Civic Talk: Peers, Politics, and the Future of Democracy (Social Logic of Politics)

Civic Talk

In the series The Social Logic of Politics edited by Alan Zuckerman

Also in this series: Alan S. Zuckerman, ed., The Social Logic of Politics: Personal Networks as Contexts for Political Behavior James H. Fowler and Oleg Smirnov, Mandates, Parties, and Voters: How Elections Shape the Future Simon Bornschier, Cleavage Politics and the Populist Right: The New Cultural Conflict in Western Europe

Casey A. Klofstad

Civic Talk Peers, Politics, and the Future of Democracy

TEMPLE UNIVERSITY PRESS Philadelphia

TEMPLE UNIVERSITY PRESS

Philadelphia, Pennsylvania 19122 www.temple.edu/tempress Copyright © 2011 Temple University All rights reserved Published 2011 Library of Congress Cataloging-in-Publication Data Klofstad, Casey A. (Casey Andrew), 1976– Civic talk : peers, politics, and the future of democracy / Casey A. Klofstad. p. cm. Includes bibliographical references and index. ISBN 978-1-4399-0272-1 (cloth : alk. paper) ISBN 978-1-4399-0274-5 (electronic) 1. Communication in politics—United States. 2 Communication—Political aspects—United States. 3. Political participation—United States. I. Title. JA85.2.U6K56 2010 320.97301'4—dc22

2010018213

The paper used in this publication meets the requirements of the American National Standard for Information Sciences—Permanence of Paper for Printed Library Materials, ANSI Z39.48-1992 Printed in the United States of America 2 4 6 8 9 7 5 3 1

For Rindy

Contents

Preface

ix

Acknowledgments

xi

1

Introduction

1

2

Civic Talk and Civic Participation

11

3

Does Civic Talk Cause Civic Participation?

29

4

Why Does Civic Talk Cause Civic Participation?

51

5

Do You Matter?

71

6

Do Your Peers Matter?

91

7

The Significant and Lasting Effect of Civic Talk

109

8

Peers, Politics, and the Future of Democracy

127

APPENDIX A: The Collegiate Social Network Interaction

Project (C-SNIP)

143

APPENDIX B: C-SNIP Panel Survey Questions and Variables

151

APPENDIX C: Matching Data Pre-processing

161

References

165

Index

175

Preface

uring my first two years of graduate school, I became interested in the sociological antecedents of public opinion and political behavior. At that time, the tools I had at my disposal to address this topic were publicly available survey data and a variety of methods of statistical analysis. Using these tools, I convinced myself that engaging in political discussion with friends, family, and other peers—what I call “civic talk”—led people to become more active in civil society. As soon as I had written my first paper on this subject, however, I was challenged in a research workshop with a number of indictments, such as “endogeneity,” “reciprocal causation,” and “selection bias.” In plain English, with survey data from one period in time, I was unable to show whether our peers influence our behavior or our patterns of behavior influence how we select and interact with our peers. This experience left me wondering how I could possibly study a phenomenon that is shrouded in such deep analytical complexity. It was at that point that I took a step back from my research question and appealed to the simplest tenets of the scientific method. I had a hypothesis that civic talk causes civic participation. All I needed now was a way to test the hypothesis with an experiment. Ideally, to do this I would randomly assign people to new peer groups, some of whom engaged in civic talk and others of whom did not. Random assignment would ensure that any observed effects would be attributable to civic talk and not any other factor.

D

x

Preface

Unfortunately, unless you are the producer of a reality television show, this ideal research design is not feasible. As much as we may want to, we cannot simply pick a representative sample of individuals and randomly assign them to new social circles. So, instead, I found a situation in nature that approximated my ideal: I conducted a panel study of first-year college students who had been randomly assigned their dormitory roommates. I also conducted a series of focus groups to gain a richer qualitative understanding of civic talk and its effect on civic participation. As the following pages show, these data give us an unprecedented level of confidence in the notion that civic talk has a meaningful and lasting effect on an individual’s patterns of civic participation. Given that citizens’ involvement in civil society is essential to the survival of democracy, this book shows that our peers have a critical role to play in maintaining this form of governance. Moreover, these results illustrate that extant theories of civic participation are incomplete because they focus on individual-level antecedents to the exclusion of sociological ones. Humans may not be Aristotelian political animals, but we certainly are social animals. This simple fact needs to be incorporated into our understanding and study of democracy.

Acknowledgments

number of people provided invaluable feedback and advice as I completed this project. First, I thank David Campbell. His comments at the Harvard University Political Psychology and Behavior Workshop led me to design the study presented in this book. I am also indebted to the members of my dissertation committee for their guidance and training: Barry Burden, Theda Skocpol, and Sidney Verba. My sincere thanks go to the many friends, colleagues, anonymous referees, and conference participants who have commented on this project. I specifically thank Ben Bishin (who put me in contact with Temple University Press), Merike Blofield, Louise Davidson-Schmich, Jeff Drope, Bob Huckfeldt, Kosuke Imai, Gary King, Lisa Klein, Scott McClurg, David Nickerson, Meredith Rolfe, Tony Smith, Hillel Soifer, Anand Sokhey, John Stevenson, Laura Stoker, Elizabeth Stuart, and Kathy Walsh. I also am very grateful to Mary Rouse, former dean of students and former head of the Morgridge Center for Public Service at the University of Wisconsin, Madison, for her insight and encouragement. Finally, I will always be indebted to Mike Fischer, Jack Dennis, and the University of Wisconsin Survey Center for giving me my start in this field. A number of agencies supplied monetary and research support for this project. I am deeply indebted to the University of Wisconsin Survey Center for administering the studies presented in this book. Grants from the American National Election Studies, the Harvard University Center for

A

xii

Acknowledgments

American Political Studies, and the University of Miami helped to defray the cost of this research. I am also indebted to the National Science Foundation, the Harvard University Center for American Political Studies, the Harvard University Graduate School of Arts and Sciences, and the University of Miami for financial support through various fellowships and awards. My thanks go to Louis Menand for permission to reprint a portion of his article “The Unpolitical Animal,” which appeared in the August 30, 2004, edition of the New Yorker (published by Condé Nast). I am also grateful to the following journals for permission to reprint previously published material: Social Forces (with permission from the University of North Carolina, Chapel Hill) for Casey A. Klofstad, “The Lasting Effect of Civic Talk on Civic Participation: Evidence from a Panel Study,” vol. 88 (2010): 2353–2375; Public Opinion Quarterly (with permission from Oxford University Press) for Casey A. Klofstad, Scott McClurg, and Meredith Rolfe, “Measurement of Political Discussion Networks: A Comparison of Two ‘Name Generator’ Procedures,” vol. 73 (2009): 462–483; American Politics Research (with permission from Sage Publications) for Casey A. Klofstad, “Civic Talk and Civic Participation: The Moderating Effect of Individual Predispositions,” vol. 37 (2009): 856–878; and Political Research Quarterly (with permission from Sage Publications) for Casey A. Klofstad, “Talk Leads to Recruitment: How Discussions about Politics and Current Events Increase Civic Participation,” vol. 60 (2007): 180–191. Finally, my sincere thanks go to the 2003–2004 and 2007–2008 firstyear undergraduate classes of the University of Wisconsin, Madison, as well as to my editor, Alex Holzman, and the rest of the staff at Temple University Press. Without their help, this book would not have been possible.

Civic Talk

1 Introduction

he process of selecting the forty-fourth president of the United States began in the State of Iowa. On January 3, 2008, tens of thousands of Iowans braved the below-zero cold of winter, and took time away from their jobs and families, to gather in town halls and school gymnasiums to express their preference for the nominees of the Democratic and Republican parties. At the end of the caucuses, Senator Barack Obama of Illinois emerged as the Democratic Party’s winner, with 38 percent of the vote. In doing so, Senator Obama beat out two better-known candidates: John Edwards, former senator from North Carolina and vice presidential candidate in 2004, and Hillary Clinton, senator from New York and former first lady of the United States. Senator Obama’s victory in the 2008 Iowa caucuses was an extraordinary achievement. Despite the legacy of slavery and a history of contentious race relations in the United States, a state with a white population of more than 90 percent chose to give a plurality of its votes to an African American candidate. Moreover, Obama entered the 2008 race for the presidency as a relatively outsider candidate. A poll taken in Iowa in March 2007 by the University of Iowa showed Obama to be running a distant third behind Clinton and Edwards. Even on the eve of the caucuses, a poll taken in the State of Iowa by the American Research Group

T

2

Chapter 1

showed Obama losing to Clinton—at that time the presumptive Democratic Party nominee—by 9 percentage points. Nonetheless, Obama came out on top in Iowa. On their own, the Iowa caucuses did not secure Barack Obama’s place in history as the first-ever African American president of the United States. They were, however, a significant critical juncture in the 2008 elections. Winning in Iowa gave Obama the electoral credibility, media exposure, and campaign funds that he desperately needed to continue to compete throughout the rest of the country against his more favored opponents. While we will never know what would have happened if Obama had lost the Iowa caucuses, suffice it to say that the victory propelled him toward the White House. How did this extraordinary moment in history occur? The story of Rex Boyd gives us insight into how Senator Obama overcame such steep odds in Iowa on his road to the presidency.1 Rex went with his wife, Nell, to his local caucus expecting to cast his vote for Senator Clinton. However, after chatting with people from his community about the candidates, he switched his support to Senator Obama. Had Rex lived in a state with a primary election system instead of a caucus, he probably would have cast his vote for Senator Clinton, as he had originally intended. Instead, by participating in the caucus, he was exposed to the views of the people in his community. These interactions with his fellow citizens caused Rex to switch his support to Senator Obama and, as a consequence, helped shift the course of the 2008 presidential election. Rex Boyd’s experience at the 2008 Iowa caucuses is what this book is about: the influence that citizens have on each other and on the political system when they discuss politics and current events. I call these types of discussions “civic talk.” In the chapters that follow, I use a number of unique sources of data to show how and why civic talk affects how we participate in civil society and, as such, how civic talk plays a central role in maintaining the strength of participatory democracy. In this chapter, I lay down the foundation of this argument by showing how our understanding of the relationship between civic talk and civic participation can be made clearer with new sources of evidence. I then preview the results that will be presented in subsequent chapters.

1 For

the full version of this story, see Zeleny 2008.

Introduction

3

A Research Agenda Focused on the Individual Without civic participation there is no democracy. Popular sovereignty is based on the ability of citizens to freely express their wishes to the government and, if necessary, prevent that government from committing acts of tyranny. Along with protecting popular sovereignty, civic participation facilitates democratic governance. Citizens who are active in the public sphere demand, and subsequently tend to receive, better governance from elected officials and political institutions. Because the active involvement of citizens in the processes of governance is so essential to the survival of democracy, civic participation has been heavily studied. Most of what political scientists know about whether an individual chooses to enter the public sphere is based on studies of individual-level characteristics. Arguably, one of the best examples of this brand of scholarship is Verba and colleagues’ (1995) seminal volume Voice and Equality. Of the myriad factors that might influence an individual to participate in civic activities, Verba and his colleagues focus their attention on resources (i.e., one’s available free time, income, and civic skills) and civic engagement (i.e., one’s interest in politics and current events).2 Using survey data collected in the United States, they show that individuals with more resources and higher levels of engagement with politics and current events are more likely to participate in civic activities. Resources and engagement function in this manner because they make civic participation less costly and more beneficial. For example, the more money one has, the less costly it is to make a donation to an interest group or political candidate. Or the more interested one is in politics, the more enjoyable it is to participate in that process. While political scientists’ focus on individual-level characteristics has significantly increased our understanding of how and why citizens choose to engage in civic activity, we are nonetheless left with many questions about this form of human behavior. One of the best examples of why civic participation continues to be a mysterious phenomenon is the relationship between educational attainment and civic participation. On an individual level, there is an extremely tight correlation between education and civic 2 Verba

and his colleagues do include recruitment—being asked by someone else to participate—as a third component of their model of civic participation. However, this factor is given tertiary status in their analysis. As they state it, recruitment “plays an important role, but participation can, and does, take place in the absence of specific requests for activity” (Verba et al. 1995, 270).

4

Chapter 1

90 Education (high school or more) Voter Turnout

Percentage

80

70

60

50

40 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008

FIGURE 1.1

Education and voter turnout over time in the United States

Sources: Education was measured with data from the American National Election Studies Time Series Study. Voter turnout is calculated as the number of voters divided by the voting age population. Turnout figures were compiled from the U.S. Election Assistance Commission and the U.S. Election Project.

participation; the better educated you are, the more likely you are to participate in civil society. Like income, education is a resource that makes participation in civic activities easier for the individual. For example, the better educated you are, the more you know about the political system and the easier it is for you to become involved in that system. However, while the “positive relationship between education and political participation is one of the most reliable results in empirical social science” (Lake and Huckfeldt 1998, 567), the data in Figure 1.1 show that this relationship is not nearly as clear-cut on the aggregate level. As documented by many scholars, citizens’ involvement in civil society has been in decline over the past half-century (see, e.g., Putnam 2000). As an example, Figure 1.1 shows that voter turnout in the United States has been in a steady decline since 1960 (albeit with a recent rebound in the 2000s). However, Figure 1.1 also shows that the American public’s mean level of educational attainment has increased over the same time period. If an individual’s level of education is one of the best predictors of whether he or she will participate in civil society, then why has civic participation declined over time as the level of education in the general public has increased?

Introduction

5

In this book I show that, to solve puzzles like this, we need to include social-level variables such as civic talk in our study of participatory democracy. The data I present show how, why, and under what circumstances civic talk causes individuals to participate more actively in civil society. However, to make these findings, a number of analytical pitfalls needed to be overcome.

Shifting the Agenda to Include Social-Level Variables: The Need for New Evidence Despite the fact that focusing on individual-level variables has left political scientists with an incomplete understanding of why people choose (or do not choose) to participate in civic activities, social-level variables such as civic talk have been overlooked by most political scientists. Sociological studies of participatory democracy have been relegated to the background of the field because it is difficult to determine whether the people in our social environment influence us, or whether our own patterns of behavior influence how we choose and act with the people around us (see, e.g., Klofstad 2007; Laver 2005; Nickerson 2008). For example, a number of scholars have shown that the amount of civic talk occurring in an individual’s social network correlates with his or her level of civic participation, even after controlling for a host of alternative explanations (Campbell and Wolbrecht 2006; Huckfeldt and Sprague 1991, 1995; Huckfeldt et al. 1995; Kenny 1992, 1994; Lake and Huckfeldt 1998; McClurg 2003, 2004; Mutz 2002). However, we cannot conclude that civic talk is causing civic participation with this type of evidence. One problem associated with this form of analysis is reciprocal causation; an equally plausible explanation for the relationship between civic talk and civic participation is that being civically active causes an individual to talk about politics and current events with his or her peers. Another problem is selection bias. Individuals who are more active in civic activities might consciously choose to associate with people who are more interested in talking about politics and current events. Finally, some factor that has not been accounted for could be causing people to both talk about politics and participate in civic activities (i.e., endogeneity or omitted variable bias). How can we overcome these problems and subsequently increase our understanding of participatory democracy? One way would be to randomly assign individuals to new peer groups and then track how their

6

Chapter 1

patterns of behavior change over time in response to interacting with this new set of peers. This ideal research design would allow us to determine causation because it follows the logic of a controlled experiment. First, an individual enters a new randomly assigned social setting. Second, some of these peer groups are randomly selected to engage in civic talk (i.e., to be “treated” with civic talk), while the remainder of the peer groups being studied are not allowed to engage in such discussions. Finally, the effect of being exposed to civic talk on subsequent patterns of behavior is measured by comparing the behaviors of individuals who were exposed with the behaviors of those who were not. Random assignment to treatment allows us to be confident that the outcomes of the study are actually being caused by civic talk and not any other observed or unobserved factors. I designed the Collegiate Social Network Interaction Project (C-SNIP) in line with this ideal research design. The cornerstone of the study is a panel survey I conducted on the 2003–2004 first-year class of students at the University of Wisconsin, Madison. Random assignment is incorporated into the study design because the participants were assigned to their college dormitories based on a lottery. Participants in the C-SNIP Panel Survey initially completed two survey questionnaires over the course of the 2003–2004 school year—one at the beginning of the school year before they engaged in civic talk with their randomly assigned roommate, and a second at the end of the school year after they engaged in civic talk. During the first wave of the study, each student was asked about his or her patterns of civic participation during high school. During the second wave of the study, students were asked about their civic activities in college, as well as about their randomly assigned roommate. These students also completed a follow-up survey during their fourth (and likely final) year of college in 2007. While most of the evidence I present in this book comes from the C-SNIP Panel Survey, those results are verified with data that I gathered through a series of focus groups conducted on the 2007–2008 first-year class at the University of Wisconsin, Madison. These students, like the 2003– 2004 cohort of students who participated in the C-SNIP Panel Survey, were also randomly assigned to their first-year dormitory roommates.

Overview of the Book The remainder of this book proceeds as follows. Chapter 2 begins my argument for the need to include civic talk in the civic participation research

Introduction

7

agenda through an examination of the extant literature. The conclusion reached from this discussion is that new evidence is needed to substantiate and explain the causal relationship between civic talk and civic participation. The C-SNIP Panel Survey and Focus Group Study are presented as a next-best alternative to a fully controlled laboratory experiment because they leverage random assignment to peer groups and document change in behavior over time. Chapter 3 begins with a descriptive assessment of civic talk. My data show that civic talk is less frequent than other forms of discussion, but it is still pervasive. The data also show that civic talk typically occurs in response to, and with regard to, an election, what issues are being covered in the news, and other such current events. Most important, the C-SNIP data also show that discussing politics and current events has a positive effect on participation in a variety of civic activities. For example, C-SNIP Panel Survey respondents who engaged in civic talk were 38 percent more active in voluntary civic organizations and were 7 percentage points more likely to have voted in the 2004 presidential primary. Chapter 4 is dedicated to answering why civic talk influences our patterns of civic participation. Talk, in and of itself, cannot be what is causing individuals to take action. So what is it about these conversations that leads us to participate? The answer is recruitment and engagement. When we converse about politics and current events with our peers, they explicitly ask us to get involved. In addition, engaging in civic talk correlates with enhanced interest in politics and current events. Chapters 5 and 6 dig more deeply into the C-SNIP data sets by examining the factors that might enhance or mitigate the positive relationship between civic talk and civic participation. Chapter 5 focuses on the characteristics of the individual who is engaging in civic talk. Evidence from the C-SNIP studies shows that not all individuals get the same participatory boost from discussing politics and current events. Instead, those of us who are already predisposed to participate in civil society—for example, those with prior participatory experience—get the most out of engaging in civic talk. Consequently, when asking who among us gains from engaging in civic talk, the answer is that only the politically savvy among us do. Chapter 6 shifts focus from the individual’s characteristics to the characteristics of his or her peers. One hypothesis tested is whether peer groups with greater levels of political knowledge and interest (“civic expertise”) are better equipped to motivate individuals to participate in civic activities. The C-SNIP data show that this is the case; individuals are more

8

Chapter 1

likely to participate in civic activities when they engage in civic talk with people who have civic expertise. This chapter also engages the unresolved debate on whether political disagreement among peers depresses civic participation (see, e.g., Huckfeldt et al. 2004; Mutz 2006). The C-SNIP data show that individuals who agree with their peers get more of a participatory boost out of engaging in civic talk compared with those who disagree with their peers. However, the data also show that exposure to disagreement does not depress civic participation. In conclusion, Chapter 6 shows that social intimacy increases the effect of civic talk. For example, the more you trust your peers, the more influence they have over you when engaging in civic talk. Chapter 7 answers three additional questions about the effect of civic talk on participatory democracy. First, given the extant literature’s focus on individual-level antecedents of civic participation, how does the effect of civic talk compare with the effect of one’s individual characteristics? The C-SNIP Panel Survey data show that the effect of civic talk is typically equal to or greater than the effect of individual-level antecedents of civic participation. Second, while civic talk has a significant effect on civic participation, does it have an effect on other attitudes and behaviors? The C-SNIP Panel Survey data show that civic talk has a significant effect on other civically relevant attitudes and behaviors, such as knowledge about and psychological engagement with politics. Finally, does the effect of civic talk last beyond the initial point of exposure? The third and final wave of the C-SNIP Panel Survey is used to answer this question. These data show that the effect of civic talk lasts beyond the initial point of exposure—in this case, three years into the future. Further analysis shows that the boost in civic participation initially after engaging in civic talk is the mechanism by which the effect of civic talk lasts into the future. In whole, this final set of results further illustrates the significant and lasting effect that conversations about politics and current events have on participatory democracy. Chapter 8 concludes this examination of civic talk with an assessment of the normative implications of the relationship between civic talk and civic participation, as well as a discussion of the future agenda of researchers of and practitioners in civil society. The main conclusion reached from this discussion is that, while civic talk cannot answer all of the questions researchers and practitioners face, this form of social influence has a powerful effect on participatory democracy and is thus worthy of our continued attention.

Introduction

9

Conclusion Because civic participation is so integral to the performance of democracy, the question of what causes a person to step out of his or her private life and enter the public sphere has been a subject of constant study in the social sciences. Of the numerous explanations that have been generated for why and how individuals choose to participate in the processes of democratic governance, no one theory has a monopoly on the truth. One thing we do know, however, is that the people around us have a place on this list of explanations. Just like Rex Boyd at the 2008 Iowa caucuses, we experience politics with and through the individuals in our social environment. Against this seemingly logical presumption, research on civic participation has been dominated by theories and research that negate sociological factors and instead focus on individual-level characteristics to explain civic behavior. This said, a growing number of studies assert that social context can have a meaningful impact on how we participate in civil society. More specifically, many have suggested that civic talk can cause people to participate in civic activities. However, political scientists have heavily criticized this line of argument because we have been unable to accurately measure the causal relationship between civic talk and civic participation. Consequently, the question of how civic talk and participatory democracy are related to each other has remained unresolved. In the pages that follow I provide an answer to this question.

2 Civic Talk and Civic Participation It seems at least plausible that those explicitly named by respondents as people with whom they discuss politics may be a biased selection of those with whom politics is actually discussed. —Michael Laver (2005, 934)

n this chapter I lay down the foundation of my argument by examining the existing scholarship on civic talk and civic participation. I then show how our understanding of the relationship between these two phenomena can be made clearer with new evidence. In the chapters that follow, I use a number of different sources of data to show how and why civic talk affects how we participate in civil society.

I

What Is Civic Talk? Civic talk is the informal discussion of politics and current events that occurs within a social network of peers: the friends, colleagues, family members, and other individuals who are present in our social environment. A variety of examples of this type of behavior exist, including talking about the news of the day over a family dinner, discussing the economy during a coffee break at work, chatting among patrons at a bar about the current election, and other such informal discussions. Typically, these types of interactions are not purposively sought. Instead, civic talk is usually an unintended byproduct of people going about their normal daily routine (Downs 1957; Klofstad et al. 2009; Walsh 2004). For example, a husband and wife might sit down to dinner together and end up discussing

12

Chapter 2

the issues that were covered in the news that day, or a group of friends socializing at a party might end up talking about the current election, or co-workers might end up discussing the state of the local school system during a lunch break. Because civic talk is “accidental,” it is important to underscore that it is distinct from another form of political discussion that is examined in the political science literature: deliberation. In contrast to the informal nature of civic talk, deliberation is a more formal process where citizens are brought together for the expressed purpose of formulating government policy (Barabas 2004; Conover et al. 2002; Delli Carpini et al. 2004; Mendelberg 2002; Page and Shapiro 1992). Civic talk is not as purposive or formal as deliberation. If civic talk is a natural part of daily life, how often do we engage in it? Politics and current events may not be at the top of everyone’s list of favorite discussion topics, but most of us do engage in civic talk at least from time to time. For example, respondents to the 2008–2009 American National Election Studies Panel Study were asked, “During a typical week, how many days do you talk about politics with family or friends?” About 91 percent of respondents said they engaged in such discussions at least once a week. On average, respondents reported that they engaged in civic talk about three days per week.1 If most people do engage in civic talk, at least sometimes, then who do they talk to? Because civic talk is a byproduct of going about our normal everyday lives, people do not typically seek out specific individuals with whom to talk about politics and current events. Instead, we tend to talk about these matters with the same set of people with whom we discuss other topics (Huckfeldt and Sprague 1995; Klofstad et al. 2009; McClurg 2006; Mutz 2002; Walsh 2004). Table 2.1 offers an illustration of the overlap between political and non-political discussion networks. These data come from the 1996 Indianapolis–St. Louis Study conducted by Huckfeldt and Sprague (2000). That study is uniquely tailored to answering the question of who we discuss politics and current events with because it used two different methods to collect information on civic talk. Half of the respondents in this study were randomly selected to provide information on the individuals with whom they discussed “important matters” in their life. The other half of the respondents were asked to 1 These

figures are taken from wave 9 of the study, collected in September 2009. Responses to this question collected in waves 10 (October 2009) and 11 (November 2009) yield comparable results.

Civic Talk and Civic Participation

TABLE 2.1

13

Composition of political and core discussion networks

Political discussion network

Spouses (%) Other family members (%) Coworkers (%) “Close” friends (%)

13 25 23 66

Subset of “important matters” “Important matters” network that discussion network discusses politics

16 30 15 73

17 30 15 74

Sources: 1996 Indianapolis–St. Louis Study (Huckfeldt and Sprague 2000). This table is a partial reproduction of Table 2 in Klofstad et al. (2009).

provide information on the individuals with whom they specifically “discuss politics.” 2 The data presented in Table 2.1 show that political discussion networks consist of more or less the same people with whom we discuss other matters. The data in the first two columns of the table suggest that “important matters” discussion networks are, perhaps, a bit more socially intimate. Political discussants are less likely to be spouses (t = −2.45, p = .01) or family members (t = −3.20, p < .01), and more likely to be coworkers (t = 6.63, p < .01) when compared with “important matters” discussants. And while members of the political discussion network are as likely as “important matters” discussion partners to be considered “friends,” they are less likely to be considered “close friends” (t = −4.36, p < .01). However, while statistically significant, most of these differences are not substantively large. Moreover, comparison of the second and third columns of Table 2.1 shows that the individuals we choose to discuss politics with in our core “important matters” social networks look like the rest of the individuals in that network. Otherwise stated, in going about our daily lives we tend to engage in civic talk with the same people with whom we discuss other topics.

What Is Civic Participation? The dependent variable of interest in this book is civic participation. “Civic participation” is a term that is used frequently in academic writings and the popular press these days. This is especially the case because many 2 For

a detailed discussion of these two methods of soliciting data on civic talk, see Klofstad et al. 2009.

14

Chapter 2

The State

Political Participation (e.g., voting, contacting elected officials, etc. …)

Civil Society Non-Political Participation

Private Life

(e.g., participating in voluntary civic organizations)

FIGURE 2.1 What is civic participation?

observers feel that we are in an era of civic disengagement in the United States (see, e.g., Putnam 2000). However, while “civic participation” has become a common term in everyday language, it is important to clarify what this phenomenon is and why it is important for our understanding of how democracy works. Simply stated, civic participation is activity that draws individuals out of their private lives and into civil society. The diagram in Figure 2.1 presents this definition graphically. The left-hand side of the figure represents a simple society, showing government at the top, and each individual’s private life—or the “private sphere”—at the bottom. In between the state and private life we find civil society, or the “public sphere.” Civil society is the space where citizens are able to step out of their private lives and associate with one another. This space is both literal, as in the local town square where citizens congregate, and figurative, as in the rights to assemble and speak freely while you are in the town square. The upward-facing arrow in Figure 2.1 illustrates the ways in which civil society upholds popular sovereignty. The arrow symbolizes the ability of citizens to articulate their views to the government by voting, engaging in protests, contacting their legislators, and other such activities. The upward-pointing arrow also symbolizes the fact that participation in public life gives citizens the means to push back against the government to prevent tyranny. In other words, civil society serves as both a bridge and a barrier between the state and the private sphere (Foley and Edwards 1996; Gellner 1995; Hall 1995).3 3 It

is worth noting that civil society is a necessary, although not sufficient, condition for democracy. For example, civil society was extremely vibrant in the German Weimar Republic. However, because the state was unresponsive to the people during this time period, the

Civic Talk and Civic Participation

15

In addition to protecting popular sovereignty, civic participation is important because it facilitates democratic governance, symbolized by the downward-pointing arrow in Figure 2.1 (Mayhew 1974; Putnam 1993, 2000). For example, Putnam (1993) shows that, despite having similar institutional structures, regional governments in the north of Italy provide better services to their citizens than do those in the south. Even after accounting for a host of alternative explanations, Putnam finds that this variation in governmental performance can be explained by the fact that northern Italians are more active in civil society than are their neighbors to the south (what he calls the strength of the “civic community” or “social capital”). In a more recent study, Putnam (2000) also makes the same claim about the United States. Civil society facilitates democratic governance because civically active citizens demand, and thus tend to get, better policies from the state. Moreover, organizations within civil society can aid in the development and implementation of policy solutions—for example, a soup kitchen can help the state fight hunger and homelessness, an after-school program can help the government combat juvenile delinquency, and the like. Many different types of behavior fit under this broad definition of civic participation as participation in civil society. The right-hand side of Figure 2.1 splits this large set of activities into two categories: political and nonpolitical civic participation. Under this rubric, civic activities are distinguished based on whether the activity in question directly influences the democratic governing process (Putnam 2000; Tocqueville 2000; Verba et al. 1995; Zukin et al. 2006). Political civic participation is activity that is aimed at directly influencing the government (see, e.g., Brady 1999; Putnam 2000; Verba et al. 1995). Examples of political civic activities include voting, participating in parties and interest groups, contacting elected officials about important issues, marching or protesting, and working on political campaigns. On the other side of the civic participation coin, we find non-political civic activities. These are activities that pull individuals out of their private lives and into the public sphere but that have no intentional influence on the processes of democratic governance. In practice, non-political civic activity is conceived of and measured as participation in voluntary membership organizations (Putnam 2000; Skocpol 2003; Tocqueville 2000; Verba et al. 1995). Examples include professional associations, philanthropic door was left open for the Nazis to gain control of the country (Berman 1997). Moreover, the preferences of individuals that choose to participate in civil society are not always representative of those of the wider public (see, e.g., Putnam 2000; Verba et al. 1995).

16

Chapter 2

organizations, civic leadership groups such as the Elks and the Knights of Columbus, education organizations such as Parent–Teacher Organizations and Parent–Teacher Associations, religious groups, and the like. In thinking about the distinction between political and non-political civic activities it is important to note that non-political civic activities could have unintended consequences on the governing process. For example, by volunteering in a soup kitchen an individual could have an effect on how the government addresses issues such as homelessness and unemployment. Or by attending a meeting of a religious organization, an individual could be exposed to requests to support a political candidate or party. These examples highlight the fact that the dividing line between political and non-political civic activities is not perfectly definitive. This is especially the case in the United States, where, compared with other industrialized democracies, government is more limited and civil society is more vibrant (Tocqueville 2000; Verba et al. 1995). This said, some method of classifying and distinguishing between the vast numbers of civic acts is needed, and the political–non-political taxonomy serves this purpose. Moreover, as subsequent chapters will show, civic talk can have different effects on how active individuals choose to be in political and non-political civic activities.

A Tautology? Based on these definitions of “civic talk” and “civic participation,” a question some will have is whether treating civic talk and civic participation as distinct variables is tautological. More specifically, one might argue that informal discussion among peers about politics and current events is itself an act of civic participation. However, while civic talk and civic participation are closely related—this is the central theme of this book—they are distinct concepts. With reference to Figure 2.1, civic talk is defined and measured as informal discussions that occur in the private sphere. In contrast, civic participation is activity that occurs in the public sphere. The existing body of political science scholarship supports this distinction. For example, through an extensive survey of existing scholarship, Brady (1999, 737; emphasis in the original) concludes that political participation “requires action by ordinary citizens directed towards influencing some political outcomes.” Thus, since “political discussion is not an activity aimed—directly or indirectly—at influencing the government” (Verba et al. 1995, 362), civic talk is not in itself an act of civic participation.

Civic Talk and Civic Participation

17

Social-Level Antecedents of Civic Participation Now that the independent and dependent variables of interest in this study have been defined, we need to ask what we already know about their relationship to each other. Admittedly, this is not a new research question. Observers as far back as Alexis de Tocqueville, in his seminal study Democracy in America that was conducted in the 1830s, have noted that there is a strong relationship between how individual citizens interact with one another and how well democracy functions.4 In this same spirit, a number of different lines of contemporary social science research have addressed the effect that social-level factors have on various attitudes and behaviors. For example, works in sociology and social psychology show that individuals emulate the attitudinal and behavioral norms of their social network (Festinger et al. 1950; Latané and Wolf 1981; Michener and DeLamater 1999). Economists and sociologists have shown that one’s college roommate can influence a variety of behaviors, such as scholastic achievement (Sacerdote 2001) and binge drinking (Duncan et al. 2005). Research on households shows that people living under the same roof can influence one another to vote (Nickerson 2008). As discussed earlier, the literature on public deliberation shows that individuals become more informed about politics through the process of formulating policy options with other citizens (Barabas 2004; Delli Carpini et al. 2004; Page and Shapiro 1992; Mendelberg 2002). Works on social capital and cooperation illustrate that interacting with fellow citizens causes individuals to have a greater sense of attachment to community, which leads to more frequent participation in civic activities (Dawes et al. 1990; Putnam 2000; Sally 1995). Research on political communication, opinion formation, the mass media, and political socialization shows that the individuals around us influence how we learn about politics because civically engaged individuals provide the rest of us with information about politics (Alwin et al. 1991; Barker 1998; Dawson et al. 1977; Downs 1957; Lazarsfeld et al. 1968; Newcomb 1943; Newcomb et al. 1967; Silbiger 1977; Stimson 1990; Zaller 1992).5 4 Or,

as Tocqueville (2000, 492) stated it, “Among the laws that rule human societies there is one that seems more precise and clearer than all the others. In order that men remain civilized or become so, the art of associating must be fully developed and perfected among them.”

5 Of

special relevance in this set of citations is Theodore Newcomb’s “Bennington Studies” (Alwin et al. 1991; Newcomb 1943; Newcomb et al. 1967). Like Newcomb, I assess the socializing effect that college life has on young adults.

18

Chapter 2

With regard to research specifically on civic talk, a number of works have examined the relationship between political discussion and civic participation (Campbell and Wolbrecht 2006; Huckfeldt and Sprague 1991, 1995; Huckfeldt et al. 1995; Kenny 1992, 1994; Lake and Huckfeldt 1998; McClurg 2003, 2004; Mutz 2002). These studies suggest that civic talk has an influence on how individuals view and participate in politics. For example, using a national survey of the United States, Lake and Huckfeldt (1998) show that the amount of political discussion occurring in an individual’s social network correlates with his or her level of political participation, even after controlling for a host of alternative explanations. Beyond showing a positive relationship between civic talk and civic participation, more recent works on social networks have attempted to identify the mechanisms that allow individuals to translate discussion into action. For example, in an analysis of Huckfeldt and Sprague’s (1995) data set from South Bend, Indiana, McClurg (2003) shows that peers are an important source of information on politics and current events (also see Downs 1957). Information motivates participation because it increases civic competence (the ability to participate) and civic engagement (having an interest in participating in the first place).

Difficulties Producing Evidence of Causation Based on the large amount of research that has already been conducted on sociological antecedents of civic participation, it is fair to ask why additional study of this phenomenon is needed. While there is a growing body of literature on civic talk and other related sociological phenomena, the vast majority of political scientists have discounted this line of research. Scholars have expressed serious skepticism about the role of civic talk in encouraging civic participation because conclusive evidence of a causal relationship between these two variables has not been found. Causation has not been shown because it is difficult to determine whether the people in our social environment influence us or whether our own patterns of behavior influence how we choose and act with the people around us (Klofstad 2007; Laver 2005; Nickerson 2008).6 An example helps illustrate why it is difficult to show a causal relationship between civic talk and civic participation. Existing works show a 6 In

reviewing an edited volume dedicated to the study of social networks, one critic even went as far as to say that “few of us really know what to do about implementing rigorous models of complex political interactions with endogenous preferences” (Laver 2005, 934).

Civic Talk and Civic Participation

19

strong correlation between how much a person talks with his or her peers about politics and how active that person is in civic activities (Campbell and Wolbrecht 2006; Huckfeldt and Sprague 1991, 1995; Huckfeldt et al. 1995; Kenny 1992, 1994; Lake and Huckfeldt 1998; McClurg 2003, 2004; Mutz 2002). However, we cannot conclude that a person’s social context is causing him or her to be more civically active with this evidence. One reason is reciprocal causation; an equally plausible explanation for the relationship between talk and participation is that being civically active causes an individual to talk about politics with his or her peers. Another problem is selection bias. Individuals who are more active in civic activities might consciously choose to associate with people who are more interested in talking about politics. Finally, we also need to consider the possibility of endogeneity bias (or, alternatively worded, omitted variable bias). Some factor that we have not been able to be account for could be causing people to both talk about politics and participate in civic activities. Because of these serious analytical biases, the voluminous political science literature on civic participation has been dominated by theories that focus on individual-level explanations of civic participation. For example, of the thirty-one articles published on civic participation in the American Political Science Review between 2000 and 2009, only four studies directly examined the influence of sociological antecedents.7 Individual-level explanations of civic participation carry more weight in the field of political science because individual-level determinants of behavior are easier to study than social-level factors. Individual-level data are readily available from social surveys such as the American National Election Studies, and standard methods of statistical analysis are well suited to this type of data. In contrast, sociological factors such as civic talk have been relegated to the background of the field because definitive evidence of a causal relationship between social interactions and individual behavior has been hard to come by using existing data sources and methods of analysis.8 7 This

figure was determined by examining articles whose abstracts contained the following keywords: “civic participation,” “political participation,” or “voter.” The search was conducted with the Cambridge Journals online database, available at http://journals.cambridge .org.

8A

quintessential example of this focus on individual-level characteristics is the Michigan School of political behavior (see, e.g., Campbell et al. 1960). Research in this tradition focuses on partisan identification and other individual-level antecedents and negates the influence of sociological factors (Zuckerman 2004). In fact, the founders of the Michigan School went as far as to say, “By and large we shall consider external conditions as exogenous to our theoretical system” (Campbell et al. 1960, 27).

20

Chapter 2

Consequently, we are left with an incomplete understanding of participatory democracy.

