View Frontmatter as PDF (315KB)
PARTISAN FAMILIES
People decide about political parties by taking into account the preferences, values, expectations, and perceptions of their family, friends, colleagues, and neighbors. As most persons live with others, members of their households influence each other’s political decisions. How and what they think about politics and what they do are the outcomes of social processes.
Applying varied statistical models to data from extensive German and British household surveys, this book shows that wives and husbands influence each other; young adults influence their parents, especially their mothers. Wives and mothers sit at the center of households: their partisanship influences the partisanship of everyone else, and the others affect them.
Politics in households interacts with competition among the political parties to sustain bounded partisanship. People ignore one of the major parties and vary their preference of its major rival over time. Election campaigns reinforce these choices.
Alan S. Zuckerman is a professor of political science at Brown University and a research professor at the DIW (German Institute of Economic Research). He has also served as a visitor at the University of Essex (British Academy Fellow); Tel-Aviv University (Fulbright Professor and Visiting Professor); the Hebrew University of Jerusalem (Lady Davis Visiting Professor); the University of Pisa (Fulbright Professor); the Istituto di Scienze Umane, Florence, Italy (Visiting Professor); and Stanford University (visiting associate). Professor Zuckerman has also published many books on diverse topics in political and social science and many articles in the leading journals of these fields.
Josip Dasović is an adjunct lecturer in political science at the University of Richmond. He is completing his Ph.D. in political science at Brown University. Prior to graduate school, he worked extensively in the former Yugoslavia with, among other organizations, the Croatian Helsinki Committee for Human Rights, at which he was a human rights activist and the organization’s liaison to the International Criminal Court for the Former Yugoslavia (ICTY).
Jennifer Fitzgerald (Ph.D. Brown University, M.A. University of Chicago, B.A. Indiana University) is assistant professor of political science at the University of Colorado at Boulder. She has also been a visiting researcher at the DIW in Berlin, Germany, and the Centre d’étude et de recherche sur la vie locale (CERVL) in Bordeaux, France. Her research interests center around the influence of social context on political behavior and the social dimensions of immigration politics in Western Europe.
CAMBRIDGE STUDIES IN PUBLIC OPINION
AND POLITICAL PSYCHOLOGY
Series Editors
Dennis Chong, Northwestern University
James H. Kuklinski, University of Illinois, Urbana-Champaign
Cambridge Studies in Public Opinion and Political Psychology publishes innovative research from a variety of theoretical and methodological perspectives on the mass public foundations of politics and society. Research in the series focuses on the origins and influence of mass opinion, the dynamics of information and deliberation, and the emotional, normative, and instrumental bases of political choice. In addition to examining psychological processes, the series explores the organization of groups, the association between individual and collective preferences, and the impact of institutions on beliefs and behavior.
Cambridge Studies in Public Opinion and Political Psychology is dedicated to furthering theoretical and empirical research on the relationship between the political system and the attitudes and actions of citizens.
Books in the series are listed on the page following the Index.
PARTISAN FAMILIES
The Social Logic of Bounded Partisanship
in Germany and Britain
ALAN S. ZUCKERMAN
Brown University
JOSIP DASOVIĆ
University of Richmond
JENNIFER FITZGERALD
University of Colorado at Boulder
CAMBRIDGE UNIVERSITY PRESS
Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo, Delhi
Cambridge University Press
32 Avenue of the Americas, New York, NY 10013–2473, USA
www.cambridge.org
Information on this title: www.cambridge.org/9780521874403
© Alan S. Zuckerman, Josip Dasović, Jennifer Fitzgerald 2007
This publication is in copyright. Subject to statutory exception
and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without
the written permission of Cambridge University Press.
First published 2007
Printed in the United States of America
A catalog record for this publication is available from the British Library.
Library of Congress Cataloging in Publication Data
Zuckerman, Alan S., 1945–
Partisan families : the social logic of bounded partisanship in Germany and Britain /
Alan S. Zuckerman, Josip Dasović, Jennifer Fitzgerald.
p. cm – (Cambridge studies in public opinion and political psychology)
Includes bibliographical references and index.
ISBN 978-0-521-87440-3 (hardback) – ISBN 978-0-521-69718-7 (pbk.)
1. Political participation – Social aspects – Germany. 2. Political participation – Social
aspects – Great Britain. 3. Political socialization –Germany. 4. Political socialization –
Great Britain. 5. Political sociology. I. Dasović Josip.
II. Fitzgerald, Jennifer, 1972– III. Title. IV. Series.
JN3971.A91Z828 2007
306.20941 – dc22 2006036503
ISBN 978-0-521-87440-3 hardback
ISBN 978-0-521-69718-7 paperback
Cambridge University Press has no responsibility for
the persistence or accuracy of URLs for external or
third-party Internet Web sites referred to in this publication
and does not guarantee that any content on such
Web sites is, or will remain, accurate or appropriate.
