikson, R., and J. H. Goldthorpe. 1992. The Constant Flux: a Study of Class Mobility in Industrial Societies. Oxford: Clarendon Press. Esping-Andersen, G. 1990. The Three Worlds of Welfare Capitalism. Cambridge, UK: Polity Press. Fenger, H. J. M. 2007. “Welfare Regimes in Central and Eastern Europe: Incorporating Post-Communist Countries in a Welfare Regime Typology.” Contemporary Issues and Ideas in the Social Sciences 3: 1–30. Finseraas, H. 2008. “Immigration and Preferences for Redistribution: an Empirical Analysis of European Survey Data.” Comparative European Politics 6: 407–31. Goldthorpe, J. H., and G. Marshall. 1992. “The Promising Future of Class Analysis—a Response to Recent Critiques.” Sociology—the Journal of the British Sociological Association 26: 381–400. Granberg, D., and S. Holmberg. 1988. The Political System Matters: Social Psychology and Voting Behavior in Sweden and the United States. Cambridge, UK, and New York: Cambridge University Press. Iversen, T., and D. Soskice. 2001. “An Asset Theory of Social Policy Preferences.” American Political Science Review 95: 875–93. Jæger, M. M. 2006. “Welfare Regimes and Attitudes Towards Redistribution: The Regime Hypothesis Revisited.” European Sociological Review 22: 157–70. Kitschelt, H., and P. Rehm, P. 2006. “New Social Risk and Political Preferences.” In The Politics of Post-Industrial Welfare States. Adapting Postwar Social Policies to New Social Risks, edited by K. Armingeon and G. Bonoli, 52–82. London: Routledge. Kumlin, S. 2007. “The Welfare State: Values, Policy Preferences, and Performance Evaluations.” In The Oxford Handbook of Political Behavior, edited by R. J. Dalton and H.-D. Klingemann, 362–82. Oxford and New York: Oxford University Press. Kumlin, S., and S. Svallfors. 2007. “Social Stratification and Political Articulation: Why Attitudinal Class Differences Vary Across Countries.” In Social Justice, Legitimacy and the Welfare State, edited by S. Mau and B. Veghte, 19–46. Aldershot, UK: Ashgate Publishers Ltd. Lieberson, S. 1985. Making It Count: the Improvement of Social Research and Theory. Berkeley: University of California Press. Lipset, S. 1959. “Democracy and Working-Class Authoritarianism.” American Sociological Review 24: 482–501. ———. 1996. American Exceptionalism: a Double-Edged Sword. New York: W. W. Norton.