A Solution: A Quasi-experimental Panel Study Traditionally, “two-stage” regression models are used to overcome the types of analytical biases described above. In such specifications, the independent variable of interest (in this case, engaging in civic talk) is modeled with instrumental variables that do not correlate with the outcome variable being predicted (in this case, civic participation). This form of analysis is inappropriate for assessing the relationship between political discussion and political participation, however, because it is difficult to identify variables that reliably predict how often one engages in civic talk that are not correlated with one’s level of civic participation.9 If traditional methods for overcoming analytical bias are not an appropriate way to study civic talk, then what is? As illustrated in Table 2.2, one way to get past the biases that have handicapped this line of research would be to conduct an experiment by randomly assigning one group of individuals to be exposed to civic talk (the treatment group) and another group of like individuals not to be exposed to civic talk (the control group). Random assignment to treatment allows us to be confident that the outcomes of the study are actually being caused by civic talk instead of any other observed or unobserved factors.10 Ideally, data on the study population would also be collected over multiple points in time. Panel data further reduce analytical biases if the two phenomena are collected at separated points in time (with talk measured before participation). But how can we collect data that look like this? A researcher cannot randomly pull people off of the street, force some to engage in civic talk and others not to, and see what happens as these new social networks evolve over time (at least, not very easily). Thus, to execute this research design we need to find a situation that naturally approximates it. Fortunately, such an environment exists: colleges where students are randomly assigned to dormitories. 9 However,

this type of analysis has been employed when the independent variable of interest is behavior, such as the voting choices of one’s peers, instead of discussion (see, e.g., Kenny 1992).

10 Nickerson

(2008) uses this type of experimental research design to test whether individuals living in the same household influence one another to vote. However, this study does not examine whether civic talk is the causal mechanism behind civic participation.

Civic Talk and Civic Participation

TABLE 2.2 Analytical problem

Overcoming analytical biases associated with the study of civic talk Solution

Selection bias

Endogeneity bias

Reciprocal causation

21

Random assignment to peer group

Measure patterns of behavior before and after exposure to civic talk

Explanation

The individual is no longer able to select his or her peer group. Any explanation of civic participation that is not accounted for is still orthogonal. Controlling for past patterns of behavior allows causation to be inferred if there is a relationship between past instances of civic talk and current patterns of behavior.

C-SNIP execution

Study participants were randomly assigned to their firstyear college roommate.

Patterns of civic participation were measured before and after study participants encountered their new peers at college.

An example of such an environment is the University of Wisconsin, Madison. Random assignment is incorporated into the university’s housingassignment process because students are assigned to first-year college dormitory roommates based on a lottery. Incoming first-year dormitory residents rank the sixteen dormitories in order of where they want to live.11 Students are then “randomly sorted by computer” (University of Wisconsin, Madison, Division of University Housing 2004) to determine the order in which they will be assigned to dormitories. If space is available in the student’s first housing choice at the time his or her name is reached in the randomly sorted list, the student is randomly placed in a room in that dormitory. If space is not available, an attempt is made to place the student in his or her second choice of dormitory, and so on. To gather information on this population, three surveys were administered between 2003 and 2007: the Collegiate Social Network Interaction Project (C-SNIP) Panel Survey. Survey participants initially completed two survey questionnaires: one at the beginning of the 2003–2004 school 11 While

this ranking procedure had an effect on the dormitory to which students were assigned, students were still randomly assigned to roommates within that dormitory. Moreover, fixed effects control variables for dormitory assignment, as well as a series of variables that measure why students ranked the dormitories in the way that they did, are included in the analysis.

22

Chapter 2

year, before they were affected by their college roommates, and a second at the end of the school year. During the first wave of the study, students were asked about their patterns of civic participation during high school. During the second wave of the study, students were asked how civically active they were during their first year of college, as well as about their relationship with their randomly assigned roommates. In the spring of 2007, during their fourth year of college, students who had participated in the 2003–2004 C-SNIP surveys were re-interviewed.

Additional Analytical Precision via Data Pre-processing While the process of assigning C-SNIP participants to dormitory roommates was random, these students were allowed to discuss politics and current events with their roommates as much or as little as they wished. Because of this deviation from random assignment, factors that are out of my control could affect both the treatment (the amount of civic talk to which each student was exposed) and the outcome of interest (civic participation) (Achen 1986; Dunning 2008). For example, students who had been active in voluntary civic organizations before they came to college were more likely to discuss politics and current events with their new roommates (r = .19, p < .01). Prior experience participating in voluntary civic organizations also increased subjects’ likelihood of choosing to participate in such activities during their first year of college (r = .37, p < .01). A seemingly logical way to address this feature of the data would be to add offending factors such as past patterns of civic participation to the analysis as control variables. Unfortunately, this approach is not a sufficient solution. Including variables in the analysis that are strongly related to both the independent and dependent variables being examined greatly reduces the precision of the analysis (Achen 1986). However, this feature of the C-SNIP study can be accounted for by pre-processing the data with a “matching” procedure (see, e.g., Dunning 2008; Ho et al. 2007a, 2007b, 2007c). Under this procedure, the effect of engaging in civic talk is measured by comparing the civic participation habits of survey respondents who are similar to one another except that one engaged in civic talk and the other did not. By comparing the participatory habits of similar individuals who did and did not engage in civic talk, we can be confident that any observed difference in civic participation between them is

Civic Talk and Civic Participation

23

unrelated to the factors on which the respondents were matched and, as such, is a consequence of civic talk.12 More detail on how this procedure was conducted is included in Appendix C.

Additional Qualitative Evidence While survey data allow for systematic study of political phenomena, this form of inquiry is limited by the fact that numbers alone cannot give us a full picture of what actually occurs when peers discuss politics and current events. Richer qualitative data are needed to construct a more holistic picture of how civic talk might influence participatory democracy. Therefore, the quantitative findings presented throughout this book are verified with data collected through a series of focus groups that were conducted with the 2007–2008 first-year class of students at the University of Wisconsin, Madison. This group of students is different from the students examined in the panel survey, which allows me to test the claims I am making with the survey data with another case. However, like the C-SNIP Panel Survey participants, the focus group participants were randomly assigned to their first-year dormitories under the same lottery procedure. Thus, while the qualitative and quantitative data presented in this book come from two separate cases, they are comparable cases. In total, four focus groups were conducted with eight students in each.13 Students were assigned to one of the four groups based on how civically active they had been in high school (“high” versus “low”) and how much civic talk they engaged in with their randomly assigned roommates (“high” versus “low”). Structuring the focus groups in this way allows me to examine not just whether civic talk causes civic participation, but also 12 Matching

is less precise than in a controlled experiment because the procedure does not account for unobserved differences between individuals who did and did not engage in civic talk (Arceneaux et al. 2006). However, given the extensive set of pre-treatment covariates that were used in the matching procedure (see Appendix C), it is difficult to think of any meaningful unobserved factors that are not accounted for in the analysis. Moreover, unobserved differences between individuals who did and did not engage in civic talk are likely to correlate with observed differences and thus are accounted for by proxy in the matching procedure (Stuart and Green 2008). As such, and given that a true experiment is an extremely difficult (if not impossible) research design to execute for this research question, matching (in concert with panel data and quasi-random assignment to treatment) is arguably a next-best alternative.

13 Budget

constraints prevented me from conducting more than four groups. Nonetheless, subsequent chapters show that there is striking agreement between the results of the focus groups and the panel survey.

24

Chapter 2

how the effect might vary from person to person and from peer group to peer group. For example, individuals with prior experience participating in civic activities may be better able to translate civic talk into civic participation than their less experienced cohorts. This and other related questions are addressed in Chapters 5 and 6.

Case Selection and External Validity In any scientific study, researchers need to be concerned with how representative their findings are vis-à-vis the case or cases that they have selected to study. The number of cases and the method in which they are selected determine how confident we can be in the results that are generated and in how applicable those results are to the wider world. Keeping this in mind, it is necessary to note that the main sources of evidence presented in this book come from two student populations at one university. Nonetheless, the insights we can gain from these data are of broader import to students of participatory democracy for two reasons: for their internal validity (i.e., the ability to make causal inferences) and because collegiate social networks are a “crucial” case of peer influence (i.e., if we are unable to show causation in this case, we are unlikely to find it in other cases).

The Tradeoff between Internal and External Validity Figure 2.2 illustrates a necessary tradeoff that is made in any scientific study. The figure shows a taxonomy of studies classified on two dimensions: external and internal validity. Ideally, we would like our data to be representative of the entire population in which we are interested (externally valid) while simultaneously allowing us to make definitive claims about causation (internally valid). However, this unique mix of strengths is very difficult, if nearly impossible, to create in one study. Because it is difficult to maximize both internal and external validity, we are left with the choice of maximizing either one or the other. As addressed earlier, by relying on large-scale social surveys the existing literature on civic talk has already maximized external validity. With these data, we can say with certainty that the types of people who engage in civic talk also participate in civic activities. However, with this type of evidence we have been unable to show that civic talk actually causes civic participation. In response to this persistent problem, I used a different approach to gather the evidence presented in this book: increase internal

Civic Talk and Civic Participation

25

High

Low

High

Difficult

Extant Literature

Low

EXTERNAL VALIDITY

INTERNAL VALIDITY

C-SNIP Studies

Undesirable

FIGURE 2.2

Comparing analytic strengths and weaknesses

validity to make more definitive causal claims. The quasi-experimental nature of the data presented in this book allows me to make more reliable causal inferences about the relationship between civic talk and civic participation, an undertaking that had not yet been accomplished. To restate this point using slightly different language: Those who are critical of research on social networks cannot “have it both ways.” 14 On one hand, skeptics have observed that while extant data sets on civic talk are representative of the wider public (i.e., high external validity), these data cannot be used to make causal inferences about social influence (i.e., low internal validity). In this we are in agreement; this persistent problem is what is motivated me to write this book. On the other hand, critics cannot dismiss research solely because of lower external validity when a scholar collects data that can be used to assess causation more accurately. As illustrated in Figure 2.2, the tradeoff between internal and external validity is inevitable. Given that the current goal of social network scholars is (or, at least, should be) to show whether a causal relationship exists between civic talk and civic participation, it is necessary to focus our energy on examining internally valid cases. Once causation is (or is not) established in such cases, we can then tackle issues of external validity by examining additional cases to validate, or reject, previous findings.

College Students as a “Crucial” Case of Peer Influence In addition to increasing internal validity, the case of first-year college students is also an ideal setting in which to study civic talk because it 14 I

thank Scott McClurg for this language.

26

Chapter 2

represents a “crucial” case of peer influence.15 College is such a case for two reasons. First, college is a “most likely” case of peer influence (Gerring 2001). When a young person leaves his or her family to begin life as an independent adult, peers become highly influential in his or her life (Beck 1977; Campbell et al. 1960). Consequently, we should expect to find evidence of peer effects in this environment. In other words, college is a crucial case to study because if we do not find evidence of a causal relationship between civic talk and civic participation in this environment, we are not likely to find it in other contexts, where peers may be less influential. But if the data presented in this book do show that civic talk has an effect on this collegiate population, we have reason to invest resources in investigating whether this phenomenon occurs in other cases. Possible methods for conducting such further research are discussed in Chapter 8. In a related vein, a person’s first year of college is also a crucial case because it is a “paradigmatic” case of peer influence (Gerring 2001). A paradigmatic case is one that illustrates the theoretical or conceptual importance of the phenomena being studied. For example, one would not want to study communism without examining the case of the former Soviet Union (Gerring 2001, 219). This case has come to define what communism is and thus is necessary to study when examining that form of governance. In the same vein, collegiate peers define what peer influence is because they are a central facet of the individual’s life as he or she begins adulthood. Moreover, collegiate peers illustrate the importance of peer influence because they are likely to influence the patterns of civic participation that young people carry with them through the rest of their lives. Otherwise stated, and to place this discussion of case selection in a more normative context, steady declines in civic participation over the past half-century have left many wondering whether this dangerous disengagement from public life will continue. If it does, the fate of democracy is in serious peril (Putnam 2000; but see also McDonald and Popkin 2001). Thus, it is incumbent on us to understand how the current generation of young citizens is learning (or not learning) to participate in civic activities. This knowledge will be critically important as scholars and practitioners continue to study and attempt to maintain the strength of civil society. 15 Crucial

cases are those that are, “for one reason or another, crucial to a concept or to a broader body of theory” (Gerring 2001, 219; see also Eckstein 1975).

Civic Talk and Civic Participation

27

Conclusion The questions addressed in this book are not new. Because of the critical need for citizens to be involved in the processes of democratic governance, social scientists have always been interested in studying civic participation. Moreover, there is a growing literature on the influence that civic talk within social networks has on civil society. However, the extant literature has not shown a causal relationship between civic talk and civic participation. For this reason, this line of research has been heavily discounted, and our understanding of civic participation is incomplete because it centers on individual-level antecedents of human behavior. Moreover, the inability to establish causation means that second-order questions, such as which causal mechanisms drive the civic talk effect and whether the effect varies under different circumstances, have been understudied. In response, what is new and innovative in this study is the way I have chosen to gather my data. By leveraging a situation in nature that approximates an experiment, collecting data over time, making use of data preprocessing, and verifying results from survey data with rich qualitative focus group data, I present results in the following chapters that add to our understanding of how participatory democracy works. In the next chapter, I begin this task with a descriptive examination of civic talk and civic participation. I then show that a causal link exists between these two variables.

3 Does Civic Talk Cause Civic Participation? Political discussions with roommates and floormates have allowed me to see my own political ignorance and have made me want to read [and] learn more about current events. —C-SNIP Panel Survey respondent

he experience this student had during her first year in college is a textbook example of what we would see if civic talk has a meaningful effect on how individuals choose to participate in civil society. The student came to college with a given set of characteristics and patterns of behavior. She was then placed into a new social setting in her dormitory, where the interactions she had with her randomly assigned roommate had an influence on how she looked at and participated in civil society. In other words, engaging in civic talk led the student to have what we could call a civic “awakening,” a moment that led her to become more engaged with the processes of democratic governance. Chapter 2 introduced the concept of civic talk and showed how quasiexperimental panel survey data and rich qualitative focus group data can be used to overcome analytical biases that have remained unresolved in the literature on civic participation. In this chapter, I use these two data sets to test whether the experience described above is the exception or the rule. I begin by describing how civic participation and civic talk are measured. I then test whether these two phenomena are causally related to each other. The results of the analysis show that civic talk leads individuals to participate in civic activities. These data also show that the magnitude and certainty of the civic talk effect varies based on the civic

T

30

Chapter 3

act in question. In short, the evidence shows that civic talk encourages individuals to participate in civic activities, but only in those activities in which they are interested in engaging.

Measuring Civic Talk The primary measure of civic talk used in this study is based on the C-SNIP Panel Survey question, “When you talk with your roommate, how often do you discuss politics and current events: often, sometimes, rarely, or never?” Use of self-reports is standard practice in this area of research (see, e.g., Campbell and Wolbrecht 2006; Huckfeldt and Sprague 1991, 1995; Huckfeldt et al. 1995; Kenny 1992, 1994; Lake and Huckfeldt 1998; McClurg 2003, 2004; Mutz 2002). An alternative to relying on students’ self-reports about how much civic talk occurred between roommates, however, would be to use a more exogenous measure: the report supplied by each subject’s roommate. This strategy depends on correctly identifying roommate pairs. Based on the small number of subjects who were willing to report their dormitory addresses, only eight-four roommate pairs could be identified. Unfortunately, this sample is not large enough to conduct a thorough investigation of the relationship between civic talk and civic participation. That said, this small amount of data suggests that the use of self-reports is a valid approach. An analysis of the amount of civic talk reported by these roommate pairs shows that roommates agree on how much civic talk they engaged in during their first year of college (t = −1.14, p = .16). Thus, in this population self-reports of civic talk behavior are likely to be observationally equivalent to an exogenous measure of civic talk.

Frequency of Civic Talk In line with measurements taken in adult populations (see the “What Is Civic Talk?” section in Chapter 2), civic talk is pervasive but not at the top of everyone’s list of discussion topics in the two C-SNIP study populations. On average, C-SNIP Panel Survey respondents reported that they conversed with their randomly assigned roommates somewhere between “sometimes” and “often” (an average of 2.4 on the 0–3 scale ranging from “never” to “often”). In comparison, when specifically asked how often they engaged in civic talk, C-SNIP respondents reported that they participated in these types of conversations somewhere between “rarely” and “sometimes” (an average of 1.4 on the 0–3 scale ranging from “never” to “often”).

Does Civic Talk Cause Civic Participation?

31

Participants in the C-SNIP Focus Group Study painted a similar picture of how frequently civic talk occurred. When asked to list the topics that they discussed with their roommate, politics and current events were always on the list, although never at the top of the list. Instead, classes and course work, music, television shows, celebrity gossip, sports, plans for the future, and dating were among the various discussion topics that were mentioned sooner and more frequently throughout the course of each of the focus group sessions. In addition, when asked to directly compare how often civic talk occurred relative to the discussion of other topics, the vast majority of the focus group participants reported that they discussed politics and current events with their roommate less frequently than other subjects. Through further discussion, the focus group participants expressed two reasons for why civic talk is less frequent than other types of conversations. One reason is conflict avoidance: In each of the focus groups, the desire to avoid disagreements or arguments with one’s roommate was a common explanation for why civic talk was infrequent, and sometimes actively avoided. This type of exchange was typical in all four of the focus groups: PARTICIPANT: I think my roommate has the opposite view that I

do. I don’t know, because I don’t really talk to her, but I get that impression. So I figure I will just avoid it just to save time so we don’t fight about it or, like, I don’t know, get in a disagreement. MODERATOR: That’s interesting. So it’s a way to avoid conflict? PARTICIPANT: Yeah.

Moreover, the focus group sessions revealed that some students would avoid engaging in civic talk even when they had similar political preferences to their roommates’. As one participant stated: We’re both pretty liberal. But I still disagree with him a lot, and so, like, every once in a while we’ll talk about it. But it’s different between me and, like, a friend where we can talk about it and, like, not get along and then leave and see each other again, and it wouldn’t be that big of a deal. But if you kind of get in, like, a fight with your roommate, it kind of sucks. Statements of conflict avoidance like these are not, perhaps, very surprising given that these students had an incentive to maintain positive

32

Chapter 3

relationships with the person they shared small dormitory rooms with for an entire academic year.1 The second most commonly stated reason for why civic talk is not as prevalent as other types of discussions is a lack of interest in politics, either on the part of the individual or on the part of his or her roommate. When directly asked why politics and current events were not frequently discussed, a number of participants in each of the focus groups made statements such as, “My roommate doesn’t like politics,” “It’s almost like a boring topic,” and “Politics usually gets old pretty quickly.” One participant went as far as to say, “My roommate’s more or less mystified by anyone interested in politics.” Another participant summarized his experience by saying, “I asked my roommate once if he was interested in politics or the election or anything, and he said no, so that was the extent of our political conversation.”

Subject Matters and Sources of Civic Talk Perhaps not surprisingly, the focus groups revealed that civic talk conversations typically occurred in response to current events. A variety of topics were discussed by roommates, including the Iraq war, a student who had recently been murdered near the university campus, the recent election of a student to the City Council, genocide in Darfur, global climate change, and the 2008 presidential primaries. Given that the focus group sessions were conducted in the early spring of 2008, the presidential primaries were the most common topic of conversation. The election was especially in the forefront of conversations between roommates because the State of Wisconsin had just held a presidential primary election on February 19, and each of the major candidates from both parties had held campaign rallies on or near the University of Wisconsin campus. Arguably, the most salient of these events was the large rally held by Barack Obama on February 13, the night of his victories in the “Potomac Primaries” in Virginia, Maryland, and the District of Columbia. The vast majority of the focus group participants reported that they had talked about this rally with their roommates, and many reported having attended (or having attempted to attend) the event.2 1 The

effect that disagreement has on the relationship between civic talk and civic participation is examined in Chapter 6.

2 Senator

Obama filled the university’s main indoor stadium to capacity, with more than 19,000 attendees, many of whom were students at the University of Wisconsin, Madison. His main opponent at the time, Senator Hillary Clinton, held a much smaller rally of

Does Civic Talk Cause Civic Participation?

33

Along with highlighting the specific topics discussed when engaging in civic talk, the focus groups revealed two conduits through which politics and current events were brought up during conversations between roommates. One was news media consumption. A majority of the students mentioned that civic talk would occur during normal everyday conversation in relation to what was in the news that day. Statements such as these were common in all of the focus groups: I would tend to just like read the news on CNN, and so, like, if something pops up usually that I’m opinionated about, then that’s when we starting talking. So it just kind of depends on that. Well, like, we read the paper, so usually we, like, discuss what’s in the paper. It usually doesn’t come up unless, like, there’s something [on] TV on it, and then we’re, like, “Oh, that’s kind of interesting,” and then we’ll start talking about it. The second most often mentioned sources of civic talk were the personal interests and extracurricular activities of individuals and their roommates. Statements such as these were common in each of the focus groups: We talk a lot more about social issues, too, and like I said before, [my roommate is], like, a really big environmentalist, so she always has something to say about a candidate and what they think about recycling and stuff like that. So we talk a lot about that. I always talk about Barack Obama, because I’m a huge fan. We talk a lot about [health maintenance organizations] and doctors and things because she’s going into med[ical] school now as a doctor, so I think we just relate on it. And we talk because I am really passionate about Darfur in Africa, and so we can talk about that.

Measuring Civic Participation As discussed in Chapter 2, civic participation is a form of activity that pulls an individual out of his or her private life and into civil society. This approximately 2,000 attendees at an off-campus location on February 18. None of the focus group participants mentioned that they had attended the Clinton event or discussed it with their roommates.

34

TABLE 3.1

Chapter 3

Patterns of civic participation during high school and college Mean activity level

Percentage inactive

High school

1st year of college

High school

1st year of college

Participation in voluntary civic organizations (0–21 point activity scale)

6.60

2.43

5

35

Participation in political activities (0–6 point activity scale)

1.16

.56

44

68

Civic activities

Voter turnout (2004 presidential primary) (%)

51

49

Source: C-SNIP Panel Survey. Note: As is common in studies of voter turnout, the figures are likely inflated. For example, data from the U.S. Census Bureau’s Current Population Survey show that in the 2004 presidential election, only 47 percent of citizens between eighteen and twenty-four turned out to vote (Faler 2005).

large set of activities can be broken down into two categories: political civic activity and non-political civic activity. The analyses in this chapter examine three such forms of behavior: voter turnout, participation in other types of political activities, and membership in voluntary civic organizations. A summary of how active students in the C-SNIP Panel Survey were in these activities during high school and their first year of college is presented in Table 3.1.

Voter Turnout and Participation in Other Political Activities The C-SNIP Panel Survey collected information on two forms of political participation. One measure was derived from how active each student reported being in three different activities: contacting an elected official about an issue, participating in a march or protest, and working for a political campaign at any level of government (including student government campaigns). For each type of activity, students were asked to rate how many times they had participated over the previous year: “never,” “once,” or “more than once.” Participation in these activities was coded as the sum of the three activity scales. A second measure of political participation was based on each student’s self-reported voter turnout in the 2004 presidential primary. Voting is similar to the other political activities described above in that each of

Does Civic Talk Cause Civic Participation?

35

these acts is motivated by the individual’s desire to influence the government. However, voting is treated separately in this analysis because it is the only behavior that allows citizens to have a direct voice in selecting their leaders. Moreover, despite declines in voter turnout over the past fifty years, voting differs from other political activities because it is the least costly and most widely participated in political act in the United States (Verba et al. 1995, 51).

Non-Political Civic Activity: Participation in Voluntary Membership Organizations In total, this analysis accounts for seven different types of voluntary civic group affiliations: charitable and voluntary service; leadership and civic training; groups that “take stands on political issues or current events”; partisan groups; student government; student publications such as newspapers; and speech clubs and teams (e.g., forensics, debate). For each of these types of organizations, students were asked to rate how active they were on a 0–3 point scale, ranging from “not at all active” to “very active.” Participation is coded as the total amount of organizational activity in which each student engaged (i.e., the sum of the seven 0–3-point scales). These voluntary civic organizations may from time to time engage in politically relevant activities. For example, members of the university’s student government might lobby the state legislature to provide more resources to the university, or a service organization dedicated to helping the homeless might lobby the local government to provide more shelters. Nonetheless, it is important to underscore that the act of participating in one of these voluntary civic organizations is distinct from engaging in one of the political activities described above. As discussed in Chapter 2, voluntary civic associations may be politically relevant. However, they are not classified as political activities because they do not directly influence the processes of governance (i.e., they do not involve citizens explicitly attempting to directly affect decisions made by the government or determine who is selected to run the government by influencing electoral outcomes).

Trends in Civic Participation across Activities The data in Table 3.1 show that C-SNIP Panel Survey respondents participated in political activities far less frequently than in voluntary civic organizations. Overall, 5 percent of students reported that they had not participated in any voluntary civic membership organizations during high

36

Chapter 3

school, compared with 44 percent who claimed that they had not participated in any political activities. This trend continued into the first year of college, where 35 percent of students reported not participating in any voluntary organization activities, compared with 68 percent who reported not participating in any political activities. Participants in the C-SNIP Focus Group Study provided a similar picture of the gap between political and non-political civic participation. When asked to describe the civic participation in which they had engaged during their first year of college, the vast majority of the activities mentioned by the focus group participants were non-political in nature. The one notable exception to this trend, however, was voting. Many of the focus group participants spontaneously mentioned that they had voted in the 2008 presidential primary, and the vast majority of participants reported that they had voted when specifically asked whether they had. As mentioned in a previous section, the majority of the focus group participants also reported that they had attended (or had attempted to attend) the large rally staged by Senator Barack Obama on February 13, 2008. The prevalence of voting and engagement with the 2008 primary elections makes sense, given the close proximity of the Wisconsin primary to when the focus groups were conducted and the broad appeal of the Obama campaign among young people. Engagement with the 2008 election aside, what explains the sizeable gap between political participation and participation in voluntary civic organizations in these two student populations? A likely explanation for why these students are not politically active is because they are not politically engaged. For example, in the C-SNIP Panel Survey, subjects were asked, “In general, which do you think is the better way to solve important issues facing the country, through political involvement (for example, voting, working for political candidates, and the like) or through community involvement (for example, volunteering in the community, and the like)?” Students vastly preferred community involvement to political involvement in both high school (77%) and during their first year of college (72%).3 Not surprisingly, political engagement and political participation are related to each other. Students who cited “community involvement” as their preferred mode of civic activity were less active in political activities both in high school (t = −2.07, p = .04) and during their first year of college (t = −2.03, p = .05). 3 These

figures are commensurate with national samples of college students: see, e.g., Harvard University Institute of Politics 2000.

Does Civic Talk Cause Civic Participation?

37

This said, while this student population was politically apathetic, it still adhered to civic-minded norms (for similar findings, see Zukin et al. 2006). For example, while political disengagement correlates with low levels of political participation, panel study respondents were equally likely to participate in voluntary civic organizations in high school (t = −.17, p = .87) and during the first year of college (t = −1.12, p = .27), regardless of whether they cited “political involvement” or “community involvement” as their preferred mode of civic expression. Moreover, when asked in wave 2 of the panel study, “How important do you think it is for people like you to be active and interested in politics and current events: very, somewhat, not very, not at all?” 91 percent of the respondents said that civic participation was somewhat or very important.4 In short, the C-SNIP Panel Survey respondents understood the virtues of participating in civil society. They did not, however, see politics as a desirable venue for such activity. The focus group participants offered a similar assessment of the gap between political and non-political activity. A number of students mentioned that they were less engaged with politics than with non-political matters. These two comments especially illustrate this attitude: I’m not exactly really involved in a politically affiliated group as far as, like, a party goes, but I do feel like it’s important to at least represent yourself in a number form [sic] with like humanitarian kind of things. The reason I don’t get involved in a lot of these political groups, is I think it’s, like, really bureaucratic and not that effective. And as much as I care about it, I really don’t want to, you know, walk around posting fliers or handing out stickers or [sit] in a booth at whatever event. It seems, like, really boring, really boring. In a similar vein, some students mentioned that they felt unqualified or unprepared to participate in political activities. As one student put it, “We don’t talk about political activities, because we don’t know about anything.” Or, as articulated in this exchange between the moderator and two participants: MODERATOR: And how about the more political activities? PARTICIPANT 1: I wish I could get involved in more. It’s just kind

of overwhelming at first. There’s just so much stuff going on and 4 This

question was not asked of students in the high school questionnaire.

38

Chapter 3

stuff, so I’m interested in politics and stuff and even, like, the other side of politics. I don’t know. Hopefully, next year I’ll be a little bit more involved with that. PARTICIPANT 2: I like to know what other people think about it,

but I don’t know anything about it, so I feel stupid. Low levels of political participation, engagement, and efficacy are not very surprising, given the relatively young age of the populations examined in the C-SNIP studies. As shown by the high number of politically inactive students reported in Table 3.1, these young people did not have extensive prior experience participating in political activities before coming to college.5 In contrast, the vast majority of the students in the study did have some form of prior experience participating in non-political civic activities—that is, only 5 percent reported that they had not participated in any civic organizations before coming to college, compared with 44 percent who reported that they had never participated in political activities before coming to college. Past patterns of civic behavior are meaningful because civic participation, like any other form of behavior, is habitual (Brady et al. 1999; Burns et al. 2001; Fowler 2006; Gerber et al. 2003; Plutzer 2002; Putnam 2000; Rosenstone and Hansen 1993; Verba et al. 1995). Since the students examined in this study did not gain extensive experience participating in political activities during high school, it is not surprising that they were not very active in or engaged with politics during their first year of college.

Trends in Civic Participation over Time A second notable trend in Table 3.1 is that this population of students was less civically active during the first year of college than in high school. Data from the focus groups lead to a similar conclusion. When specifically asked to compare how civically active they had been in high school with how active they had been during their first year of college, the vast majority of the focus group participants reported that they had been more active during their high school years. One explanation for this marked drop in participation between high school and the first year of college is that the first year of college is a period of significant transition in one’s life. After leaving home and family 5 This

lack of experience is sometimes legally mandated, as in the case of the right to vote being extended at age 18.

Does Civic Talk Cause Civic Participation?

39

for college, young people are forced to adjust to a new and very different lifestyle as independent adults. Consequently, participation in civic activities should be expected to become less of a priority when a person needs to spend his or her time and energy learning how to navigate a new town, make new friends, learn how to study, and the like. Existing research in the fields of higher education, psychology, and political science offer support for this hypothesis. The transition to college has been documented to be a stressful and life-changing process as the student leans how to find his or her place in a new social and academic environment (Boyer 1987; Compas et al. 1986; Cutrona 1982; Takahashi and Majima 1994; Terenzini et al. 1994). Moreover, studies have shown that civic engagement and participation drop during times of transition and relocation (Putnam 2000, 204). Data from the C-SNIP Panel Survey also offer evidence in support of this hypothesis. For example, in the first and second waves of the study respondents were asked, “How much free time have you had during the average week to participate in the types of organizations you answered questions about at the beginning of this survey: a lot of time, a moderate amount of time, very little time, or no time at all?” Survey participants reported having significantly less free time to dedicate to civic participation during their first year of college than they had in high school (t = −19.22, p < .01). Respondents were also asked, “Generally speaking, how much would you say that you know about politics and current events: a great deal, some, or not much?” Comparing the responses to this question given during the first and second waves of the survey shows that the students felt they were more informed about politics and current events during high school than they had been during the first year of college (t = 7.28, p < .01). In the first and second waves of the panel survey students were also asked, “How many days per week, on average, have you read or watched the news to learn about politics and current events?” Compared with their habits during high school, this population consumed significantly less news about politics and current events during the first year of college (t = −7.36, p < .01). Do these measures help explain the participation gap between high school and the first year of college? Logically, we would expect that people with less free time would be less civically active. Time is a resource that is requisite for participation in civil society (Putnam 2000; Verba et al. 1995). However, the results in Table 3.2 show that available free time does not correlate with how civically active these students chose to be during their first year of college. The data do show, however, that political

40

Chapter 3

Free time, political knowledge, media use, and patterns of civic participation during the first year of college (correlations)

TABLE 3.2

Civic activities

Free time

Knowledge about politics and current events

News media usage

Participation in voluntary civic organizations Participation in political activities Voter turnout (2004 presidential primary)

.05 .05 .05

.32*** .31*** .22***

.24*** .30*** .20***

Source: C-SNIP Panel Survey. ***p ≤ .01.

knowledge and news media usage are highly correlated with all three forms of civic participation. For this reason, declines in knowledge and media use between high school and the first year of college help explain declines in civic participation over this same time period. The idea that civic participation becomes less of a priority as people transition to life on their own at college was also a pervasive theme in the focus groups. Statements such as these were typical throughout the sessions: In the beginning of the year, I was just kind of stressed with school and getting used to everything here, and I didn’t really pay attention to any of that kind of stuff. I felt for the first, maybe first, semester, I was completely removed. I stopped reading the news. I was just figuring stuff out and became busy. . . . [I]n addition, . . . a large part of my political involvement was, like, you know, talking with my family at dinner. So mixing that up in a new environment [at college], it really caused me to do a lot less. At the beginning of the year when we first got here, they had a club fair, but I was just so busy meeting people that I didn’t really think of it then. While civic disengagement is a satisfying explanation for the drop in participation between high school and college, an additional explanation is the prevalence of service learning opportunities in American primary and secondary schools. These programs lower the costs of civic participation by providing prepackaged opportunities for students to become active. Moreover, the programs can raise the costs associated with not

Does Civic Talk Cause Civic Participation?

41

participating, because participation is often mandated as a requirement for graduation or course credit. Such programs became increasingly popular at the time that the C-SNIP Panel Survey participants were completing elementary and secondary school. On the federal level, these programs started receiving support when President George H. W. Bush enacted the National and Community Service Act of 1990, which created the Commission on National and Community Service (Public Law 101-610). The commission’s main duty was to create and support service learning programs. President Bill Clinton extended federal support for service learning programs by reauthorizing the National and Community Service Act under the National and Community Service Trust Act of 1993 (Public Law 103-82). This legislation merged a number of different agencies focused on civic involvement into the Corporation for National and Community Service. The act also established Learn and Serve America, a program that supports service learning programs. In response to actions taken on the federal level, the American states have increased their commitment to providing service learning opportunities to school-age children. This has especially been the case in the State of Wisconsin, where 72 percent of the C-SNIP Panel Survey participants attended high school. For example, the state’s Department of Public Instruction established the Cooperative Educational Service Agency to promote service learning programs and to act as a liaison between schools and the federal Learn and Serve America program. The state also allows school districts to “require a pupil to participate in community service activities in order to receive a high school diploma” (Wisconsin Statute 118.33). In summary, the populations studied in this book completed primary and secondary school as service learning education was growing in popularity across the United States. Moreover, the vast majority of the students had attended high school in a state where service learning is a requirement for graduation. Participating in mandatory service learning programs likely led these students to take part in an abnormally high number of activities during high school because civic activities are easier to engage in when opportunities to act are arranged for you or if sanctions are placed on those who choose not to participate (Olson 1965; Rosenstone and Hansen 1993; Verba et al. 1995). On entering college, these students no longer had built-in incentives to participate. Consequently, it is not surprising that participation levels fell after leaving high school. The focus group participants corroborated the notion that civic participation was easier to engage in during high school because it was required

42

Chapter 3

or because structured opportunities were provided. A number of participants made comments such as these: [I’m] a lot less involved now, I guess, just because in high school it was much more convenient, because it was either before or after class, so you’re already right there. It was just more of a clear-cut commitment. I miss the structure. I think it’s in a way hurt me. Like, I have a lot of free time, and I don’t know what to do with it. I think during high school I put more time into it only because I was in government class and we had hours required for that. In a related vein, a number of students in the focus groups also appeared to be using their time at college to take a break from the civic participation requirements they were obliged to fulfill in high school. I guess I kind of lost motivation for it. It used to be a mandatory part of my life to volunteer and be a part of things for school or other organizations. But now it’s, like, up to me, and I’d rather not. I was student council president and stuff like that. I actually had a class hour dedicated to filling out forms and talking to people higher up than me, going and talking to everyone in their school and stuff. And I had to organize one volunteer thing a month and stuff like that. So I was very happy to get rid of all that and have time off from that. I haven’t done as much this year, but I think it’s more like for my first year, I’m a freshman, and so I just, like, want to get settled in and not have to worry about that stuff as much. I figure I could do it later if I wanted to.

Does Civic Talk Cause Civic Participation? Survey Evidence Participation in Voluntary Civic Organizations To what extent does civic talk influence how active a person chooses to be in civil society? I begin to answer this question with a regression analysis of how active the C-SNIP Panel Survey respondents were in voluntary

Does Civic Talk Cause Civic Participation?

43

The effect of civic talk on participation in voluntary civic organizations during the first year of college (regression analysis)

TABLE 3.3

Matched (1)

Unmatched (2)

Civic talk

.81** (.30)

.86*** (.18)

Participation in voluntary civic organizations during high school

.22*** (.03)

.25*** (.02)

Constant Adjusted R2

1.67 (1.76)

.78 (1.20)

.13

.18

Source: C-SNIP Panel Survey. Model Type: Ordinary least squares (Imai et al. 2007c). Note: Dormitory assignment fixed effects were included in the analysis but are omitted from the table (none of the coefficients were statistically significant). Standard errors are in parentheses. N = 1,044. *p ≤ .10; **p ≤ .05; ***p ≤ .01.

civic organizations during their first year of college (Table 3.3). To assess the effect of the matching data pre-processing procedure, the table presents results for both the matched and unmatched data sets. Also, to increase the precision of the analysis, each model controls for how active subjects had been in voluntary civic organizations during high school, before they engaged in civic talk with their roommate (i.e., a lag of the dependent variable), as well as for how the dormitory assignment process was executed (i.e., a fixed effect for each dormitory; these coefficients are omitted from the table because none were statistically significant). Columns 1 and 2 of Table 3.3 show that subjects who engaged in civic talk with their randomly assigned roommates were more likely to participate in voluntary civic organizations during their first year of college. Substantively, the matched data set (column 1) shows that participation rates among subjects who engaged in civic talk were 38 percent higher than that of subjects who did not engage in civic talk (an increase from 2.1 to 2.9 on the voluntary organization participation scale).6 The estimated civic talk effect in the unmatched data set (column 2) is 45 percent (an increase from 2.0 to 2.9 on the voluntary organization participation 6 This

and other substantive interpretations of regression coefficients presented in this book were calculated with the setx and sim procedures in the Zelig package for R (Imai et al. 2007a, 2007b). Unless otherwise indicated, throughout this book all other factors included in the analysis are held at their means when calculating such estimates.

44

Chapter 3

Expected Change in Participation

4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0

Civic Talk

Past Participation

Comparing the effects of civic talk and past participation on participation in voluntary civic organizations during the first year of college

FIGURE 3.1

Source: C-SNIP Panel Survey Notes: The line on each bar represents the 95 percent confidence interval about the estimate. Figures are based on the matched data regression analysis in Table 3.3. The effect of civic talk is calculated as the level of participation in civic organizations estimated to have been engaged in by individuals who were exposed to civic talk, minus that of those who were not exposed, all other factors in the model held at their means. The effect of past participation is calculated by comparing the estimated levels of participation for subjects who had average levels of prior experience with those who had the maximum level of prior experience, all other factors in the model held at their means.

scale), indicating that the influence of civic talk would have been overestimated if the matching process had not been applied to this analysis.7 Moreover, closer comparison of the civic talk coefficients in columns 1 and 2 shows that the matched data set produced a larger standard error. This shows that the matched data set produced a less certain, and therefore more conservative, estimate of the civic talk effect. To understand the magnitude of the effect that civic talk has on participation in voluntary civic organizations more clearly, Figure 3.1 compares the effect of civic talk to the effect of having participated in voluntary 7 This

said, it is worth noting that the matching procedure did not drastically affect the results; the civic talk effect was statistically significant regardless of whether the matching procedure was used. Since the procedure worked correctly by significantly improving the similarity between individuals who did and did not engage in civic talk (see Appendix C), this could mean one of two things (or both). First, the use of panel data and random assignment of roommates, on their own, could have reduced the biases that are assumed to exist in studies of social networks. Alternatively, the small differences between the matched and unmatched results could indicate that biases thought to be present in social network studies are not as virulent as critics believe they are.