Contents
| List of Figures and Tables | page ix | ||
| Acknowledgments | xiii | ||
| Preface: The Theoretical Approach, the Question, and the Data | xv | ||
| 1. | The Social Logic of Partisanship: A Theoretical Excursion | 1 | |
| 2. | Bounded Partisanship in Germany and Britain | 32 | |
| 3. | A Multivariate Analysis of Partisan Support, Preference, and Constancy | 47 | |
| 4. | Bounded Partisanship in Intimate Social Units: Husbands, Wives, and Domestic Partners | 71 | |
| 5. | Bounded Partisanship in Intimate Social Units: Parents and Children | 91 | |
| 6. | Partisan Constancy and Partisan Families: Turnout and Vote Choice in Recent British Elections | 123 | |
| Conclusion: Family Ties, Bounded Partisanship, and Party Politics in Established Democracies | 142 | ||
| Appendix | 161 | ||
| References | 167 | ||
| Index | 183 | ||
List of Figures and Tables
FIGURES
| 2.1 Aggregate Party Preferences in Germany | 34 | ||
| 2.2 Aggregate Party Preferences among Germans in All Waves | 35 | ||
| 2.3 Aggregate Party Preferences in Britain | 36 | ||
| 2.4 Aggregate Party Preferences among Britons in All Waves | 37 | ||
| 2.5 Aggregate Levels of Political Interest in Germany | 39 | ||
| 2.6 Aggregate Levels of Political Interest in Britain | 39 | ||
| 2.7 Partisan Constancy in Germany | 43 | ||
| 2.8 Partisan Constancy in Britain | 43 | ||
| 3.1 Household Influence on Partisan Choice in Germany | 56 | ||
| 3.2 Household Influence on Partisan Choice in Britain | 62 | ||
| 3.3 Household Influence on Partisan Constancy in Germany | 67 | ||
| 3.4 Household Influence on Partisan Constancy in Britain | 68 | ||
| 4.1 Reciprocal Effects of Partisan Choice in Germany | 77 | ||
| 4.2 Reciprocal Effects of Partisan Choice in Britain | 79 | ||
| 4.3 Joint Partisan Constancy in Germany | 82 | ||
| 4.4 Joint Partisan Constancy in Britain | 82 | ||
| 4.5 Postestimation Probabilities: Years of Marriage and Partisan Agreement in Germany | 86 | ||
| 4.6 Postestimation Probabilities: Years of Marriage and Partisan Agreement in Britain | 88 | ||
| 5.1 Trends in Aggregate Partisan Support in German Households | 96 | ||
| 5.2 Trends in Aggregate Partisan Support in British Households | 97 | ||
| 5.3 Trends in Partisan Support in German Households (among persons in all waves) | 98 | ||
| 5.4 Trends in Partisan Support in British Households (among persons in all waves) | 99 | ||
| 5.5 Bounded Partisan Choice among Young Persons in Germany | 99 | ||
| 5.6 Bounded Partisan Choice among Young Persons in Britain | 100 | ||
| 5.7 Postestimation Probabilities of Child’s Partisan Support in Germany | 105 | ||
| 5.8 Postestimation Probabilities of Child’s SPD and CDU/CSU Choice | 106 | ||
| 5.9 Postestimation Probabilities of Child’s Partisan Support in Britain | 109 | ||
| 5.10 Postestimation Probabilities of Child’s Labour and Conservative Choice | 109 | ||
| 5.11 Political Influence in Households | 121 | ||
| 6.1 Parental Influence on Child’s Labour Vote | 133 | ||
| 6.2 Influence of Husband and Child on Mother’s Labour Vote | 134 | ||
| 6.3 Influence of Wife and Child on Father’s Labour Vote | 134 | ||
| 6.4 Parental Influence on Child’s Conservative Vote | 137 | ||
| 6.5 Influence of Husband and Child on Mother’s Conservative Vote | 137 | ||
| 6.6 Influence of Wife and Child on Father’s Conservative Vote | 138 |
TABLES
| 2.1 Partisan Change between any Two Adjacent Years | 41 | ||
| 2.2 The Electorates Distributed by Partisan Choice between Two Points in Time | 42 | ||
| 3.1 Partisan Choice for SPD – Heckman Probit Selection Model | 52 | ||
| 3.2 Partisan Choice for CDU/CSU – Heckman Probit Selection Model | 54 | ||
| 3.3 Partisan Choice for Labour – Heckman Probit Selection Model | 58 | ||
| 3.4 Partisan Choice for Conservatives – Heckman Probit Selection Model | 60 | ||
| 3.5 Partisan Choice and Constancy in Germany (1985–2001) – Zero-Inflated Negative Binomial Regression Model | 64 | ||
| 3.6 Partisan Choice and Constancy in Britain (1991–2001) – Zero-Inflated Negative Binomial Regression Model | 66 | ||
| 4.1 Reciprocal Effects on SPD Choice – Instrumental Variable Probit Model | 75 | ||
| 4.2 Reciprocal Effects on CDU/CSU Choice – Instrumental Variable Probit Model | 76 | ||
| 4.3 Reciprocal Effects on Labour Choice – Instrumental Variable Probit Model | 78 | ||
| 4.4 Reciprocal Effects on Conservative Choice – Instrumental Variable Probit Model | 79 | ||
| 4.5 Partisan Agreement between Husband and Wife for the SPD – Time-Series Logit Model | 84 | ||
| 4.6 Partisan Agreement between Husband and Wife for the CDU/CSU – Time-Series Logit Model | 85 | ||
| 4.7 Partisan Agreement between Husband and Wife for Labour – Time-Series Logit Model | 87 | ||