238

Svallfors

Manza, J., and F. L. Cook. 2002. “A Democratic Polity? Three Views of Policy Responsiveness in the United States.” American Politics Quarterly 30: 630–67. Mau, S. 2003. The Moral Economy of Welfare States. Britain and Germany Compared. London: Routledge. Mau, S., and B. Veghte, eds. 2007. Social Justice, Legitimacy and the Welfare State. Hampshire, UK: Ashgate Publishers Ltd. Milner, H. 2002. Civic Literacy: How Informed Citizens Make Democracy Work. Hanover, MA: University Press of New England. Myles, J. 2006. “Welfare States and Public Opinion: Comment on Brooks and Manza.” American Sociological Review 71: 495–98. Ragin, C. C. 1987. The Comparative Method: Moving Beyond Qualitative and Quantitative Strategies. Berkeley: University of California Press. Rehm, P. 2007. “Who Supports the Welfare State? Determinants of Preferences Concerning redistribution.” In Social Justice, Legitimacy and the Welfare State, edited by S. Mau and B. Veghte, 47–72. Aldershot, UK: Ashgate Publishers Ltd. Roller, E. 1995. “The Welfare State: The Equality Dimension. In The Scope of Government, edited by O. Borre and E. Scarbrough, 165–97. New York/ Oxford: Oxford University Press. Shalev, M. 2007. “Limits and Alternatives to Multiple Regression in Comparative Research.” Comparative Social Research 24: 261–308. Soroka, S., and C. Wlezien. 2004. “Opinion Representation and Policy Feedback: Canada in Comparative Perspective.” Canadian Journal of Political Science 37: 531–59. ———. 2005. “Opinion-Policy Dynamics: Public Preferences and Public Expenditure in the United Kingdom.” British Journal of Political Science 35: 665–89. Svallfors, S. 1997. “Worlds of Welfare and Attitudes to Redistribution: A Comparison of Eight Western Nations.” European Sociological Review 13: 283–304. ———. 2003. “Welfare Regimes and Welfare Opinions: a Comparison of Eight Western Countries.” Social Indicators Research 64: 495–520. ———. 2004. “Class, Attitudes and the Welfare State: Sweden in Comparative Perspective. Social Policy & Administration 38: 119–38. ———. 2006. The Moral Economy of Class: Class and Attitudes in Comparative Perspective. Stanford, CA: Stanford University Press. ———, ed. 2007. The Political Sociology of the Welfare State: Institutions, Social Cleavages, and Orientations. Stanford, CA: Stanford University Press. ———. 2008. “The Generational Contract in Sweden: Age-Specific Attitudes to Age-Related Policies.” Policy and Politics 36: 381–96. ———. 2010. “Public Attitudes.” In The Oxford Handbook of the Welfare State, edited by F. G. Castles, S. Leibfried, J. Lewis, H. Obinger, and C. Pierson, 241–51. Oxford: Oxford University Press.



Welfare Attitudes in Context

239

Taylor-Gooby, P. 2005. “Is the Future American? Or, Can Left Politics Preserve European Welfare States from Erosion through Growing ‘Racial’ Diversity?” Journal of Social Policy 34: 661–72. Van Oorschot, W. 2000. “Who Should Get What, and Why? On Deservingness Criteria and the Conditionality of Solidarity Among the Public.” Policy & Politics 28: 33–48.

inde x

Italic page numbers indicate material in tables or figures. Adorno, T., 124 African Americans, 203–5, 204, 213 Age. See Elderly; Government responsibilities, attitudes toward Analyses, first-generation, 5–7 Analytical framework, 10–11 Anglo-Saxon welfare regime: described, 64, 65; high-risk groups, attitudes among, 75–78; high-risk groups, attitudes among: government responsibilities, 70, 71; high-risk groups, attitudes among: path analysis, 70, 72, 73, 74; high-risk groups, attitudes among: perceived future risk, 67–68, 68; high-risk groups, attitudes among: welfare state performance evaluation, 68–69, 69; normative welfare legitimacy, 103–5, 104. See also High-risk groups, attitudes among Authoritarianism: collective, 129, 143, 145; conception of, 124–25; in moral order, 89–90, 95; normative welfare legitimacy and, 99, 106, 107, 108, 109; welfare chauvinism and, 125–26, 127– 28, 139, 140, 141, 147. See also Normative approach to welfare attitudes Beliefs, generally, 81, 96–98, 110, 114n, 162, 187n3 Beliefs in Government project, 7 Blacks, 203–5, 204, 213 Calzada, I., 28 Categorical differentiation, 86, 87, 87

Change, question of, 194–97 Chauvinism, welfare. See Welfare chauvinism Child care, 205, 206, 211 Class and welfare chauvinism, 121–26, 127, 139, 140. See also Government responsibilities, attitudes toward Collective authoritarianism, 129, 143, 145 Comparative Manifestos data set, 186 Conceptual framework, 12, 12–13 Continental welfare regime: described, 64, 65; high-risk groups, attitudes among, 75–78; high-risk groups, attitudes among: government responsibilities, 70, 71; high-risk groups, attitudes among: path analysis, 70, 72, 73, 74; high-risk groups, attitudes among: perceived future risk, 67–68, 68; high-risk groups, attitudes among: welfare state performance evaluation, 68–69, 69; normative welfare legitimacy, 103–5, 104. See also High-risk groups, attitudes among Coughlin, Richard, 5 Cross-disciplinary tensions, 231–32 Cultural approaches to welfare attitudes, 82–84 Cultural diversity, 224–26 Denmark, 178–80, 179 Dependent variables, generally, 224 Deservingness theory, 36–38, 54n4 Differentiation between social categories, 86–87, 87