Does Civic Talk Cause Civic Participation?

45

civic organizations in high school before engaging in civic talk in college (the lagged dependent variable). These results illustrate that, while the effect of civic talk is statistically significant, it is not as substantively large as that of prior participatory experience. In fact, the results in Figure 3.1 show that the effect of engaging in civic talk is less than half of the effect of having above average patterns of prior participatory experience. More detailed discussion of how the effect of civic talk compares with other antecedents of civic participation is taken up in Chapter 7.

Participation in Political Activities While civic talk has a significant effect on participation in voluntary civic organizations, the results in Table 3.4 show that such conversations have a less reliable influence over whether a person participates in political activities. Comparing columns 1 and 2, we see that civic talk did not have a statistically significant effect on how politically active subjects were during their first year of college, regardless of whether the matched or unmatched data were used. What explains this difference between participation in voluntary civic organizations and participation in political activities? The less reliable influence of civic talk on political participation is likely caused by the fact that the C-SNIP population is not politically active or engaged. As documented in Table 3.1, only 35 percent of the subjects reported that they The effect of civic talk on participation in political activities during the first year of college (regression analysis)

TABLE 3.4

Matched (1)

Unmatched (2)

Civic talk among roommates

−.02 (.09)

.05 (.07)

Participation in political activities during high school

.37*** (.04)

.35*** (.02)

Constant

.55 (.76)

.50 (.46)

.24

.21

Adjusted R2 Source: C-SNIP Panel Survey. Model Type: Ordinary least squares (Imai et al. 2007c).

Note: Dormitory assignment fixed effects were included in the analysis but are omitted from the table (none of the coefficients were statistically significant). Standard errors are in parentheses. N = 1,044. *p ≤ .10; **p ≤ .05; ***p ≤ .01.

46

Chapter 3

did not participate in any voluntary civic organizations during their first year of college, while 68 percent reported that they did not participate in any political activities that year. Also recall that, with regard to their sense of political engagement, C-SNIP participants vastly preferred community involvement to political involvement as a means to solve important policy issues. Moreover, students who cited community involvement as their preferred mode of civic activity were less active in political activities, both during high school and during the first year of college. Keeping this in mind, it makes sense that civic talk has a less reliable influence on participation in political activities. If an individual is not willing to engage in an activity, influence from his or her social environment will likely have little effect on his or her behavior (Verba et al. 1995). This notion will be examined in greater detail in Chapter 5.

Voter Turnout Table 3.5 provides an analysis of the effect of civic talk on voter participation. In this analysis, participation in other political activities is used as the measure of past experience because a sizeable portion of this student population was not eligible to vote before coming to college. In contrast to the

The effect of civic talk on voter turnout during the first year of college (regression analysis)

TABLE 3.5

Matched (1)

Unmatched (2)

Civic talk among roommates

.31† (.19)

.45*** (.14)

Participation in political activities during high school

.14** (.06)

.10** (.05)

Constant

−2.11 (2.20)

−.61 (1.01)

Akaike’s information criterion (AIC)

1,283

1,407

Source: C-SNIP Panel Survey. Model Type: Logistic regression (Imai et al. 2007d). Note: Dormitory assignment fixed effects were included in the analysis but are omitted from the table (none of the coefficients were statistically significant). AIC is twice the number of parameters in the model, minus twice the value of the model’s log-likelihood. This diagnostic statistic is presented for logistic regression models throughout the book because the Zelig statistical computing package in R (Imai et al. 2007a) does not produce an R2 statistic for logistic models. Standard errors are in parentheses. N = 1,044. †p = .12; *p ≤ .10; **p ≤ .05; ***p ≤ .01.

Does Civic Talk Cause Civic Participation?

47

Expected Change in Likelihood of Voting

0.30 0.25 0.20 0.15 0.10 0.05 0

Civic Talk

Past Participation

Comparing the effects of civic talk and past participation on voter turnout during the first year of college

FIGURE 3.2

Source: C-SNIP Panel Survey Notes: The line on each bar represents the 95 percent confidence interval about the estimate. Figures are based on the matched data regression analysis in Table 3.5. The effect of civic talk is calculated as the likelihood of voter turnout estimated among individuals who were exposed to civic talk, minus that of those who were not exposed, all other factors in the model held at their means. The effect of past participation is calculated by comparing the estimated levels of participation for subjects who had average levels of prior experience with those who had the maximum level of prior experience, all other factors in the model held at their means.

relationship between civic talk and participation in other political activities, the results show that civic talk had a positive effect on voter turnout. Looking at the results from the matched data set in column 1, subjects who engaged in civic talk were 7 percentage points more likely to have voted in the 2004 presidential primary.8 The effect of civic talk on turnout increases to 11 percentage points if the unmatched data are used.9 Mirroring the analysis presented in Figure 3.1, Figure 3.2 compares the effect that civic talk had on voter participation to the effect of having engaged in political participation in high school, the lagged dependent variable used in the analysis presented in Table 3.5. Compared with the 7 percentage point boost in turnout in 2004 due to civic talk, the estimated increase in turnout due to past participatory experience is estimated to be 8 While

the significance of the civic talk coefficient falls just below the 90 percent confidence threshold, the expected increase in turnout is estimated to be significant at the 95 percent level (see Figure 3.2).

9 Again,

it is worth noting that the matching procedure did not drastically affect the results. See fn. 7 for an explanation of why this is the case.

48

Chapter 3

15 percentage points. However, after taking the uncertainty about these estimates into account, the data show that civic talk and past experience participating in political activities had the same magnitude of effect on the likelihood of turning out to vote.

Does Civic Talk Cause Civic Participation? Focus Group Evidence In line with the quantitative survey data, each of the focus group sessions showed that there is a meaningful relationship between civic talk and civic participation. However, most of the participants were not explicitly aware of this relationship. At the end of each focus group, participants were directly asked whether engaging in civic talk with their roommates had an effect on how civically active they were. The modal response was no; the most common explanation offered for why these students chose to participate in civic activities was their own internal interest and motivation. A number of focus group participants made statements such as, “My choices were because of my own choices,” “If I’m going to involved, I’ll just, like, do it myself,” and “Because what [my roommate] does seems really boring to me, I don’t want to get involved.” In a related vein, other participants stated that they felt that their roommates did not have an effect on them because they did not agree with their roommates’ views on politics and current events. As one student summarized, “We both have different opinions, and I’m going to do what I want; she’s going to do what she wants. So . . . we don’t really influence each other so much.” In thinking about why many of the focus group participants did not see the causal link between civic talk and civic participation, it is worth noting that directly asking people whether their social environment has an influence on how they behave may be a biased method for measuring the effect of civic talk. This is the case because human beings tend to attribute positive behaviors to their own qualities (and, conversely, negative behaviors to their environment). This is an example of what social psychologists refer to as the “self-serving motivational attribution bias” (see, e.g., Bernstein et al. 1979; Bradley 1978; Lau and Russell 1980; Reifenberg 1986; Ross and Fletcher 1985). For example, studies show that when asked to explain the grades they received, high-achieving students cite their hard work and intelligence, while students attribute low grades to external factors such as the difficulty of the test or “bad luck” (Bernstein et al. 1979; Reifenberg 1986). Thus, since civic participation is generally seen as a positive behavior in which to engage—recall that 91 percent of the C-SNIP

Does Civic Talk Cause Civic Participation?

49

Panel Survey respondents said that civic participation was important— it is not surprising that most of the focus group participants wanted to attribute their civic participation to their own desires rather than to the influence of their social environment. This said, while the modal focus group participant did not explicitly acknowledge the link between civic talk and civic participation, a vocal minority of focus group participants did report that discussing politics and current events with their roommates had a direct effect on how civically active they were during their first year of college. For example, participants in each of the focus groups mentioned that engaging in civic talk caused them to become more interested in politics and current events. Other students mentioned that talking about the state of the environment with their roommates caused them to become more conscious about recycling and energy conservation. Other participants reported that their roommates had reminded them to vote in the 2008 presidential primary. One student even reported that he felt he might have caused his roommate to shift his choice in the primary election from a Republican candidate to Barack Obama. In addition, before being directly asked whether civic talk with their roommates influenced how civically active they were during the first year of college, many of the participants in the focus groups indicated that this was the case without explicitly saying so. For example, many of the focus group participants reported that either they or their roommates had asked the other to get involved in a civic activity. Based on the proximity of the focus group sessions to the 2008 presidential primary election in Wisconsin, the most commonly mentioned events that roommates encouraged each other to go to or attended together were the Obama rally on February 13, 2008, and the presidential primary election. A number of participants also reported that they had attempted to persuade their roommates in some way about politics and current events. Again, based on the proximity in time of these groups to the 2008 presidential primary in the Wisconsin, most of the persuasion discussions were about the election and the candidates.

Conclusion The purpose of this chapter is to document the causal relationship between civic talk and participation in civil society. Using quantitative panel survey data, in concert with qualitative focus group data, the evidence shows that civic talk can have a causal influence on how citizens

50

Chapter 3

participate in the processes of self-governance. This is the case even after accounting for how civically active subjects were before they engaged in civic talk, arguably one of the best measures of an individual’s predilection to participate in civic activities. While civic talk has a meaningful effect on civic participation, however, it is essential to underscore that the results of this study do not suggest that sociological explanations of civic participation are a substitute for individual-level explanations. On the contrary: The C-SNIP data show that, depending on the act in question, the effect of civic talk on civic participation can be less than that of having prior participatory experience. The focus group data also show that while civic talk has an effect on how active individuals are in civil society, most people are not explicitly aware that this relationship exists. Instead, the majority of people attribute their patterns of behavior to their personal attributes instead of to the nature of their social context. Thus, the results presented in this chapter show that, to understand how contemporary participatory democracy functions, both social-level and individual-level antecedents of civic participation need to be considered. Neither factor on its own is a sufficient explanation of why an individual chooses to participate in civil society. It is also important to note that while the main purpose of this chapter is not to assess how the influence of civic talk varies under different conditions, the results lend themselves to some initial conclusions. Both C-SNIP studies show that civic talk has a larger and more reliable effect on behavior in which individuals are predisposed to engage. For example, because the subjects in the study were not politically active or engaged, civic talk had no detectable influence on whether they participated in political activities. This question of how the relationship between civic talk and civic participation varies under different circumstances will be addressed in greater detail in Chapters 5 and 6. In addition, while the evidence presented in this chapter shows that there is a causal relationship between civic talk and civic participation, it does not show why this is the case. That is the subject of the next chapter.

4 Why Does Civic Talk Cause Civic Participation? [My roommate] was voting on the primary day and I, like, wasn’t really planning on it just because I felt like I didn’t really know enough. I mean, I knew, I guess; I knew enough. But I was just, like, “No. I don’t even want to.” But she encouraged me to, and I did. I feel a lot more interested. I really didn’t talk much politics in high school with anybody, and coming here, it’s, like, “Holy cow! We’re in the capital [city of Wisconsin], and everyone talks about politics all the time where I’m living.” So I think I’ve gained interest from talking to people. —C-SNIP Focus Group Study participants

he previous chapter presented evidence of the causal relationship between civic talk and civic participation. Data from the C-SNIP Panel Survey show that civic talk increased participation in voluntary civic organizations by 38 percent. These same data show that civic talk also increased the likelihood of turning out to vote by 7 percentage points. The positive effect of civic talk on civic participation appears, however, to be contingent on the individual’s motivation to participate. For example, the C-SNIP Panel Survey shows that civic talk had little effect on participation in political activities such as protesting, contacting elected officials, and political campaign work. This appears to be the case because the student population examined in this study is not politically active or engaged. Evidence from the qualitative focus groups led to the same conclusion. While the previous chapter shows that there is a meaningful relationship between civic talk and civic participation, talk in and of itself cannot be what is driving individuals to participate. Talk about politics and current

T

52

Chapter 4

events must be generating other forces that encourage us to act. Therefore, the next logical step in this analysis is to determine what these other forces are. In this chapter, evidence from the C-SNIP Panel Survey and Focus Group Study are used to show that when we talk about politics and current events with our peers, we are asked by them to participate in civil society. These data also show that civic talk makes us more engaged with politics and current events.

Why Do We Participate in Civic Activities? In attempting to answer why civic talk causes civic participation, it is necessary to ask why people choose to participate in civic activities in the first place. With this knowledge, we can determine what types of evidence are needed to assess why conversations about politics and current events lead individuals to participate in civic activities. The simplest answer to this question is that the decision to participate is determined by the costs and benefits associated with participating. If the benefits outweigh the costs, a person is more likely to choose to participate (Downs 1957; Olson 1965; Verba et al. 1995). One of the most prolific articulations of this argument appears in Anthony Downs’s analysis of voter turnout in An Economic Theory of Democracy (1957). Downs shows that an individual decides whether to vote by comparing the costs associated with voting (e.g., time and energy) to the potential benefit of his or her vote being the decisive one in the election. On that basis, Downs concludes that voting is an irrational act, since even in small electorates any one vote is unlikely to be decisive. Mancur Olson offers a similar argument in his seminal examination of interest-group politics, The Logic of Collective Action (1965). He shows that individuals are more likely to contribute their time and money to an interest group if they receive selective benefits (i.e., excludable benefits that are only given to those who contribute to the group) to offset the costs associated with participation. Without such inducements, we have an incentive to “free ride” in the hope of receiving the public (i.e., non-excludable) benefits of interest groups’ lobbying without having to incur the costs associated with contributing to the effort.1

1 For

example, if an environmental interest group successfully lobbies for clean air legislation, everyone can enjoy the benefits of cleaner air regardless of whether they contributed to the cause.

Why Does Civic Talk Cause Civic Participation?

53

While the calculus of deciding whether to participate in civil society can easily be distilled down to a question of costs and benefits, it is necessary to dig deeper and ask what factors determine how costly and how beneficial it is to engage in civic activities. The extant literature on civic participation suggests four factors that drive the costs and benefits associated with civic participation: resources, civic engagement, recruitment, and norms.2

Resources To illustrate how resources might help explain the relationship between civic talk and civic participation, consider the various factors that determine whether a person decides to vote. Information about the candidates and the issues at stake in the election is the most basic resource that this person would need to decide whether to vote (Popkin 1995). This is the case because when such informational resources are readily available, the act of voting becomes less costly. For example, the task of deciding who to vote for is easier if you already know the policy positions of the candidates in the race. Moreover, information can make the task of voting more beneficial. Information allows voters to cast their ballots for candidates who best support their preferences. Thus, if the candidate a voter selected wins an election and faithfully pursues the policy agenda on which he or she campaigned, this is beneficial to the voter. Individuals can obtain civically relevant information from a number of sources. For example, our hypothetical voter could gather information by monitoring the news media, attending political rallies, sifting through pieces of direct mail sent by candidates and interest groups, and the like. However, this individual might also obtain information through conversations about politics and current events with the members of his or her social network (Conover et al. 2002; Downs 1957; Huckfeldt et al. 2000; Huckfeldt and Sprague 1995; Lazarsfeld et al. 1968; McClurg 2003; Popkin 1995; Walsh 2004). One benefit associated with seeking information in one’s social network is cost reduction. For example, talking about the current election over coffee with a friend is an easier way for an individual to obtain information than taking the time to read a newspaper. In 2 Three

of these factors—resources, engagement, and recruitment—make up Verba and colleagues’ (1995) “Civic Voluntarism Model.” Using survey data on the participatory habits of American citizens, Verba and his colleagues found that each of these three factors has an independent effect on the likelihood of becoming civically active.

54

Chapter 4

addition to being lower in cost, obtaining information from our peers is likely to be enjoyable. For example, an individual is more likely to enjoy discussing politics over a dinner out with friends than going to a candidate’s rally filled with strangers.3

Civic Engagement Being civically engaged—having an interest in politics and current events and a sense of political efficacy—also motivates civic participation. This is the case even after accounting for how rich in resources a person is (Verba et al. 1995). Consider our hypothetical voter again. Regardless of how much information this person has on the campaign and the candidates, this citizen is unlikely to vote if he or she is not interested in politics and current events or if he or she does not feel that his or her vote will matter (Verba et al. 1995). Civic engagement makes voting and other forms of civic participation easier to engage in largely by increasing the benefits associated with participation. For example, because any one vote is rarely decisive in an election, the tangible benefits of voting are largely non-existent (Downs 1957). However, if individuals are imbued with a strong sense of civic engagement, they will nonetheless find psychological satisfaction in voting because they are fulfilling their sense of civic duty. Moreover, because of their interest in politics and sense of political efficacy, the civically engaged among us are less likely to be concerned about the costs associated with participating in civic activities. For example, the opportunity cost of lost time at work in order to vote is likely to be perceived as less severe by individuals who are interested in politics and current events. As with information, civic talk could increase an individual’s level of civic engagement. For example, talking about politics and current events in an informal social setting could lead an individual to learn about and become more interested in participating in civic activities (McClurg 2003). Extant research also shows that experiences in the family and in the school help solidify the political interest and efficacy that individuals carry with them for the rest of their lives (see, e.g., Verba et al. 1995). Perhaps interactions in collegiate peer groups, such as those documented in the C-SNIP data sets, do the same. 3 Otherwise

stated, and as discussed in Chapter 2, the political content we get from conversations with our peers is typically a “byproduct” of social interactions based on non-political goals such as socializing.

Why Does Civic Talk Cause Civic Participation?

55

Recruitment Even the most engaged and resource-rich individuals among us will be more likely to participate in civic activities if they are mobilized by someone else to act (Gerber and Green 2000; Nickerson 2008; Rosenstone and Hansen 1993; Verba et al. 1995). Consider our hypothetical voter again. This individual will be more likely to vote if he or she is motivated to do some by someone else (Gerber and Green 2000; Nickerson 2008; Rosenstone and Hansen 1993).4 Recruitment is effective at eliciting civic participation because it makes the process of taking action less costly. For example, if a political party volunteer offers to give you a ride to the polls, the task of voting becomes much easier for you. Moreover, in some circumstances recruitment might affect the benefits associated with civic participation. For example, if a religious leader asked his or her parishioners to vote, those individuals might feel obliged to do so to maintain the benefit of membership in good standing in the church. Existing research suggests that the face-to-face style of recruitment that one would expect to arise out of peer networks may be especially effective at eliciting civic participation (Brady et al. 1999; Gerber and Green 2000; Godwin and Mitchell 1984; Klofstad 2007). For example, through an experimental study of voters in New Haven, Connecticut, Gerber and Green (2000) show that door-to-door neighborhood canvassing is more effective at stimulating voter participation than less direct methods such as phone calls and pamphlets sent through the mail. In a similar vein, Brady and his colleagues (1999, 157) hypothesize that “recruiters who have a close relationship to their prospects should have two advantages: They are more likely to have information about the activity potential of the target and more likely to have the leverage that makes acquiescence to a request probable.” Their analysis offers support for this hypothesis; recruitment to participate in political activities was estimated to be 28 percent more effective if the target was already acquainted with the recruiter.

Social Norms Finally, social norms—“beliefs about which behaviors are acceptable and which are unacceptable for specific persons in specific situations” 4 In

fact, Rosenstone and Hansen (1993) show that about 50 percent of the drop-off in voter turnout in the United States between the 1960s and the 1980s was caused by declines in face-to-face mobilization, such as neighborhood canvassing, by the political parties.

56

Chapter 4

(Michener and DeLamater 1999, 64)—can have a powerful effect on individual behavior (Crandall 1988; Festinger et al. 1950; Latané and Wolf 1981; Mendelberg 2002; Michener and DeLamater 1999; Putnam 2000; Schachter 1959). For example, consider once again a hypothetical voter who is fully stocked with resources, engagement, and recruitment. This individual is already going to be quite likely to vote but will be even more likely to do so if he or she adheres to civic-minded norms. For example, Putnam (2000) shows that Americans were more civically active when norms of social trust and reciprocity were more prevalent in society. Such norms, he argues, lead individuals to be integrated into events and issues outside their own personal sphere. In other words, pro-civic norms make civic participation more beneficial, since the individuals that adhere to them feel the need to contribute to wider society. Moreover, when norms dictate that civic participation is a necessary activity in a society, violating those rules by failing to participate is costly to the social deviant (e.g., one experiences “cognitive dissonance”—a sense of psychological discomfort— over having violated the norm). Norms are especially germane to the study of civic talk among peers because, along with the family and the school, the peer group is one of the most important transmitters of social norms that an individual encounters during his or her lifetime (Beck 1977; Dawson et al. 1977; Michener and DeLamater 1999; Silbiger 1977; Walsh 2004). Findings from the field of social psychology suggest why this is the case. Research by Latané and Wolf (1981) on social impact theory shows that social influence is stronger when the socializing agent is close and sustained—what they term “immediate”— the way a peer group is. In a similar vein, Festinger and his colleagues (1950, 163) show in their seminal examination of social interaction in a student housing unit at the Massachusetts Institute of Technology that “it is through the small face-to-face groups that many attitudes and ideologies which affect our behavior are transmitted.” Thus, because social networks are typically composed of close and sustained social relationships, they can be expected to have normative influence on their members. If peers transmit civic-minded norms during civic talk discussions, looking for these norms in peer networks might help explain how civic talk leads individuals to participate in civic activities.

Why Does Civic Talk Cause Civic Participation?

57

Why Does Civic Talk Cause Civic Participation? Survey Evidence Measures RESOURCES

To see if the transfer of informational resources occurs among peers, C-SNIP Panel Survey respondents were asked, “How many times have your roommates given you any information about how to become active in politics and current events: often, sometimes, rarely, or never?” On average, students reported exposure to such information somewhere between “never” and “rarely” (an average of .4 on the 0–3 scale ranging from “never” to “often”). ENGAGEMENT

Increased civic engagement as a consequence of civic talk was first measured by asking, “Thinking about how interested you were in politics and current events before you came to the University of Wisconsin, Madison, has talking with your roommates increased your interest in politics and current events: very much, somewhat, not that much, or not at all?” On average, students reported enhanced engagement somewhere between “not at all” and “not that much” (an average of .7 on the 0–3 scale ranging from “not at all” to “very much”). A second measure of civic engagement is based on each student’s level of political efficacy, measured with the survey question, “How much would you agree or disagree with this statement: ‘People like me don’t have any say about what the government does’: strongly agree, somewhat agree, neither agree nor disagree, somewhat disagree, or strongly disagree?” On average, students reported that they neither agreed nor disagreed with the idea that they do not have a say in what the government does (an average of 3.4 on the 1–5 scale ranging from “strongly agree” to “strongly disagree”). RECRUITMENT

Peer-to-peer recruitment was measured by asking, “How many times have your roommates asked you to participate in an event or organization related to politics and current events: often, sometimes, rarely, or never?” On average, students reported enhanced engagement somewhere between “never” and “rarely” (an average of .3 on the 0–3 scale ranging from “never” to “often”).

58

Chapter 4

SOCIAL NORMS

While norms are rules of social conduct, they typically are not formally codified. This leaves us in danger of misusing the concept as a “garbage can” to explain a host of different phenomena that may or may not actually be attributable to norms. Therefore, to develop a multidimensional picture of the potential impact of norms on the relationship between civic talk and civic participation, this analysis uses two different measurement strategies: behavioral modeling and civic-minded attitudes. Norms are often learned by observing and modeling the behavior of others (Crandall 1988; McClurg 2004; Michener and DeLamater 1999). Accordingly, one measure of norms used in this analysis is the individual’s perception of his or her peers as measured by the question, “How active and interested do you think your roommate is in politics and current events: very, somewhat, not very, or not at all?” Under the assumption that individuals learn how to behave by observing the actions of others, this measure serves as a proxy for civic norms by capturing how civically active an individual’s roommate is. On average, students reported that they thought their roommates were “not very” to “somewhat” active (a mean of 1.4 on the 0–3 scale ranging from “not at all” to “very”). A second measure of norms focuses on attitudes rather than behaviors. This measure was calculated as each respondent’s response to the question, “How important do you think it is for people like you to be active and interested in politics and current events: very important, somewhat important, not very important, or not at all important?” The expectation is that if an individual begins to feel that civic participation is important as a consequence of engaging in civic talk, he or she may feel pressure to become more active in civil society. On average, respondents felt that civic engagement and participation is somewhere between “somewhat” and “very” important (a mean of 2.3 on the 0–3 scale ranging from “not at all” to “very”).

Basic Correlations If resources, engagement, recruitment, and norms help explain how civic talk influences us to participate in civic activities, these factors should correlate with both the amount of civic talk occurring in the peer network and the amount of civic participation in which a person engages. The results in Table 4.1 support this expectation. Civic talk correlates with information transfers, increased psychological engagement with politics and current events, enhanced political efficacy, instances of recruitment, and

Why Does Civic Talk Cause Civic Participation?

TABLE 4.1

59

Explaining the civic talk effect (correlations) Civic talk among roommates

Participation in voluntary civic organizations

Political participation

Voter turnout

Resources

.38***

.15***

.09***

.10***

Engagement Interest in politics and current events Political efficacy

.45*** .11***

.13*** .14***

.10*** .14***

.12*** .11***

Recruitment

.31***

.16***

.12***

.09***

Norms Perceived activity/interest level of roommate Perceived importance of civic participation

.40***

.03

.18***

.20***

Causal mechanisms

−.01 .22***

.07*** .16***

Source: C-SNIP Panel Survey. ***p ≤ .01.

the acceptance of pro-civic norms. In turn, resources, engagement, recruitment, and norms correlate with higher levels of civic participation. Some of these correlations are not extraordinarily large, but they are all positive and statistically significant. The only exception in Table 4.1 is the perception of how active and interested one’s roommate is in civic activities. These results show that students who engaged in civic talk became aware that their roommates were active and interested in civic activities, which makes sense. If someone you know frequently discusses a certain topic, it is safe to assume that he or she is interested in that subject. However, knowing that their roommates are active and interested in civic activities appears to have little influence over how active students are in civil society. This suggests that behavioral modeling is not likely to be a mechanism that governs the relationship between civic talk and civic participation.

Multivariate Analysis: “Explaining Away” Civic Talk with Causal Mechanisms The correlations listed in Table 4.1 are necessary evidence for resources, engagement, recruitment, and norms to be viable explanations of the civic talk effect. If there are no bivariate statistical relationships among these four factors, civic talk, and civic participation, it is unlikely that they are acting as the mechanisms that cause individuals to translate talk into

60

Chapter 4

action. However, to validate these findings with more sufficient evidence, it is useful to turn to a multivariate analysis.5 To test whether resources, engagement, recruitment, and norms help explain the civic talk effect, measures of these concepts were added to the regression analyses presented in Chapter 3. The goal of adding these variables to the model is to “explain away” the civic talk effect. If resources, engagement, recruitment, and norms explain how we translate discussions about politics and current events with peers into civic participation, the civic talk coefficient should drop in value after these variables are added to the model. This will only occur if resources, engagement, and recruitment account for the variance in civic participation that was once accounted for by civic talk.6 The result of this analysis on participation in voluntary civic organizations is presented in Table 4.2. The results show that resources, engagement, recruitment, and civic-minded norms help explain the civic talk effect. For each causal mechanism, the civic talk regression coefficient drops in value after adding the causal mechanism variable to the analysis. However, unlike the civic talk coefficient, the lagged dependent variable coefficient is not affected when the causal mechanism variables are added. As anticipated, then, in this analysis the addition of resources, engagement, recruitment, and norms accounts only for the variance once explained by civic talk. Closer examination of the causal mechanism coefficients in Table 4.2 gives us greater insight into how individuals translate civic talk into civic participation. Moving from the left to the right in the table, adding resources to the analysis in column 2 led to the second-largest drop in the value of the civic talk coefficient (15 percent). However, the direct effect of resources on participation in voluntary organizations is uncertain, since the resource coefficient is not statistically significant. Column 3 shows that adding interest in politics and current events to the analysis only led to a 5 percent drop in the value of the civic talk coefficient. Moreover, the interest coefficient itself is not statistically significant. The efficacy measure of civic engagement (column 4) also led to a 5 percent decline in the civic talk coefficient. However, unlike interest, the efficacy coefficient is 5 The

perceived activity/interest level of one’s roommate was omitted from this analysis because this variable does not correlate significantly with how active one chooses to be in most civic activities (see Table 4.1).

6 Campbell

and Wolbrecht (2006) use the same technique to explain the relationship between greater numbers of women serving in elected office and greater levels of political engagement among female adolescents. As a germane side note, Campbell and Wolbrecht find that the mechanism that links these two phenomena together is political discussion within the family.

Why Does Civic Talk Cause Civic Participation?

61

Explaining the effect of civic talk on participation in voluntary civic organizations during the first year of college (regression analysis)

TABLE 4.2

(1)

(2)

(3)

(4)

(5)

(6)

Civic talk among roommates

.81** (.30)

.69* (.33)

.77** (.31)

.77** (.31)

.67* (.33)

.76** (.32)

Participation in voluntary civic organizations during high school

.22*** (.03)

.22*** (.03)

.22*** (.03)

.21*** (.03)

.22*** (.03)

.21*** (.03)

Causal mechanisms Resources

.23 (.20)

Engagement Interest in politics and current events Political efficacy

.06 (.16) .26** (.11)

Recruitment

.50* (.25)

Norms (perceived importance of civic participation) Constant Adjusted R2 Percent change in civic talk coefficient

.72*** (.23) 1.67 (1.76) .13

1.66 (1.78)

1.65 (1.76)

.94 (1.85)

1.53 (1.74)

−.15 (1.88)

.13

.13

−15

−5

.14

.14

.15

−5

−17

−6

Source: C-SNIP Panel Survey. Model Type: Ordinary least squares (Imai et al. 2007c). Columns: (1) Base model. (2)–(6) Base model with addition of individual causal mechanisms as indicated within the body of the table. Note: Dormitory assignment fixed effects were included in the analysis but are omitted from the table (none of the coefficients were statistically significant). The matched data set is used in this analysis (see Appendix C). Standard errors are in parentheses. N = 1,044. *p ≤ .10; **p ≤ .05; ***p ≤ .01.

statistically significant. Adding the recruitment measure to the regression analysis in column 5 led to the largest decline in the value of the civic talk coefficient (17 percent). Moreover, the recruitment coefficient itself is statistically significant. Finally, adding civic-minded norms in column 6 only caused a 6 percent decline in the value of the civic talk coefficient. However, the coefficient is statistically significant, showing that norms have a more stable relationship with participation than do resources and interest in politics and current events.7 7 These

results hold if all of the causal mechanism variables are added to the model simultaneously.

62

Chapter 4

Table 4.3 applies the same style of analysis to the case of voter turnout. Again, the results show that adding the four causal mechanisms helps explain the effect of civic talk on civic participation. In fact, the causal mechanism variables appear to do a better job of explaining the relationship between civic talk and voter turnout. The average reduction in the

Explaining the effect of civic talk on voter turnout during the first year of college (regression analysis)

TABLE 4.3

(1)

(2)

(3)

(4)

(5)

(6)

Civic talk among roommates

.31† (.19)

.23 (.21)

.19 (.19)

.28 (.18)

.27 (.18)

.29 (.19)

Participation in political activities during high school

.14** (.06)

.14** (.06)

.14** (.06)

.13** (.06)

.14** (.06)

.12** (.06)

Causal mechanisms Resources

.17 (.12)

Engagement Interest in politics and current events Political efficacy

.19* (.11) .19*** (.07)

Recruitment

.16 (.17)

Norms (perceived importance of civic participation)

.32* (.16)

Constant

−2.11 (2.20)

−2.12 (2.20)

−2.18 (2.21)

−2.67 (2.29)

−2.16 (2.19)

−2.87 (2.19)

Akaike’s information criterion (AIC)

1,283

1,282

1,280

1,276

1,283

1,278

−26

−39

−10

−13

−6

Percent change in civic talk coefficient

Source: C-SNIP Panel Survey. Model Type: Logistic regression (Imai et al. 2007d). Columns: (1) Base model. (2)–(6) Base model with addition of individual causal mechanisms, as indicated within the body of the table. Note: AIC is twice the number of parameters in the model, minus twice the value of the model’s log-likelihood. Dormitory assignment fixed effects were included in the analysis but are omitted from the table (none of the coefficients were statistically significant). The matched data set is used in this analysis (see Appendix C). While the significance of the civic talk coefficient in column 3 falls just below the 90 percent confidence threshold, the expected increase in turnout is estimated to be significant at the 95 percent level (see Figure 3.2). Standard errors are in parentheses. N = 1,044. †p ≤ .12; *p ≤ .10; **p ≤ .05; ***p ≤ .01.

Why Does Civic Talk Cause Civic Participation?

63

civic talk coefficient is 19 percent in the analysis of voter turnout, compared with 10 percent in the analysis of participation in voluntary civic organizations. As with Table 4.2, closer examination of each column in Table 4.3 gives greater insight into how each of these causal mechanisms functions. Moving from left to right in the table, adding informational resources in column 2 reduced the civic talk coefficient by 26 percent. However, the information coefficient is not significant. In column 3, adding interest in politics and current events to the analysis led to the largest decline in value of the civic talk coefficient (39 percent). Moreover, the interest coefficient meets a minimal standard for statistical significance. The direct influence of political efficacy on voter turnout (column 4) is also statistically significant. However, efficacy has far less of an effect on the civic talk coefficient (a 10 percent decline in value) than interest in politics and current events. Adding recruitment to the analysis in column 5 leads to a 13 percent decline in the value of the civic talk coefficient. However, the recruitment coefficient is not statistically significant. Finally, as in the analysis of voluntary civic memberships, adding norms to the analysis in column 6 led to only a 6 percent decline in the voter turnout civic talk coefficient. However, the norms coefficient is statistically significant, showing that norms have a more systematic impact on voter turnout than resources or recruitment.8

Differences between Voluntary Civic Organizations and Voting All four causal mechanisms offer some level of explanation of the relationship between civic talk and civic participation. However, in the case of participation in voluntary organizations, recruitment plays the largest and most consistent role in explaining this relationship. In contrast, civic engagement carries more of the explanatory load for voter turnout. This difference makes sense considering the uniqueness of the vote as a civic act. Since the probability that any one vote will affect the outcome of an election is small (Downs 1957), very few tangible benefits are associated with voting. Consequently, a prospective voter must be motivated by a sense of civic engagement to see the benefit of participating. In contrast, participation in the voluntary civic organizations examined in this 8 Again,

these results hold if all of the causal mechanism variables are added to the model simultaneously.

64

Chapter 4

study (i.e., student groups) comes with tangible benefits. For example, students receive social (i.e., “solidary”) benefits because they get to socialize with each other as they participate in student organizations. Participation in voluntary student organizations also allows individuals to build their résumés. Thus, it makes sense that being recruited to participate in this form of civic activity caries more explanatory weight than being civically engaged.9

Why Does Civic Talk Cause Civic Participation? Focus Group Evidence Resources The focus group sessions revealed that information transfers were pervasive during civic talk discussions. As discussed in Chapter 3, the vast majority of the information about politics and current affairs being shared by roommates was germane to current events. Given the proximity of the focus group sessions to the 2008 presidential primary in Wisconsin, most of these conversations centered on the campaign (and especially on the candidacy of Senator Barack Obama). As also discussed in Chapter 3, the majority of these information transfers were caused by the consumption of news media content. Most of the participants in the focus groups mentioned that they or their roommates would initiate a civic talk conversation to relay factual information, and to articulate opinions, about the events of the day after watching or reading the news. However, in line with the survey results showing that, on average, respondents were exposed to information about how to become civically active somewhere between “never” and “rarely,” only a minority of the focus groups’ participants reported that they shared information with their roommates about the civic activities in which they engage. Most of the participants in each of the focus groups reported that they had discussed attending candidate rallies or voting during the 2008 presidential primary season. A much smaller group of participants, however, mentioned that either they or their roommates discussed participation in student organizations, such as Habitat for Humanity; religious organizations; occupational or field-of-study groups; campus-based partisan organizations; 9A

related explanation is that the measure of civic engagement used in this analysis is interest in politics and current events. Since the voluntary civic organizations assessed in this study are largely a-political, it makes sense that interest in politics and current events has less of an effect on participation in voluntary civic organizations than on voting.

Why Does Civic Talk Cause Civic Participation?

65

and the like. A few focus group participants also reported that they or their roommates had explicitly mentioned how to become involved in those activities.10 A number of focus group participants also reported that they relayed information to their roommates to persuade them about politics and current events. Again, based on the proximity of the focus groups to the 2008 presidential primary in the Wisconsin, it is not surprising that most of these persuasion conversations centered on the election and the candidates. These types of statements were typical in all four of the focus group discussions: I’ve been in that situation. I’ve always tried to persuade my roommate that it’s a really good idea to be a Democrat, and care about politics, and hate George Bush on principle. My roommate always tries to persuade me. She used to be an Obama fan, but now she’s Hillary [sic]. She always tries to persuade me to vote for her, too. I kind of tried to convince my roommate about Obama, . . . because I think she was . . . I think she liked [John] McCain, and I convinced her to go up to the Obama rally. However, while such persuasion attempts were pervasive, many participants in each of the focus groups mentioned that they actively avoided trying to persuade their roommates about politics and current events. In line with the results presented in Chapter 3, these students reported that they chose to not engage in persuasion to avoid conflict with their roommates. As one participant succinctly stated, “I don’t try to persuade him. I just leave it alone.”

Engagement When directly asked whether civic talk conversations with roommates increased their interest in politics and current events, the participants in each of the focus groups provided a mix of responses between “yes” and “no.” This result is in line with the survey results showing that, on average, students reported enhanced engagement via civic talk somewhere between “not at all” and “not that much.” 10 This

phenomenon is considered in more detail in the section on evidence of recruitment gathered in the focus group sessions.

66

Chapter 4

A notable number of participants in each of the focus groups did, however, report that their civic engagement was enhanced as a consequence of interacting with roommates. Many of these statements were linked to the desire to increase civic knowledge and competence: The conversation might travel into a gray area, where it’s like, “Hey, I don’t know as much about this, and now I want to, because I want to be able to carry on these conversations and really feel like I have things to back up my opinion with.” I’d say it increases [my interest]. Just like wanting to learn more and, like, look at all the different points of view and see what other people think. If someone talks about something a lot, you can, like, think twice about it, maybe. Further evidence of enhanced engagement vis-à-vis civic talk is also present when the different focus groups are compared with one another. Two of the focus groups consisted of students who said that they frequently engaged in civic talk with their roommates, while the other two consisted of students who reported that they discussed politics and current events infrequently. Statements regarding enhanced engagement were more prevalent when civic talk was more frequent.