| 4.8 Partisan Agreement between Husband and Wife for Conservatives – Time-Series Logit Model | 87 | ||
| 5.1 Agreement among Mothers, Fathers, and Children in Germany – Polychoric Correlation Coefficients | 93 | ||
| 5.2 Agreement among Mothers, Fathers, and Children in Britain – Polychoric Correlation Coefficients | 93 | ||
| 5.3 Partisan Choice across Generations in German and British Households | 95 | ||
| 5.4 Choice of SPD among Young Persons in Germany – Heckman Probit Selection Model | 103 | ||
| 5.5 Choice of CDU/CSU among Young Persons in Germany – Heckman Probit Selection Model | 104 | ||
| 5.6 Choice of Labour among Young Persons in Britain – Heckman Probit Selection Model | 107 | ||
| 5.7 Choice of the Conservatives among Young Persons in Britain – Heckman Probit Selection Model | 108 | ||
| 5.8 Agreement on SPD Choice in Germany among Mothers, Fathers, and Children – Three-Stage Regression Model | 112 | ||
| 5.9 Agreement on CDU/CSU Choice in Germany among Mothers, Fathers, and Children – Three-Stage Regression Model | 114 | ||
| 5.10 Agreement on Labour Choice in Britain among Mothers, Fathers, and Children – Three-Stage Regression Model | 117 | ||
| 5.11 Agreement on Conservative Choice in Britain among Mothers, Fathers, and Children – Three-Stage Regression Model | 118 | ||
| 5.12 Predicted Probabilities of Partisan Choice | 119 | ||
| 6.1 Labour Vote – Heckman Probit Selection Model | 128 | ||
| 6.2 Conservative Vote – Heckman Probit Selection Model | 129 | ||
| 6.3 Family Effects on Labor Vote for Mothers, Fathers, and Children – Three-Stage Regression Model | 132 | ||
| 6.4 Family Effects on Conservative Vote for Mothers, Fathers, and Children – Three-Stage Regression Model | 136 | ||
| 6.5 Participation in British General Elections – Logit Model | 140 | ||
| 6.6 Frequency of Participation in British Local Elections – Ordered Logit Model | 140 |
Acknowledgments
Many persons assisted our efforts to complete this research. It is a pleasure and a privilege to be able to thank them publicly.
Alan S. Zuckerman is grateful to DIW (German Institute for Economic Research) for an appointment as Research Professor, which facilitated two research trips to Berlin and fostered his collaboration on a related project with Martin Kroh. These visits enabled him to more deeply understand the tortured history and the routine present that make up Germany. Jennifer Fitzgerald too benefited from a research stay at DIW, and she would like to thank Deborah Bowen, Anita Drever, Bettina Isengard, Gundi Kniess, and Thorsten Schneider. We also want to thank other colleagues at DIW who are involved with the German Socio-Economic Panel Study (GSOEP), especially Gert Wagner, the director, Christina Kurka, the administrator, and Martin Kroh, a first-rate political scientist.
Zuckerman also takes pleasure in thanking the British Academy for a research fellowship that allowed him to work at the Institute for Social and Economic Research at the University Essex. This grant enabled him to benefit from Malcolm Brynin’s deep familiarity with the British Household Panel Survey (BHPS). For assistance with the British data, we are also pleased to thank Jonathan Gershuny, director, and John Brice, who always answered our long-distance and sometimes panicky messages.
Our colleagues at GSOEP and BHPS bear no responsibility for how we have used the data that they have kindly made available to us.
Readers of this book encounter many numbers, models, and graphs. When these presentations succeed, it’s because of the help that we received from Scott Allard, Andrew Foster, Rachel Friedberg, Joseph Harkness, James Heckman, Michael Herron, Joseph Hogan, Gary King, Martin Kroh, Scott Long, Michael White, and Ezra Zuckerman. We are honored to be able to thank them. When our analyses do not measure up, it’s our fault.
This book began as a graduate seminar in the department of political science, Brown University. Over time, including many long days peering together at the computer in Zuckerman’s office, we developed into a team. Each of our contributions was necessary to turn the hope of successful research into a book in which we take delight and pride. During that time too, we benefited from the assistance of our colleagues and friends in the department. In addition to those already mentioned, we are very pleased to thank Roger Cobb, Patti Gardner, and Wendy Schiller.
We are also very happy to thank Lew Bateman, our editor at Cambridge University Press, for his efficiency and encouragement, as well as Sara Black, our copy editor, Ernie Haim, production editor, and Shelby Peak, associate production controller, for their help in bringing our work to light.