241

242

Index

Distrust, 90, 95, 99, 107, 108, 109. See also Normative approach to welfare attitudes Diversity, 87, 91, 95, 151n4, 224–26 Duckitt, John, 124–25, 133, 151n2 Eastern European welfare regime: described, 64–65; high-risk groups, attitudes among, 75–78; high-risk groups, attitudes among: government responsibilities, 70, 71; high-risk groups, attitudes among: path analysis, 70, 72, 73, 74–75; high-risk groups, attitudes among: perceived future risk, 67–68, 68; high-risk groups, attitudes among: welfare state performance evaluation, 68–69, 69; normative welfare legitimacy, 103–5, 104; welfare chauvinism, 136–37, 137. See also High-risk groups, attitudes among Economic crisis, impact of, 202–3, 205, 206, 207, 233–34 Economic performance of a country, 142, 143, 144 Edlund, J., 28 Education level: normative beliefs as function of, 96–98, 110, 114n; as social position measure, 93–94; welfare chauvinism and, 127, 146, 147 Egalitarianism, 92, 95, 98–99, 108–9. See also Normative approach to welfare attitudes Elderly: old-age dependency ratio and attitudes toward labor market policies, 177, 178; standard of living of, 34–36, 35, 42, 43, 46, 47, 210. See also Government responsibilities, attitudes toward; Welfare performance and welfare support Empirical indicators and theoretical concepts, indeterminate relation between, 232 Erikson-Goldthorpe class schema, 32, 131–32, 169 Esping-Andersen, G., 3, 8–9 Ethnic diversity, 151n4, 224–26 Ethnocentrism, 91, 95, 99, 106, 107, 108, 109. See also Normative approach to welfare attitudes European Social Survey, 12, 13, 168 Explanatory mechanisms, new, 226–27

Feedback effects, 11 Feldman, S., 83–84 Finland, 169–70, 170, 171 First-generation analyses, 5–7 Framing theory, welfare attitudes, and the United States case, 193–221; about, 19–20, 193–94; analyses and results, 203–13; analyses and results: cross-national perspective, 207, 208, 209, 209, 210, 211, 212; analyses and results: economic crisis, impact on opinion formation, 205, 206, 207; analyses and results: framing impacts, magnitude of, 212–13; analyses and results: target group effects, 203–5, 204; change, question of, 194–97; change, question of: institutionalist approaches, 194–96; change, question of: policy responsiveness, 196–97; conclusion, 214–16; economic crisis experiments, 202–3; empirical application, 200–203; framing theory, 197–200; survey experiments and research design, 201; The Survey of American Policy Attitudes (2009), 200–203, 216–17; target group experiments, 201–2; welfare state scholarship, implications for, 199–200 Free market (conception of social order), 87, 90–91, 95 Fridberg, T., 64–65, 78n2 Government responsibilities, attitudes toward, 158–92; about, 18–19, 158–61; age and class differences across countries, 170–74, 171, 172–73, 178–80, 179; age categories, 168–69; age differences in attitudes, 177–78, 178, 179, 180; categories, mechanisms, and contexts, 161–63; class differences in attitudes, 176–77, 179, 179–80; class indicators, 169; conclusion, 180–84; findings, 175–78, 177, 178; Finland, 169–70, 170, 171; high-risk groups, 70, 71; human values dimensions/ items, 184–85; indicators and analytical strategies, 166–70, 170; macrolevel factors, 163–66; macro-variables, 186, 186–87; micro-level mechanisms, 174–75; Scandinavia versus Eastern Europe comparison, 178–80, 179