Recruitment A large number of students in each of the focus group sessions reported that recruitment attempts did occur while roommates engaged in civic talk. Also, comparing the “low” and “high” civic talk focus groups showed that recruitment attempts were more common among the participants in the “high talk” groups, indicating a positive relationship between civic talk and recruitment. Based on the proximity of the focus group sessions to the 2008 presidential primary in the Wisconsin, most of these recruiting attempts concerned attending rallies for candidates (specifically the large rally for Obama held on the night of February 13, 2008) or turning out to vote in the presidential primary. A smaller group of students reported that either they or a roommate had made recruitment attempts for nonpolitical student organizations. Many participants in each of the focus groups reported that these recruitment attempts were successful, especially concerning attending candidates’ rallies and turning out to vote in the presidential primary.

Why Does Civic Talk Cause Civic Participation?

67

It is necessary to highlight, however, that while recruitment was pervasive, it was not universally effective. Chapter 3 showed that the relationship between civic talk and political participation is weak because the C-SNIP Panel Survey student population was not interested in participating in political activities. In the same vein, the focus group data indicate that recruitment attempts were effective only when the students had some basic level of interest in becoming active in civil society. This exchange is especially illustrative of this trend in the data: MODERATOR: Going back to conversations you have with your

roommates about events and current events and things like that, do you ever talk with your roommates about getting involved in civic activities or campus activities of any kind? PARTICIPANT: Such as? MODERATOR: Any sorts of political or campus organizations, polit-

ical events that may be going on. A couple of you mentioned the rallies and things like that, things of that nature. PARTICIPANT: Like a student government something or not? MODERATOR: Yeah, absolutely. PARTICIPANT: Because my roommate is, like, seriously involved in

student government. MODERATOR: Really? PARTICIPANT: Yeah. He’s as high as a freshman could get in [stu-

dent government], and he’s in [the dormitory’s] student government. Like, he’s super-involved, so he’s really active, and he’s, like, tried to get me to go to stuff, and I don’t. MODERATOR: What kind of stuff does he . . . PARTICIPANT: I don’t know. He’s like, “You should come, like, to

the meetings and, like, be a part of it.” And I was like, “Yeah, maybe.” And then I didn’t. And then he was like, “You should come to the kick-off, because, like, you know, just to learn about [it].” And I was, like, “Yeah, maybe.” And then I didn’t. And then he was, like, “You know, you should like join an e-mail list.” And I’m, like, “Ahhh . . . maybe.” MODERATOR: So he’s very explicitly asking you to get involved in

this stuff. PARTICIPANT: Yeah, he’s really into it.

68

Chapter 4

MODERATOR: And what would you say is your reasoning? Why do you choose not to get involved in it? PARTICIPANT: . . . Basically because I don’t really care about school

or local politics, and so I don’t want to get involved. This idea that civic talk has an effect only on individuals who are already primed to participate in civil society is examined in detail in Chapter 5.

Social Norms In the C-SNIP Panel Survey analysis, civic-minded social norms were operationalized in two ways: modeling the behavior of one’s roommate and adhering to the idea that civic participation is an important activity in which to engage. With regard to behavioral norms, the focus group participants were not directly asked whether they knew how civically active their roommates were. However, it appears that most of the participants had some notion of their roommates’ patterns of civic participation. For example, in the discussion of whether recruitment occurred during civic talk conversations, many participants revealed that they knew that their roommates either were or were not participating in civic activities. As also discussed in Chapter 3, when specifically asked what topics they discussed when engaging in civic talk, a number of participants mentioned that they talked about the civic activities in which they or their roommates engaged. In line with the correlations presented in Table 4.1, however, the focus groups did not reveal any direct evidence that participants felt compelled to model the civic behavior of their roommates. As discussed in the analyses of information sharing and recruitment, a number of focus group participants revealed that they had been recruited or persuaded to participate in civic activities by their roommates (especially with regard to the presidential primaries in 2008). This said, none of the focus group participants stated that they had witnessed their roommates participating and subsequently chose to model that behavior. Moreover, as also discussed, some focus group participants were actually “turned off” to civic participation as a consequence of engaging in civic talk with civically active roommates. With regard to social norms as civic-minded attitudes, the focus group participants were not asked directly whether civic talk increased their sense of obligation to participate in civic activities. However, as documented in the analysis of civic engagement, a number of focus group participants reported that they became more interested in politics and current

Why Does Civic Talk Cause Civic Participation?

69

events as a consequence of engaging in civic talk. This enhanced level of interest likely translates into a stronger sense of civic-mindedness. Evidence from the C-SNIP Panel Survey corroborates this assumption. Students who became more interested in politics and current events as a result of engaging in civic talk also felt more strongly about the need to participate in civic activities (r = .14, p < .01).

Conclusion The purpose of this chapter has been to explain why civic talk causes civic participation. I began with a review of the factors that are known to motivate individuals to participate in civil society. Simply stated, individuals are more likely to participate in civic activities when the costs are low and the benefits are high. Specifically, four factors affect the costs and benefits associated with participating in civil society: resources (information), engagement (interest and efficacy), recruitment (being asked to participate), and civic-minded social norms. An analysis of the C-SNIP Panel Survey and focus group data shows that these four factors help to explain how individuals translate civic talk into civic participation. Of the four, recruitment (in the case of participation in voluntary civic organizations) and enhanced civic engagement (in the case of voter turnout) appear to carry the most explanatory weight. However, in line with findings in Chapter 3, this exploration of causal mechanisms reveals that the relationship between civic talk and civic participation is not constant across all conditions. For example, the C-SNIP Panel Survey data show that, when we discuss politics and current events, we are asked by our peers to participate in civic activities, and these instances of recruitment make us more likely to participate in civil society. However, an examination of the focus group data shows that while civic talk does lead to recruitment, these attempts are not always successful. Instead, the data show that the targets of recruitment need to have some basic level of interest in participating. If they do not, recruitment will have no effect no matter how much they are asked to participate by their peers. In response to the growing amount of evidence suggesting that the relationship between civic talk and civic participation is not constant across all conditions, the next two chapters dig more deeply into this phenomenon by examining how the civic talk effect varies under different circumstances.

5 Do You Matter? MODERATOR: Do you guys ever talk with your roommates about

getting involved in student organizations or political groups? PARTICIPANT: Yes. MODERATOR: And what? Can you tell . . . PARTICIPANT: Well, he mostly talks. He prompts me to get me

involved, and I usually tell him I’m not interested. —C-SNIP Focus Group Study participant

n the preceding chapters, I showed that the people in our social environment have a meaningful impact on how we choose to participate in the processes of democratic governance. Evidence from a quasi-experimental panel survey and a series of qualitative focus groups shows that, when we discuss politics and current events with our peers, we are more likely to participate in civic activities. This relationship between civic talk and civic participation exists because of resources, engagement, recruitment, and norms. Conversations about politics and current events provide us with information about how to participate in civic activities, enhance our sense of psychological engagement with civil society, subject us to blandishments to participate in civic activities, and lead us to greater acceptance of civic-minded attitudes. Further analysis shows that of these four factors, engagement and recruitment carry the most weight when attempting to explain the causal link between civic talk and civic participation. While the preceding chapters have shown that the influence of civic talk on participatory democracy is meaningful, however, we cannot assume that the process of translating civic talk into civic participation will work in the same way in all cases. Two intervening factors need to be considered when studying the effects of engaging in civic talk: our own characteristics and those of our peers. The role of individual characteristics is addressed in this chapter; Chapter 6 examines how the characteristics of our peers affect the relationship between civic talk and civic participation.

I

72

Chapter 5

Relying again on the C-SNIP data sets, the evidence presented in this chapter offers a number of insights into how our own characteristics can both enhance and mitigate the influence of civic talk.1 More specifically, the preponderance of the data shows that individuals who are willing and able to participate in civic activities experience a larger boost in their level of civic participation after engaging in civic talk than do their less willing and able counterparts.

Who Should Get More Out of Civic Talk? Individuals with Weaker Ability to Participate As depicted in Figure 5.1, from the simplest of perspectives, human behavior can be viewed as a product of our personal characteristics (P), our environment (E), and the interaction between these two factors (P × E). Consequently, as P becomes weaker in this model, one’s environment will have more of an influence in determining one’s patterns of behavior. Conversely, as P becomes stronger, the forces in one’s environment will have less influence in determining one’s patterns of behavior. In other words, this model predicts that individuals with weaker predilections to participate in civic activities should be more influenced by discussing politics and current events with their peers. Research in the field of social psychology offers evidence that supports this hypothesis (see, e.g., Festinger 1954; Marsden and Friedkin 1993; Meyer 1994; Michener and DeLamater 1999; Petty and Cacioppo 1986; Silbiger 1977). For example, in his work on social comparison theory, Festinger (1954) found that human beings have an innate need to compare their own opinions and behaviors to those of the members of their social network. We feel this overwhelming need to compare ourselves with our peers to verify that our actions and attitudes are “correct.” Moreover, through their development of the Elaboration Likelihood Model of persuasion, Petty and Cacioppo (1986) found that individuals are more likely to mimic the attitudes and behaviors of their peers when their own attitudes are uncertain. In short, what these and other studies suggest is that individuals with less ability and less desire to participate in civic activities will be more likely to be influenced by their social network of peers.

1 Individual-level

characteristics also intervene in how social networks affect their members in non-human animals (see, e.g., Pike et al. 2008).

Do You Matter?

73

Personal Characteristics (P) Interaction (P×E)

Human Behavior

Environment (E) FIGURE 5.1 A simple model of human behavior

Individuals with Stronger Ability to Participate While research in social psychology implies that individuals with weaker ability and desire to participate in civic activities will be more affected by civic talk, the opposite could be true if we take into consideration how individuals weigh the costs and benefits associated with their actions. To account for this facet of decision making, Figure 5.2 complicates the model presented in Figure 5.1 by breaking down the process of human behavior into two stages: a data gathering phase and a data analysis phase. In the data gathering phase, as originally depicted in Figure 5.1, people collect information about how to behave from themselves and from their environment. However, to act on the data they have gathered, Figure 5.2 shows

Data Gathering Phase

Data Analysis Phase

Personal Characteristics (P) Interaction (P×E)

Consideration of Costs and Benefits

Environment (E) FIGURE 5.2

Modified simple model of human behavior

Human Behavior

74

Chapter 5

that people will first weigh the costs and benefits associated with taking action. In this model, only when costs are low and benefits are high will people choose to act on what they have learned after consulting with themselves and taking cues from their environment (Downs 1957; Olson 1965; Verba et al. 1995). But what determines the costs and benefits associated with participating in civic activities? As discussed in Chapter 4, the extensive body of literature on civic participation has come to a consensus that individuals are not automatically equipped to participate in civil society. Instead, a person needs a number of different resources and motivations to participate in civic activities. For example, Verba and his colleagues (1995) show that a person will find the costs of participating in civil society low enough, and the benefits high enough, when he or she is equipped with resources (time, money, and civic skills), has a sense of psychological engagement with politics and current events, and has been recruited by someone, such as a member of his or her social network, to participate. However, the analyses conducted by Verba and colleagues show that, of these three factors, resources and engagement are the most important in determining whether the benefits of civic participation outweigh the costs. As they state it, “Participation can, and does, take place in the absence of specific requests for activity. In contrast, it is hard to imagine activity without at least a modicum of resources and some political engagement” (Verba et al. 1995, 270). In other words, if we are not willing or able to participate in civic activities, no amount of cajoling by our social network of peers will force us to act. McClurg (2003) also addresses the question of how personal characteristics might influence the effect of civic talk through an analysis of data on social networks collected in South Bend, Indiana, by Huckfeldt and Sprague (1991, 1995). In line with the social psychology literature, McClurg hypothesizes that individuals with low levels of education will experience more of a participatory boost from civic talk than their more educated counterparts. As he states it, “Social interaction should make up the absence of personal resources and we should see a meaningful increase in the propensity to participate among low status individuals who discuss politics” (McClurg 2003, 457). However, the data from South Bend lead to the opposite conclusion: Less-well-educated individuals actually experienced less of a participatory boost from engaging in civic talk than their better-educated counterparts. In fact, McClurg finds that the civic participation gap between the well educated and the poorly educated actually increases as both groups engage in higher amounts of civic talk (McClurg 2003, 459, table 4).

Do You Matter?

75

In summation, what these and other studies suggest is that individuals need to be equipped with certain personal characteristics to be able to respond to stimuli in the environment. In contrast to the social psychology literature, then, political science research on civic participation predicts that if a person lacks the ability or desire to participate in civil society, civic talk should have less of an effect on his or her patterns of civic participation.

Who Gets More Out of Civic Talk? Survey Evidence Previous Experience with Civic Participation Do our predilections to engage in civic activity mitigate or enhance the effect of engaging in civic talk? To address this question, I start by examining how active C-SNIP Panel Survey subjects were in civic activities before engaging in civic talk (i.e., civic participation in high school). The data show that subjects who participated in civic activities at an above average rate in high school were more civically active during their first year of college (participation in voluntary civic organizations: t = 10.42, p < .01; voter turnout: t = 3.30, p < .01). Thus, past experience might have an intervening influence on the civic talk effect. What is unclear, however, is how an individual’s repertoire of past experience participating in civil society will interact with the amount of civic talk in which he or she engages. To examine the effect of prior participatory experience on the relationship between civic talk and civic participation, C-SNIP Panel Survey respondents were split into two groups based on how civically active each subject had been in high school, either above average or below average. The regression analysis in Chapter 3 (see Tables 3.3 and 3.5), which showed the causal effect of civic talk on civic participation for the entire pool of C-SNIP Panel Survey respondents, was then conducted separately on the two subsets of respondents. This procedure allows the civic talk effect to be estimated for individuals with above average and below average levels of prior experience participating in civic activities. The results of this analysis are in Figure 5.3. The data in Figure 5.3 show that prior experience participating in civic activities enhances the relationship between civic talk and civic participation. Focusing on the top portion of the figure, the data show that the effect of civic talk on participation in voluntary civic organizations increases as the individual’s level of prior experience increases. In

76

Chapter 5

Expected Change in Participation

2.5

Voluntary Civic Organizations

2.0 1.5 1.0 0.5 0

Below Average

–1.0

Prior Participatory Experience

Voter Turnout

0.25

Expected Change in Likelihood of Voting

Above Average

–0.5

0.20 0.15 0.10 0.05 0

Below Average –0.05

Above Average

Prior Participatory Experience

FIGURE 5.3 The intervening effect of participatory experience on the relationship

between civic talk and civic participation Source: C-SNIP Panel Survey Notes: The line on each bar represents the 95 percent confidence interval about the estimate. The effect of civic talk is calculated as the level of civic participation estimated to have been engaged in by individuals who were exposed to civic talk, minus that of those who were not exposed, all other factors in the model held at their means.

fact, the civic talk effect is only statistically significant when the student has an above average level of prior experience participating in civic activities. The results in the bottom portion of Figure 5.3 tell a similar story for voter turnout: Only the above average cohort experienced a statistically significant increase in voter turnout due to engaging in civic talk.

Do You Matter?

77

Along with experience participating in civic activities, an individual’s upbringing provides a window into his or her prior experience with politics and current events. The extant literature on political socialization shows that children who were raised by civically active and engaged parents are civically active and engaged as adults (for a summary of this literature, see McIntosh et al. 2007). Data from the C-SNIP Panel Survey corroborate these findings. Subjects raised by parents who were active and interested in politics and current events were also more civically active during their first year of college (participation in voluntary civic organizations: t = 2.81, p < .01; voter turnout: t = 2.42, p = .02). Employing the same procedure used to produce Figure 5.3, Figure 5.4 presents the intervening effect that home life during high school has on the relationship between civic talk and civic participation during the first year of college. These data mirror the results in Figure 5.3.2 The effect of engaging in civic talk on participation in civic organizations and voter turnout is larger for subjects who grew up with parents who were interested and active in politics and current events. In fact, only subjects who were exposed to above average levels of parental civic interest or activity at home before coming to college experienced a statistically significant increase in civic participation due to engaging in civic talk during the first year of college. Further analysis of the data presented in Figure 5.4 shows that there is a gendered component to the relationship between home life during high school and patterns of civic participation in college. For example, the relationship between the total amount of parental civic activity and interest to which one was exposed (pooling mother’s and father’s activity and interest, as was done in Figure 5.4) and one’s level of participation in voluntary civic organizations during the first year of college is roughly the same for male and female students (male students: r = .09, p = .10; female students: r = .08, p = .04). However, when this result is broken down by gender of parent, we see that these students were more likely to be affected by the parent who shared their gender. More specifically, male respondents appeared to be more likely to mimic the attitudes and behaviors of their fathers than those of their mothers (father: r = .11, p = .06; mother: r = .03, p = .62), while female respondents appeared to be more responsive to their mothers than to their fathers (father: r = .05, p = .18; mother: r = .10, p = .01). 2 Similar

results are found when examining how much civic talk each subject was exposed to in the home before coming to college.

78

Chapter 5

Expected Change in Participation

1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0 –0.2

0.16

Expected Change in Likelihood of Voting

Voluntary Civic Organizations

Below Average Above Average Parental Interest and Activity

Voter Turnout

0.14 0.12 0.10 0.08 0.06 0.04 0.02 0 –0.02 –0.04

Below Average

Above Average

Parental Interest and Activity

FIGURE 5.4 The intervening effect of parents’ civic activity and interest on the rela-

tionship between civic talk and civic participation Source: C-SNIP Panel Survey Notes: The line on each bar represents the 95 percent confidence interval about the estimate. The effect of civic talk is calculated as the level of civic participation estimated to have been engaged in by individuals who were exposed to civic talk, minus that of those who were not exposed, all other factors in the model held at their means.

Civic Engagement As discussed in Chapter 4, individuals who are interested in politics and current events are more likely to participate in civic activities (Verba et al. 1995). Data from the C-SNIP Panel Survey corroborate this; respondents with above average levels of interest during high school were more

Do You Matter?

79

civically active during their first year of college (participation in voluntary civic organizations: t = 8.80, p < .01; voter turnout: t = 5.37, p < .01). As discussed in the previous chapter, people who feel that their actions will make a difference—those with a sense of efficacy—are also more likely to participate in civil society (Verba et al. 1995). Data from the C-SNIP Panel Survey corroborate this as well; subjects with above average levels of political efficacy during high school were more likely to participate in civic activities in college (participation in voluntary civic organizations: t = 3.87, p < .01; voter turnout: t = 2.48, p = .02). To test whether civic engagement enhances the effect of civic talk in the same way that prior experience does, the split-sample style of analysis presented in Figures 5.3 and 5.4 was used to take each subject’s level of political interest and efficacy before engaging in civic talk into account. Figure 5.5 shows the intervening effect of prior interest in politics and current events. The results more or less mimic those presented in the previous figures. The higher the subject’s level of interest in politics and current events was before being engaging in civic talk, the larger the effect of civic talk is for participation in voluntary civic organizations. Moreover, the relationship between civic talk and civic participation for individuals with below average levels of civic engagement is not statistically significant. The results for voter turnout in the bottom half of Figure 5.5 are not as conclusive because the effect of civic talk was not statistically significant for the above average half of the sample. However, the data at least suggest a trend of a stronger relationship between civic talk and voter turnout as prior interest in politics and current events increases. Figure 5.6 (see page 81) tells a somewhat different story for past levels of political efficacy. For participation in voluntary civic organizations, the estimated effect of civic talk is positive and statistically significant in both the below average and above average efficacy cohorts, and the error bars for the two estimates overlap. Thus, these results show that efficacy neither enhances nor mitigates the effect of civic talk on participation in voluntary civic organizations. In contrast, the bottom half of Figure 5.6 shows that the effect of civic talk on voter turnout increases as efficacy increases. Moreover, the civic talk effect is not statistically significant for those individuals with below average levels of political efficacy. A likely explanation for the difference between the top and bottom portions of Figure 5.6 is that efficacy, as it was measured in the C-SNIP Panel Survey, is more germane to political acts than it is to participation in voluntary civic organizations. The C-SNIP Panel Survey question used to measure efficacy asked respondents, “How much would you agree or

Chapter 5

Expected Change in Participation

80

1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0 –0.2 –0.4

Expected Change in Likelihood of Voting

0.25

Voluntary Civic Organizations

Below Average

Above Average

Prior Political Interest

Voter Turnout

0.20 0.15 0.10 0.05 0

Below Average

Above Average

–0.05 –0.10

Prior Political Interest

FIGURE 5.5 The intervening effect of interest in politics and current events on the

relationship between civic talk and civic participation Source: C-SNIP Panel Survey Notes: The line on each bar represents the 95 percent confidence interval about the estimate. The effect of civic talk is calculated as the level of civic participation estimated to have been engaged in by individuals who were exposed to civic talk, minus that of those who were not exposed, all other factors in the model held at their means.

disagree with this statement: ‘People like me don’t have any say about what the government does?’” Thus, it could be the case that subjects’ attitudes about the government were not important in their decision to participate in non-political student organizations. In contrast, the feeling that one’s actions will (or will not) have a tangible effect on the actions

Do You Matter?

81

Voluntary Civic Organizations

Expected Change in Participation

1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0

Below Average Above Average Prior Political Efficacy

Voter Turnout

Expected Change in Likelihood of Voting

0.25 0.20 0.15 0.10 0.05 0

Below Average –0.05

Above Average

Prior Political Efficacy

FIGURE 5.6 The intervening effect of political efficacy on the relationship between

civic talk and civic participation Source: C-SNIP Panel Survey Notes: The line on each bar represents the 95 percent confidence interval about the estimate. The effect of civic talk is calculated as the level of civic participation estimated to have been engaged in by individuals who were exposed to civic talk, minus that of those who were not exposed, all other factors in the model held at their means.

of the government is more logically related to the decision of whether to vote. It is therefore not surprising that political efficacy affects the relationship between civic talk and voter turnout but does not affect the relationship between civic talk and participation in voluntary civic organizations.

82

Chapter 5

Political Preferences The strength of one’s political preferences is also an indicator of his or her willingness to participate in civil society (Verba et al. 1995). As with civic engagement, individuals with strong political preferences are more likely to feel the need to participate in civil society. Data from the C-SNIP Panel Survey corroborate this. For example, subjects who reported stronger-than-average ideological identifications in high school were more active in civic activities during their first year of college (participation in voluntary civic organizations: t = 2.37, p < .05; voter turnout: t = 5.47, p < .01).3 But what effect, if any, does this correlation between strength of political preferences and frequency of civic participation have on the effect of engaging in civic talk? To answer this question, Figure 5.7 presents the relationship between civic talk and civic participation for individuals with above and below average ideological preferences. For both participation in voluntary civic organizations and voter turnout, strength of political preferences enhances the effect of civic talk. In fact, the relationship between civic talk and civic participation is not statistically significant for individuals who adhered to weaker political preference before engaging in civic talk.4 The results in Figure 5.7 document the intervening effect of the strength of one’s political preferences, but what about the direction of those preferences? Do liberals experience a larger boost in civic participation than conservatives after engaging in civic talk, or do conservatives get more out of civic talk than liberals? Or is civic talk “value-free,” in the sense that liberals and conservatives experience similar effects after engaging in such discussions? These questions are of special importance to the C-SNIP study and the case of the University of Wisconsin, Madison. Throughout history, the State of Wisconsin and the University of Wisconsin, Madison, have adhered to a culture of progressive political thought.5 This culture is reflected in the C-SNIP Panel Survey data. On a five-point scale, running from “very conservative” to “very liberal,” subjects on average scored 3.4 during their first year of college (moderate, yet leaning toward the liberal side of the 3 Similar

results appear in an analysis of strength of partisan identity.

4 Similar

results appear in an analysis of strength of partisan identity.

5 This

tradition is exemplified by prolific political leaders such as the Progressive Party’s “Fighting Bob” La Follette and by dark moments in history, such as the violent Vietnam War protests on the University of Wisconsin, Madison, campus during the late 1960s and early 1970s (Bates 1993).

Do You Matter?

83

Expected Change in Participation

2.0 1.5 1.0 0.5 0

Below Average –0.5

0.20

Expected Change in Likelihood of Voting

Voluntary Civic Organizations

Above Average

Prior Ideological Strength

Voter Turnout

0.15 0.10 0.05 0

Below Average

Above Average

–0.05 –0.10 –0.15

Prior Ideological Strength

FIGURE 5.7 The intervening effect of strength of ideological preferences on the

relationship between civic talk and civic participation Source: C-SNIP Panel Survey Notes: The line on each bar represents the 95 percent confidence interval about the estimate. The effect of civic talk is calculated as the level of civic participation estimated to have been engaged in by individuals who were exposed to civic talk, minus that of those who were not exposed, all other factors in the model held at their means.

spectrum).6 Moreover, the more liberal C-SNIP subjects were before they came to college, the more civically active they were during the first year of college (participation in voluntary civic organizations: r = .05, p = .08; voter turnout: r = .16, p < .01).7 The fact that the correlation between 6 Similar

results appear in an analysis of partisan identity.

7 Similar

results appear in an analysis of partisan identity.

84

Chapter 5

Expected Change in Participation

2.0

Voluntary Civic Organizations

1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0

Conservative Liberal Subject’s Ideology

Expected Change in Likelihood of Voting

0.16

Voter Turnout

0.14 0.12 0.10 0.08 0.06 0.04 0.02 0 –0.02

Conservative Liberal Subject’s Ideology

FIGURE 5.8 The intervening effect of ideology on the relationship between civic

talk and civic participation Source: C-SNIP Panel Survey Notes: The line on each bar represents the 95 percent confidence interval about the estimate. The effect of civic talk is calculated as the level of civic participation estimated to have been engaged in by individuals who were exposed to civic talk, minus that of those who were not exposed, all other factors in the model held at their means.

liberalism and civic participation is stronger for voter turnout than for participation in voluntary civic organizations makes sense because the liberal–conservative ideological spectrum is more germane to the realm of politics than it is to non-political civic activity.

Do You Matter?

85

Does the correlation between liberalism and civic participation in the C-SNIP Panel Survey population affect the relationship between civic talk and civic participation? In the same mode as the previous split-sample analyses, Figure 5.8 breaks out the estimated effect of engaging in civic talk on civic participation for liberals and conservatives.8 The top portion of the graph shows that liberals and conservatives alike experienced a significant increase in participation in voluntary civic organizations. In contrast, the bottom portion of Figure 5.8 shows that civic talk does not have a significant effect on voter turnout among conservatives but does among liberals. Mirroring the results for political efficacy presented in Figure 5.6, then, these results shows that ideology plays an intervening role in the relationship between civic talk and the explicitly political act of voting, but not between civic talk and participation in non-political voluntary civic organizations.

Knowledge and Education One final factor that may affect the relationship between civic talk and civic participation is civically relevant knowledge. Individuals who are more knowledgeable about politics and current events are better equipped, and feel more need, to participate in civic activities (Verba et al. 1995). Data provided by the C-SNIP subjects verify this. Individuals who were more knowledgeable about politics and current events before coming to college were more civically active in college (participation in voluntary civic organizations: t = 6.88, p < .01; voter turnout: t = 3.26, p < .01).9 The results in Figure 5.9 show that knowledge about politics and current events affects the relationship between civic talk and civic participation. However, instead of individuals with higher levels of knowledge experiencing a larger civic talk effect, subjects with below average levels of knowledge appeared to experience a larger and more statistically reliable bump in civic participation after engaging in civic talk.10 With regard to participation in voluntary organizations, the civic talk effect is only statistically significant among those with below average levels of knowledge. With regard to voter turnout, the effect of civic talk on those with 8 Similar

results appear in an analysis of partisan identity.

9 Similar

results occur if one examines news media usage in high school as a proxy for knowledge about politics and current events.

10 Similar

results occur if one examines news media usage in high school as a proxy for knowledge about politics and current events.

86

Chapter 5

Voluntary Civic Organizations

Expected Change in Participation

2.5 2.0 1.5 1.0 0.5 0

Below Average –0.5

Prior Knowledge about Politics and Current Events

Voter Turnout

0.25

Expected Change in Likelihood of Voting

Above Average

0.20 0.15 0.10 0.05 0

Below Average

Above Average

–0.05 –0.10 –0.15

Prior Knowledge about Politics and Current Events

FIGURE 5.9 The intervening effect of knowledge on the relationship between civic

talk and civic participation Source: C-SNIP Panel Survey Notes: The line on each bar represents the 95 percent confidence interval about the estimate. Figures are based on the matched data regression analyses in Tables 3.3 and 3.5. The estimated effect of civic talk on civic participation is calculated as the level of civic engagement estimated to have been engaged in by individuals who were exposed to civic talk, minus that of those who were not exposed, all other factors in the model held at their means.

below average knowledge is relatively large but falls just outside the minimal threshold of statistical significance (p = .14). In contrast, the effect of civic talk on voter turnout among those with above average knowledge is relatively small and does not come close to a recognized

Do You Matter?

87

level of statistical significance (p = .98). Consequently, contrary to the results presented thus far in this chapter, the evidence in Figure 5.9 is more in line with the expectations of social psychology scholars that individuals with less civically relevant knowledge will be more influenced by their social environment.11 It is also worth examining the potentially intervening effect of education. Educational attainment works in the same way as knowledge about politics and current events: Those with more of it tend to engage in more civic activity (Lake and Huckfeldt 1998; Verba et al. 1995). While the students in the C-SNIP Panel Survey were at the same level of education when the surveys were administered, variance in academic achievement can be assessed by examining the educational background of respondents’ parents, under the assumption that the educational attainment of parents affects the educational prospects of offspring (Croll 2004; Gimpel et al. 2003). While parents’ education does not correlate with how civically active C-SNIP participants were during their first year of college (participation in voluntary civic organizations: t = −1.35, p = .19; voter turnout: t = −.54, p = .54), the results in Figure 5.10 show that it can play an intervening role in the relationship between civic talk and civic participation. As with civically relevant knowledge, subjects who were raised by parents with below average levels of education appear to have gained more from civic talk than those raised by better-educated parents. In light of the results in Figure 5.9, however, this is not surprising because subjects who were raised by less-well-educated parents were less knowledgeable about politics and current events when they arrived for their first year of college (t = −1.96, p = .05).

Who Gets More Out of Civic Talk? Focus Group Evidence In short, the qualitative focus group data also show that individuals with greater ability and desire to participate in civic activities are more affected by engaging in civic talk. As discussed in Chapter 4, the focus groups showed that recruitment attempts during civic talk conversations were only effective when a student has some basic level of interest in becoming 11 These

results echo the focus group data presented in Chapters 3 and 4, which show that the majority of civic talk conversations center on sharing civically relevant knowledge—that is, facts and opinions about politics and current events, typically in response to what is being covered in the news media.

88

Chapter 5

Expected Change in Participation

2.0 1.5 1.0 0.5 0

Below Average –0.5

0.25

Expected Change in Likelihood of Voting

Voluntary Civic Organizations

Above Average

Parents’ Education

Voter Turnout

0.20 0.15 0.10 0.05 0

Below Average –0.05

Above Average

Parents’ Education

FIGURE 5.10 The intervening effect of parents’ educational attainment on the relationship between civic talk and civic participation Source: C-SNIP Panel Survey Notes: The line on each bar represents the 95 percent confidence interval about the estimate. The effect of civic talk is calculated as the level of civic participation estimated to have been engaged in by individuals who were exposed to civic talk, minus that of those who were not exposed, all other factors in the model held at their means.

active in civil society. As one student said, in response to being asked whether civic talk with his roommate caused him to participate in civic activities, “I want to get involved, but sometimes I just lose my motivation. Either I don’t care enough or I would much rather waste my time doing other things.”

Do You Matter?

89

Conclusion This chapter examined the intervening role that one’s characteristics have in the relationship between civic talk and civic participation. The extant literature suggests two competing hypotheses. Social psychologists suggest that individuals with weaker participatory predilections should be more affected by their social setting. In contrast, political science research on civic participation shows that individuals are not automatically equipped to participate in civil society. Thus, the influence of the social network is expected to be larger when the individual has the desire and ability to participate in civic activities. The majority of the evidence presented in this chapter shows that individuals who are predisposed to participate in civil society before engaging in civic talk get a larger participatory boost out of engaging in civic talk with their peers. In fact, in most cases, the effect of engaging in civic talk is not statistically significant among those individuals who are less predisposed to engage in civic activities. The two notable exceptions are knowledge about politics and current events and parents’ levels of education. In these cases, individuals with below average knowledge and who were raised by parents with below average levels of educational attainment actually gained more from talking to their peers about politics and current events. So in some cases, the effect of civic talk promotes civic participation among individuals who might not do so unless they are pushed into it by their social network of peers. This said, from a normative perspective the results presented in this chapter should serve as a cautionary example to scholars of social influence. In this current era of civic engagement in the United States, some have argued that increased social connectedness may be a means by which to revive civil society. For example, Putnam (1994, 2000) found that communities with higher levels of societal interaction and connectedness have higher levels of citizen involvement in public affairs and, as a consequence, are governed more effectively. In the same spirit, the data presented throughout this book confirm that a causal link exists between interpersonal interaction and the strength of participatory democracy. However, the evidence in this chapter shows that, by and large, only those of us who are predisposed to be civically active reap the benefits of engaging in civic talk. Thus, these results suggest that civic talk is not a panacea for civic disengagement. This issue, and the broader question of how we might use social networks to maintain and strengthen participatory democracy, will be given deeper consideration in Chapter 8.

6 Do Your Peers Matter? We both have different opinions, and I’m going to do what I want, and she’s going to do what she wants. So I don’t—we don’t—really influence each other so much. —C-SNIP Focus Group Study participant

he evidence presented in the previous chapter shows that one’s personal characteristics mediate the effect that civic talk has on civic participation. Specifically, the preponderance of the data shows, perhaps disappointingly to those concerned with civic disengagement, that individuals who are willing and able to participate in civic activities will be more positively affected by engaging in civic talk. This issue, and its normative consequences, will be considered in greater detail in Chapter 8. Given that who we are matters, the next question to address is whether the characteristics of our peers matter. The data presented in this chapter show that the relationship between civic talk and civic participation varies based on the characteristics of the people with whom we engage in discussion. In this analysis I examine three types of peer characteristics: social intimacy, disagreement, and civic expertise. With regard to intimacy, the C-SNIP data show that trust between peers enhances the effect of civic talk. The evidence also shows that civic talk is more influential when peers are similar to one another, both in general and with regard to political preferences. Finally, the C-SNIP data show that civic talk has a larger influence on our behavior when we think that the person we are engaged in discussion with is knowledgeable about politics and current events.

T

92

Chapter 6

Should the Characteristics of Our Peers Matter? Social Intimacy Existing scholarship in political science, sociology, and social psychology shows that for one person to influence another, the two must share a certain level of social connectedness. Moreover, these literatures suggest that the deeper that social connection is, the greater the effect peers will have on an individual’s patterns of behavior. One reason social influence is enhanced by social intimacy is that individuals are more likely to comply with the preferences of a peer group to continue to be a part of that group (Dawson et al. 1977; Mendelberg 2002; Putnam 2000). For example, in a review of the extant literature on deliberation, Mendelberg (2002) finds that when individuals identify emotionally with the group they are attempting to deliberate with, they are more willing to cooperate and accept the will of the group during the deliberation process. Social intimacy facilitates compliance with the group because the tighter the social bond is between an individual and his or her peer group, the more the individual has to lose from the dissolution of his or her membership in the group. In turn, the more he or she has to lose, the greater the individual’s incentive is to comply with the will of the group. This is especially the case in small groups, such as peer networks, where social bonds are more intimate and individuals have a greater capacity to monitor the actions of their peers to make sure they comply with group norms. The role of information is another way in which social intimacy might affect the relationship between civic talk and civic participation (Brady et al. 1999; Brehm et al. 2002). For example, in a study of how political parties and interest groups recruit individuals to work on their behalf, Brady and his colleagues (1999) show that individuals are 28 percent more likely to agree to become political activists if they know the person who is attempting to recruit them. In explaining this result, Brady and colleagues contend that the more intimate the social relationship is between two people, the more perfect the information is about one another. For example, if you and I have a close social relationship, I am likely to have a general idea of how likely you are to want to participate in civic activities and what types of activities you might desire to participate in. As such, I have knowledge about how to persuade you to adopt a certain way of behaving. The extant literature also shows that even simple physical proximity can make social persuasion more effective (Festinger et al. 1950; Kopstein

Do Your Peers Matter?

93

and Reilly 2000). A classic example of this phenomenon is in Festinger and colleagues’ (1950) seminal study of a student dormitory at the Massachusetts Institute of Technology. Among the numerous discoveries made about social interaction and human relations in that study, one of the most interesting was that individuals are friendlier with, and tend to share the preferences and patterns of behavior of, the people to whom they live near. Festinger and colleagues surmised that physical proximity facilitates social influence by facilitating social interaction. Otherwise stated, the more you cross paths with a person, the more likely you are to form a social bond with that person. In turn, that social bond can facilitate social influence over attitudes and behavior.

Homogeneity The existing literature also suggests that homogeneity in the peer group should have an effect on peers’ influence. The discussion of social intimacy above suggested that, for one person to influence another, the two must have a certain level of social connectedness. The deeper that connection is, the greater the effect of civic talk should be. Not surprisingly, human beings stereotypically prefer to have these types of close social relationships with individuals who are more like them (Alesina and La Ferrara 2002; Brehm et al. 2002; Costa and Kahn 2003). This tendency is reflected in the C-SNIP Panel Survey data. For example, how similar subjects feel they are to their roommates correlates with how much they trust their roommates (r = .49, p < .01). So, assuming that individuals are more likely to heed the words of people they trust, homogeneous peer groups may be more effective at influencing individuals to participate in civil society than heterogeneous ones (Lupia and McCubbins 1998).1 Beyond basic similarities and differences between peers, one specific form of social network homogeneity that is of special interest to this study is (dis)agreement about politics and current events. Some existing works suggest that political disagreement among peers will lead an individual to withdraw from civil society. For example, Mutz (2002, 2006) examined this question through a series of studies on “cross-cutting” peer networks. In the case of these studies, disagreement is operationalized as having electoral preferences that are different from those of your peers. Mutz’s 1 As

Lupia and McCubbins (1998, 74) put it, “People also use knowledge of others’ backgrounds and knowledge of their circumstances to evaluate their character and incentives. These strategies lead all of us to rely more on the testimony of family, friends, and colleagues and less on the testimony of strangers.”