Portions of Chapter 1 were published in Alan S. Zuckerman, ed., The Social Logic Politics by Temple University Press; they are reprinted here by permission of Temple University Press.
Alan Zuckerman wants also to thank his wife Ricki, for being Ricki, and to dedicate this and so much more to her.
Josip Dasović dedicates this book to his parents, Marko and Dragica Dasović, and also to Ivana.
Jennifer Fitzgerald offers this as a small token of appreciation to her amazing parents, John and Fay.
Preface: The Theoretical Approach, the
Question, and the Data
Social relationship: The behavior of a plurality of actors insofar as, in its meaningful content, the action of each takes account of that of the others and is oriented in these terms.
Max Weber, Economy and Society I, 26
To those [guards] who do not talk to you…you dare not speak. If you are fortunate enough to have someone next to you with whom you have a common language, good for you, you’ll be able to exchange your impressions, seek counsel, let off steam, confide in him; if you don’t find anyone, your tongue dries up in a few days, and your thought with it.
Primo Levi, The Drowned and the Saved, 71
We offer a simple story. How do people decide about political parties? Much as they make other decisions. They take into account the preferences, values, expectations, and perceptions of their family, friends, colleagues, and neighbors.1,2 People affect one another, and so any one decision responds to the particular mix of views in a person’s social networks. As Max Weber, a founder of social science, taught, and as Primo Levi witnessed in the Holocaust: people live and experience their lives and their thoughts in social relationships. How and what they think about politics and what they do are the outcomes of social processes.
These simple claims entail complex analyses. Political influence, like other outcomes of social relationships, is usually reciprocal. Absent pure dominance of one member of a dyad over another, the logic that implies that A influences B also maintains that B affects A, and so on for each of the additional dyads in the social network. In order to detail the patterns of political influence, evidence needs to be applied to appropriate analytical models. Furthermore, political influence is probabilistic, not determined; people retain the possibility to go their own ways, and these relationships must be appropriately modeled as well. As we seek to account for partisan choices, we will specify the relative strength and causal flows among the relationships that we observe.
As most persons live with others, members of their households – usually husbands, wives, parents, and children – influence each other’s political decisions.3 This general statement directly implies critical points for our analysis. Frequency of interaction affects the probability of influence. Family members affect each other, in part at least, because they see each other frequently and so they send and receive many cues. Wives and husbands influence each other; it is wrong to assume that, in principle at least, one partner dominates the other. Similarly, children, especially those who are adults, influence their parents; the flow of influence need not extend solely from the older to the younger generation. By implication, therefore, political interest is not the source of influence within families. The results of our empirical analyses place wives/mothers at the center of households: their partisanship influences the partisanship of everyone else, and the others affect them; neither statement applies to husbands/fathers or children.
What do people decide about political parties?4 They decide whether or not to support any political party (partisan support), which party to name (partisan preference or choice), and they make these decisions again and again over time (support and preference constancy). Most people think about a party as an object of support or preference, and most consider whether or not to vote for its candidates; few take part in the party’s activities. In established democracies, partisanship and electoral behavior encompass much of the way that most people relate to politics. Even as people consider the political parties and cast ballots in elections, politics is not one of their daily activities. Taken in the aggregate, however, these decisions carry enormous power, conditioning the behavior of government leaders.
What kind of choice is the preference for a political party? All choices are not the same. Some are perceived to entail few if any consequences for the actor; some are not so understood. On this account, philosophers distinguish “picking,” which applies to “small” decisions, from “choosing,” which entails reasons and preferences (see Ullmann-Margalit and Morgenbesser 1977; Ullmann-Margalit 2005). The consequences of the choice may be distinguished with regard to the effect on the actor. Some choices are “big”; they are meant to change lives and may be classified as “opting” or “converting,” where “drifting” implies that the decision will not transform the individual (Ullmann-Margalit 2005). Closer to a small than a big one, partisan choice is a preference; it is a decision for a reason.
When people choose a political party, are they seeking to advance an interest? In this study, we apply the principles of bounded rationality, not optimization, to partisan choice. In chess, for example, players seek to win; that is their interest. This “macroscopic driving force” (Aumann 2005) affects decisions about strategy and tactics. Maximizing expected utility guides economic choices, where “making money” serves as a shorthand for each person’s interest. Similarly, “controlling power” may guide the analysis of the decisions of politicians. In all these cases, people decide how best to advance their interest by calculating with regard to clearly defined rules and the strategic choices of opponents. When a person announces a partisan preference or casts a ballot, there is no immediate and direct gain or loss. Selecting a party does not affect one’s ability to advance an interest the way that capturing an opponent’s queen, getting the best deal on a purchase, or landing a cabinet office does. Blais (2000) details the limited value of interest-based rational choice analyses of voting and by implication partisanship. Partisan preference involves choices about distant and abstract objects. The hope of somehow influencing government policies that might affect the actor’s life does not provide the driving force for these decisions. Instead, we apply the principles of social learning to the effort to make the best possible choice in an arena without clearly defined interests.