Health Care and Support for Elderly index, 188n10 High-risk groups, attitudes among, 58–80; about, 16, 58–60, 60; conclusion, 75–78; methods, 67; research questions, 66–67; results, 67–75; results: government responsibilities, 70, 71; results: path analysis for different welfare regimes, 70, 72, 73, 74–75; results: perceived future risk, 67–68, 68; results: welfare state performance evaluation, 68–69, 69; studies, previous, 60–63; variables, 67, 78n4; welfare regimes and, 63–65 Identity effect, 145 Immigrants, 203–4, 204, 209. See also Welfare chauvinism Improvement/overburden reactions, 29–30 Information problem in cross-country comparisons, 232 Institutionalist scholarship, 194–96 Interaction effects, 11 Interest theory, 44 Intergenerational contract, welfare state as, 159 International Social Survey Program, 234 Interpretive effects, 11 Israel, 137, 138 Kangas, O., 64–65, 78n2 K index, as summary measure, 169–70, 170, 171, 189nn12–13 Korpi, Walter, 121 Kumlin, S., 28–29 Labor market, 122–23, 143, 144–45, 177, 178. See also Unemployed Luxembourg Income Study, 189n17 Macro-micro interplay, methodological contributions on modeling, 229–31 Macro-to-micro effects, 11 Macro-variables, generally, 10–11, 227–28 Main effects, 11 Material risk. See Risk perceptions Material vulnerability, 94, 96–98, 110, 114n Mechanisms, generally, 11, 161

Index

243

Mediation model, 98, 100, 100–102, 101, 115n2 Middle-class Americans, 203–4, 204, 205, 206, 207 Moral economy, 10 Moral order, 87, 89–90, 95 National surveys, 5 Nordic welfare regime: described, 64, 65; high-risk groups, attitudes among, 75–78; high-risk groups, attitudes among: government responsibilities, 70, 71; high-risk groups, attitudes among: path analysis, 70, 72, 73, 74, 75; high-risk groups, attitudes among: perceived future risk, 67–68, 68; highrisk groups, attitudes among: welfare state performance evaluation, 68–69, 69; normative welfare legitimacy, 103–5, 104. See also High-risk groups, attitudes among Normative approach to welfare attitudes, 81–118; about, 16–17, 81–82; analytic model, 92–93, 93; conclusion, 109–14; cultural approaches to welfare attitudes, 82–84; data and measures, 93–96, 95; mediation model, 98, 100, 100–102, 101, 115n2; model underlying, 84–89, 87; multilevel approach to country-level variation, 105–6, 106–7, 108–9, 112–14; normative beliefs as function of education level and vulnerability, 96–98, 110, 114n; social government intervention, normative legitimacy of, 98–100; social order, conceptions of, 86–88, 87, 89–92, 95, 110; welfare regimes and, 102–5, 104 Normative beliefs, 81, 96–98, 110, 114n Normative differentiation, 86–87, 87 Normative mechanisms, 11 Old-age dependency ratio, 177, 178. See also Elderly Persuasion, 217n3 Pierson, P., 11, 195 Poland, 178–80, 179 Policy responsiveness, 196–97, 233 Political Manifesto data, 181 Political-sociological perspective, 231–32 Portugal, 137, 138