94

Chapter 6

analyses suggest that disagreement is relatively infrequent in social networks, presumably because humans tend to associate with people who are similar to them. Moreover, her studies suggest that individuals who disagreed with their peers about who should win the 1992 presidential election were less active in campaign activities during that election. Mutz argues that disagreement suppressed participation because individuals who disagreed with their peers were more likely to have ambivalent political views and subsequently were less interested in participating in the campaign. In addition to the potentially negative effects of disagreement-driven ambivalence on participatory democracy, Mutz (2002, 2006) suggests that disagreement can lead to conflict in the social network. In turn, this potential for conflict could cause the individual to feel pressure to keep harmony in the social network by not participating in politics. Evidence that supports this assumption can also be found in social psychological studies of attitudes. Studies on cognitive consistency show that when an individual encounters information that goes against his or her own beliefs, a psychological state of negative arousal develops (Fazio et al. 1977; Fink et al. 1983; Fiske and Taylor 1991; Heider 1958; Michener and DeLamater 1999; Petty et al. 1997; Wood 2000). To put an end to these feelings of “cognitive dissonance,” individuals often attempt to minimize the salience of the situation that is causing the negative affect (Fiske and Taylor 1991; Heider 1958; Michener and DeLamater 1999). In other words, individuals who interact with peers who espouse beliefs unlike their own may withdraw from civic activities to reduce their own state of cognitive dissonance and maintain harmony in the peer group. This said, while a growing body of evidence suggests that disagreeable civic talk suppresses civic participation, not all studies lead to the same conclusion. For example, Huckfeldt and his colleagues (2002, 2004) found that political disagreement is actually quite frequent in social networks. They suggest that disagreement can occur among peers because social groups are not closed systems; even if some of your peers disagree with you about politics and current events, someone else in your social circle is likely to agree with your views. Moreover, in direct contrast to the findings presented by Mutz (2002, 2006), Huckfeldt and colleagues (2002, 2004) found that exposure to disagreement did not have a negative effect on voter turnout in the 1996 and 2000 presidential elections.2 2 However,

in line with Mutz (2002, 2006), Huckfeldt and colleagues (2002, 2004) show that political ambivalence—in this case, operationalized as having both positive and negative feelings toward candidates running for office—increases when an individual is exposed to disagreement in his or her social network.

Do Your Peers Matter?

95

Ironically, as with the argument put forth by Mutz, the argument forwarded by Huckfeldt and colleagues (2002, 2004) also has a basis in the social psychology literature on cognitive dissonance. In response to disagreement about politics with one’s peers, a person can choose to withdraw from civic activity to reduce dissonance and maintain social harmony in the social network. However, an alternative strategy for soothing dissonance is to remain active in civic activities while simultaneously filtering out, discounting, or avoiding the political views of disagreeable peers (Heider 1958).3 In other words, most of us have friends, colleagues, and relatives with whom we disagree about politics. Rather than withdrawing from civil society, a common response is just to avoid discussing politics with those people to maintain stable social relationships.

Civic Expertise While there is a great deal of disagreement in the literature on the role of disagreement in social networks, there is consensus on the role of expertise. Specifically, existing research suggests that the more our peers know about politics and current events, the more likely we are to listen to and act on what they say. To illustrate why civic expertise matters, it is helpful to think about peer groups as a set of interpersonal relationships governed by a system of “social exchange” (Brehm et al. 2002). We enter into and maintain social relationships because our peers supply us with goods that we cannot or do not wish to supply for ourselves. Otherwise stated, in a complex world we cannot be specialists in every aspect of life. Thus, we often defer to the judgments and opinions of experts in our social circles when creating our own judgments and opinions. Our natural deference to experts is especially acute in the realm of politics and current events (Bar-Tal and Saxe 1990; Beck et al. 2002; Huckfeldt et al. 2000; Lupia and McCubbins 1998; Lake and Huckfeldt 1998; Mendelberg 2002; Popkin 1995). For example, in his oft-cited study of information shortcuts and public opinion, Popkin (1995) shows that the

3 Campbell

(2006) makes a similar argument in his study of community heterogeneity and voter turnout. This analysis shows a “U”-shaped relationship between community heterogeneity and turnout. Highly homogeneous communities have high levels of turnout because such communities are politically unified and “civically motivated.” Highly heterogeneous communities, by contrast, have high turnout due to political motivations; individuals who disagree with one another are motivated to be active to respond to the political challenges posed by the opposition.

96

Chapter 6

public is largely inattentive to politics and current events. However, when attention is deemed necessary—say, when the government is about to make a major change in public policy—the public tends to base its opinions on the opinions of political experts (Popkin 1995). In a similar vein, Bar-Tal and Saxe (1990) show that the public pays more attention to, and puts more stock in, the opinions of political experts. So, just as we would go to the expert on poker in our social group to learn what “rivering a belly buster” is (Bellin 2002), we might turn to the resident expert on politics and current events in our social circle for advice on such matters.

Do Your Peers Matter? Survey Evidence Social Intimacy Do the characteristics of our peers affect the relationship between civic talk and civic participation? To address this question, I start by examining the role of social intimacy. Specifically, I start by assessing the effect that trust has on the relationship between civic talk and civic participation. Trust is built through repeated interactions between individuals that, if positive, bring those individuals closer to each other. Thus, social trust is a particularly good measure of social intimacy because it indicates how bonded individuals are to each other. Moreover, and of specific interest to the study of civic participation, social trust correlates with higher levels of civic participation (Putnam 2000). The C-SNIP Panel Survey data show that subjects were relatively trusting of their roommates; on average, subjects trusted their roommates somewhere between “somewhat” and “very much” (a mean of 2.4 on a 0–3 scale ranging from “not at all” to “very much”). The data also show that trust increased the likelihood that roommates would engage in civic talk (t = 6.29, p < .01). Thus, the next logical question to ask is whether this form of social intimacy enhances the influence that peers have over one another. In line with the analyses presented in Chapter 5, C-SNIP Panel Survey participants were split into two groups based on how trusting each respondent was of his or her roommate; specifically, respondents were categorized as either having above average or below average levels of trust in their roommates. The regression analyses in Chapter 3 (see Tables 3.3 and 3.5), which showed the causal effect of civic talk on civic participation for the entire C-SNIP Panel Survey sample, were then conducted separately on the two subsets of subjects. This procedure allows the influence of civic talk to be estimated for roommate pairs with different levels of trust. The results of the analysis are presented in Figure 6.1.

Do Your Peers Matter?

Expected Change in Participation

3.5

Voluntary Civic Organizations

3.0 2.5 2.0 1.5 1.0 0.5 0

Below Average –0.5

0.25

Expected Change in Likelihood of Voting

97

Above Average

Trust in Roommate

Voter Turnout

0.20 0.15 0.10 0.05 0

Below Average

Above Average

–0.05 –0.10

Trust in Roommate

FIGURE 6.1 The intervening effect of trust on the relationship between civic talk

and civic participation Source: C-SNIP Panel Survey Notes: The line on each bar represents the 95 percent confidence interval about the estimate. The effect of civic talk is calculated as the level of civic participation estimated to have been engaged in by individuals who were exposed to civic talk, minus that of those who were not exposed, all other factors in the model held at their means.

Figure 6.1 shows that social trust enhances the effect of civic talk on civic participation. The civic talk effect for the portion of the sample with above average trust is larger than that for the portion of the sample with below average trust. In fact, the figure shows that only subjects who trusted their roommates experienced a statistically significant increase in civic participation due to engaging in civic talk.

98

Chapter 6

In addition to trust, the C-SNIP Panel Survey contains a measure of how much each subject enjoyed spending time with his or her roommate. On average, subjects “somewhat” enjoyed spending time with their roommates (a mean of 2.0 on a 0–3 scale ranging from “not at all” to “very much”).4 As with social trust, enjoying spending time with one’s roommate correlates with more frequent instances of civic talk discussions between roommates (t = 11.26, p < .01). But does enjoying time with one’s roommate enhance the effect of civic talk the way trust does? Figure 6.2 answers this question using the same method of analysis as that used to produce Figure 6.1. Looking first at the bottom of Figure 6.2, enjoying spending time with one’s roommate enhanced the relationship between civic talk and voter turnout. The increase in turnout due to being exposed to civic talk is much larger in the cohort who reported above average levels of enjoyment. Moreover, the relationship between civic talk and voter turnout is not significant in the cohort with below average levels of enjoyment. The evidence is less clear, however, when we look at the top portion of Figure 6.2. These results show that enjoying the company of one’s roommate does not have an effect on the relationship between civic talk and participation in voluntary civic organizations; both the above average and below average cohorts experienced a significant increase in participation in such groups due to being exposed to civic talk (the regression coefficient for civic talk in the above average cohort is significant at p < .10), and the error about the two estimates overlap. This said, a looser interpretation of these results suggest a trend whereby those who did not enjoy spending time with their roommates actually participated in more civic organizations due to being exposed to civic talk.5 What explains this anomaly? We might expect trusting and enjoying time with one’s peers to be more or less exchangeable feelings. So why might they function differently when it comes to the relationship between civic talk and participation in voluntary civic organizations? While trust and enjoyment of time spent with one’s roommate are highly correlated

4 This

figure is roughly in line with national figures. For example, data from the 1992 CrossNational Election Studies (Beck et al. 1995) show that, of the people we choose discuss politics with in our social networks, we consider 92 percent of them to be close friends (see Klofstad et al. 2009, table 2).

5 Similar

results appear when one examines the overall amount of conversation that the subject had with his or her roommate and how interested the respondent felt the roommate was in his or her life.

Do Your Peers Matter?

Expected Change in Participation

3.0

Voluntary Civic Organizations

2.5 2.0 1.5 1.0 0.5 0

Below Average –0.5

0.20

Expected Change in Likelihood of Voting

99

Above Average

Enjoys Spending Time with Roommate

Voter Turnout

0.15 0.10 0.05 0

Below Average

Above Average

–0.02 –0.04

Enjoys Spending Time with Roommate

FIGURE 6.2 The intervening effect of enjoying spending time with one’s roommate

on the relationship between civic talk and civic participation Source: C-SNIP Panel Survey Notes: The line on each bar represents the 95 percent confidence interval about the estimate. The effect of civic talk is calculated as the level of civic participation estimated to have been engaged in by individuals who were exposed to civic talk, minus that of those who were not exposed, all other factors in the model held at their means.

in this sample, they are not perfectly so (r = .66, p < .01). It is therefore possible that, under certain circumstances, these two forms of social intimacy will affect the relationship between civic talk and civic participation in different ways. As such, one way to interpret these results is that individuals who do not enjoy spending time with their roommates have a

100

Chapter 6

greater incentive to leave their dorm and join voluntary organizations for a social outlet, while those who do enjoy their roommates’ company are happy socializing in the dorm.6 This said, it is important to underscore that the results in the top portion of Figure 6.2 only suggest a trend; the results are not definitive.

Peer Group Homogeneity When asked how similar they felt to their roommates, the C-SNIP Panel Survey subjects, on average, felt that they were between “somewhat similar” and “somewhat dissimilar” to their roommate (a mean of 1.6 on a 0–3 scale ranging from “not at all similar” to “very similar”). This relatively low level of homogeneity is not surprising, given that roommate pairings were made at random. As with social intimacy, C-SNIP Panel Survey participants who reported feeling similar to their roommates were more likely to engage in civic talk (t = 7.15, p < .01). The intervening role of this general measure of homogeneity on the relationship between civic talk and civic participation is presented in Figure 6.3. In line with expectations, the effect of civic talk on both measures of civic participation is larger, and more statistically reliable, for individuals who felt that they were similar to their roommates (the effect of civic talk on participation in voluntary civic organizations for the below average cohort comes close to, but does not reach, statistical significance at p = .11). The results in Figure 6.3 show that a general sense of similarity to one’s peers enhances the effect of civic talk. But what effect do the political attitudes of one’s peers have on the relationship between civic talk and civic participation? On average, the C-SNIP Panel Survey subjects felt that their own views on politics and current events fell somewhere between “somewhat similar” and “somewhat dissimilar” to those of their roommates (a mean of 1.6 on a 0–3 scale ranging from “not at all similar” to “very similar”). Again, this relatively low level of homogeneity is not surprising, given that roommates were paired at random.7 The C-SNIP Panel Survey data also show that roommates who shared similar views about politics and current events engaged in more civic talk (t = 6.16, p < .01). 6 In

contrast, because voting is an a-social activity, the same logic does not apply to voter turnout.

7 This

figure is roughly in line with national figures. For example, data from the 1992 CrossNational Election Studies (Beck et al. 1995) show that, of the people we choose discuss politics with in our social networks, about half of them agree with us about politics and current events (see Klofstad et al. 2009, table 3).

Do Your Peers Matter?

Expected Change in Participation

1.8

101

Voluntary Civic Organizations

1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0 –0.2

Below Average

Above Average

Overall Similarity with Roommate

Expected Change in Likelihood of Voting

0.25

Voter Turnout

0.20 0.15 0.10 0.05 0 –0.05

Below Average

Above Average

–0.10 –0.15 –0.20

Overall Similarity with Roommate

FIGURE 6.3 The intervening effect of overall similarity with one’s roommate on the

relationship between civic talk and civic participation Source: C-SNIP Panel Survey Notes: The line on each bar represents the 95 percent confidence interval about the estimate. The effect of civic talk is calculated as the level of civic participation estimated to have been engaged in by individuals who were exposed to civic talk, minus that of those who were not exposed, all other factors in the model held at their means.

The results in Figure 6.4 show that the effect of peer group homogeneity on the relationship between civic talk and civic participation is magnified when we move our attention from a general measure of similarity to a more specific measure of political similarity. Relative to the results in Figure 6.3, the gap in the size of the civic talk effect between the above and below average political agreement cohorts is larger.

102

Chapter 6

Expected Change in Participation

2.0 1.5 1.0 0.5 0

Below Average –0.5

0.20

Expected Change in Likelihood of Voting

Voluntary Civic Organizations

Above Average

Political Similarity with Roommate

Voter Turnout

0.15 0.10 0.05 0

Below Average

Above Average

–0.05 –0.10

Political Similarity with Roommate

FIGURE 6.4 The intervening effect of political similarity with one’s roommate on

the relationship between civic talk and civic participation Source: C-SNIP Panel Survey Notes: The line on each bar represents the 95 percent confidence interval about the estimate. The effect of civic talk is calculated as the level of civic participation estimated to have been engaged in by individuals who were exposed to civic talk, minus that of those who were not exposed, all other factors in the model held at their means.

In summation, the data presented in Figures 6.3 and 6.4 offer a different perspective on political disagreement compared with the extant literature on this subject. Existing research shows that the peer group homogeneity enhances the effect of civic talk, and the C-SNIP data corroborate this. The story becomes more complicated, however, when we

Do Your Peers Matter?

103

shift our attention to heterogeneity. Half of the existing scholarship on this topic suggests that exposure to political disagreement suppresses civic participation (Mutz 2002, 2006), while the other half says that it does not (Huckfeldt et al. 2002, 2004). The C-SNIP Panel Survey data suggest a split of the difference between these two perspectives. These data do not show that exposure to disagreeable peers suppresses civic participation. For this to be the case, the “below average” bars presented in Figures 6.3 and 6.4 would need to be negative and statistically significant. In contrast, the evidence shows that exposure to disagreeable peers had no detectable effect on civic participation.

Civic Expertise Finally, along with social intimacy and peer group homogeneity, one would expect the civic expertise of our peers to affect how much we act on what they say. One measure of expertise in the C-SNIP Panel Survey was based on how much each subject felt that his or her roommate knew about politics and current events. On average, subjects felt that their roommates were “somewhat” knowledgeable about civic matters (a mean of 1.9 on a 1–3 scale ranging from “not much” to “a great deal”).8 C-SNIP Panel Survey subjects also reported that they were more likely to engage in civic talk when they perceived their roommates to be knowledgeable about politics and current events (t = 8.18, p < .01). The results in Figure 6.5 conform to the expectation that individuals with greater levels of expertise in politics and current events will have more sway over their peers when engaging in civic talk. Focusing first on the top portion of the figure, the magnitude of the effect of civic talk on participation in voluntary civic organizations is approximately the same regardless of how knowledgeable one’s roommate is about politics and current events. However, the effect of being exposed to civic talk is statistically significant only for the cohort whose roommates had above average political knowledge. Likewise, looking at the bottom portion of the figure, only those subjects in the above average knowledge cohort experienced a significant increase in voter turnout due to being exposed to civic talk (the estimated effect of engaging in civic talk is significant at p < .10 for the above average cohort). 8 This

figure is roughly in line with national figures. For example, data from the 1992 CrossNational Election Studies (Beck et al. 1995) show that of the people we choose discuss politics with in our social networks, more than half of them are knowledgeable about politics and current events (see Klofstad et al. 2009, table 3).

104

Chapter 6

Expected Change in Participation

2.5 2.0 1.5 1.0 0.5 0

Below Average –0.5

0.25

Expected Change in Likelihood of Voting

Voluntary Civic Organizations

Above Average

Roommate’s Political Knowledge

Voter Turnout

0.20 0.15 0.10 0.05 0 –0.05

Below Average

Above Average

–0.10 –0.15 –0.20 –0.25

Roommate’s Political Knowledge

FIGURE 6.5 The intervening effect of roommate’s knowledge about politics and

current events on the relationship between civic talk and civic participation Source: C-SNIP Panel Survey Notes: The line on each bar represents the 95 percent confidence interval about the estimate. The effect of civic talk is calculated as the level of civic participation estimated to have been engaged in by individuals who were exposed to civic talk, minus that of those who were not exposed, all other factors in the model held at their means.

The C-SNIP Panel Survey also asked subjects how active and interested they thought their roommates were in politics and current events. On average, subjects lived with roommates whom they felt were somewhere between “not very” and “somewhat” active and interested in civic activities (a mean of 1.4 on a 0–3 scale ranging from “not at all” to “very”). As was

Do Your Peers Matter?

105

the case with knowledge, subjects with civically engaged roommates were exposed to more civic talk (t = 10.80, p < .01).9 This result makes sense, given that the focus group results presented in Chapter 3 show that civic talk most often occurs in response to activities, such as consumption of news media content or participation in civic activities, engaged in by peers. The intervening role of roommates’ civic activity and interest is examined in Figure 6.6. Beginning with the bottom half of the figure, the data show that roommates’ civic engagement enhances the effect of civic talk on voter turnout. In fact, the below average cohort’s pattern of voting does not appear to have been affected at all by being exposed to civic talk. In contrast, the top half of Figure 6.6 shows the opposite result for participation in voluntary civic organizations; only the below average cohort experienced a statistically significant increase in participation in civic organizations due to being exposed to civic talk. What explains this difference? In line with results presented in Chapter 5, it could be a product of the less political nature of participation in voluntary organizations. The activities included in this measure of participation were civically oriented student groups such as student publications and charitable and community service groups. Although some of these organizations no doubt engage in politically relevant activities, these forms of civic participation are not overtly aimed at affecting the government. In contrast, voting is an explicitly political act. Consequently, civically engaged individuals should be expected to have a positive effect on the voter turnout rates of their roommates but not necessarily on their participation in non-political civic organizations. But what explains the positive effect that peers with below average levels of civic engagement have on the relationship between civic talk and participation in voluntary civic organizations? Recall that the same phenomenon was found in Figure 6.2, where not enjoying spending time with one’s roommate had a positive effect on the relationship between civic talk and participation in voluntary civic organizations. Also recall that the C-SNIP Panel Survey participants saw great virtue in civic participation: As earlier chapters documented, 91 percent of the respondents reported that civic participation was somewhat or very important. Consequently, perhaps C-SNIP respondents did not enjoy spending time with civically 9 Subjects

were also marginally more civically active if their roommates were active and interested in politics and current events (voluntary civic organizations: t = 1.92, p = .07; voter turnout: t = 1.94, p = .06), suggesting that some behavioral modeling is occurring between roommates. (See Chapter 4 for a discussion of social norms in the context of civic talk among peers.)

106

Chapter 6

Expected Change in Participation

1.8

Voluntary Civic Organizations

1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0 –0.2

Below Average

Above Average

Roommate’s Civic Interest and Activity

Expected Change in Likelihood of Voting

0.35

Voter Turnout

0.30 0.25 0.20 0.15 0.10 0.05 0 –0.05

Below Average

Above Average

–0.10 –0.15

Roommate’s Civic Interest and Activity

FIGURE 6.6 The intervening effect of roommate’s civic interest and activity on the

relationship between civic talk and civic participation Source: C-SNIP Panel Survey Notes: The line on each bar represents the 95 percent confidence interval about the estimate. The effect of civic talk is calculated as the level of civic participation estimated to have been engaged in by individuals who were exposed to civic talk, minus that of those who were not exposed, all other factors in the model held at their means.

disengaged roommates and thus had a greater incentive to leave their dorm rooms and participate in voluntary civic organizations as a social outlet. The C-SNIP Panel Survey data offer evidence in favor of this hypothesis, as respondents with civically disengaged roommates were less likely to say that they enjoyed spending time with their roommates than respondents with engaged roommates (t = −5.65, p < .01).

Do Your Peers Matter?

107

Do Our Peers Matter? Focus Group Evidence One of the benefits of focus groups is that they are semi-structured. While the moderator has a predefined path of questions to go through during the session, participants are given free rein to weave their way through those questions as they see fit. This method of data collection allows rich qualitative data to be generated on topics of discussion that are the most salient to the focus groups’ participants. The potential cost associated with the freeform nature of focus groups, however, is the prospect that not every subject of interest to the researcher will be addressed by the participants. This ended up being the case with the subject of peer characteristics: The focus group participants did not volunteer much evidence on the impact that their roommates’ characteristics had on the relationship between civic talk and civic participation. This said, as addressed in Chapter 3, participants in each of the focus groups provided a great deal of insight on the relationship between political disagreement and civic talk. In each of the focus groups, the desire to avoid disagreements or arguments with roommates was a common explanation for why civic talk was infrequent and sometimes even actively avoided. This type of exchange was typical in all four of the focus groups: PARTICIPANT: I think my roommate has the opposite view that I

do. I don’t know, because I don’t really talk to her, but I get that impression. So I figure I will just avoid it just to save time so we don’t fight about it or, like, I don’t know, get in a disagreement. MODERATOR: That’s interesting. So it’s a way to avoid conflict? PARTICIPANT: Yeah.

Unlike the C-SNIP Panel Survey data, these types of statements do not give us much insight on how disagreement might directly mitigate or enhance the effect of civic talk. They do show, however, that most individuals feel a desire to avoid civic talk with peers who do not share their political preferences. Consequently, the focus group data show that disagreement has an indirect impact on the civic talk effect by decreasing the amount of civic talk that one chooses to engage in.

Conclusion Building on the results presented in Chapter 5, the purpose of this chapter was to shift focus from ourselves to the characteristics of our peers. Overall, the results presented in this chapter show that the characteristics of

108

Chapter 6

our peers matter a great deal. Social intimacy tends to enhance the positive effect of civic talk on civic participation. Peers who are more similar to us, both in general and with specific regard to political preferences, also have more influence over whether civic talk leads us to participate in civic activities. Finally, the data show that civic expertise usually enhances the civic talk effect. In light of these findings, it is worth commenting on one way the results presented in this chapter might differ from those generated by more naturally occurring social relationships. As discussed in Chapters 1 and 2, when constructing a social network of peers we select the individuals we wish to socialize with from the set of people who are available in our immediate social environment. In contrast to this process of network formation, the peer groups examined in this book were exogenously forced onto the study subjects. As discussed in Chapter 2, while this process does not perfectly mirror how networks form in nature, it presents a unique way to estimate the impact that peer characteristics have on the civic talk effect. For example, think about the people in your inner circle of friends, loved ones, and colleagues. No doubt you feel a high level of social intimacy with these individuals, which either led you to select them as peers in the first place or developed over time as you forged relationships with them. In more technical terms, there is probably very little variance in the level of social intimacy in your social network. Consequently, without variance on this variable, we cannot get an accurate estimate of how intimacy affects the relationship between civic talk and civic participation. In contrast, by randomly assigning individuals to social networks, the C-SNIP studies created variance on intimacy, as well as the other peer characteristics examined in this chapter, allowing me to more accurately estimate how those characteristics affect the relationship between civic talk and civic participation. In the next chapter, the focus of this study shifts away from the immediate effect of civic talk on civic participation to address three additional questions. First, given the extant literature’s focus on individual-level antecedents of civic participation, how does the effect of civic talk compare with the effect of one’s individual characteristics? Second, while civic talk has a significant effect on civic participation, does it have an effect on other politically relevant attitudes and behavior? And finally, does the effect of civic talk last beyond the initial point of exposure?

7 The Significant and Lasting Effect of Civic Talk

n the previous chapters I showed how, why, and under what conditions civic talk affects one’s patterns of civic participation. While these findings extend our knowledge on participatory democracy, three important questions have been left unanswered. First, given the extant literature’s focus on individual-level antecedents of civic participation, how does the magnitude of the civic talk effect compare to that of one’s individual characteristics? Second, while civic talk has a significant effect on civic participation, does it have an effect on other politically relevant attitudes and behaviors? Finally, does the relationship between civic talk and civic participation last beyond the initial point of exposure to such conversations? To answer the first two questions, I take a closer look in this chapter at the first two waves of the C-SNIP Panel Survey. With regard to the relative significance of civic talk, these data show that the effect of civic talk is typically equal to or greater than the effect of individual-level antecedents of civic participation. The C-SNIP data also show that civic talk has a significant effect on other politically relevant attitudes and behaviors, such as political knowledge and civic engagement. Finally, to assess whether the effect of civic talk lasts beyond the point of initial exposure I examine the third wave of the C-SNIP Panel Survey collected during the study population’s fourth year in college. These data show that the effect of civic talk lasts beyond the initial point of exposure—in this case, three years into the future. Further analysis

I

110

Chapter 7

shows that the boost in civic participation initially after engaging in civic talk is the mechanism by which the effect of civic talk lasts into the future— that is, causing an initial increase in civic participation is the mechanism by which the effect of civic talk lasts.

The Effect of Civic Talk Relative to Other Antecedents of Civic Participation The main focus of this study has been to show how, why, and under what circumstances civic talk leads individuals to participate in civil society. However, as discussed throughout this book, the preponderance of research on civic participation focuses on individual-level characteristics and largely excludes social-level factors such as civic talk. This begs the question of how the effect of civic talk compares with individual-level antecedents of civic participation. Figure 7.1 presents a first cut at this question by examining the effect of civic talk relative to, and while simultaneously controlling for, a number of different measures of one’s prior experience with civic participation. For the sake of simplicity, the results presented are only for participation in voluntary civic organizations as the dependent variable. The results are similar, however, for voter turnout. As in previous chapters, the effect of civic talk (white bars) is calculated as the level of participation engaged in by individuals who were exposed to civic talk, minus that of those who

Civic Talk

Prior Experience

Expected Change in Participation

4 3 2 1 0

Past Participation

Parental Activity/Interest

Civic Talk at Home

–1 FIGURE 7.1 The effect of civic talk relative to prior experience Source: C-SNIP Panel Survey Notes: The line on each bar represents the 95 percent confidence interval about the estimate. Figures are based on regression analyses of matched data (see Appendix C).

The Significant and Lasting Effect of Civic Talk

Civic Talk

111

Civic Engagement

Expected Change in Participation

1.5

1.0

0.5

0

Interest in Politics

Efficacy

FIGURE 7.2 The effect of civic talk relative to civic engagement Source: C-SNIP Panel Survey Notes: The line on each bar represents the 95 percent confidence interval about the estimate. Figures are based on regression analyses of matched data (see Appendix C).

were not exposed, all other factors in the model held at their means. The effect of prior experience (shaded bars) is calculated between the mean and maximum, all other factors held at their means. The analysis also controls for a lag of the dependent variable (i.e., civic participation in high school) and dormitory fixed effects (none of which were statistically significant). As shown in Chapter 3 (Figure 3.1), the bars on the far left-hand side of Figure 7.1 show that one’s past experience participating in civic activities has a larger effect on civic participation than civic talk does. This said, the effect of civic talk is still significant, even after controlling for prior participatory experience. The middle of the figure shows that civic talk has a statistically significant effect on civic participation, while the attitudes and behavioral patterns of one’s parents do not. The far right-hand side of the figure shows that the effect of engaging in civic talk with one’s peers is about on par with the effect of engaging in such conversations with one’s parents before leaving home for college; the size of the effects for both factors are roughly equal, and the error about these two estimates overlap. The results presented in Figure 7.2 extend this analysis to civic engagement, measured as one’s interest in politics and current events and one’s sense of political efficacy. The left-hand side of the figure shows that the effect of civic talk is on par with that of one’s interest in politics, even while simultaneously controlling for interest in politics. The results on the right-hand side of the figure are similar for political efficacy.

112

Chapter 7

The data presented in Figure 7.3 show the influence of civic talk relative to the strength of one’s political preferences. The C-SNIP data show that individuals with stronger political preferences are significantly more likely to participate in voluntary civic organizations (partisan strength: r = .14, p < .01; ideological strength: r = .14; p < .01). Nonetheless, the data in Figure 7.3 show that the effect of civic talk on civic participation is on par with the strength of one’s political preferences. As a final set of analyses, Figure 7.4 presents the effect of engaging in civic talk relative to that of civic expertise, one’s level of preparedness to participate in civic activities. At the far left-hand side of the figure, the data show that civic talk and one’s self-reported level of knowledge about politics and current events have the same effect on civic participation. Focusing on the middle of the figure, the same can be said when we compare the effect of civic talk to how often one watches or reads the news for information about politics and current events. The far right-hand side tells the same story when we examine the educational attainment of respondents’ parents (the proxy for education used in Chapter 5, since each respondent had the same degree of educational attainment when the C-SNIP Panel Survey was conducted).

Civic Talk

Preference Strength

Expected Change in Participation

1.5

1.0

0.5

0

Partisan Strength

Ideological Strength

FIGURE 7.3 The effect of civic talk relative to strength of political preferences Source: C-SNIP Panel Survey Notes: The line on each bar represents the 95 percent confidence interval about the estimate. Figures are based on regression analyses of matched data (see Appendix C).

The Significant and Lasting Effect of Civic Talk

113

Civic Talk

Civic Expertise

Expected Change in Participation

1.5

1.0

0.5

0

Knowledge

Media Use

Parents’ Education

FIGURE 7.4 The effect of civic talk relative to civic expertise Source: C-SNIP Panel Survey Notes: The line on each bar represents the 95 percent confidence interval about the estimate. Figures are based on regression analyses of matched data (see Appendix C).

The Effect of Civic Talk on Other Civically Relevant Phenomena Another way to illustrate the substantive significance of civic talk is to shift the focus away from civic participation to other civically relevant phenomena. I begin this examination by looking at the effect of civic talk on various measures of civic engagement. The data in Figure 7.5 show the percent change in civic engagement as a consequence of engaging in civic talk. In contrast to previous analyses in this book, I examine percent change in various measures of civic engagement instead of change in the raw scale of each factor because the variables are scaled differently. Each of the analyses presented in Figure 7.5 controls for the respondent’s level of civic engagement before being engaging in civic talk (i.e., a lag of the dependent variable) and dormitory fixed effects (none of which were significant). The effect of civic talk is calculated as the level of civic engagement among those who were exposed to civic talk, minus that of those who were not exposed, all other factors held at their means. The results show that, even after controlling for how civically engaged one was before engaging in civic talk, being exposed to such discourse still increases one’s level of engagement with politics and current events.

114

Chapter 7

Expected Percentage Change in Engagement

14 12 10 8 6 4 2 0

Political Interest

Political Involvement

Efficacy

FIGURE 7.5 The effect of civic talk on civic engagement Source: C-SNIP Panel Survey Notes: Error bars are not presented because each effect is statistically significant at p ≤ .05. Figures are based on regression analyses of matched data (see Appendix C).

Moving from left to right on the graph, we see that students who engaged in civic talk increased their level of interest in politics and current events by about 13 percent. The middle of the graph shows that civic talk also led subjects to be 13 percent more likely to say that they would engage in political activities, as opposed to community activities, as a way to address

Expected Percentage Change in Expertise

14 12 10 8 6 4 2 0

Political Knowledge

News Media Use

FIGURE 7.6 The effect of civic talk on civic expertise Source: C-SNIP Notes: Error bars are not presented because each effect is statistically significant at p ≤ .05. Figures are based on regression analyses of matched data (see Appendix C).

The Significant and Lasting Effect of Civic Talk

115

important political issues. Finally, the far right-hand side of the graph shows that civic talk increased one’s sense of political efficacy by close to 6 percent. The data presented in Figure 7.6 show similar results for civic expertise. Again, even after controlling for past levels of expertise, the left-hand side of the figure shows that civic talk increased one’s self-reported level of knowledge about politics and current events by about 8 percent. The right-hand side of the graph shows that individuals who engaged in civic talk were 12 percent more likely to watch or read the news media for information on politics and current events. While civic talk has a meaningful effect on a number of civically relevant phenomena, however, it is important to discuss the various factors on which civic talk does not have an effect. Examples of these are presented in Table 7.1. After controlling for one’s attitudes prior to engaging in civic talk (i.e., a lag of the dependent variable), the first two columns of Table 7.1 show that civic talk has no effect on the intensity or direction of this population’s political preferences. The last two columns of the table also show that civic talk had no effect on trust in government or one’s sense that the government is run for the benefits of all citizens as opposed to the benefit of the better heeled and more powerful among us. TABLE 7.1

Variables on which civic talk has no effect (regression analysis) Partisan strengtha (1)

Partisanshipa (2)

Government run for allb (4)

−.02 (.05)

.09 (.17) 2.15*** (.19)

Civic talk among roommates

.03 (.06)

Attitude in high school

.67*** (.23)

.80*** (.03)

.53*** (.03)

Constant

1.92 (.66)

1.52** (.56)

.81** (.38)

.23

.66

.30

Adjusted R2 Akaike’s information criterion (AIC)

>−.01 (.05)

Trust in governmenta (3)

−2.33 (3.17)

1,091

Source: C-SNIP Panel Survey. Model Type: aOrdinary least squares (Imai et al. 2007c). bLogistic regression (Imai et al. 2007d). Note: The matched data set is used in this analysis (see Appendix C). AIC is twice the number of parameters in the model, minus twice the value of the model’s log-likelihood. Dormitory assignment fixed effects were included in the analysis but are omitted from the table (none of these coefficients were statistically significant). An analysis of ideology and ideological strength shows results that are similar to those for partisanship and partisan strength. Standard errors are in parentheses. N = 1,044. *p ≤ .10; **p ≤ .05; ***p ≤ .01.

116

Chapter 7

The Lasting Effect of Civic Talk The previous two sections illustrate the significant effect that civic talk has on participatory democracy. Not only does the effect of civic talk stack up to individual-level antecedents of civic participation, but it also has a positive effect on other civically relevant attitudes and behavior. All of these findings, however, focused on the initial effect of engaging in civic talk. What about the future? Does the effect of civic talk last beyond the initial moments after such conversations? To address this question, I examine data from the third and final wave of the C-SNIP Panel Survey, collected during this populations’ fourth year in college.

Patterns of Civic Participation over Time Before examining whether the effect of civic talk lasts beyond the initial point of exposure, it is instructive to look at how and why the patterns of civic participation in the study population changed as the students progressed from high school to their fourth year of college. These trends are listed in Table 7.2. The first two columns of Table 7.2 show that the participants in the C-SNIP Panel Survey became far less civically active once they arrived at college. As discussed in Chapter 3, one explanation for this decline in civic activity is that, after leaving home for college, civic participation becomes less of a priority as a person spends his or her time and energy learning how to navigate life as an independent adult. Data from the C-SNIP Panel Survey offer evidence that supports this hypothesis. For example, survey respondents felt that they became less informed about politics and current

TABLE 7.2

Patterns of civic participation over time (mean activity levels) 1st year of college (2)

High school (1)

4th year of college (3)

Participation in voluntary civic organizations (0–21 point activity scale)

6.6

>

2.4

<

3.2

Participation in political activities (0–6 point activity scale)

1.2

>

.6

<

.8

51

<

64

Voter turnout (%)

Source: C-SNIP Panel Survey. > or < indicates a significant paired difference of means at p < .01.

The Significant and Lasting Effect of Civic Talk

TABLE 7.3

117

Changes in civically relevant characteristics during college (means)

Interest in politics and current events (1–4 point scale) Knowledge about politics and current events (1–3 point scale) Strength of partisan identity (1–3 point scale) Strength of ideology (1–3 point scale) Attention to news media (days per week that news is watched or read)

1st year of college

4th year of college

2.7 1.9 2.0 2.0

3.0 2.0 2.1 2.2

3.3

4.0

Source: C-SNIP Panel Survey. p < .01 for differences in these characteristics between the first and fourth years of college (paired t-tests).

events (t = −7.28, p < .01), and were less likely to consume news about politics and current events (t = −7.36, p < .01) after moving to college. As also discussed in Chapter 3, another explanation for the decline in civic participation between high school and college is the prevalence of service learning opportunities at the time that this cohort of students was progressing through primary and secondary school. Participation in these programs may have led these students to participate in an abnormally high number of civic activities during high school. Data from the C-SNIP Focus Group study corroborate this hypothesis. In contrast to this trend of decreased civic participation between high school and the first year of college, comparing the second and third columns in Table 7.2 shows that the C-SNIP Panel Survey participants became significantly more civically active between their first and fourth years of college. Table 7.3 offers a suggestion for why these students became more civically active as they progressed through college. During their fourth year of college, study subjects were more interested in and knowledgeable about politics and current events, held stronger partisan and ideological beliefs, and were more likely to use the news media to keep up to date on politics and current events than in their first year of college. These trends documented in Tables 7.2 and 7.3 echo the extant literature on civic participation, which shows that individuals become more civically active and engaged as they age (e.g., Putnam 2000).1 1 An

additional explanation is that young people have an incentive to be civically active in high school to build their résumé and get into college. This incentive may be weak early in one’s college career but strengthens as one approaches graduation and moving on to graduate school or the workforce.

118

Chapter 7

Does the Civic Talk Effect Last? The evidence presented thus far shows that civic talk has a significant effect on civic participation during the initial point of exposure. But what role, if any, did civic talk during the first year of college play in increasing the civic activity of the C-SNIP participants during the remainder of their tenure in college? To address this question, I used a style of regression analysis similar to that used in previous chapters. In the previous analyses, I examined the immediate effect of civic talk on civic participation (civic participation measured during the second wave of the C-SNIP Panel Survey as the dependent variable). In the analyses that follow, I look at the effect of civic talk that occurred during the first year of college on patterns of civic participation during the fourth year of college (civic participation measured during the third wave of the C-SNIP Panel Survey as the dependent variable). I begin the analysis in Figure 7.7 with an examination of the effect of civic talk on participation in voluntary civic organizations. As done previously, the analysis included civic talk, civic participation in high school (i.e., a lag of the dependent variable), and dormitory fixed effects (none of which were significant). The effect of civic talk is calculated as the level of participation engaged in by those who were exposed to civic talk, minus that of those who were not exposed, all other factors in the model held at their means.

Expected Change in Participation

1.4 1.2 1.0 0.8 0.6 0.4 0.2 0

1st Year of College

4th Year of College

FIGURE 7.7 The effect of civic talk on participation in voluntary civic organizations

over time Source: C-SNIP Panel Survey Notes: The line on each bar represents the 95 percent confidence interval about the estimate. Figures are based on regression analyses of matched data (see Appendix C).