At any given election, partisans usually vote for their party’s candidates. This relationship is an empirical regularity, not a tautology. First, different survey questions define these concepts: one asks for self-reflection and another describes behavior, and so they differ in practice as well as principle. In the same survey, these responses may overlap. Asked again and again over time, as in the surveys that we examine, temporal and analytical distance separates the answers about the political parties and reports of voting decisions. Equally as important, we show that the constancy of partisan choice varies, and this variation influences electoral decisions. Partisanship and electoral behavior are distinct clusters of concepts.
We detail the elements of partisanship, providing evidence of choices taken by persons in large longitudinal sample surveys in Germany (1985–2001) and Britain (1991–2001). Examining partisanship in only one year without regard to other years provides a limited and, we think, distorted picture of partisan behavior. Our German and British panel data enable us to address the more complex issue of choices made at different points in time.
The citizens of Germany and Great Britain are bounded partisans. Over time, most of them effectively turn their back on one of the major parties by never choosing to support it, and they vary their choice of the other large party – sometimes choosing it and sometimes not. They are generally more constant with regard to the party that they do not name than with their preferred party. Also, the responses of very few persons take them from party to party. People construct a choice set from the full list of political parties, eliminating one of the major parties and establishing the other as a possible object of support.
When we claim to explain partisanship, what do we mean? How do we account for the elements of partisanship and electoral behavior? First, drawing on the social logic of politics, we propose a set of explanatory mechanisms that emphasize the centrality of family and household interactions in partisan decisions. In the first chapter, we provide the intellectual history of these social mechanisms by examining the work of anthropologists, economists, decision theorists, social psychologists, and sociologists, as well as political scientists. Where appropriate and possible, we contrast our perspective with the expectations of other more widely used approaches: the claim of classical rational choice that people support political parties that are in line with their personal interests; the view that partisan choice is a form of social identification in which people develop psychological attachments to political parties; and the hypothesis that partisan support derives from attachments to a social class or ethnic group and the political party with which it is associated. We also test our claims against the processes and events that characterize Germany and Britain, our two cases for empirical analysis. We control for the effects of a secular decline in partisanship, which has characterized European electorates over the past two decades, and for particular events, such as recurrent elections, German Reunification, and the rise of Labour to power in Britain. We explain by justifying and testing a theory of the social logic of partisanship.
Our empirical analyses that follow revolve around probabilities. They specify the presence and strength of the relationships between predictor or explanatory variables and outcome or dependent variables. They compare the results to the mean probability in the sample for each of the dependent variables in our different models: partisan support, choice, constancy, turnout, and vote choice. They examine these in the context of the absolute values as well: the probability of naming or not naming a party, of agreeing with someone else’s partisan choice, and so forth. We detail the impact of others in the household, net of the effects of other predictor variables, on the probability that a person is above or below the mean probability of the particular outcome.
As a result, a quick glance ahead to the pages of our book finds many tables and graphs, as we structure the empirical analysis around the results of statistical models. Naming a particular party rests on whether or not a person supports any party. Similarly, vote choice rests on the decision to go to the polls, turnout. Some of our models, therefore, analyze together partisan support and choice or join turnout and vote choice (Heckman Probit Selection models) and some link partisan choice and constancy (Zero-Inflated Negative Binomial models). When we examine the reciprocal relationships within households, we use instrumental probit and three-stage linear probability models for systems of simultaneous equations. These use instrumental variables and two-stage models, where we examine household partners, and three-stage models, where we focus on wives/mothers, husbands/fathers, and children. In turn, these models require that we examine one dependent variable at a time, and so we look only at the party named or vote choice, not partisan support or turnout. Each model has significant strengths and some weaknesses. Taken together with our theory, they offer a compelling account of the social logic of partisanship. And so, sometimes, we analyze with words, sometimes with numbers presented in tables, and sometimes with figures and graphs that help to interpret the statistical analyses.
We examine two established democracies, Germany and Britain. No matter Germany’s checkered, brutal, and tragic political history, by the middle of the 1980s, four decades after the end of the Second World War, Germans had become accustomed to democratic rule. Happily, little remains of their past to challenge the claim that on the dimensions of our analysis their political system is not much different from that of the British. Indeed, we show again and again that at least with regard to partisanship Germany and Britain are very similar.
Overflowing with data on family members but absent information on anyone else, we apply the social logic of politics to political decisions made by members of German and British households. This focus responds to the data that we have; it does not imply that only family members matter with regard to political decisions, but it does suggest that they do affect each other. We use survey responses to access both partisanship and electoral behavior, so for each concept we, like other scholars of mass political behavior, rely on what people say they think and do. Because of the importance of these surveys, the German Socio-Economic Panel Study (GSOEP) and the British Household Panel Survey (BHPS), we describe them here in the book’s Preface.