244

Index

Power resource theory, 121 Priming, 198–99 Realistic conflict theory, 123, 128–29, 151n4 Resource and incentive effects, 11 Reward/punishment reactions, 29, 30 Risk perceptions: of future risk, 67–68, 68; government responsibilities, attitudes toward, and, 167, 174–75; by high-risk groups, 67–68, 68; of material risk, 143, 145, 146, 147, 148; welfare chauvinism and, 143, 145, 146, 147, 148 Scandinavian welfare regime. See Nordic welfare regime Self-interest, 83, 159–60, 231–32 Slovakia, 178–80, 179 Social diversity (conception of social order), 87, 91, 95 Social expenditure, 89, 105, 106, 106–7, 108 Social government intervention, normative legitimacy of, 98–100. See also Government responsibilities, attitudes toward Social order, conceptions of, 86–88, 87, 89–92, 95, 110 Social-psychological theory, 54n3, 231–32 Social representations theory, 84–85 Sociocultural approach to welfare chauvinism, 120, 124–26 Socioeconomic approaches to welfare chauvinism, 120, 122–23, 126 Socioeconomic status, 127 Southern European welfare regime: described, 64, 65; high-risk groups, attitudes among, 75–78; high-risk groups, attitudes among: government responsibilities, 70, 71; high-risk groups, attitudes among: path analysis, 70, 72, 73, 74; high-risk groups, attitudes among: perceived future risk, 67–68, 68; highrisk groups, attitudes among: welfare state performance evaluation, 68–69, 69; normative welfare legitimacy, 103–5, 104. See also High-risk groups, attitudes among Spain, 28, 137, 138 Split labor market theory, 122–23

Standard of living: of elderly, 34–36, 35, 42, 43, 46, 47, 210; perceived welfare performance and, 31–32; of unemployed, 36–38, 37, 42, 43, 46–47, 48, 211 State of the art, current, 4–10; Beliefs in Government project, 7; comparative data sets, 7; data, lack of, 4–5; findings from early works, common key, 5–6; first-generation analyses, 5–7; national surveys, 5; “worlds of welfare” categorization, 8–9 Structural inequality (conception of social order), 87, 91–92, 95 Survey of American Policy Attitudes, The (2009), 200–203, 216–17 Surveys, national, 5 Sweden, 28, 137, 138, 178–80, 179 Theoretical concepts and empirical indicators, indeterminate relation between, 232 Thomas theorem, 26 “Three worlds of welfare capitalism,” extending analyses beyond, 228–29 Transfer classes, 61 Transfer-dependent groups, 61–62. See also High-risk groups, attitudes among 2009 Survey of American Policy Attitudes, The, 200–203, 216–17 Unemployed: normative welfare legitimacy and, 89, 105, 106, 106, 107, 108–9; standard of living of, 36–38, 37, 42, 43, 46–47, 48, 211; welfare chauvinism and, 141–42, 143. See also Labor market; Welfare performance and welfare support United States. See Framing theory, welfare attitudes, and the United States case Values, 83–84, 162–63, 168, 174, 184– 85, 187n4, 189n15 Van der Waal, J., 125–26 Variables, generally, 10–11, 224, 227–28 Vulnerability, material, 94, 96–98, 110, 114n Welfare attitudes, dynamics of, 233–35 Welfare attitudes in context, 222–35; about, 20, 222–23; cross-disciplinary

tensions, 231–32; dependent variables, broadening and specification of, 224; diversity as new challenge, 224–26; economic crisis, impact of, 233–34; explanatory mechanisms, new, 226–27; findings, thematic, 223–31; information problem in cross-country comparisons, 232; macro-micro interplay, methodological contributions on modeling, 229–31; macro-variables, impact of, 227–28; policy responsiveness to attitudes, 233; regime typologies, dealing with, 232; self-interest, role in attitude formation, 231–32; tensions and new challenges, 231–35; theoretical concepts and empirical indicators, indeterminate relation between, 232; “three worlds of welfare capitalism,” extending analyses beyond, 228–29; welfare attitudes, dynamics of, 233– 35 Welfare chauvinism, 119–57; about, 17–18, 119–21; authoritarianism and, 125–26, 127–28, 139, 140, 141, 147; class and, 121–26, 127, 139, 140; collective authoritarianism and, 129, 143, 145; conclusion, 148–50; country-level determinants of, 141–42, 143, 144–45, 152n9; data and method, 129–36; data and method: covariates, 131–35; data and method: method, 135–36; data and method: response variable, 130–31; defined, 120, 122; economic performance of a country and, 142, 143, 144; education level and, 127, 146, 147; hypotheses, country-level, 128–29; hypotheses, individual-level, 126–28; individual-level determinants of, 139, 140, 141; labor force and, 143, 144–45; perceived material risk and, 143, 145, 146, 147, 148; results, 136–48, 137; results: country-level determinants of welfare chauvinism,