The Significant and Lasting Effect of Civic Talk

119

The effect of civic talk on voter turnout and political participation over time (regression analysis)

TABLE 7.4

Voter turnout

Political participation

1st year of college (1)

4th year of college (2)

1st year of college (3)

4th year of college (4)

Civic talk

.31‡ (.19)

−.25 (.25)

−.02 (.09)

.13† (.10)

Participation in political activities during high school

.14** (.06)

.19*** (.07)

.37*** (.04)

.31*** (.03)

Constant Akaike’s information criterion (AIC)

−2.11 (2.20)

2.47 (4.50)

.55 (.76)

.33 (.86)

1,283

1,187

.24

.14

Source: C-SNIP. Model Type: Logistic regression (Imai et al. 2007d). Note: The matched data set is used in this analysis (see Appendix C). AIC is twice the number of parameters in the model, minus twice the value of the model’s log-likelihood. Dormitory assignment fixed effects were included in the analysis but are omitted from the table (none of the coefficients were statistically significant). Standard errors are in parentheses. N = 1,044. †p = .19; ‡p = .12; *p ≤ .10; **p ≤ .05; ***p ≤ .01.

The bar at the far left of Figure 7.7 shows the initial effect that civic talk had on civic participation. As documented in Chapter 3, civic talk increased participation in voluntary civic organizations during the first year of college by 38 percent (an increase from 2.1 to 2.9 on the voluntary organization participation scale). The bar on the right-hand side of the graph shows that civic talk caused a 20 percent increase in civic participation three years later (an increase from 3.0 to 3.6 on the voluntary organization participation scale). While these results show that the influence of civic talk is lasting, they suggest that the effect may diminish over time. However, the confidence intervals around the two estimates overlap. This indicates that, on average, the positive effect of civic talk on participation in voluntary civic organizations did not significantly decrease, despite the passage of three years. Table 7.4 extends this analysis to voter turnout and political participation. The first two columns of the table examine the effect of civic talk on voter turnout. As shown in Chapter 3 (Table 3.5), the first column of Table 7.4 indicates that civic talk has an immediate effect on voter turnout (as shown in Figure 3.2, the expected increase in turnout is estimated to be significant at p < .05). In the second column of the table, however, the civic talk coefficient for voter turnout during one’s fourth year of college

120

Chapter 7

is not statistically significant. This shows that conversations about politics and current events did not have a lasting effect on voter turnout in this population. The final two columns of Table 7.4 examine the effect of civic talk on political participation. As shown in Chapter 3 (Table 3.4), the third column of the table indicates that the relationship between civic talk and political participation was not significant in the immediate term. The fourth column of Table 7.4 also shows that civic talk did not have a statistically significant effect on political participation in the long run (although the size of the coefficient increases and comes relatively close to a minimal threshold for statistical significance at p = .19).

Why Does the Effect Last? What explains the lasting effect of civic talk on civic participation? Despite the lack of extant research on this question, the literature on path dependence offers a theoretical framework for why phenomena persist over time. Simply stated, “path dependence” means that the past plays a role in what will happen in the future. More precisely, path dependence is a process of self-reinforcement “in which preceding steps in a particular direction induce further movement in the same direction” (Pierson 2000, 252). Self-reinforcement occurs because of increasing returns: Once a course of action is initiated, it becomes increasingly costly to change its direction over time (Pierson 2000).2 A process of increasing returns such as this one is initiated by a formative moment in time referred to as a “critical juncture” (Collier and Collier 1991; Pierson 2000). These critical moments may seem insignificant when they occur and yet be traced forward to large and significant outcomes in the future (Pierson 2000).3 While the concept of path dependence traditionally has been applied to studies of institutional development and policymaking, one could argue that civic participation is also a self-reinforcing phenomenon. For example, a number of studies show that individuals who have voted in the past are more likely to vote in the future (Fowler 2006; Gerber et al. 2003; Plutzer 2002). Additional research suggests that other forms of civic activity may also be self-reinforcing (see, e.g., Brady et al. 1999; Burns et al. 2001; 2 For

example, after more than two hundred years of conducting congressional elections in the United States under the system of single-member district plurality, it would be extremely difficult (logistically and politically) to change to a system of proportional representation.

3 For

example, while Rosa Parks was only one woman on one bus in one Southern city, her act of civil disobedience in 1955 was a catalyst for the rise of the Civil Rights Movement in the United States.

The Significant and Lasting Effect of Civic Talk

121

Putnam 2000; Rosenstone and Hansen 1993; Verba et al. 1995). For example, Verba and his colleagues (1995) found that individuals who participate in civic activities through their church or a voluntary civic organization also tend to be active in other civic activities, such as volunteering for a political campaign. Research on political socialization also shows that past patterns of civic participation, especially the experiences one has during adolescence and young adulthood, are highly influential in determining how civically active a person will be in the future (see, e.g., Campbell 2006; Jennings and Niemi 1981). Civic participation is self-reinforcing in this way because the more civically active an individual is today, the easier it becomes for him or her to participate in the future.4 One reason that civic participation is subject to such increasing returns is that individuals require resources (such as knowledge about how to participate) and psychological motivations (such as civic engagement) to participate in civic activities. These prerequisites can be obtained by participating in civic activities (Verba et al. 1995). Otherwise said, we take the experiences we acquire through participating in civic activities today and apply them to participation in the future. Civic participation is also self-reinforcing because citizens who are mobilized to participate in civic activities tend to have been civically active in the past. This is the case because agents of civic mobilization, such as political parties and other civic organizations, are “rational prospectors” (Brady et al. 1999) who want their mobilization efforts to be effective. Thus, they target their blandishments at individuals who are already participating in civil society. Assuming that civic participation is a self-reinforcing behavior, past patterns of participation will help determine future patterns of participation. Consequently, if engaging in civic talk causes an individual to become more active in civil society, that initial effect should be felt after the point of exposure as the individual parlays his or her current participatory experiences into future participation in civic activities. In other words, causing an initial increase in civic participation could be the mechanism by which the positive effect of civic talk lasts into the future. Table 7.5 offers a test of this hypothesis. Since the long-term effect of civic talk was not statistically significant for voter turnout and political participation, I restrict this analysis to participation in voluntary civic organizations. 4 To

clarify, my use of path dependence theory in this context varies somewhat from traditional use of the theory. Typically, processes are seen as path-dependent if they become more costly to change over time. In contrast, I am suggesting that civic participation is pathdependent because as a person becomes more active in such activities, it because less costly to participate over time.

122

Chapter 7

Explaining the effect of civic talk during the first year of college on participation in voluntary civic organizations during the fourth year of college (regression analysis)

TABLE 7.5

(1)

(2)

Civic talk

.62* (.31)

.33 (.31)

Participation in voluntary civic organizations during high school

.17*** (.03)

.09** (.04)

Participation in voluntary civic organizations during 1st year of college Constant Adjusted R2

.36*** (.04) 2.50 (2.16)

1.88 (1.91)

.09

.24

Source: C-SNIP Panel Survey. Model Type: Ordinary least squares (Imai et al. 2007c). Columns: (1) Analysis controlling for the amount of civic participation in which each respondent engaged before engaging in civic talk. (2) Analysis as in column 1, with added consideration of the amount of civic activity each subject engaged in immediately after engaging in civic talk. Note: The matched data set is used in this analysis (see Appendix C). Dormitory assignment fixed effects were included in the analysis but are omitted from the table (none of the coefficients were statistically significant). Standard errors are in parentheses. N = 1,044. *p ≤ .10; **p ≤ .05; ***p ≤ .01.

The first column of Table 7.5 shows the lasting effect of civic talk on participation in voluntary civic activities while controlling for the amount of civic participation in which each respondent engaged before being exposed to civic talk; this result was presented graphically on the right-hand side of Figure 7.7. In the second column of Table 7.5, however, the analysis accounts for the amount of civic activity each subject engaged in immediately after engaging in civic talk (the dependent variable in previous analyses). As with the analysis of causal mechanisms in Chapter 4, the goal of adding this variable to the analysis is to “explain away” the effect of civic talk. If the initial boost in civic participation caused by civic talk during one’s first year in college explains why the effect of civic talk lasts into one’s fourth year in college, the civic talk coefficient should no longer be statistically significant after a measure of civic participation during one’s first year in college is added to the model. This will occur only if participation during the first year of college accounts for the variance in civic participation during the fourth year of college that was originally accounted for by civic talk. The results in column 2 of Table 7.5 show that this is the

The Significant and Lasting Effect of Civic Talk

123

case. Once participation during one’s first year in college is added to the analysis, the civic talk coefficient is no longer statistically significant. Thus, these results show that causing an initial increase in civic participation is the mechanism by which the effect of civic talk lasts into the future.5

Why Does the Effect Not Last? If civic talk has a lasting effect on participation in voluntary civic organizations, then why does the effect not last for political participation and voter turnout? With regard to political participation, perhaps the best answer is that this form of civic activity was not very popular in this student population. During these students’ first year of college, 35 percent were not active in voluntary civic organizations, compared with 68 percent who were not active in political activities. During the students’ fourth year of college, the ratio improved a bit; 15 percent of students were not active in civic organizations, compared with 45 percent who were not active in political activities. Nonetheless, the survey participants still were not very active in political activities in their fourth year of college. In line with findings in Chapter 5 that civic talk has an effect only on individuals who are willing and able to become civically active, the lack of a lasting relationship between civic talk and political participation make sense. If a person is unwilling or unable to participate in a specific type of civic activity, no amount of conversation about politics and current events is going to prompt him or her to participate. In other words, because civic talk did not have an initial effect on political participation, the mechanism that allows the effect of civic talk to last over time was not present in the case of political participation. It is worth noting, however, that while the long-term effect of civic talk on political participation did not reach a recognized level of statistical significance (p = .19), we can at least be 81 percent certain that there was a longer-term effect. This presents a puzzle. Why would civic talk not have an immediate effect on political participation but, perhaps, a marginal effect on such behavior three years later? This type of phenomenon is referred to as a “sleeper effect” (Campbell 2006; Jennings and Stoker 2004), a situation in which a stimulus does not have an immediate effect on behavior but does so later in a person’s life. Tables 7.2 and 7.3 show that the C-SNIP Panel Survey participants became significantly more active in and engaged with politics between their first and final years of college. 5A

two-stage model leads to the same conclusion.

124

Chapter 7

Enhancements in the willingness and desire to participate in politics over time could be driving this sleeper effect. Again, it is necessary to underscore that this result is not statistically significant at an accepted level of confidence. Nonetheless, the data in Table 7.4 suggest a trend that civic talk can sometimes have a delayed effect on an individual’s patterns of political participation. While a lack of interest in participating in political activities helps explain the lack of a statistically significant long-run relationship between civic talk and political participation, the same cannot be said of voter turnout. As documented in Table 7.2, a majority of the students in the study voted in both 2004 and 2006. While the turnout figures are likely a bit inflated (Faler 2005), it is still safe to say that voting was more popular among these students than was participating in other types of political activities. To explain this result, then, it is useful to ask how voting differs from other forms of civic participation. One key way is that it is far more sporadic. For example, citizens can be active in political activities and voluntary civic organizations at any point in time. One can only vote, however, when an election is held. As a consequence, talking about politics and current events shortly before the 2004 presidential primary encouraged this student population to vote in that election. Since the next election was the 2006 midterm election, however, there was no opportunity to parlay discussion about politics and current events immediately into more voter participation. In other words, the relative infrequency of elections makes initiation of a self-reinforcing pattern of increased voter turnout vis-à-vis civic talk harder to achieve.6

Conclusion The purpose of this chapter has been to expand the analysis beyond the immediate effect that civic talk has on civic participation. In doing so, I examined three questions that were left unanswered in previous chapters. First, in reference to the extant literature’s focus on individual-level antecedents of civic participation, I showed that the effect of civic talk is typically as large as, if not larger than, that of one’s own characteristics. That is, neither social-level nor individual-level factors on their own sufficiently explains how active a person chooses to be in civil society. Both factors are important. 6 Another

explanation may be that discussions about politics and current events that occurred around the 2004 presidential primary were not germane to the 2006 midterm elections.

The Significant and Lasting Effect of Civic Talk

125

Second, I examined the effect that civic talk has on other civically relevant phenomena. In switching the focus of this study to additional dependent variables, I found that, along with increasing civic participation, civic talk has a positive effect on how civically engaged and knowledgeable about politics and current events people are. In contrast, however, the analysis shows that civic talk does not have an effect on the strength or direction of one’s political preferences or views on the legitimacy of the government. Taken together, and in line with the focus group data presented in Chapter 3, these results suggest that civic talk is more about relaying information and less about changing minds. Finally, I examined the potentially lasting effect of civic talk on civic participation. The data show that for certain types of civic activities, the effect of civic talk is lasting. Asking whether the civic talk effect lasts beyond three years is outside the scope of the C-SNIP Panel Survey. It is worth underscoring, however, that the data show that civic talk engaged in by first-year college students had a significant effect on how they behaved when they were fourth-year students (i.e., likely graduating seniors). In other words, as these young adults prepared to leave college and become full-fledged citizens, the civic talk in which they engaged three years earlier still shaped how they choose to participate in the processes of democratic governance. Given that the attitudes and patterns of behavior that we develop during our college years help determine the attitudes and patterns of behavior we carry with us for the rest of our lives, it is likely that the effect of civic talk experienced by the C-SNIP study population will last for many years into the future—and, arguably, over their lifespan.

8 Peers, Politics, and the Future of Democracy Man may not be a political animal, but he is certainly a social animal. Voters do respond to the cues of commentators and campaigners, but only when they can match those cues up with the buzz of their own social group. . . . Voters go into the booth carrying the imprint of the hopes and fears, the prejudices and assumptions of their family, their friends, and their neighbors. For most people, voting may be more meaningful and more understandable as a social act than as a political act. —Louis Menand (2004, 96)

It is time to take the next step and ask some larger questions. When you are socialized by peers what, if anything, does it portend for the performance of a political system? —Sara Silbiger (1977, 189)

n his assessment of what political scientists know about predicting elections, the famed cultural historian Louis Menand came to the conclusion that there is much about civic participation that we still do not understand. Of the myriad explanations that exist for why we vote and whom we choose to vote for, no single theory has a monopoly on the truth. However, one thing that is certain is that social context has a place on this list of explanations. We may not be Aristotelian political animals, but we are social animals; we experience politics and current events with and through our peers. In this book, I have used two new and innovative data sources to demonstrate empirically what we already know from experience: social context has a significant influence on participatory democracy. Specifically, I have shown that engaging in civic talk—discussing politics and current events with peers—leads people to become more civically active. But, as Sara Silbiger asked more than thirty years ago, what does it matter if civic

I

128

Chapter 8

talk has an impact on civil society? The purpose of this chapter is to conclude my argument by presenting answers to this question. This chapter begins with an overview of the findings presented in this study. I then examine what the civic talk effect might portend for the future of participatory democracy. This discussion shows that civic talk plays a crucial role in maintaining the performance of democratic political systems. Thus, research on this subject needs to continue, and practitioners in civil society need to keep using civic talk and peer networks as resources to strengthen participatory democracy.

Summary of the Results A great deal of scholarship already exists on civic participation, and the body of political science literature on social networks has been growing over the past twenty years. However, extant scholarship suffers from a critical problem: an inability to show causation. Existing works on civic talk show a strong correlation between how much a person talks with his or her peers about politics and how active that individual is in civic activities (see, e.g., Campbell and Wolbrecht 2006; Huckfeldt et al. 1995; Huckfeldt and Sprague 1991, 1995; Kenny 1992, 1994; Lake and Huckfeldt 1998; McClurg 2003, 2004; Mutz 2002). However, we cannot conclude with this evidence that civic talk is causing individuals to be more civically active. An equally plausible explanation for the relationship between talk and participation is that being civically active causes a person to talk about politics (reciprocal causation). Another problem is that individuals who are active in civic activities might consciously choose to associate with people who are interested in talking about politics (selection bias). Finally, we also need to consider the possibility some factor that has not been accounted for could be causing people to both talk about politics and participate in civic activities (endogeneity or omitted variable bias). As discussed in Chapters 1 and 2, existing data sets and methods of analysis cannot account for these analytical biases. To address this problem, I collected panel survey data at three separate points in time from a cohort of students at the University of Wisconsin, Madison. As discussed in Chapters 1 and 2, the design of the study allowed me to account for analytical biases by examining change in behavior over time in a population of individuals who were randomly assigned to their peer groups (i.e., their dormitory roommates). In addition, the survey data were run through a matching pre-processing procedure to make the results more like those that would be generated by a controlled laboratory experiment.

Peers, Politics, and the Future of Democracy

129

Four focus groups were also conducted with a separate cohort of students at the University of Wisconsin, Madison, to verify the findings from the C-SNIP Panel Survey. Chapters 3 and 4 discussed what these two new data sets can tell us about the causal relationship between civic talk and civic participation. Chapter 3 showed that our peers do have influence over how we behave; C-SNIP Panel Survey subjects who engaged in civic talk were 38 percent more active in voluntary civic organizations and 7 percentage points more likely to have voted in the 2004 presidential primary. Chapter 4 showed that the relationship between civic talk and civic participation is a product of four factors. Conversations about politics and current events provide us with informational resources, increase our engagement with politics and current events, subject us to opportunities to be recruited into civic activities, and expose us to civic-minded attitudes and norms of behavior. Of these four factors, recruitment and civic engagement carry the most explanatory weight. With the causal relationship between civic talk and civic participation quantified and explained, the middle portion of this book assessed how this form of social influence varies under different circumstances. The results presented in these chapters show that the effect of civic talk varies significantly depending on who you are and with whom you associate. With regard to one’s own characteristics, Chapter 5 showed that individuals who are predisposed to participate in civic activities get more of a participatory boost out of engaging in civic talk with their peers. With regard to the characteristics of the members of our social networks, Chapter 6 showed that peers with whom we are socially intimate, peers who have expertise on politics and current events, and peers who are similar to us (in general and specifically with regard to shared political preferences) have more sway over our patterns of civic participation vis-à-vis civic talk. Chapter 7 answered three additional questions. First, the C-SNIP Panel Survey data show that the effect of civic talk is typically equal to or greater than the effect of individual-level antecedents of civic participation. Consequently, when we study civic participation we need to take more than just individual-level characteristics into consideration; we also need to pay attention to social-level determinants of individual behavior. Second, the C-SNIP Panel Survey data show that civic talk has a significant effect on other civically relevant attitudes and behavior, such as knowledge about and psychological engagement with politics and current events. Finally, to assess whether the effect of civic talk lasts beyond the point of initial exposure, I examined the third and final wave of the C-SNIP

130

Chapter 8

Panel Survey collected during the study population’s fourth year in college. These data show that the effect of civic talk lasts beyond the initial point of exposure—in this case, three years into the future. Further analysis showed that the boost in civic participation initially after engaging in civic talk is the mechanism by which the effect of civic talk lasts into the future.

Normative Considerations The evidence presented in this book shows that civic talk has the capacity to significantly influence participatory democracy. However, the normative implications of civic talk have not yet been given due consideration. Civic talk may bring people into civil society, but does it always lead to positive outcomes for participatory democracy?

Deliberation as a Benchmark for Civic Discourse To gain a better understanding of the normative implications of the civic talk effect, it is useful to examine the literature on deliberation (see, e.g., Barabas 2004; Conover et al. 2002; Delli Carpini et al. 2004; Mendelberg 2002; Page and Shapiro 1992). As discussed in Chapter 2, civic talk is not as purposive or formal as deliberation. However, deliberation scholars have carefully considered the implications of citizens’ discussing politics and current events with one another, making this research agenda a useful lens through which to examine the normative consequences of civic talk. To use deliberation in this way, it is helpful to begin by reviewing what this form of discourse is. Two relatively recent reviews of the literature on this subject show that consensus is growing on what deliberation entails (Delli Carpini et al. 2004; Mendelberg 2002). Delli Carpini and his colleagues (2004, 319) conclude that deliberation is a form of “discursive” discourse. Discussions are discursive when participants have the opportunity “to develop and express their views, learn the positions of others, identify shared concerns and preferences and come to understand and reach judgments about matters of public concern.” Mendelberg (2002, 153) offers a similar assessment of the literature, concluding that “the promise of deliberation is its ability to foster the egalitarian, reciprocal, reasonable, and open-minded exchange of language.” Theoretically, the value of this form of dialogue is that, by participating in a deliberation, individuals gain a greater appreciation of the needs of others and are subsequently able to formulate public policies that favor the collective good of society over one’s self-interests.

Peers, Politics, and the Future of Democracy

131

How Deliberative Is Civic Talk? Although discussing politics and current events in informal peer networks is not the same as the formal process of deliberation, does civic talk approximate this benchmark of discursive discourse? To answer this question, it is useful to begin by reviewing how common these types of conversations are. While we are by no means constantly engaged in political discourse, civic talk is pervasive. As discussed in Chapter 2, data from the 2008–2009 American National Election Studies Panel Study show that about 91 percent of the American public engages in civic talk at least once a week. This said, civic talk is not purposively sought out by individuals. Instead, these types of discussions are an unintended byproduct of people going about their normal daily routine (Downs 1957; Klofstad et al. 2009; Walsh 2004). For example, a husband and wife might discuss issues covered in the news over dinner, or a group of friends at a party might end up talking about the current election. So we engage in at least some amount of civic talk as we go about our daily lives. But what exactly goes on when people talk to one another about politics and current events? One might assume that the level of discourse occurring in peer networks is low in quality because most people are inattentive to, and have low levels of knowledge about, politics and current events. For example, data from the 2008 American National Election Studies Time Series Study show that a striking 55 percent of the American public reported being only “somewhat” or “not much” interested in the 2008 campaigns.1 Nonetheless, the evidence presented in this book shows that civic talk is still a useful way to engage the public with politics and current events. For example, the C-SNIP Focus Group Study data in Chapter 3 show that most of what goes on during civic talk discussions is the relaying of useful information about politics and current events. Moreover, a number of participant-observation studies show that individuals are more willing to discuss politics in a meaningful way when they are in supportive venues such as peer groups that are protected from the scrutiny and normative constraints of wider society (Eliasoph 1998; Harris-Lacewell 2004; Walsh 2004). In short, while the content of civic talk may not always meet our idealized expectations of deliberative discourse, these conversations are an easy way to keep a largely inattentive public engaged with the processes of democratic governance. If deliberation is our benchmark for quality discourse, it is also necessary to assess how cooperative and egalitarian these informal discussions 1This

figure is based on question A1a (variable number V083001a) in the pre-election questionnaire.

132

Chapter 8

are. Social networks certainly have the potential to meet these two ideals. First, numerous studies show that cooperation is more likely to occur when individuals repeatedly meet face to face, as they do in peer groups (Dawes et al. 1990; Delli Carpini et al. 2004; Mendelberg 2002; Putnam 2000; Sally 1995). For example, in a meta-analysis of more than one hundred experimental studies of the “Prisoners’ Dilemma,” a task that requires individuals to collude with one another to reach an optimal outcome, Sally (1995, 78) found that a “dilemma with discussion before each round would have 40% more cooperation than the same game with no discussion.” Second, since most people have a social circle with which they affiliate, this discussion forum is relatively egalitarian because all but the misanthropes among us have the opportunity to engage in civic talk. Finally, in comparing civic talk with deliberation, it is necessary to ask whether informal conversations about politics and current events actually have policy implications the way deliberations (in theory) do. Admittedly, it is hard to envision that an informal conversation among peers, no matter how enlightened, egalitarian, and cooperative, will lead directly to the implementation of public policy. As this study has shown, however, civic talk can have a meaningful impact on policy outcomes by leading individuals to articulate their preferences to the government by voting. Moreover, I have shown that civic talk encourages individuals to engage in non-political forms of civic participation. Participation in such activities can also affect public policy. For example, volunteering at a soup kitchen has a direct effect on hunger and homelessness; donating supplies to a local school has a direct effect on education; and organizing an after-school program has a direct effect on juvenile delinquency.

Four Caveats Civic talk lives up to the ideals of deliberative discourse in many ways. But is civic talk always inherently desirable from a normative standpoint, as has been assumed throughout this book? The answer is “not always.” First, the central thesis of this book is that civic talk encourages civic participation. To put this phenomenon into a broader normative context, however, it is useful to examine who among us participates in civil society. Unfortunately, in the past half-century, traditionally disadvantaged members of society such as the poor and the undereducated have been withdrawing from public life at higher rates than the more advantaged among us. For example, the data in Figure 8.1 show that, while voter turnout in U.S. presidential elections has remained more or less stable among the

Peers, Politics, and the Future of Democracy

0 to 16th percentile

133

96th to 100th percentile

100

Turnout Percentage

90

80

70

60

50

40 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008

FIGURE 8.1 Voter turnout in presidential elections by income Source: American National Election Studies Time Series Study

richest Americans, it has steadily declined among the poorest (albeit with a rebound in turnout over the past two elections). Declines in civic participation are a threat to popular sovereignty, regardless of the demographics of the individuals who are withdrawing from civil society. However, the class bias in this decline is especially dangerous to participatory democracy, because those with the most acute needs in American society have the weakest civic voice. Can civic talk solve the problem illustrated in Figure 8.1? Probably not on its own. While civic talk encourages individuals to participate in civic activities, the strength of this relationship varies under different circumstances. Specifically, this study has shown that civic talk pulls the more advantaged among us into civil society; those individuals who are already predisposed to become civically active get the most benefit from engaging in civic talk. In other words, while civic talk has the capacity to pull more of us into civil society, the political interests and preferences of this larger pool of participants may not be representative of the public at large (and of the socioeconomically disadvantaged among us in particular). Second, while peers groups have the capacity to be egalitarian and cooperative, group discussions can sometimes be dominated by individuals with higher levels of expertise on the issue at hand (Delli Carpini et al.

134

Chapter 8

2004; Mendelberg 2002). These types of “alpha” individuals tend to have stronger opinions about the issues being discussed and thus may be less willing to engage in compromise or listen to alternative points of view, thereby reducing the value of the discourse. Conversely, peer groups may be overly cooperative. This phenomenon is referred to as “groupthink”— when individuals are so concerned with maintaining the group’s cohesion that they rush to consensus and subsequently fail to assess critically the issues being discussed. As with too little cooperation, too much cooperation reduces the value of discourse because alterative points of view are not articulated or considered by the group. Moreover, groupthink may be a uniquely potent threat to the quality of civic talk discussions that occur in peer networks because the phenomenon is more common in small groups, in which unity and conflict avoidance are essential to maintaining social cohesion (Janis 1972). Third, in critically assessing the normative consequences of civic talk, it is necessary to consider what these conversations do and do not affect. Again, when a deliberation is run effectively, participants view the outcomes of the discussion as legitimate because they were directly involved in the decision-making process. This sense of legitimacy increases one’s faith in the processes of democratic governance (Delli Carpini et al. 2004; Mendelberg 2002). In contrast to this ideal, the analyses presented in Chapter 7 show that civic talk has no effect on one’s trust in government or sense that the government is run for the benefit of all citizens. One reason this could be the case is that, unlike deliberations, civic talk discussions have no direct link to government institutions and policymaking. Thus, while informal discussions about politics and current events do not erode institutional legitimacy vis-à-vis one’s attitudes toward the government, it is questionable whether such discussions enhance it. Finally, echoing the discussion of social intimacy in Chapter 6, civic talk discussions in peer groups may be beneficial to the democratic system because the deliberative process works more effectively when the participants share a social bond and have common viewpoints (Delli Carpini et al. 2004; Mendelberg 2002), as the members of most peer groups do. At the same time, however, when individuals are heavily integrated into a social circle, they are less likely to cooperate with or trust those outside the group (Bornstein 1992; Dawes et al. 1990; Delli Carpini et al. 2004; Mendelberg 2002; Putnam 2000; Walsh 2004).2 Thus, discussion of poli2 For

example, through a participant observation study of a small-town coffee klatch, Walsh (2004) shows that interpersonal interactions among peers strengthen a peer group’s identity

Peers, Politics, and the Future of Democracy

135

tics and current events within one’s peer group might work well in addressing small-scale issues but not large-scale issues. For example, in a small-scale environment, such as a neighborhood, individuals are likely to be able to resolve issues by working with people whom they already know and trust and with whom they share preferences. Moreover, the need to engage with unfamiliar outsiders will be less likely due to the smaller scope of the issue at hand. In contrast, small-scale peer group discussions will not be as effective when large-scale issues, such as how to reform the country’s financial sector, are at stake. In such cases, small groups would need to collaborate with one another to devise policy solutions. Such collaboration may not be effective because of the likelihood of intergroup conflict between peer groups.3

Peers, Politics, and the Future of Democracy Where do we go from here? The evidence presented in this book shows that civic talk has a significant and largely positive effect on the function of participatory democracy. Thus, it is incumbent on academics and practitioners in civil society to incorporate this fact into their agendas.

Changing the Civic Participation Research Agenda Although the C-SNIP data allowed me to make unprecedented causal inferences about civic talk, and although college students are a crucial case of peer influence (see Chapter 2), the results presented in this book should be tested under other circumstances. Large-scale social surveys are one way we can continue to collect data on the effects of social networks. Without innovations in the way this type of data is collected and analyzed, however, observational studies will continue to be plagued by analytical biases (or, at least, panned as such by critics of this line of research). One such innovation used in this study was the matching data pre-processing procedure (see, e.g., Ho et al. 2007a, 2007b, 2007c). Matching is a powerful at the cost of generating hostility toward outsiders. In an experimental study, Bornstein (1992, 604) also shows that “efforts to mobilize collective action within the groups are liable to interfere with the resolution of the conflict between them.” 3 This

said, debate and dissent is necessary for a democratic system to function. Indeed, under the right circumstances, disagreement among citizens can lead to learning and the creation of informed consensus (Barabas 2004; Huckfeldt et al. 2004). The point is that policy solutions are more easily reached when there is some level of agreement and understanding among the decision makers.

136

Chapter 8

tool for examining complex causal phenomena such as civic talk because it allows researchers to make results generated with observational data more approximate to those generated by a controlled experiment. Moreover, collecting variables before, during, and after individuals are exposed to civic talk, as done in the C-SNIP Panel Survey, allows researchers to maximize the benefits of matching. Specifically, variables that were collected before individuals are exposed to civic talk can be used to match people together who were similar before they did or did not engage in civic talk, and the effect of engaging in civic talk can be measured with variables collected after such conversations have occurred. The results I have presented in this book also illustrate the utility of using alternative methods of data collection that allow complex causal dynamics to be studied with greater precision. Scholars interested in social networks should make further use of quasi and natural experiments to study peer effects. Ideally, this research agenda should also be pushed to include fully controlled experiments. This ideal type of research design is difficult, although maybe not impossible, to implement. For example, it is hard to envision how a researcher could construct new social networks, force a randomly selected set of those groups to engage in civic talk, and prevent the rest of the groups from engaging in such discussions. Nonetheless, such a study would be a powerful test of the civic talk effect. In addition to experiments, we should make use of other research designs that are tailored to studying complex causal phenomena. Participant observation and focus group studies are especially powerful in this regard (Eliasoph 1998; Harris-Lacewell 2004; Walsh 2004). Like experiments, these methods come at the cost of reduced ability to draw generalizable conclusions but have the benefit of being able to provide a rich understanding of how citizens interact and influence one another. Recent research also shows that formal models (Siegel 2009) and agent-based models (Johnson and Huckfeldt 2004) can provide valuable predictions about how and why social influence occurs. At their best, these types of studies also use data to test these theoretical predictions empirically. In addition to thinking about the methods we use to collect and analyze data on civic talk, we need to consider the types of information that we collect. Specifically, future research should give additional consideration to both the content of civic talk discussions and the complex mechanisms that govern the relationship between civic talk and civic participation. For example, we know that one’s social network is an important source of information on politics and current events and that information motivates participation because it increases civic competence and civic engagement.

Peers, Politics, and the Future of Democracy

137

The analysis presented in this book, however, did not account for whether the quantity or quality of information being exchanged during civic talk discussions affects how active one chooses to be in civic activities. For example, is one conversation sufficient to encourage participation, or is more sustained interaction necessary? Does the answer to this question change if we are talking about the long-term effects of civic talk? In a similar vein, future studies should address the potentially dynamic relationship between civic talk and civic participation. In Chapter 7, I showed that the initial positive effect of civic talk on civic participation is the mechanism that governs the lasting relationship between these two variables—that is, talk today begets participation today, which begets participation in the future. However, it could also be the case that civic talk encourages more civic talk, which in turn encourages civic participation in the future. Data on subsequent civic talk interactions, as well as a more complex modeling approach, would be necessary to test this proposition. Finally, it is important to reiterate that this book has focused on relationships between college roommates (i.e., “dyads”). While this research design has yielded unprecedented results, most individuals are embedded in larger and more complex social networks. Consequently, future research should add more discussion partners to the research design. Ideally, this would entail mapping out the complete (i.e., “full”) social network of a specific set of individuals who live or congregate in the same place, such as a neighborhood, a church, or a school. This method of collecting data is laborious and expensive because it entails tracing all of the social connections between all of the people in a given social environment. However, sociologists have been using the full network research design with great success for years (see, e.g., Burt 1992) and have developed a number of methods to collect and assess such data (Scott 1991; Wasserman and Faust 1994). Political scientists should learn from their experience and apply this knowledge to the study of politically relevant social networks. Given that this study of roommates has shown that even one person in the social environment can have a significant effect on one’s civically relevant attitudes and behaviors, it will be intriguing to see how larger and more complex discussion networks affect their members.

An Agenda for Practitioners Over the past half-century, a staggering number of Americans have withdrawn from nearly every form of civic participation, from community voluntarism to voting in presidential elections (Putnam 2000; but see also

138

Chapter 8

Party contact

Socializing with friends

80 70

Percentage

60 50 40 30 20 10

19

74 19 76 19 78 19 80 19 82 19 84 19 86 19 88 19 90 19 92 19 94 19 96 19 98 20 00 20 02 20 04 20 06 20 08

0

FIGURE 8.2

Contact with political parties versus contact with peers

Sources: American National Election Studies (ANES) Time Series Study; General Social Survey (GSS) Notes: The ANES question asked, “Did anyone from one of the political parties call you up or come around and talk to you about the campaign?” The GSS question asked how often the respondent spent “a social evening with friends who live outside the neighborhood.” Respondents were coded as having socialized with friends if they responded “about once a month” or more frequently.

McDonald and Popkin 2001). This has left nongovernmental organizations, political campaigns, and others who work in civil society worried about the state of participatory democracy and wondering how to go about strengthening it. In response, the results presented in this book show that civic talk can be used as a resource to re-engage the public with civil society. This said, a fair question to ask is, “Why civic talk?” What is it about this potential lever for strengthening participatory democracy that is unique? To answer this question, recall the results presented in Chapter 7 comparing the effect of engaging in civic talk with other antecedents of civic participation. Overall, civic talk is as effective as, if not more effective than, these other antecedents. But why is this? The data in Figure 8.2 offer a suggestion. These results show that Americans socialize with their peers more often than they interact with representatives of political parties. This result is almost too banal to present; of course we socialize with our friends

Peers, Politics, and the Future of Democracy

139

more frequently than we interact with representatives of political parties! Nonetheless, the data illustrate why peer-based mobilization can be so effective in soliciting civic participation. In contrast with more remote antecedents of civic participation such as agents of political parties, we are in more frequent contact with our social network of peers. This makes peer-based mobilization efforts “cheaper” and often more effective (Klofstad et al. 2009). In other words, civic talk is special because it is not special. The fact that these interactions occur naturally as we go about our normal routines it what makes them powerful. Political actors are beginning to realize this. Before 2000, political parties and candidates focused most of their resources on the “air war,” using the mass media and direct mail to present their messages to voters. In the past two or three election cycles, however, candidates have realized that they need to put more resources into the “ground war,” or face-to-face mobilization of voters through canvassing. Moreover, candidates have also begun to use social networks as a way to get out the vote. The contrasting campaign tactics of the George W. Bush and John Kerry during the 2004 election illustrate this. Both candidates made heavy use of face-to-face canvassing but took very different approaches to this method of campaigning (Bai 2004). Kerry’s campaign largely used professional canvassing organizations that brought in outsiders (frequently progressive college students and other paid employees) to mobilize voters. In contrast, Bush’s camp recruited local volunteers to canvass in their own communities. Bush’s method—using volunteers people already knew from their communities rather than strangers who were paid for their service—was vastly more effective and helped propel him to a second term as president. Needless to say, his victory in 2004 continues to have widespread effects across the globe. Although peer-based civic mobilization has been, and can continue to be, used as a powerful resource for practitioners in civil society, it is important to underscore that civic talk is not a panacea. Perhaps the most critical point to remember is that civic talk does not work on everyone. As this study shows, civic talk is the most effective for those who are already predisposed to participate in civic activities. Consequently, practitioners need to make use of civic talk as just one part of a multifaceted strategy to increase the public’s engagement with civil society, especially when they are attempting to mobilize those of us who are less engaged with politics and current events. Since individuals are more responsive to civic talk when they have the means and wherewithal to participate in civil society, practitioners need

140

Chapter 8

to engage in activities that help the public develop these skills and motivations to more effectively harness the power of civic talk. Specifically, because civic participation is habitual, a logical suggestion for practitioners is to encourage young people to become civically engaged and active as early in life as possible. Models of such efforts already exist. For example, as discussed in Chapter 3 and Chapter 7, the young people in the C-SNIP Panel Survey were much more civically active in high school than in college. A likely explanation for this participation gap is that these students participated in service learning programs during high school that were no longer available to them during college. This suggests that we need to do a better job of offering opportunities for young people to become civically active after they leave high school. Data from the C-SNIP studies suggest that such efforts would be welcome on college campuses. As one student commented at the end of the third wave of the C-SNIP Panel Survey: “I think it would benefit the majority of the student population if there were a required course regarding politics and government at some point during undergraduate study.” The focus group sessions made it clear, however, that designing such outreach programs is not a straightforward task. Most of the participants commented that they would like pre-packaged opportunities to participate in civic activities like the ones they had experienced in high school. The students also made the point, however, that they do not want to be over-burdened with requests to participate. As one student said: “I don’t want it to be shoved down my throat. I’d rather say, ‘Oh, look, here’s a list of groups that I think [are] interesting and that I’ll choose to go [to].’” Such statements were typical in each of the focus group sessions, suggesting that college faculty and administrators have to walk a tightrope when attempting to create institutions that promote civic participation. The task they face is how to provide opportunities for students to become civically active without being too forceful. In thinking about how to increase the civic competence of young people, it is also important for practitioners to consider that this segment of the population is far more interested and active in non-political civic activities, such as community voluntarism and student organizations, than they are in political activities. For example, when asked, “In general, which do you think is the better way to solve important issues facing the country, through political involvement (e.g., voting, working for political candidates, and the like) or through community involvement (e.g., volunteering in the community, and the like)?” C-SNIP respondents vastly preferred community involvement to political involvement during their first year of

Peers, Politics, and the Future of Democracy

141

college (72 percent). Consequently, practitioners need to do a better job of encouraging young people to become engaged with politics. Such efforts can be made in college, when young people become eligible to vote. Additional efforts need to be made earlier in the educational career of the student, as well, to instill a sense of political engagement as soon as possible in the lifecycle.4

Conclusion This book began with the story of Rex Boyd and the 2008 Democratic Party caucuses in Iowa (Zeleny 2008). Before he went to his local caucus, Rex had intended to support Senator Hillary Clinton. After discussing the merits of each of the candidates with his friends and neighbors, however, he shifted his support to Senator Barack Obama. While neither this one vote nor this one caucus gave Obama the presidency, they were critical to the eventual outcome of the 2008 election. The win in Iowa gave Obama the political momentum he needed to win in future primaries, and the Obama campaign used this momentum to build a grassroots organization that reached out to the entire country—“red” and “blue” states alike. In line with the results presented in this book, this massive organization made extensive use of social networks to spread the message, using tactics as old as face-to-face canvassing and as new as e-mail, text messaging, and social networking websites. Rex Boyd and Barack Obama’s 2008 presidential campaign are just two examples of how the power of civic talk can be channeled to significantly affect the function of participatory democracy. Countless others certainly exist. The purpose of this book has been to demonstrate this point empirically and to encourage scholars and practitioners to more actively pursue the influence of social networks in their work.