OUR DATA
Like almost all other studies of partisanship and voting, our evidence comes from nationally representative surveys. Several characteristics distinguish the ones that we use. Each covers many adjacent years, offering an extensive panel that provides information on many more years than any other survey with information on partisanship. These years are not restricted to periods with elections, and so our analysis of partisanship is not limited to moments of competition among the parties. Focusing on German and British households, these surveys also interview all persons over the age of 15 in a household. The surveys do not rely on reports of only one member of the household; they also do not ask respondents to remember the partisan preferences of their parents. Because all family members usually live in the same household, we use the terms interchangeably. GSOEP and BHPS enable us to explore the reciprocal effects of family members on each other as they decide about partisan support and choice at a single point in time and over time, and as they decide about casting ballots in elections.
In our estimation, these are the best surveys available for the study of partisanship. Even so, they have limitations. Designed primarily for the use of labor economists, demographers, and scholars of community health, they do not contain the extensive battery of political questions that one finds in election surveys. Even as the data allow us to elaborate a social logic of politics, they do not directly permit us to confront other theoretical perspectives within political science.
There are also some technical considerations. These are unbalanced panel surveys, meaning that some persons respond in many, or even all, of the years, and some answer only once. Second, the respondents are usually persons who live together, and so we have a priori reason to expect their answers to resemble each other. We address these concerns in several ways. First, where appropriate, we apply controls that produce robust standard errors, thereby accounting for the interdependence of observations within clustered responses. Some of our models control directly for autocorrelation, occasioned by time-series data. In addition, the count models allow us to pay particular attention to persons who respond in all waves. Because these people are most likely to remain in the most stable social relationships, we expect them to display particularly high rates of partisan constancy. Here, we purposefully study a subset of the population whose responses one might expect to be systematically different from the full sample. Finally and most obviously, the reciprocal models do not seek to erase the effect of household influences on partisanship and voting. To the contrary, they seek to specify these relationships. That, after all, is the point of the book.
GERMAN SOCIO-ECONOMIC PANEL STUDY
The German Socio-Economic Panel Study5 interviews a large and representative sample of the German population. Beginning in 1984 with West Germans and immigrants and adding the East German sample in the months just before Reunification, it continues its interviews each year into the foreseeable future. We look at the years 1985–2001.6 No other panel survey encompasses so many waves or so extensive a period of time; all others contain much smaller samples, and GSOEP provides a national – not just local – study of family networks. It is an exemplary panel survey, also because it follows persons who move into new households and then asks the battery of questions of all adults present there as well.7 Of particular importance for our analysis is the fact that it interviews everyone present in the household over the age of 15, without relying on the reports of only one member, such as the head of household, as do many household surveys: 12,031 persons are interviewed at least once; the mean number of years for each person in the survey is 9.6. Combining the number surveyed each year with the number of waves produces 115,372 person-years. In turn, 2,997 respond in each and every wave. Because of the large sample, it is also possible to aggregate the responses into descriptions of the German states, the Bundesländer. We study Germans living in what was West Germany at the initiation of the survey (Zuckerman and Kroh 2005 extend the analysis to East Germans and immigrants, and see also Schmitt-Beck, Wieck, and Christoph 2006). Examining these Germans (and not all those in GSOEP’s samples) enables us to examine partisanship over the greatest number of years and provides a sample that is most comparable to the respondents in BHPS. GSOEP offers an extensive and detailed array of data for the exploration of micropartisanship, as well as many other elements of German social, economic, and political life.
Consider how GSOEP defines and measures partisanship. The English-language translation of the relevant questions reads: “Many people in the Federal Republic of Germany [Germany, after 1990] are inclined to a certain political party, although from time to time they vote for another political party. What about you: Are you inclined – generally speaking – to a particular party?” Those who respond “Yes” – we define as party supporters. They are then asked, “Which one?” and handed a card that lists all parties with seats in the Bundestag. This defines party preference (or choice), and we focus almost all of our attention on those who name the two dominant parties, the Social Democrats (SPD) and the Christian Democrats or the Christian Socials in Bavaria (CDU/CSU).8
The protocol repeats the set of questions in each and every one of GSOEP’s waves. Because the opening question names no parties, it avoids problems of instrumentation that are associated with the traditional measure. 9 It does not presume an answer by offering the names of the parties to those who would not otherwise be able to do so. Because it only asks the respondents to describe themselves, it does not prejudge the issue of psychological attachment. After all, one may support, prefer, or incline towards a party without identifying with it. Because GSOEP regularly taps party choices during and between electoral periods, political campaigns do not induce most of the responses, as they might in election surveys. These questions offer reliable and internally valid measures of partisan support in Germany.
The survey offers unparalleled opportunities to study the social logic of partisanship. By providing answers offered by each person in the household, the data enable us to focus on the social unit characterized by relatively high levels of interaction, trust and dependence, shared information and values, and the unit in which political discussion – both direct and verbal and indirect and nonverbal – is most likely to occur. Because family members vary in political interest, we can also see whether or not these asymmetries underpin variations in political influence within households. Offering information on the distribution of partisanship in the Bundesländer in which each respondent resides, GSOEP presents a measure of political context. In addition, questions about voluntary organizations, social contacts, and trade union and other social memberships tap social ties beyond the household. There are several indicators of social class (household income; education; occupation graded by Goldthorpe measures10); and a direct question about religious self-identification and attendance. Furthermore, there are assessments of worries about the economy. These data allow us to describe and model the partisanship of Germans.