Index

245

141–42, 143, 144–45, 152n9; results: descriptive findings, 136–39; results: individual-level determinants of welfare chauvinism, 139, 140, 141; results: probabilities, predicted, 145–46, 147, 148; sociocultural approach to, 120, 124–26; socioeconomic approaches to, 120, 122–23, 126; socioeconomic status and, 127; unemployment and, 141–42, 143 Welfare dependency, 83, 91, 95, 99, 109. See also Normative approach to welfare attitudes Welfare expenditure, 89, 105, 106, 106–7, 108 Welfare Index, 13–14, 14, 21n Welfare performance and welfare support, 25–57; about, 15–16, 25–27; conclusion, 50–53; data and methods, 30–34, 31; findings, 34–50; findings: exploratory analyses, 34–38, 35, 37; findings: multilevel analyses of perceived performance, 47–50; findings: multilevel analysis of welfare support, 38, 39–41, 42, 43, 44–47, 47; highrisk groups, attitudes among, 68–69, 69; perceived versus actual welfare performance, 31–32; research hypotheses, 29–30; research questions, 26–27; statistical models, 34; studies, previous, 27–29 Welfare regimes: defining, 9; high-risk groups, attitudes among, 63–65; normative welfare legitimacy and, 102–5, 104; path analysis for high-risk groups in, 70, 72, 73, 74–75; typologies of, 3, 232. See also specific welfare regimes Working class, 205, 206, 207. See also Welfare chauvinism “Worlds of welfare” categorization, 8–9, 228–29 Zaller, J., 83–84

St udies i n Soci a l I n equa l i t y Improving Learning Environments: School Discipline and Student Achievement in Comparative Perspective Edited by Richard Arum and Melissa Velez 2012 The New Gilded Age: The Critical Inequality Debates of Our Time Edited by David B. Grusky and Tamar Kricheli-Katz 2012 Broke: How Debt Bankrupts the Middle Class Edited by Katherine Porter 2012 Making the Transition: Education and Labor Market Entry in Central and Eastern Europe Edited by Irena Kogan, Clemens Noelke, and Michael Gebel 2011 Class and Power in the New Deal: Corporate Moderates, Southern Democrats, and the Liberal-Labor Coalition By G. William Domhoff and Michael J. Webber 2011 Social Class and Changing Families in an Unequal America Edited by Marcia J. Carlson and Paula England 2011 Dividing the Domestic: Men, Women, and Household Work in Cross-National Perspective Edited by Judith Treas and Sonja Drobnicˇ 2010

Gendered Trajectories: Women, Work, and Social Change in Japan and Taiwan By Wei-hsin Yu 2009 Creating Wealth and Poverty in Postsocialist China Edited by Deborah S. Davis and Wang Feng 2008 Shifting Ethnic Boundaries and Inequality in Israel: Or, How the Polish Peddler Became a German Intellectual By Aziza Khazzoom 2008 Boundaries and Categories: Rising Inequality in Post-Socialist Urban China By Wang Feng 2008 Stratification in Higher Education: A Comparative Study Edited by Yossi Shavit, Richard Arum, and Adam Gamoran 2007 The Political Sociology of the Welfare State: Institutions, Social Cleavages, and Orientations Edited by Stefan Svallfors 2007 On Sociology, Second Edition Volume One: Critique and Program Volume Two: Illustration and Retrospect By John H. Goldthorpe 2007 After the Fall of the Wall: Life Courses in the Transformation of East Germany Edited by Martin Diewald, Anne Goedicke, and Karl Ulrich Mayer 2006