4 Kids

Voting USA is an example of such a program. Another idea would be to allow students to fulfill service learning requirements by participating in political activities.

APPENDIX A

The Collegiate Social Network Interaction Project (C-SNIP)

C-SNIP Panel Survey The population surveyed was all 4,358 first year students at the University of Wisconsin, Madison, living in university housing during the 2003–2004 academic year (82 percent of the 5,322 students who entered the university in 2003). Study participants initially completed two questionnaires over the Internet during the 2003–2004 academic year: one at the beginning of the year (October–November 2003), and a second at the end of the year (March–April 2004). A third questionnaire was administered between April and May 2007. Lack of access to the Internet can bias survey response rates (Best et al. 2001; Couper 2000; Zhang 2000). However, this was not an issue in this study because subjects had free access to the Internet. During each wave of the study, three attempts were made by e-mail to recruit the sample to fill out a questionnaire. E-mail addresses were obtained from the University of Wisconsin, Madison, Office of the Registrar and from publicly accessible student directories. Unique login names and passwords were assigned to each respondent in these e-mails to prevent subjects from completing more than one questionnaire. To increase participation from a broad cross-section of the population under study, each student who completed a questionnaire was entered into a prize

144

Appendix A

drawing for one of fifty twenty-dollar prizes. (This procedure was used in all three waves of the study.) The recruitment e-mails were also worded to make the prospect of participating in the study appealing to a wide audience. In total, 23 percent of the study population completed the first two questionnaires (N = 999). Just under 24 percent of the population completed at least some portion of the first two questionnaires (N = 1,044). Of the 1,044 students who at least partially completed both of the first two questionnaires, 53 percent fully completed the 2007 follow-up questionnaire (N = 557), and 57 percent at least partially completed the 2007 questionnaire (N = 598). These response rates exclude subjects who were excluded from the analysis to reduce analytical bias: subjects who moved from the dormitory room to which they were initially assigned (N = 95), subjects who chose their roommates (N = 464), and subjects who had no roommates (N = 11). To account for missing data, the data were preprocessed using the Amelia II multiple imputation package for R (Honaker et al. 2007; see also King et al. 2001). The data set was imputed five times. To aid in the imputation process, the tolerance level was set to .001, and a ridge prior of 5 percent of the cases in the data set was used. All dichotomous variables were imputed using the nominal transformation; no other transformations were used. Descriptive statistics are listed in Appendix B. While imputation compensates for missing data, it is still important to address the issue of response rate. Although the recruitment procedures for the C-SNIP study were designed to attract a broad cross-section of participants from the population, it still could be the case that certain types of individuals—say, those who are more interested in the subject of the study—chose to participate at higher rates. For example, after participating in the first survey, each respondent knew that the subsequent two waves of the study would address politics and current events. This might have prompted individuals who were more engaged with and active in these matters to complete the study. If this was the case, the civic talk effects presented in this book could be inflated—that is, those who are participating in the study are more civically active and more likely to engage in civic talk than their non-respondent counterparts. Table A.1 offers an empirical test of this proposition by examining the characteristics of respondents and non-respondents in each of the three waves of the C-SNIP Panel Survey. The top portion of the table examines basic demographics of respondents and non-respondents. Fortunately, these three measures were available for the entire population surveyed;

= =

> > < 6.41 1.17

27.29 .49 .12

6.47 1.10

28.12 .60 .07

<

= =

= <

> > <

.44

2.21 .53

6.43 1.20

27.21 .50 .13

> or < indicates a significant difference between means at p ≤ .10; = indicates an insignificant difference between means at p > .10 (two-tailed t-tests).

Note: The figures for ACT score, gender, and race are for all first-year students, regardless of whether they lived in university dormitories, as indicated in enrollment files supplied the Department of the Registrar, University of Wisconsin, Madison. Voter turnout was not measured in the wave 1 questionnaire.

Source: C-SNIP Panel Survey.

.52

6.48 1.15

27.91 .62 .09

Non-respondents

Wave 3 (4th year of college) Respondents

Voter turnout (2004 primary)

27.18 .44 .13

Non-respondents

Wave 2 (1st year of college) Respondents

2.39 .55

> > <

Non-respondents

Wave 2 civic participation Voluntary civic organizations Political participation

Wave 1 civic participation Voluntary civic organizations Political participation

27.76 .60 .10

Respondents

Wave 1 (high school)

Characteristics of respondents and non-respondents (means)

ACT score Gender (female) Race (non-white)

TABLE A.1

146

Appendix A

thus, we can use them to assess non-response bias in all three waves of the study. These data show that when compared with non-respondents, respondents scored higher on their American College Testing (ACT) entrance exam, were more likely to be female, and were less likely to belong to a racial or ethnic minority group. Although these differences are statistically significant, in the case of ACT score and race, the substantive difference between respondents and non-respondents is small. Moreover, all three of these demographic characteristics were included in the matching data pre-processing procedure (see Appendix C). Thus, any differences between respondents and non-respondents on these variables are automatically accounted for in the analysis—that is, additional weighting of the data to correct for non-response bias is not required. The middle portion of Table A.1 uses survey responses provided in the first wave of the C-SNIP study to gauge differences between respondents and non-respondents in the second and third waves of the study. Respondents to wave 3 were more politically active during high school than were non-respondents. While this difference is statistically significant, it is substantively small (only .10 on a six-point activity scale). Moreover, each of the civic participation variables collected during the first wave of the C-SNIP study were included in the matching data preprocessing procedure, and as such any differences between respondents and non-respondents on these variables are accounted for in the analysis (again, see Appendix C). The bottom portion of Table A.1 examines differences in responses given during the second wave of the C-SNIP study between respondents and non-respondents in the third wave of the study. In this set of results we can now examine voter turnout (since most members of this student population could not vote while they were in high school, the first wave of the C-SNIP study did not collect information about voter participation). With the exception of voter turnout, there are no significant differences between respondents and non-respondents in the final wave of the study. Admittedly, the gap in voter turnout between respondents and non-respondents in 2004 is somewhat high, at 8 percentage points. However, this gap is not worrisome for two reasons. First, as addressed in Appendix C, given that the matching data pre-processing procedure accounted for how civically active subjects were in high school (participation in voluntary organizations and political participation), we can be reasonably confident that the analysis accounts for each subject’s baseline predilection to vote. Second, if this gap in voter turnout between respondents and non-respondents is biasing the results of wave 3, we

The Collegiate Social Network Interaction Project (C-SNIP)

147

would expect a strong relationship between civic talk during one’s first year of college and voter turnout during one’s fourth year of college. However, as documented in Chapter 7, this is not the case.

C-SNIP Focus Group Study During the second week of April 2008, a series of four focus groups were conducted with 2007–2008 first-year students at the University of Wisconsin, Madison. Each group was composed of eight students. These students, like the 2003–2004 cohort that participated in the C-SNIP Panel Survey, were randomly assigned to their first-year dormitories. Students were recruited to participate in dormitory cafeterias and lobbies. Potential participants were asked to complete a questionnaire that asked about their patterns of civic participation before coming to college, the amount of civic talk they engaged in with their randomly assigned roommates, times and dates of availability, and contact information. Based on their responses to the recruitment questionnaire, students were assigned to one of the four groups based on two factors: how civically active they had been in high school (“high” or “low”), and how much civic talk with their randomly assigned roommates they were exposed to (“high” or “low”). Selected participants were contacted by phone and mail to remind them of the date and time of their session. When they arrived at the focus group session, the participants completed standard consent forms and were invited to help themselves to refreshments. Each session lasted approximately ninety minutes and was conducted by a trained focus group moderator. The content of the sessions was audiotaped and later transcribed for analysis. At the completion of the focus group, each student was presented with a forty-dollar incentive. The moderator guided each focus group through the following set of discussion questions: 1. When you hear the words “politics and current events,” what different kinds of things come to mind? 2. Let’s think about the experiences you have had this year living with your roommate at college. We know that roommates talk about a lot of different kinds of things. What are the all of the different sorts of topics do you usually discuss with your roommate? We want to hear about all of the different things you discuss (including, but not only, those related to politics and current events).

148

Appendix A

3. How often do you discuss politics and current events with your roommate? Follow-Up Question: •

If you don’t typically discuss politics and current events with your roommate, what would you say are the main reasons that you don’t discuss these kinds of issues?

4. Who initiates these conversations between you and your roommate? 5. When you talk about politics and current events with your roommate, what do you talk about? What kinds of topics do you discuss? Follow-Up Questions: • • • • •

Have you been talking about the 2008 elections? Do you talk about local, state, national, or world politics? Do you talk about what you might have read or heard about in the news? Do you share information about how to get involved in a student group or political activities? Does your roommate ever explicitly ask you to get involved in civic activities?

6. How much do you typically find that you share the same opinions with your roommates when you discuss politics and current events? Follow-Up Questions: •

•

When you have these kinds of conversations with your roommate, how often does one of you try to persuade the other person about political or current events topics? Have you ever had what you would consider to be a heated discussion or an argument with your roommate about politics and current events?

7. Think about all of the different topics you discuss with your roommate (and you can reference the list we made together earlier). How frequently would you say that you discuss politics and current events with your roommate compared with the other topics you discuss? 8. There are lots of different kinds of civic activities that students sometimes choose to get involved in on campus. These

The Collegiate Social Network Interaction Project (C-SNIP)

include things like participating in student groups, volunteering, working on political campaigns, protesting, and voting. What experiences have you had participating in these types of activities this year at the university? 9. What are some of the reasons that you chose to get involved in these civic activities? Follow-Up Questions: • •

•

What were some of the factors that caused you to not get involved in civic activities? Overall, how much would you say that you take part in these kinds of civic activities compared with last year (your senior year of high school)? What are some of the reasons that your involvement in these activities changed from last year to this year?

10. Overall, what effect do you feel discussing politics with your roommate has had on your decision to become active in these types of activities? Follow-Up Questions: • • • •

What effect did sharing information about politics and current events have on your decision? What effect did their asking you to get involved have on your decision? What effect would you say that these conversations have had on your interest in politics and current events? Do you feel like your roommate might be disappointed in you if you don’t participate in civic activities?

11. Finally, what was this experience like for you tonight? Was this an easy or a difficult topic to discuss?

149

APPENDIX B

C-SNIP Panel Survey Questions and Variables

Civic Participation Participation in Voluntary Civic Organizations (Table B.1) “How active were you in the following types of organizations [during high school/during your first year here at the University of Wisconsin/at the University of Wisconsin this year]: very active, somewhat active, not very active, or not at all active?” • • •

•

• • • 1 The

“Student government (for example, [student council/ASM], etc.)”1 “Partisan political groups (e.g., [Young/College] Republicans or Democrats, etc.)” “Organizations that take stands on political issues or current events (e.g., a group interested in protecting the environment, etc.)” “Charitable or voluntary service organizations (i.e., working in some way to help others without pay and not for course credit)” “Leadership training or civic organizations (e.g., community service organizations, etc.)” “Student publications (e.g., yearbook, newspaper, etc.)” “Forensics, debate, or other speech clubs or teams”

Associated Students of Madison (ASM) is the student government body at the University of Wisconsin, Madison.

152

TABLE B.1

Appendix B

Participation in voluntary civic organizations (descriptive statistics) Minimum

Maximum

Mean

SD

High school Full participation scale Student government Partisan “Take stands” Charitable/service Leadership/civic Student publications Forensics, etc.

−.76 .00 −.46 −1.05 .00 −1.70 −.63 −.91

19.00 3.00 3.00 3.00 3.51 3.13 3.29 3.00

6.60 1.14 .27 .65 1.97 .89 .91 .76

3.96 1.19 .63 .92 1.02 1.15 1.12 1.07

1st year of college Full participation scale Student government Partisan “Take stands” Charitable/service Leadership/civic Student publications Forensics, etc.

−6.54 −.93 −1.48 −1.99 −1.44 −1.77 −.80 −1.16

21.00 3.00 3.00 3.16 3.24 3.17 3.00 3.28

2.43 .24 .27 .41 .89 .42 .11 .08

2.91 .57 .66 .80 1.04 .84 .43 .36

4th year of college Full participation scale Student government Partisan “Take stands” Charitable/service Leadership/civic Student publications Forensics, etc.

−4.98 −1.48 −1.40 −2.11 −1.87 −2.34 −1.56 −.87

14.00 3.00 3.00 3.54 4.57 3.20 3.00 3.00

3.20 .25 .32 .65 1.33 .42 .18 .05

2.65 .60 .62 .94 1.13 .93 .58 .34

Source: C-SNIP Panel Survey. Note: The values in this table are means calculated from five imputed data sets. The minimums of some variables are negative because the range for ordinal variables (i.e., a “range prior”) was not specified in the imputation algorithm. SD, standard deviation. N = 1,044.

Participation in Political Activities (Table B.2) “How many times did you contact an elected official about a problem or issue that you were concerned about [during high school/during your first year here at the University of Wisconsin/at the University of Wisconsin this year]: never, once, or more than once?” “How many times did you participate in a protest, march, or demonstration [during high school/during your first year here at the University of Wisconsin/at the University of Wisconsin this year]: never, once, or more than once?”

C-SNIP Panel Survey Questions and Variables

TABLE B.2

153

Participation in political activities (descriptive statistics) Minimum

Maximum

Mean

SD

High school Full participation scale Contacting Protesting Campaign work

−.49 −.14 −.56 −.10

6.00 2.00 2.00 2.00

1.16 .51 .40 .25

1.41 .72 .70 .58

1st year of college Full participation scale Contacting Protesting Campaign work

−2.06 −1.17 .97 −.82

6.00 2.02 2.00 2.01

.56 .24 .20 .12

1.13 .57 .50 .40

4th year of college Full participation scale Contacting Protesting Campaign work

−2.69 −1.73 −1.52 −1.13

6.00 2.91 2.09 2.00

.81 .43 .25 .13

1.28 .73 .56 .45

Source: C-SNIP Panel Survey. Note: The values in this table are means calculated from five imputed data sets. The minimums of some variables are negative because the range for ordinal variables (i.e., a “range prior”) was not specified in the imputation algorithm. SD, standard deviation. N = 1,044.

“How many times did you work as a volunteer for a candidate running in a national, state, local, or school election [during high school/during your first year here at the University of Wisconsin/at the University of Wisconsin this year]: never, once or more than once?”

Voter Turnout (Table B.3) “Did you vote in the Wisconsin Democratic Presidential Primary that was held on Tuesday, February 17, 2004: yes or no?” “Did you vote in the 2006 midterm congressional election?”

TABLE B.3

Voter turnout (descriptive statistics) Minimum

Maximum

Mean

SD

1st year of college (2004 primary)

0

1

.51

.50

4th year of college (2006 midterm election)

0

1

.64

.48

Source: C-SNIP Panel Survey. Note: The values in this table are means calculated from five imputed data sets. SD, standard deviation. N = 1,044.

154

TABLE B.4

Appendix B

Civic talk among roommates (descriptive statistics) Minimum

Full scale Dichotomous treatment coding

Maximum

Mean

SD

−.55

4.26

1.40

.89

.00

1.00

.47

.50

Source: C-SNIP Panel Survey. Note: The values in this table are means calculated from five imputed data sets. The minimums of some variables are negative because the range for ordinal variables (i.e., a “range prior”) was not specified in the imputation algorithm. SD, standard deviation. N = 1,044.

Civic Talk

(Table B.4)

“When you talk with your roommate, how often do you discuss politics and current events: often, sometimes, rarely, or never?”

Causal Mechanisms

(Table B.5)

Resources “How many times have your roommates given you any information about how to become active in politics and current events: often, sometimes, rarely, or never?”

Engagement “Thinking about how interested you were in politics and current events before you came to the University of Wisconsin, Madison, has talking with your roommates increased your interest in politics and current events: very much, somewhat, not that much, or not at all?” TABLE B.5

Causal mechanisms (descriptive statistics) Minimum Maximum

Mean

SD

Resources

−1.38

3.00

.44

.72

Engagement Interest in politics and current events Political efficacy

−1.75 .65

3.14 6.47

.68 3.38

.85 1.10

Recruitment

−1.27

3.00

.25

.56

−.73 .00

4.27 3.87

1.36 2.32

.85 .63

Norms Perceived activity level of roommate Perceived importance of civic participation Source: C-SNIP Panel Survey.

Note: The values in this table are means calculated from five imputed data sets. The minimums of some variables are negative because the range for ordinal variables (i.e., a “range prior”) was not specified in the imputation algorithm. SD, standard deviation. N = 1,044.

C-SNIP Panel Survey Questions and Variables

155

“How much would you agree or disagree with this statement: ‘People like me don’t have any say about what the government does’: strongly agree, somewhat agree, neither agree nor disagree, somewhat disagree, or strongly disagree?”

Recruitment “How many times have your roommates asked you to participate in an event or organization related to politics and current events: often, sometimes, rarely, or never?”

Norms “How active and interested do you think your roommates are in politics and current events: very, somewhat, not very, or not at all?” “How important do you think it is for people like you to be active and interested in politics and current events: very important, somewhat important, not very important, or not at all important?”

Individual-Level Characteristics Used in Chapter 5 (Table B.6) Parental Civic Interest and Activity “How active and interested would you say your [mother/father] is in politics and community affairs: very active and interested, somewhat active and interested, not very active and interested, or not at all active and interested?”

Interest in Politics and Current Events “How interested were you in politics and current events during high school: very interested, somewhat interested, not very interested, or not at all interested?”

Political Efficacy “How much would you agree or disagree with this statement, ‘People like me don’t have any say about what the government does:’ strongly agree, somewhat agree, neither agree nor disagree, somewhat disagree, or strongly disagree?”

156

Appendix B

TABLE B.6 Individual-level pre-treatment (high school) intervening variables (descriptive statistics) Minimum Maximum

Mean

SD

Parents’ civic interest/activity Full scale (mother + father) Mother Father

2.00 .94 .86

8.65 4.42 4.57

5.42 2.60 2.82

1.38 .79 .83

Interest in politics and current events

1.00

4.00

2.69

.91

Political efficacy

1.00

5.00

3.28

1.12

Ideology Strength Directional scale (conservative–liberal)

1.00 1.00

3.00 5.08

1.90 3.36

.75 1.11

.99

3.00

2.08

.70

17.00

36.33

28.14

2.88

2.00 1.00 .98

12.32 6.16 6.21

7.56 3.72 3.84

1.87 1.04 1.10

Knowledge about politics and current events ACT score Parents’ education Full scale (mother + father) Mother Father Source: C-SNIP Panel Survey.

Note: The values in this table are means calculated from five imputed data sets. SD, standard deviation. N = 1,044.

Ideology “We also hear a lot about conservatives and liberals in politics these days. What were you during high school: very conservative, somewhat conservative, moderate, somewhat liberal, or very liberal?”

Knowledge about Politics and Current Events “Generally speaking, how much would you say you knew about politics and current events during high school: a good deal, some, or not that much?”

ACT Score Each respondent’s test score was provided by the University of Wisconsin, Madison, Office of the Registrar.

Parents’ Education “What is the highest academic degree held by your [mother/father]: less than high school, high school diploma or general equivalency diploma, some college, college undergraduate degree, or postgraduate degree?”

C-SNIP Panel Survey Questions and Variables

157

Peer Characteristics Used in Chapter 6

(Table B.7)

Trust between Roommates “How much do you feel you can trust your roommate: very much, somewhat, not that much, or not at all?”

Subject Enjoys Spending Time with Roommate “How much do you enjoy spending time with your roommate: very much, somewhat, not that much, or not at all?”

Roommate’s Knowledge about Politics and Current Events “Generally speaking, how much do you think your roommate knows about politics and current events: a great deal, some, or not much?”

Roommate’s Civic Interest/Activity “How active and interested do you think your roommate is in politics and current events: very, somewhat, not very, or not at all?”

Overall Similarity between Subject and Roommate “Overall, how similar would you say your roommate is to you: very similar, somewhat similar, somewhat dissimilar, or very dissimilar?”

Political Similarity between Subject and Roommate “Thinking about your own beliefs about politics and current events, how similar would you say your roommate is to you: very similar, somewhat similar, somewhat dissimilar, or very dissimilar?”

TABLE B.7

Peer characteristics (descriptive statistics) Minimum Maximum

Trust between roommates Subject enjoys spending time with roommate Overall similarity with roommate Political similarity with roommate Roommate’s civic knowledge Roommate’s civic interest/activity

−.43 −.85 −.82 −.99 −.07 −.73

4.22 4.08 3.71 3.44 3.79 4.27

Mean

SD

2.41 1.96 1.56 1.60 1.93 1.36

.82 1.00 .96 .91 .61 .85

Source: C-SNIP Panel Survey. Note: The values in this table are means calculated from five imputed data sets. The minimums of some variables are negative because the range for ordinal variables (i.e., a “range prior”) was not specified in the imputation algorithm. SD, standard deviation. N = 1,044.

158

Appendix B

Other Variables

(Table B.8)

Political versus Community Involvement “In general, which do you think is the better way to solve important issues facing the country: through political involvement (e.g., voting, working for political candidates and the like) or through community involvement (e.g., volunteering in the community and the like)?”

Perceived Importance of Civic Participation “How important do you think it is for people like you to be active and interested in politics and current events: very important, somewhat important, not very important, or not at all important?”

Available Free Time “How much free time [did you have/have you had] during the average week to participate in the types of organizations you answered questions about at the beginning of this survey: a lot of time, a moderate amount of time, very little time, or no time at all?” TABLE B.8

Other variables (descriptive statistics) Minimum Maximum

Political versus community involvement High school 1st year of college 4th year of college Perceived importance of civic participation 1st year of college 4th year of college Free time High school 1st year of college 4th year of college Knowledge about politics and current events High school 1st year of college 4th year of college News media usage High school 1st year of college 4th year of college

Mean

SD

.00 .00 .00

1.00 1.00 1.00

.23 .28 .23

.42 .45 .42

.00 −.03

3.87 4.09

2.32 2.30

.63 .64

1.00 .67 .47

4.00 4.30 4.54

3.00 2.53 2.45

.59 .66 .66

.99 .45 .16

3.00 3.70 3.88

2.08 1.94 2.02

.70 .65 .61

.97 −2.16 −1.39

7.00 8.17 9.65

3.71 3.25 3.95

1.98 1.99 2.00

Source: C-SNIP Panel Survey. Note: The values in this table are means calculated from five imputed data sets. The minimums of some variables are negative because the range for ordinal variables (i.e., a “range prior”) was not specified in the imputation algorithm. SD, standard deviation. N = 1,044.

C-SNIP Panel Survey Questions and Variables

159

Knowledge about Politics and Current Events “Generally speaking, how much would you say you knew about politics and current events [during high school/during your first year here at the University of Wisconsin/during this last year at the University of Wisconsin]: a good deal, some, or not that much?”

News Media Use “How many days per week, on average, did you read or watch the news to learn about politics and current events [during high school/during your first year here at the University of Wisconsin/during this last year at the University of Wisconsin]?”

APPENDIX C

Matching Data Pre-processing

or this analysis I used a “full matching” procedure (Gu and Rosenbaum 1993; Hansen 2004; Ho et al. 2007a, 2007b, 2007c; Rosenbaum 1991; Stuart and Green 2008). The procedure was conducted using the using the MatchIt package for R (Ho et al. 2007a, 2007b), which makes use of the Optmatch package (Hansen 2004). The C-SNIP Panel Survey data set is tailor-made for matching because subjects were surveyed about their characteristics before they engaged in civic talk with their college roommates (i.e., before they were exposed to the civic talk “treatment”). As listed in Table C.1, 109 pre-treatment variables were used in the matching procedure. Matching on a large number of pre-treatment covariates increases the validity of the final analysis, because it is likely that more alternative explanations of the civic talk effect are accounted for (Ho et al. 2007b). This set of variables included measures of civic participation in high school, measures of why each student ranked the dormitories in the way that they did, indicators of which dorm each subject was eventually placed into, information on the subject’s roommate and dormitory, demographics, measures of home life before coming to college, and civically relevant attitudes and characteristics. The full matching procedure involved three steps. First, study subjects were classified as having been “treated” or “untreated” with civic talk. Subjects who engaged in an above-average amount of civic talk with their

F

162

TABLE C.1

Appendix C

Summary of pre-treatment covariates included in matching procedure

Type

Examples

Past patterns of civic participation

Participation in voluntary civic organizations; political participation

Reasons for dormitory ranking

Subject reported seeking a social environment, academic environment, location, amenities, and the like

Dormitory placement

Indicator variables for placement in each of the fifteen dormitories

Information on roommates and dorm mates

Familiarity with roommates and dorm mates before coming to college

Demographics

Race; gender; composite ACT score; citizenship; favorite subject of study in high school; state of residence

Home life before college

Civic talk with parents; parents’ civic interest/ activity; parents’ partisanship; parents’ household income; size of hometown

Relevant attitudes/characteristics

Ideology; partisanship; interests in politics and current events; 2000 election candidate preferences; social gregariousness

Source: C-SNIP Panel Survey.

roommates were classified as having been treated (N = 495), while those who engaged in a below-average amount of civic talk were classified as untreated (N = 549). Second, the variables included in the matching procedure were used to estimate a score of one’s propensity to engage in civic talk (Hansen 2004; Ho et al. 2007a, 2007b). Third, at least one untreated subject was matched to at least one treated subject based on how close the propensity scores were between treated and untreated subjects (a process that creates “subclasses” in which more than one treated subject can be matched to an untreated subject, and vice versa). Each untreated subject was only matched to one treated subject, and vice versa (matching without replacement). Also, after a subject was initially matched, he or she could have been moved and matched to a different subject before the procedure concluded to improve the overall similarity between treated and untreated subjects in the data set—that is, the process is “optimal,” not “greedy.” Unlike other forms of matching (e.g., “exact matching”), each case in the original data set is included in the matched data set (i.e., cases are not discarded). The results of the matching procedure were incorporated into the analysis by weighting the regression models. All treated subjects were

Matching Data Pre-processing

163

TABLE C.2 Improvement in balance between treated and untreated cases Overall

99.70%

QQ plot summary statistics Median Mean Maximum

94.10% 93.42% 89.53%

Source: C-SNIP Panel Survey.

given a weight of 1. Untreated subjects were assigned a weight equal to the number of treated subjects in the subclass to which they were assigned, divided by the number of untreated subjects in the subclass to which they were assigned. For example, an untreated subject who was assigned to a subclass with ten treated subjects and one untreated subject was assigned a weight of ten, while an untreated subject who was assigned to a subclass with one treated subject and ten untreated subject was assigned a weight of .10. Consequently, applying this weight caused the regression model to pay more attention to untreated subjects who were similar to treated subjects and less attention to untreated subjects who were dissimilar to treated subjects, making the analysis a better comparison between the treated and untreated subjects than if the data had not been weighted. The results presented in Table C.2 illustrate how the matching procedure increased the similarity, or “balance” (Ho et al. 2007a, 2007b), between subjects who did and did not engage in civic talk. The first row in the table shows the overall improvement in similarity between treated and untreated subjects, as measured by the subject’s estimated propensity to engage in civic talk (i.e., the propensity score created by the matching procedure). Overall, the similarity in the propensity to engage in civic talk between subjects who did and did not engage in civic talk increased by nearly 100 percent as a result of matching. The remaining rows of the table show the summary statistics from QQ plots, which are two-dimensional graphs that plot the empirical distribution of a variable among treated subjects on one axis against the empirical distribution of that same variable among untreated subjects on the other axis. The closer this plotted line is to the forty-five-degree line on the graph, the closer treated and untreated subjects are to being perfectly balanced on that variable. The results in Table C.2 show that the median, mean, and maximum distance of the propensity-score QQ plot from the forty-five-degree line were all greatly improved due to the matching procedure.

References

Achen, Christopher. 1986. The Statistical Analysis of Quasi-Experiments. Berkeley: University of California Press. Alesina, Alberto, and Eliana La Ferrara. 2002. “Who Trusts Others?” Journal of Public Economics 85 (2): 207–234. Alwin, Duane F., Ronald L. Cohen, and Theodore M. Newcomb. 1991. Political Attitudes over the Life Span. Madison: University of Wisconsin Press. Arceneaux, Kevin, Alan S. Gerber, and Donald P. Green. 2006. “Comparing Experimental and Matching Methods Using a Large-Scale Voter Mobilization Experiment.” Political Analysis 14 (1): 37–62. Bai, Matt. 2004. “Who Lost Ohio?” New York Times, November 21, 67. Barabas, Jason. 2004. “How Deliberation Affects Policy Opinions.” American Political Science Review 98 (4): 687–701. Barker, David C. 1998. “Rush to Action: Political Talk Radio and Health Care (Un)Reform.” Political Communication 15 (1): 83–97. Bar-Tal, Daniel, and Leonard Saxe. 1990. “Acquisition of Political Knowledge: A Social-Psychological Analysis.” In Political Socialization, Citizenship Education, and Democracy, ed. Orit Ichilov, 116–133. New York: Teachers College Press. Bates, Tom. 1993. RADS: The 1970 Bombing of the Army Math Research Center at the University of Wisconsin–Madison and Its Aftermath. New York: HarperCollins. Beck, Paul Allen. 1977. “The Role of Agents in Political Socialization.” In The Handbook of Political Socialization, ed. Stanley Allen Renshon, 115–142. New York: Free Press.

166

References

Beck, Paul Allen, Russell J. Dalton, Steven Greene, and Robert Huckfeldt. 2002. “The Social Calculus of Voting: Interpersonal, Media, and Organizational Influences on Presidential Choices.” American Political Science Review 96 (1): 57–73. Beck, Paul, Russell J. Dalton, and Robert Huckfeldt. 1995. Cross-National Election Studies: United States Study, 1992 [computer file]. ICPSR version. InterUniversity Consortium for Political and Social Research, University of Michigan, Ann Arbor. Bellin, Andy. 2002. Poker Nation: A High-Stakes, Low-Life Adventure into the Heart of a Gambling Country. New York: HarperCollins. Berman, Sheri. 1997. “Civil Society and the Collapse of the Weimar Republic.” World Politics 49 (3): 401–429 Bernstein, William M., Walter G. Stephan, and Mark H. Davis. 1979. “Explaining Attributions for Achievement: A Path Analytic Approach.” Journal of Personality and Social Psychology 37 (10): 1810–1821. Best, Samuel J., Brian Krueger, Clark Hubbard, and Andrew Smith. 2001. “An Assessment of the Generalizability of Internet Surveys.” Social Science Computer Review 19 (2): 131–145. Bornstein, Gary. 1992. “The Free-Rider Problem in Intergroup Conflicts Over Step-Level and Continuous Public Goods.” Journal of Personality and Social Psychology 62 (4): 597–606. Boyer, Ernest L. 1987. College: The Undergraduate Experience in America. New York: Harper and Row. Bradley, Gifford W. 1978. “Self-Serving Biases in the Attribution Process: A Reexamination of the Fact or Fiction Question.” Journal of Personality and Social Psychology 36 (1): 56–71. Brady, Henry E. 1999. “Political Participation.” In Measures of Political Attitudes, ed. John P. Robinson, Phillip R. Shaver, and Lawrence S. Wrightsman, 737– 801. San Diego: Academic Press. Brady, Henry E., Kay Lehman Schlozman, and Sidney Verba. 1999. “Prospecting for Participants: Rational Expectations and the Recruitment of Political Activists.” American Political Science Review 93 (1): 153–168. Brehm, Sharon S., Rowland S. Miller, Daniel Perlman, and Susan M. Campbell. 2002. Intimate Relationships, 3rd ed. New York: McGraw-Hill. Burns, Nancy, Kay Lehman Schlozman, and Sidney Verba. 2001. The Private Roots of Public Action. Cambridge, MA: Harvard University Press. Burt, Ronald S. 1992. Structural Holes: The Social Structure of Competition. Cambridge, MA: Harvard University Press. Campbell, David E. 2006. Why We Vote: How Schools and Communities Shape Our Civic Life. Princeton, NJ: Princeton University Press. Campbell, Angus, Philip E. Converse, Warren E. Miller, and Donald E. Stokes. 1960. The American Voter. New York: John Wiley and Sons. Campbell, David E., and Christina Wolbrecht. 2006. “See Jane Run: Women Politicians as Role Models for Adolescents.” Journal of Politics 68 (2): 233– 247.

References

167

Collier, Ruth Berins, and David Collier. 1991. Shaping the Political Arena: Critical Junctures, the Labor Movement, and Regime Dynamics in Latin America. Princeton, NJ: Princeton University Press. Compas, Bruce E., Barry M. Wagner, Lesley A. Slavin, and Kathryn Vannatta. 1986. “A Prospective Study of Life Events, Social Support, and Psychological Symptomatology during the Transition from High School to College.” American Journal of Community Psychology 14 (3): 241–257. Conover, Pamela Johnston, Donald D. Searing, and Ivor M. Crewe. 2002. “The Deliberative Potential of Political Discussion.” British Journal of Political Science 32: 21–62. Costa, Dora L., and Matthew E. Kahn. 2003. “Understanding the Decline in Social Capital, 1952–1998.” Kyklos 56 (1): 17–46. Couper, Mick P. 2000. “Web Surveys: A Review of Issues and Approaches.” Public Opinion Quarterly 64 (4): 464–494. Crandall, Christian S. 1988. “Social Contagion of Binge Eating.” In Social Psychology Readings: A Century of Research, ed. Amy G. Halberstadt and Steve L. Ellyson, 588–598. New York: McGraw-Hill. Croll, Paul. 2004. “Families, Social Capital and Educational Outcomes.” British Journal of Educational Studies 52 (4): 390–416. Cutrona, Carolyn E. 1982. “Transition to College: Loneliness and the Process of Social Adjustment.” In Loneliness: A Sourcebook of Current Theory, Research and Therapy, ed. Letitia Anne Peplau and Daniel Perlman, 291–309. New York: Wiley-Interscience. Dawes, Robyn M., Alphons J. C. van de Kragt, and John M. Orbell. 1990. “Cooperation for the Benefit of Us—Not Me, or My Conscience.” In Beyond SelfInterest, ed. Jane Mansbridge, 97–110. Chicago: University of Chicago Press. Dawson, Richard E., Kenneth Prewitt, and Karen S. Dawson. 1977. Political Socialization, 2nd ed. Boston: Little, Brown. Delli Carpini, Michael X., Fay Lomax Cook, and Lawrence R. Jacobs. 2004. “Public Deliberation, Discursive Participation, and Citizen Engagement: A Review of the Empirical Literature.” Annual Review of Political Science 7: 315–344. Downs, Anthony. 1957. An Economic Theory of Democracy. New York: Harper and Row. Duncan, Greg J., Johanne Boisjoly, Michael Kremer, Dan M. Levy and Jacque Eccles. 2005. “Peer Effects in Drug Use and Sex among College Students.” Journal of Abnormal Child Psychology 33 (3): 375–385. Dunning, Thad. 2008. “Improving Causal Inference: Strengths and Limitations of Natural Experiments.” Political Research Quarterly 61 (2): 282–293. Eckstein, Harry. 1975. “Case Studies and Theory in Political Science.” In Handbook of Political Science, Scope and Theory, vol. 7, ed. Fred I. Greenstein and Nelson W. Polsby, 79–138. Reading, MA: Addison-Wesley. Eliasoph, Nina. 1998. Avoiding Politics: How Americans Produce Apathy in Everyday Life. Cambridge: Cambridge University Press. Faler, Brian. 2005. “Census Details Voter Turnout for 2004.” Washington Post, May 26, A10.