Obviously, partisan choices occur in general political contexts as well as immediate social circumstances. Indeed, one might think that partisan preference directly responds to political events – although there is no evidence in our data to support this conjecture. Some of these political happenings are on-going, such as the declines in partisanship that began in the early 1970s and continue to apply to Germans and many other Europeans. Some are recurrent, such as national and local elections, whose campaigns remind persons of the need to select a party. Some are episodic, and in Germany these events give one reason to expect fundamental changes in partisanship.
The Germans who respond to GSOEP’s questions lived through years of political stasis, then the sudden and rapid transformation of their state, and after that a change of governing coalition. German Reunification occurred in the fall of 1991, after two years of political turmoil. While the first five years of the survey cover a time that might be described as normal politics in an established democracy, subsequent years are anything but usual. The German Democratic Republic collapsed. Its regions became part of the German Federal Republic, and West Germany became Germany. Led by Chancellor Kohl and his Christian Democratic Union/Christian Social Union, the government made far-reaching decisions: East Germans immediately became citizens of the state, and their currency was made the equivalent of the German mark. Massive infusions of economic aid moved to the new areas of reunified Germany. By 1998, the long-governing chancellor, Helmut Kohl, lost the national election to Gerhard Schroeder and the Social Democrats, and the CDU/CSU moved to the opposition. A few years later, the euro replaced the mark as the official currency when Germany and its neighbors moved more deeply into the European Union. These political events provide conflicting cues. Whereas secular declines in partisanship indicate that citizens are distancing themselves from the parties, election campaigns focus attention on the need to choose. In turn, the transformation of the German state implies diverse factors, some of which lead to preference for one or the other of the parties and some do not, even as they would seem to rouse the attention of German citizens and their political interest. There is reason to expect variations in partisan choices to respond to these political contexts. In turn, we use these events and processes as controls for the demonstration of household effects of political influence.
BRITISH HOUSEHOLD PANEL SURVEY
Modeled on GSOEP, the British Household Panel Survey (BPS)11 also surveys large numbers of people over many years, providing an on-going representative sample of Britons. We explore eleven waves of BHPS, 1991–2001, wherein 20,235 are interviewed at least once, with a mean number of interviews of five. Multiplying the number of persons by the number of years produces a data base of 100,533 person-years. In all, 4,989 Britons are interviewed in each and every wave. These data offer material for a fine-grained portrait of British social, economic, and political life, particularly as they play out within households over many years.
Although BHPS’s three questions on partisanship differ slightly from the German survey, they too offer a measure with internal validity. Here, the questions are: “Generally speaking do you think of yourself as a supporter of any one political party?” If the answer is “No,” the survey then asks, “Do you think of yourself as a little closer to one political party than to the others?” We define those who say “Yes” to either question as party supporters. They are then asked, “Which one?” and volunteer the party’s name (defining partisan choice or preference). Again, we study persons who name the major parties, Labour and the Conservatives (Tories).12 Like their counterparts in GSOEP, these questions are asked in each and every wave. They too do not name the parties in the opening question, nor do they imply any kind of psychological identification. Like GSOEP’s questions, this is a measure appropriate to the study of partisanship.13
BHPS, but not GSOEP, also includes questions on voting, and in Chapter 6 we analyze turnout and vote choice in Britain. These questions are straightforward. BHPS asks the respondents whether or not they voted in the general election of that year, and, if so, for the candidate of which party.
The questions also permit us to assess various theoretical perspectives on partisanship and electoral behavior in Britain. They provide measures of age, social class (subjective identification, Goldthorpe measures of occupation, and education), religion (identification and attendance at services), economic perceptions and concerns, membership in trade unions and various social organizations, and political interest. Here too we assess the impact of intimate social ties on partisan choice and persistence, and the large samples allow us to aggregate responses into pictures of Britain’s regions.
The political context of partisan choice in Britain differs dramatically from that of Germany. Even so and again, we do not find evidence that particular political events influence partisan preferences. Nothing transforms the British state; there is no massive infusion of new citizens; the government does not provide extraordinary resources to a particular region of the country, and though public discussion swirls about the topic, the euro does not replace the British pound as the official currency. There is a point of similarity. In Britain the large party of the political left replaced a long-serving conservative government, as Tony Blair and the Labour Party removed John Major from the prime minister’s office and turned the Conservative Party into an ineffective opposition. The BHPS data enable us to examine the relationship between the macroevents and the microchoices about the political parties during these years.
Both GSOEP and BHPS offer measures appropriate to each polity, and the differences between the two sets of questions do not deny the claim that each and both present indicators of the partisanship concept. Experts in electoral politics in Germany devised a measure that omits the names of the parties in the initial question, uses the verb phrase “incline towards,” and then offers a list of the parties only to those who claim to support one. British scholars tap partisanship by using the verbs “support” and “close to” and never list the parties. Each is appropriate to the particular context. Together, they provide contextually proper measures of the same concept: partisanship.
These exceptionally useful national surveys (even unique, if it makes sense to claim two unique instances of anything) cover many more years and many more people than other surveys that contain information on partisanship. Because each begins with and maintains a very large sample, aggregating the responses allows us to control for the effects of particular events – national elections for one example, the fall of the Berlin wall for another – on partisan decisions. Similarly, we can observe and test for the presence of a secular decline in partisanship in both countries during these decades. Because these are panel surveys, we can explore partisan constancy and switching, not just choice at a single point in time. This enables us to see whether or not people display the persistent behavior that would characterize those who identify with a political party or the movement across the parties that would be expected of partisan consumers. Looking at the surveys as an aggregated cross section and at the panel results also enables us to disentangle the causal processes that relate to time. These surveys offer a gold mine for the study of micropolitics – the political choices and actions – of citizens in Germany and Britain and by extension, we will argue, in other established democracies.
As household surveys, they provide information on the most intimate of social units, but they offer few details on other forms of social life. There are no direct questions on other social networks or discussion groups. In Chapter 6, we show a strong association between living with others and turnout, net of the effects of a host of personal characteristics that keep people at home. BHPS asks a series of questions about the quality of relations among members of households, but we have found the responses difficult to interpret and to use as operational measures of critical concepts; all told, they are not helpful in the analysis of partisanship. GSOEP offers a few questions on policy preferences, and the British survey includes a battery of such questions and also asks about political ideology. Because of potential problems of endogeneity with partisanship, we include few of them in our analyses (Anderson, Mendes, and Tverdova 2004; Erikson 2004; Evans and Anderson 2004; Johnston et al. 2005). These surveys are wonderful resources for the analysis of households; they enable us to explore the effects of these – but only these – intimate social relations on political preferences and behavior.
THE VOLUME’S PLAN
Put simply and directly, we argue that the intimate social contexts of people’s lives influence their partisanship (support, choice, and constancy), net of other factors, like political interest and more distant and abstract social locations and identifications, like social class and religion. The strength of this relationship in a particular dyad (husband and wife in Germany and mother and child in Britain, to note two examples that we explore) depends on other variables that characterize each case. It follows that turnout and electoral choice reflect the electoral behavior of a person’s spouse or partner, parents, and children and their own partisan constancy, which, as noted, also responds to the behavior of family and household members. We also demonstrate that negative results – not naming or voting for a political party – are particularly strong. The presence of two persons in a household who do not support Party A/B ensures that the other person refrains from naming that party. The presence of two who support A/B raises the probability of preferring A/B well above the mean, but it does not guarantee it. Throughout the volume, we interweave the principles of the social logic of partisanship with the analysis of bounded partisanship in Germany and Britain.
Our study begins by presenting the general theoretical framework. In the first chapter, we present the social logic of partisanship, reviewing the intellectual history of the approach as well as detailing general and specific hypotheses that we use to analyze German and British partisanship and voting behavior in Great Britain. Chapter 2, the first empirical chapter, details partisanship: partisan support, choice, constancy, and switching in our two cases. Because there have been so many studies of partisanship, we want first to demonstrate that the data taken from GSOEP and BHPS provide a picture that is sufficiently different from what we already know to sustain an analytical puzzle. Finding that most people are bounded partisans establishes this claim. Most people do not behave as if naming a party reflects a social identification; there is little evidence that persons choose their parties after a series of calculations that seek to foster and reflect their self-interest. Here, we lay out the properties of partisanship.
Then we apply models that apply the social logic of partisanship. In Chapter 3, we use Heckman Probit Selection models in order to analyze partisan support and choice and Zero-Inflated Negative Binomial models of partisan preference and constancy. These offer a preliminary argument on behalf of the utility of the social logic of politics. We develop this part of the analysis in two stages. The next three chapters delve deeply into German and British households. In Chapter 4, we explore political relationships among household partners. How similar are they? What accounts for variations in the level of political similarity? Does one partner dominate the other? This chapter shows that household cohesion with regard to politics is a variable, not a given, and that each partner influences the other. Following that, we detail the partisanship of young persons and examine the reciprocal influence of wives/mothers, husbands/fathers, and children on partisanship and electoral behavior. Chapter 5 looks at the components of partisanship alone and displays the central role of wives/mothers in the distribution of partisan choices within households. Chapter 6 extends the analysis to voting in British elections. In that chapter, we show that both partisan constancy and the electoral decisions of household members influence turnout and electoral choices. There is, we demonstrate, a social logic to partisanship and electoral behavior.
In the concluding chapter, we move our analysis beyond the two cases at hand to examine broader implications of the social logic of politics. Whereas we expect the general principles to apply to all political choices, political structures and institutions influence the particular patterns. Two dominant parties compete in both Germany and Britain, and their presence, campaign appeals, and activity necessarily influence the partisan choices and constancy of citizens. In the conclusion, we extrapolate our findings to people in similar party systems. Our ability to generalize to other established democracies is conditioned by variations in the party systems.
Printer friendly version
Email a colleague