168

References

Fazio, Russell H., Mark P. Zanna, and Joel Cooper. 1977. “Dissonance and SelfPerception: An Integrative View of Each Theory’s Proper Domain of Application.” Journal of Experimental Social Psychology 13 (5): 464–479. Festinger, Leon. 1954. “A Theory of Social Comparison Processes.” Human Relations 7 (2): 117–140. Festinger, Leon, Stanley Schachter, and Kurt Back. 1950. Social Pressures in Informal Groups: A Study of Human Factors in Housing. New York: Harper Brothers. Fink, Edward L., Stan A. Kaplowitz, and Connie L. Bauer. 1983. “Positional Discrepancy, Psychological Discrepancy, and Attitude Change: Experimental Tests of Some Mathematical Models.” Communication Monographs 50: 413– 430. Fiske, Susan T., and Shelley E. Taylor. 1991. Social Cognition. New York: McGrawHill. Foley, Michael W., and Bob Edwards. 1996. “The Paradox of Civil Society.” Journal of Democracy 7 (3): 38–52. Fowler, James. 2006. “Habitual Voting and Behavioral Turnout.” Journal of Politics 68 (2): 335–344. Gellner, Ernest. 1995. “The Importance of Being Modular.” In Civil Society: Theory, History and Comparison, ed. John A. Hall, 32–55. Cambridge: Polity Press. Gerber, Alan, and Donald Green. 2000. “The Effects of Personal Canvassing, Telephone Calls, and Direct Mail on Voter Turnout: A Field Experiment.” American Political Science Review 94 (3): 641–652. Gerber, Alan, Donald Green, and Ron Shachar. 2003. “Voting May Be HabitForming: Evidence from a Randomized Field Experiment.” American Journal of Political Science 47 (3): 540–550. Gerring, John. 2001. Social Science Methodology: A Criterial Framework. Cambridge: Cambridge University Press. Gimpel, James G., J. Celeste Lay, and Jason E. Schuknecht. 2003. Cultivating Democracy: Civic Environments and Political Socialization in America. Washington, DC: Brookings Institution Press. Godwin, R. Kenneth, and Robert Cameron Mitchell. 1984. “The Implications of Direct Mail for Political Organizations.” Social Science Quarterly 65 (3): 829–839. Gu, Xing Sam, and Paul R. Rosenbaum. 1993. “Comparison of Multivariate Matching Methods: Structures, Distances, and Algorithms.” Journal of Computational and Graphical Statistics 2 (4): 405–420. Hall, John A. 1995. “In Search of Civil Society.” In Civil Society: Theory, History and Comparison, ed. John A. Hall, 1–31. Cambridge: Polity Press. Hansen, Ben B. 2004. “Full Matching in an Observational Study of Coaching for the SAT.” Journal of the American Statistical Association 99 (467): 609–618. Harris-Lacewell, Melissa Victoria. 2004. Barbershops, Bibles, and BET: Everyday Talk and Black Political Thought. Princeton, NJ: Princeton University Press. Harvard University Institute of Politics. 2000. “Attitudes toward Politics and Public Service: A National Survey of College Undergraduates (Top Line Data

References

169

from the 2000 National Survey of College Undergraduates).” Available online at http://www.iop.harvard.edu/pdfs/survey/2000_topline.pdf (accessed June 27, 2007). Heider, Fritz 1958. The Psychology of Interpersonal Relations. New York: John Wiley and Sons. Ho, Daniel, Kosuke Imai, Gary King, and Elizabeth Stuart. 2007a. “Matchit: Matching as Nonparametric Preprocessing for Parametric Causal Inference.” Available online at http://gking.harvard.edu/matchit (accessed May 28, 2010). ———. 2007b. “Matchit: Matching as Nonparametric Preprocessing for Parametric Causal Inference (Technical Manual).” Available online at http://gking.harvard .edu/matchit (accessed May 28, 2010). ———. 2007c. “Matching as Nonparametric Preprocessing for Reducing Model Dependence in Parametric Causal Inference.” Political Analysis 15 (3): 199– 236. Honaker, James, Gary King, and Matthew Blackwell. 2007. “Amelia II: A Program for Missing Data.” Available online at http://gking.harvard.edu/amelia (accessed May 28, 2010). Huckfeldt, Robert, Paul Allen Beck, Russell J. Dalton, and Jeffrey Levine. 1995. “Political Environments, Cohesive Social Groups, and the Communication of Public Opinion.” American Journal of Political Science 39 (4): 1025–1054. Huckfeldt, Robert, Ken’ichi Ikeda, and Franz Urban Pappi. 2000. “Political Expertise, Interdependent Citizens, and the Value Added Problem in Democratic Politics.” Japanese Journal of Political Science 1 (2): 171–195. Huckfeldt, Robert, Paul E. Johnson, and John Sprague. 2002. “Political Environments, Political Dynamics, and the Survival of Disagreement.” Journal of Politics 64 (1): 1–21. ———. 2004. Political Disagreement: The Survival of Diverse Opinions within Communication Networks. Cambridge: Cambridge University Press. Huckfeldt, Robert, and John Sprague. 1991. “Discussant Effects on Vote Choice: Intimacy, Structure, and Interdependence.” Journal of Politics 53 (1): 122– 158. ———. 1995. Citizens, Politics, and Social Communication: Information and Influence in an Election Campaign. Cambridge: Cambridge University Press. ———. 2000. Indianapolis–St. Louis Election Study, 1996–1997 [computer file]. ICPSR version. Bloomington: Indiana University Center for Survey Research and Ann Arbor: Inter-University Consortium for Political and Social Research. Imai, Kosuke, Gary King, and Olivia Lau. 2007a. “Zelig: Everyone’s Statistical Software.” Available online at http://GKing.harvard.edu/zelig (accessed May 31, 2010). ———. 2007b. “Toward a Common Framework for Statistical Analysis and Development.” Unpublished ms. Available online at http://gking.harvard.edu/ files/abs/z-abs.shtml (accessed May 31, 2010). ———. 2007c. “ls: Least Squares Regression for Continuous Dependent Variables.” In Zelig: Everyone’s Statistical Software, ed. Kosuke Imai, Gary King, and

170

References

Olivia Lau. Available online at http://gking.harvard.edu/zelig (accessed May 31, 2010). ———. 2007d. “logit: Logistic Regression for Dichotomous Dependent Variables.” In Zelig: Everyone’s Statistical Software, ed. Kosuke Imai, Gary King, and Olivia Lau. Available online at http://gking.harvard.edu/zelig (accessed May 31, 2010). Janis, Irving. 1972. Victims of Groupthink. Boston: Houghton Mifflin. Jennings, M. Kent, and Richard G. Niemi. 1981. Generations and Politics. Princeton, NJ: Princeton University Press. Jennings, M. Kent, and Laura Stoker. 2004. “Social Trust and Civic Engagement across Time and Generations.” Acta Politica 39 (1): 342–379. Johnson, Paul E., and Robert Huckfeldt. 2004. “Agent-Based Explanations for the Survival of Disagreement in Social Network.” In The Social Logic of Politics: Personal Networks as Contexts for Political Behavior, ed. Alan S. Zuckerman, 251–268. Philadelphia: Temple University Press. Kenny, Christopher. 1992. “Political Participation and Effects from the Social Environment.” American Journal of Political Science 36 (1): 259–267. ———. 1994. “The Microenvironment of Attitude Change.” Journal of Politics 56 (3): 715–728. King, Gary, James Honaker, Anne Joseph, and Kenneth Scheve. 2001. “Analyzing Incomplete Political Science Data: An Alternative Algorithm for Multiple Imputation.” American Political Science Review 95 (1): 49–69. Klofstad, Casey A. 2007. “Talk Leads to Recruitment: How Discussions about Politics and Current Events Increase Civic Participation.” Political Research Quarterly 60 (2): 180–191. Klofstad, Casey A., Scott D. McClurg, and Meredith Rolfe. 2009. “Measurement of Political Discussion Networks: A Comparison of Two ‘Name Generator’ Procedures.” Public Opinion Quarterly 73 (3): 462–483. Kopstein, Jeffery S., and David A. Reilly. 2000. “Geographic Diffusion and the Transformation of the Postcommunist World.” World Politics 53 (1): 1–37. Lake, Ronald La Due., and Robert Huckfeldt. 1998. “Social Capital, Social Networks, and Political Participation.” Political Psychology 19 (3): 567–583. Latané, Bibb, and Sharon Wolf. 1981. “The Social Impact of Majorities and Minorities.” Psychological Review 88 (5): 438–453. Lau, Richard R., and Dan Russell. 1980. “Attributions in the Sports Pages.” Journal of Personality and Social Psychology 39 (1): 29–38. Laver, Michael. 2005. “Book Review of The Social Logic of Politics.” Perspectives on Politics 3 (4): 933–934. Lazarsfeld, Paul F., Bernard Berelson, and Hazel Gaudet. 1968. The People’s Choice: How the Voter Makes Up His Mind in a Presidential Election, 3rd ed. New York: Columbia University Press. Lupia, Arthur, and Matthew D. McCubbins. 1998. The Democratic Dilemma: Can Citizens Learn What They Need to Know? Cambridge: Cambridge University Press. Marsden, Peter V., and Noah E. Friedkin. 1993. “Network Studies of Social Influence.” Sociological Methods and Research 22 (1): 127–151.

References

171

Mayhew, David. 1974. Congress: The Electoral Connection. New Haven, CT: Yale University Press. McClurg, Scott D. 2003. “Social Networks and Political Participation: The Role of Social Interaction in Explaining Political Participation.” Political Research Quarterly 56 (4): 449–464. ———. 2004. “Indirect Mobilization: The Social Consequences of Party Contacts in an Election Campaign.” American Politics Research 32 (4): 406–443. ———. 2006. “The Electoral Relevance of Political Talk: Examining Disagreement and Expertise Effects in Social Networks on Political Participation.” American Journal of Political Science 50 (3): 737–754. McDonald, Michael, and Samuel Popkin. 2001. “The Myth of the Vanishing Voter.” American Political Science Review 95 (4): 963–974. McIntosh, Hugh, Daniel Hart, and James Youniss. 2007. “The Influence of Family Political Discussion on Youth Civic Development: Which Parent Qualities Matter?” PS: Political Science and Politics (July): 495–499. Menand, Louis. 2004. “The Unpolitical Animal: How Political Science Understands Voters.” New Yorker, August 30, 92–96. Mendelberg, Tali. 2002. “The Deliberative Citizen: Theory and Evidence.” In Political Decision Making, Deliberation, and Participation, vol. 6, ed. Michael X. Delli Carpini, Leonie Huddy, and Robert Y. Shapiro, 151–193. Greenwich, CT: JAI Press. Meyer, Gordon W. 1994. “Social Information Processing and Social Networks: A Test of Social Influence Mechanisms.” Human Relations 47 (9): 1013–1047. Michener, H. Andrew, and John D. DeLamater. 1999. Social Psychology. New York: Harcourt Brace. Mutz, Diana C. 2002. “The Consequences of Cross-Cutting Networks for Political Participation.” American Journal of Political Science 46 (4): 838–855. ———. 2006. Hearing the Other Side: Deliberative versus Participatory Democracy. Cambridge: Cambridge University Press. Newcomb, Theodore M. 1943. Personality and Social Change: Attitude Formation in a Student Community. New York: John Wiley and Sons. Newcomb, Theodore M., Katherine E. Koenig, Richard Flacks, and Donald P. Warwick. 1967. Persistence and Change: Bennington College and Its Students after Twenty-Five Years. New York: John Wiley and Sons. Nickerson, David W. 2008. “Is Voting Contagious? Evidence from Two Fields Experiments.” American Political Science Review 102 (1): 49–57. Olson, Mancur. 1965. The Logic of Collective Action. Cambridge, MA: Harvard University Press. Page, Benjamin, and Robert Shapiro. 1992. The Rational Public. Chicago: University of Chicago Press. Petty, Richard E., and John T. Cacioppo. 1986. “The Elaboration Likelihood Model of Persuasion.” In Advances in Experimental Social Psychology, vol. 19, ed. Leonard Berkowitz, 123–205. New York: Academic Press. Petty, Richard E., Duane T. Wegener, and Leandre R. Fabrigar. 1997. “Attitudes and Attitude Change.” Annual Review of Psychology 48: 609–647.

172

References

Pierson, Paul. 2000. “Increasing Returns, Path Dependence, and the Study of Politics.” American Political Science Review 94 (2): 251–267. Pike, Thomas W., Madhumita Samanta, Jan Lindström, and Nick J. Royal. 2008. “Behavioural Phenotype Affects Social Interactions in an Animal Network.” Proceedings of the Royal Society B 275 (1650): 2515–2520. Plutzer, Eric. 2002. “Becoming a Habitual Voter: Inertia, Resources, and Growth in Young Adulthood.” American Political Science Review 96 (1): 41–56. Popkin, Samuel. 1995. “Information Shortcuts and the Reasoning Voter.” In Information, Participation, and Choice, ed. Bernard Grofman, 17–35. Ann Arbor: University of Michigan Press. Putnam, Robert. 1994. Making Democracy Work: Civic Traditions in Modern Italy. Princeton, NJ: Princeton University Press. ———. 2000. Bowling Alone: The Collapse and Revival of American Community. New York: Simon and Schuster. Reifenberg, Richard J. 1986. “The Self-Serving Bias and the Use of Objective and Subjective Methods for Measuring Success and Failure.” Journal of Social Psychology 126 (5): 627–631. Rosenbaum, Paul R. 1991. “A Characterization of Optimal Designs for Observational Studies.” Journal of the Royal Statistical Society 53 (3): 597–610. Rosenstone, Steven, and John Mark Hansen. 1993. Mobilization, Participation, and Democracy in America. New York: Macmillan. Ross, Mike, and G. Fletcher. 1985. “Attribution and Social Perception.” In The Handbook of Social Psychology, 3rd ed., vol. 2, ed. Gardner Lindzey and Elliot Aronson, 73–122. Reading, MA: Addison-Wesley. Sacerdote, Bruce. 2001. “Peer Effects with Random Assignment: Results for Dartmouth Roommates.” Quarterly Journal of Economics 116 (2): 681–704. Sally, David. 1995. “Conversation and Cooperation in Social Dilemmas: A MetaAnalysis of Experiments from 1958 to 1992.” Rationality and Society 7 (1): 58–92. Schachter, Stanley. 1959. The Psychology of Affiliation: Experimental Studies of the Sources of Gregariousness. Stanford, CA: Stanford University Press. Scott, John. 1991. Social Network Analysis. London: Sage. Siegel, David A. 2009. “Social Networks and Collective Action.” American Journal of Political Science 53 (1): 122–138. Silbiger, Sara L. 1977. “Peers and Political Socialization.” In The Handbook of Political Socialization, ed. Stanley Allen Renshon, 172–189. New York: Free Press. Skocpol, Theda. 2003. Diminished Democracy: From Membership to Management in American Civic Life. Norman: University of Oklahoma Press. Stimson, James A. 1990. “A Macro Theory of Information Flow.” In Information and Democratic Processes, ed. James Kuklinski and John Ferejohn, 345–368. Champaign: University of Illinois Press. Stuart, Elizabeth A., and Kerry M. Green. 2008. “Using Full Matching to Estimate Causal Effects in Nonexperimental Studies: Examining the Relationship between Adolescent Marijuana Use and Adult Outcomes.” Developmental Psychology 44 (2): 395–406.

References

173

Takahashi, Keiko, and Naomi Majima. 1994. “Transition from Home to College Dormitory: The Role of Preestablished Affective Relationships in Adjustment to a New Life.” Journal of Research on Adolescence 4 (3): 367–384. Terenzini, Patrick T., Laura I. Rendon, M. Lee Upcraft, Susan B. Millar, Kevin W. Allison, Patricia L. Gregg, and Romero Jalomo. 1994. “The Transition to College: Diverse Students, Diverse Stories.” Research in Higher Education 35 (1): 57–73. Tocqueville, Alexis de. 2000. Democracy in America, trans. and ed. Harvey C. Mansfield and Delba Winthrop. Chicago: University of Chicago Press. University of Wisconsin, Madison, Division of University Housing. 2004. University Housing Application Process. Available online at http://www.housing.wisc .edu (accessed January 17, 2007). Verba, Sidney, Kay Lehman Schlozman, and Henry E. Brady. 1995. Voice and Equality. Cambridge, MA: Harvard University Press. Walsh, Katherine Cramer. 2004. Talking About Politics: Informal Groups and Social Identity in American Life. Chicago: University of Chicago Press. Wasserman, Stanley, and Katherine Faust. 1994. Social Network Analysis: Methods and Applications. Cambridge: Cambridge University Press. Wood, Wendy 2000. “Attitude Change: Persuasion and Social Influence.” Annual Review of Psychology 51: 539–570. Zaller, John. 1992. The Nature and Origins of Mass Opinion. Cambridge: Cambridge University Press. Zeleny, Jeff. 2008. “In Obama’s Victory, Clues to the Challenges Ahead.” New York Times, January 5, A10. Zhang, Yin. 2000. “Using the Internet for Survey Research: A Case Study.” Journal of the American Society for Information Science 51 (1): 57–68. Zuckerman, Alan S. 2004. “Returning to the Social Logic of Politics.” In The Social Logic of Politics: Personal Networks as Contexts for Political Behavior, ed. Alan S. Zuckerman, 3–20. Philadelphia: Temple University Press. Zukin, Cliff, Scott Keeter, Molly Andolina, Krista Jenkins, and Michael X. Delli Carpini. 2006. A New Engagement? Political Participation, Civic Life, and the Changing American Citizen. Oxford: Oxford University Press.

Index

Page numbers in italics refer to figures and tables. American College Testing (ACT) Score, 145, 146, 156, 162 American National Election Studies, 4, 12, 19, 98n, 100n, 103n, 131, 133, 138 American Research Group, 1 Attitudes on importance of civic participation. See Perceived importance of civic participation Attributions on motivation, 50 self-serving bias in, 48–49 Bar-Tal, Daniel, 96 Beliefs on importance of civic participation. See Perceived importance of civic participation political ideology. See Ideology, political partisanship. See Partisanship Bias endogeneity (omitted variable), 5, 19, 21, 128 in motivation attribution, 48–49 reciprocal causation, 5, 19, 21, 128 selection, 5, 19, 21, 128 Brady, Henry E., 16, 55, 92 Bush, George H. W., 41 Bush, George W., 65, 139

Cacioppo, John T., 72 Campaign activities, participation in, 15, 34, 51, 149 in Obama rally, 32, 32n, 36, 49, 66 survey evidence on, 94, 121, 153 Campbell, David E., 60n, 95n Causal mechanisms, 7, 129 civic engagement in, 54, 57, 154–155 correlation of, 58–59 difficulties in research on, 18–20, 27, 128, 135–137 focus group evidence on, 48–49, 64–69 information resources in, 53–54, 57, 154 multivariate analysis of, 59–64 reciprocal, 5, 19, 21, 128 recruitment in, 155 social norms in, 55–56, 58, 155 survey evidence on, 42–48, 57–64, 154– 155 validity of research on, 24–26 Civic participation awakening moment in, 29 changes between high school and college, 34, 38–42, 116 changes during college years, 116–117 costs and benefits of, 3, 35, 40–41, 52– 56, 63–64, 69, 73–74, 120–121, 121n

176

Civic participation (continued) definition of, 13–16 differentiated from civic talk, 16 difficulties in research on, 18–20, 21, 27, 128 focus group evidence on, 48–49, 64–69, 87–88, 107, 131, 140 future research on, 135–137 individual characteristics affecting, 71– 89. See also Individual-level characteristics knowledge about politics and current events affecting, 37–38. See also Knowledge about politics and current events lasting effect of civic talk on, 8, 116–124, 129–130 measurement of, 33–42 motivation for, 52–56, 69. See also Motivation for civic participation non-political activities in. See Nonpolitical civic participation peer characteristics affecting, 91–108. See also Peers perceived importance of. See Perceived importance of civic participation political activities in. See Political activity participation predisposition to. See Predisposition to civic participation prior experience with. See Prior civic participation experiences recruitment for. See Recruitment role in democracy, 3, 9, 13–16 self-reinforcement of, 120–124 social-level antecedents of, 17–18 social norms affecting, 55–56, 58, 68–69. See also Norms survey evidence on, 42–48, 57–64, 75– 88, 96–106, 110–113, 116–124 trends across different activities, 34, 35–38 validity of research on, 24–26 in voluntary membership organizations. See Voluntary civic organization participation voting in. See Voter turnout Civic talk amount of, in social network, 7, 12, 30– 32, 131 awakening moment in, 29 and civic engagement, 7, 54, 57–66, 113–115

Index

compared to deliberation, 12, 130–135 composition of discussion network in, 12–13 conflict avoidance in, 31–32, 65, 94–95, 107, 134 definition of, 2, 11–12, 16 differentiated from civic participation, 16 difficulties in research on, 18–20, 27, 128, 135–137 duration of impact, 8, 116–124, 129–130 focus group evidence on, 48–49, 64–69, 87–88, 107, 131, 140 frequency of. See Frequency of civic talk future research on, 135–137 individual characteristics in, 71–89. See also Individual-level characteristics influence of, compared to individual level characteristics, 8, 43–47, 50, 110–113, 129 informal and accidental nature of, 11–12, 131 as information resource, 53–54, 57, 59–65 lasting effect of, 8, 116–124, 129–130 measurement of, 30–33 news media use affecting, 33, 87n, 105, 131 normative implications of, 8, 130–135 in participatory democracy, 2, 5, 8–9, 23–24, 50, 71, 89, 94, 109, 116, 127– 128, 130, 135, 138–139 peer characteristics in, 7–8, 91–108, 129. See also Peers persuasion attempts in, 49, 65, 68, 72– 75, 92–93 and political participation, 45–46, 50–51, 119–120, 123–124 potential problems in, 132–135 recruitment in. See Recruitment self-reports on, 30–33 survey evidence on, 42–48, 57–64, 75– 88, 96–106, 110–113, 116–124 topics discussed in, 32–33 validity of research on, 24–26 variables not influenced by, 115, 134 and voluntary civic organization participation, 7, 42–45, 51, 60–61, 75–88, 96–107, 118–119, 122, 129 and voter turnout, 7, 46–48, 51, 62–63, 75–88, 96–107, 119–120, 129 Civil society, 14, 14n, 15 participation in, 14. See also Civic participation

Index

Clinton, Bill, 41 Clinton, Hillary, 1–2, 32n, 65, 141 Cognitive dissonance in political disagreements, 94–95 in social norm violations, 56 College students in first year, 6, 21–22 as crucial case of peer influence, 24–26 changes in civic participation between high school and college, 34, 38–42, 116 free time of, 39, 40, 158 political activities of, 34, 35–38, 116, 153 preference for community involvement, 36–37, 46, 114–115, 140–141, 158 prior civic participation experience of, 34, 38, 75–78 priorities of, 38–40 as respondents and non-respondents in C-SNIP Panel Study, 144–147 College students in fourth year, 6, 22, 129– 130 changes in civic participation during college years, 116–117 continuing effect of civic talk on, 8, 116– 124 as respondents and non-respondents in C-SNIP Panel Study, 144–147 Collegiate Social Network Interaction Project (C-SNIP) Focus Group Study, 23n, 23–24 on causal mechanisms, 48–49, 64–69 on civic engagement, 65–66 on decreases in civic participation between high school and college, 40– 42 on disagreement and conflict avoidance, 31–32, 65, 107 on frequency of civic talk, 31–32 on individual characteristics, 87–88 on information resources, 64–65 on link between civic talk and civic participation, 48–49 on political and non-political civic participation, 36–38 questions asked in, 147–149 on recruitment attempts, 65, 66–68, 69, 87–88 research design in, 147–149 on social norms, 68–69 on topics discussed while engaging in civic talk, 32–33

177

Collegiate Social Network Interaction Project (C-SNIP) Panel Survey, 6–8, 21– 22 on causal mechanisms, 42–48, 57–64, 154–155 characteristics of respondents and nonrespondents in, 144–146 on civic engagement, 57, 58–64, 78–81, 154–155 on civic expertise of peers, 103–106 on frequency of civic talk, 30, 154 on homogeneity of peer group, 100–103 on importance of civic participation, 37, 48, 106, 158 on individual characteristics, 75–87, 155–156 on knowledge and education, 85–87, 88, 103–104, 156, 157 matching procedure in. See Matching procedure in data pre-processing on peer characteristics, 96–106, 157 on political participation, 34–38, 45–46, 116, 119–120, 152–153 on political preferences, 82–85, 156 on preference for community or political involvement, 36–37, 46, 114–115, 140–141, 158 on prior civic participation experience, 34, 38, 75–78 questions and variables in, 151–159 research design in, 143–147, 151–159 response rate in, 144 on social intimacy, 96–100 on voluntary organization participation, 7, 34, 35–36, 42–45, 51, 60–61, 75– 88, 96–107, 118–119, 122, 129, 151, 152 on voter turnout, 7, 34–35, 46–48, 51, 62–63, 75–88, 96–107, 119–120, 129, 153 wave 1, 6, 21–22, 143, 145, 146. See also High school students wave 2, 6, 21–22, 143, 145, 146–147. See also College students in first year wave 3, 6, 22, 116–124, 143, 145, 146– 147. See also College students in fourth year Commission on National and Community Service, 41 Community involvement, preference for, compared to political involvement, 36–37, 46, 114–115, 140–141, 158

178

Competence, civic, 18, 66 Compliance with group, social intimacy affecting, 92 Conflicts in political disagreements, 93–95 avoidance of, 31–32, 65, 94, 95, 107, 134 Conservative ideology, 82–85, 156 Contact with elected officials, 14, 15, 34, 51, 152, 153 survey evidence on, 152, 153 Cooperation in social networks, 131–132 and groupthink, 134 Corporation for National and Community Service, 41 Cost-benefit analysis of civic participation, 3, 35, 40–41, 52–56, 63–64, 69, 73–74, 120–121, 121n data gathering and data analysis phases in, 73–74 Crucial cases, 24–26, 26n, 135 C-SNIP Focus Group Study. See Collegiate Social Network Interaction Project (C-SNIP) Focus Group Study C-SNIP Panel Survey. See Collegiate Social Network Interaction Project (C-SNIP) Panel Survey Data matching procedure. See Matching procedure in data pre-processing Deliberation, 12, 130–135 as benchmark for civic discourse, 130 compared to civic talk, 12, 130, 131–132 policy implications of, 12, 132, 134 social level factors in, 17, 92 Democracy active involvement of citizens in, 3, 9 citizen interactions in, 17, 17n and civil society, 14, 14n costs and benefits of participation in, 3, 35, 40–41, 52–56, 63–64, 69, 73–74, 120–121, 121n disengagement affecting, 13–14, 26, 89, 91, 133, 137–138 participatory, role of civic talk in, 2, 5, 6–9, 71, 116, 127, 130, 135 peer-based civic mobilization in, 138– 139 political disagreements in, 93–94, 135n sociological studies of, 5, 17–18 Democracy in America (Tocqueville), 17 Disagreements, political, 8, 93–95 conflict avoidance in, 31–32, 65, 94–95, 107, 134 focus group evidence on, 31–32, 107

Index

and peer group homogeneity, 100–103 persuasion attempts in, 65 policy solutions in, 135, 135n Discursive discourse, 130–132 Discussions common topics in, 31–33 discursive, 130–132 frequency of civic talk in, 7, 12, 30–32 on large scale issues, 135 network of individuals included in, 12–13 Dormitory assignment, 21, 21n Downs, Anthony, 52 Duty, civic, 54 An Economic Theory of Democracy (Downs), 52 Education level and civic participation, 3–4, 74 of parents, 87–89, 112, 113, 156 Edwards, John, 1 Efficacy, political civic talk affecting, 114–115 compared to effect of civic talk, 111 and motivation for civic participation, 54 survey evidence on, 57, 59, 61, 62, 79– 81, 154–156 and voluntary organization participation, 60–61, 78–81 and voter turnout, 62, 63, 79–81 Egalitarian discussions in social networks, 131–134 Elaboration Likelihood Model of persuasion, 72 Elections ambivalence about, 94n civic mobilization in, 139 midterm congressional election (2006), 119–120, 123–124, 125, 153 political disagreements on, 94 presidential primary elections. See Presidential primary elections recruitment attempts in, 55, 62–63, 66 as topic of civic talk, 32, 64, 65 voter turnout for. See Voter turnout Endogeneity bias, 5, 19, 21, 128 Engagement, civic, 3, 7, 129 civic talk affecting, 7, 54, 57, 59, 61, 62, 65–66, 113–115 compared to effect of civic talk, 111 correlation with other causal mechanisms, 58–59 in cost-benefit analysis of civic participation, 74

Index

focus group evidence on, 65–66 knowledge on politics and current events affecting, 18, 37–38, 66 measures of, 57, 64n and motivation for civic participation, 54 multivariate analysis of, 59–63 in Obama campaign, 36 preference for community involvement affecting, 36–37 of roommate, 103–106 and sense of political efficacy, 78–81 survey evidence on, 57, 58–63, 78–81, 154–155 and voluntary organization participation, 60–61 and voter turnout, 54, 63–64, 69, 79–81 Enjoyment of time spent with roommate, 98–100, 105–106, 157 Environmental concerns as discussion topic, 33, 49 Environmental stimuli, interaction with personal characteristics, 71–89 Expertise, civic, 91, 95–96, 103–106 civic talk affecting, 114–115, 129 compared to effect of civic talk, 110–113 and influence of civic talk, 7, 91, 95–96, 103–106, 108 survey evidence on, 103–106, 108, 157 Festinger, Leon, 56, 72, 93 Focus group study in C-SNIP. See Collegiate Social Network Interaction Project (C-SNIP) Focus Group Study Free time available, 3 survey evidence on, 39–40, 158 Frequency of civic talk, 7, 12, 30–32, 131 conflict avoidance affecting, 31–32 focus group evidence on, 31 interest of roommate in politics and current events affecting, 32 self-reports of, 12, 30–32 survey evidence on, 12, 30, 154 Gender, and effect of home life on civic participation, 77 Gerber, Alan, 55 German Weimar Republic, 14n Green, Donald, 55 Groupthink, 134 Health care concerns as discussion topic, 33 Heterogeneity of community, and voter turnout, 95n

179

High school students, 6, 22 free time of, 39, 158 knowledge about politics and current events, 39, 85n, 156, 158–159 news media use of, 85n, 158–159 political activities of, 34, 36–38, 46, 47, 152–153 political ideology of, 82–85, 156 preference for community involvement, 36–37 service learning opportunities of, 40–42, 117, 140 transition to first year of college, 35–42, 116–117 variables measured in, 152, 153, 156, 158 voluntary organization membership of, 34, 35–38, 40–42, 152 Homogeneity of peer group, 91, 93–95, 107–108, 129 survey evidence on, 100–103, 157 Huckfeldt, Robert, 12, 18, 74, 94n, 94–95 Ideology, political changes during college years, 117 compared to effect of civic talk, 112–113 disagreements on, 8, 31–32, 93–95, 107 in high school, 82, 156 and homogeneity of peer group, 100– 103 minimal impact of civic talk on, 115, 125 and persuasion attempts, 65 survey evidence on, 82–85, 117, 156 Importance of civic participation, 3, 9, 14– 15 norms on, 58, 68–69, 154, 155 perceptions on. See Perceived importance of civic participation Income, 3 and voter turnout, 133 Individual-level characteristics, 3–4, 7, 9, 71–89, 129 existing research on, 19, 19n focus group evidence on, 87–88 influence of, compared to civic talk, 8, 44, 47, 110–113, 129 interaction with environmental stimuli, 71–89 survey evidence on, 75–87, 155–156 Information resources correlation with other causal mechanisms, 58–59 focus group evidence on, 64–65, 131 multivariate analysis of, 59–63

180

Information resources (continued) peers as, 18, 53–54, 57, 59, 64–65, 95– 96, 154 quantity and quality of, 137 survey evidence on, 57, 58–63, 154 and voter turnout, 53–54, 62, 63 Interest group contributions, cost-benefit analysis of, 52, 52n Interest in politics and current events, 3, 7, 131 changes during college years, 117 and civic engagement, 54, 64n, 78–79 civic talk affecting, 49, 59, 62, 63–64, 114 focus group evidence on, 65–66 in high school, 155, 156 norms on, 154, 155 of parents, 77, 78, 110–111, 155, 156 of roommate, 7, 32, 58, 59, 60n, 68–69, 103–106, 157 survey evidence on, 57, 59, 62, 63–64, 154, 156, 157 and voluntary organization participation, 59, 60–61, 63–64 and voter turnout, 59, 62, 63–64, 79, 80 Intimacy, social, 8, 91, 92–93, 108, 129 in physical proximity, 92–93 survey evidence on, 96–100 trust as measure of, 96–97 Iowa caucuses (2008), 1–2, 9, 141 Italy, regional differences in, 15 Kerry, John, 139 Kids Voting USA, 141n Knowledge about politics and current events and civic competence, 18, 66 and civic engagement, 18, 37–38, 66 civic talk affecting, 114, 129 compared to effect of civic talk, 111 compared to free time and news media use, 39–40 correlation with other causal mechanisms, 58–59 cost-benefit analysis of information sources in, 53–54 enhancing civic talk effect, 108 in high school, 39, 84n, 156, 158 of peer or roommate, 7, 91, 95–96, 103– 106, 157 source of information in. See Information resources survey evidence on, 40, 57, 85–87, 89, 103–106, 156, 158

Index

La Follette, Bob, 82n Lake, Ronald La Due, 18 Lasting effect of civic talk, 8, 116–124, 129–130 Latané, Bibb, 56 Laver, Michael, 11 Learn and Serve America, 41 Liberal ideology, 31, 82–85, 156 The Logic of Collective Action (Olson), 52 Lupia, Arthur, 93n Massachusetts Institute of Technology, 56, 93 Matching procedure in data pre-processing, 22–23, 23n, 128, 161–163 benefits of, 22–23, 135–136 pre-treatment variables used in, 161–162 on voluntary civic organization participation, 43–44, 44n on voter turnout, 46, 47, 47n MatchIt package, 161 McCain, John, 65 McClurg, Scott D., 18, 25n, 74 McCubbins, Matthew D., 93n Measurement methods on civic participation, 33–42 on civic talk, 30–33 Media. See News media use Menand, Louis, 127 Mendelberg, Tali, 92, 130 Michigan School of political behavior, 19n Mobilization, civic, 4n, 121 peer-based, 138–139 Modeling, behavioral, 58, 59, 68 Motivation for civic participation, 52–56, 69, 117, 117n attributions on, 48–49, 50 cost-benefit analysis of, 52–56, 69, 73– 75 and response to environmental stimuli, 72–75 Multivariate analysis of causal mechanisms, 59–63 Mutz, Diana, 93–94, 94n National and Community Service Act (1990), 41 National and Community Service Trust Act (1993), 41 Newcomb, Theodore, 17n News media use changes during college years, 117, 158, 159

Index

civic talk affecting, 114, 115 compared to effect of civic talk, 112, 113 compared to free time and political knowledge, 39–40 in high school, 85n, 158 as information resource, 64 as source of civic talk, 33 Non-political civic participation, 14, 15–16 compared to political participation, 16, 35–38 examples of activities in, 15–16 measurement of, 34, 35, 116 peer characteristics affecting, 97, 99, 101, 102, 104, 106, 107 prior experience with, 38 trends in, 35–42 voluntary organization membership in. See Voluntary civic organization participation Norms, 17, 55–56, 58, 129 correlation with other causal mechanisms, 58–59 definition of, 55 focus group evidence on, 68–69 on importance of civic participation, 58, 68–69, 154, 155 modeling of, 58, 68 multivariate analysis of, 59–63 survey evidence on, 58–63, 154, 155 and voluntary organization participation, 61 and voter turnout, 56, 62, 63 Obama, Barack as discussion topic, 33, 36, 49, 64, 65 Iowa caucus votes for, 1–2, 141 Wisconsin campaign rally for, 32, 32n, 36, 49, 66 Olson, Mancur, 52 Omitted variable bias, 5, 19, 21, 128 Optmatch package, 161 Panel Survey in C-SNIP. See Collegiate Social Network Interaction Project (C-SNIP) Panel Survey Paradigmatic cases, 26 Parent civic interest and activity, 77, 78, 155, 156 compared to effect of civic talk, 110–111 Parent education, 87–89, 156 compared to effect of civic talk, 112, 113 Parks, Rosa, 120n Path dependence concept, 120–121, 121n

181

Peers, 5–8, 91–108, 129 civic expertise of, 91, 95–96, 103–106, 108 in civic mobilization efforts, 138–139 cooperative and egalitarian discussions with, 131–132 disagreements with, 93–95, 100–103 focus group evidence on, 107 frequency of contact with, 138–139 information resources transferred by, 53–54, 57, 59, 64–65, 95–96 recruitment by, 55, 57, 59, 66–68, 92 roommates as. See Roommates similarity with, 91, 93–95, 100–103, 107, 129, 157 social comparison theory on, 72 social intimacy with, 91, 92–93, 96–100 social norms of, 56, 58, 59, 68–69 survey evidence on, 96–106, 157 trust between, 8, 91, 93, 96–97, 157 Perceived importance of civic participation norms on, 58, 68–69, 154, 155 survey evidence on, 37, 58, 59, 154, 155, 158 Persuasion attempts in civic talk, 49, 65 individual characteristics affecting response to, 72–75 social intimacy affecting response to, 92–93 Petty, Richard E., 72 Policy implications of civic talk and participation, 15, 16, 132, 134–135 of deliberation, 12, 132, 134 Political activity participation, 14, 15 changes during college years, 116–117, 119, 123–124 compared to non-political participation, 16, 35–38 effect of civic talk on, 45–48, 119, 123– 124 examples of, 15, 34 of high school students, 34, 36, 38, 46, 47, 116, 153 long term effect of civic talk on, 119, 123–124 measurement of, 34–35 preference for community involvement affecting, 36–37, 46, 140–141, 158 prior experience in, 38 sleeper effect of civic talk on, 123–124 survey evidence on, 34–37, 45–48, 152– 153

182

Political activity participation (continued) trends in, 35–42 voting in. See Voter turnout Political party representatives, frequency of contact with, 138–139 Political similarity with roommate, 91, 93– 95, 100–103, 107, 129, 157 Popkin, Samuel, 95–96 Predisposition to civic participation, 7, 50 benefits of civic talk in, 89, 133 data matching procedure on, 162 recruitment effectiveness in, 67–68 response to environmental stimuli in, 72–75, 89 Presidential primary elections Iowa caucuses in (2008), 1–2, 141 recruitment attempts in, 66 as topic of civic talk, 32, 64, 65 voter turnout for (2004), 34–35, 46–48, 124, 153 voter turnout for (2008), 36, 49 Prior civic participation experiences, 7, 38, 50 compared to effect of civic talk, 110–111 data matching procedure on, 162 self-reinforcement in, 120–123 survey evidence on, 75–78 in voluntary civic organizations, 43, 44, 45, 75–78 and voter turnout, 45–48, 75–78 Prisoners Dilemma task, 132 Protest activities, participation in, 14–15, 34, 51, 82n, 149, 152, 153 Putnam, Robert, 15, 56, 89 Random assignment of roommates, 6–7, 20–21, 21n, 23 Reciprocal causation, 5, 19, 21, 128 Recruitment, 3n, 7, 55, 129 correlation with other causal mechanisms, 58–59 face-to-face style of, 55, 55n focus group evidence on, 66–68, 69, 87– 88 multivariate analysis of, 59–63 by rational prospectors, 121 social intimacy affecting response to, 92 survey evidence on, 57, 58–63, 69, 154, 155 and voluntary organization participation, 61, 63–64, 69 and voter turnout, 55, 55n, 62, 63–64, 66

Index

Research design, 5–6, 20–26, 135–137 bias in, 5, 19–23, 128 crucial cases in, 25–26, 26n, 135 in C-SNIP Focus Group Study, 147– 149 in C-SNIP Panel Survey, 143–147, 151– 159 data matching procedure in. See Matching procedure in data pre-processing future agenda in, 8, 135–137 random assignment in, 5–6, 7, 20–23, 21n validity of, internal and external, 24–25 Resources in cost-benefit analysis of civic participation, 3, 53–54, 74 focus group evidence on, 64–65 free time, 3, 39–40, 158 income, 3, 133 information. See Information resources survey evidence on, 57, 58–63, 154 Roommates characteristics of, 157 civic engagement of, 103–106 civic expertise of, 103–106, 157 disagreements and conflict avoidance with, 31–32, 65, 107 enjoyment of time spent with, 98–100, 105–106, 157 frequency of civic talk with, 154 as information resource, 57, 64–65, 154 interest in politics and current events, 7, 32, 58, 59, 60n, 103–106, 157 modeling behavior of, 68 perceived activity level of, 154, 155 random assignment of, 6–7, 20–21, 21n, 23 recruitment by, 57, 58–64, 66–68 similarity with, 100–103, 108, 157 topics discussed with, 30–33 trust in, 96–97, 157 Saxe, Leonard, 96 Selection bias, 5, 19, 21, 128 Self-reinforcement process, 120–123 Self-report measures on civic participation, 34–35 on civic talk, 30–33 Service learning opportunities in high school, 40–42, 117, 140 Silbiger, Sara, 127

Index

Similarity with roommate or peer, 91, 93– 95, 100–103, 108, 129 in overall general characteristics, 91, 100, 101, 108, 129, 157 political, 91, 93–95, 100–101, 102, 108, 129, 157 Sleeper effects, 123–124 Social comparison theory, 72 Social impact theory, 56 Social intimacy. See Intimacy, social Socialization, political, 17, 77, 121 Social-level variables, 5–6, 17–18, 19 peer characteristics. See Peers Social networks, 17–18 amount of civic talk in, 5, 12, 18, 30–32 in civic mobilization, 138–139 cooperative and egalitarian discussions in, 131–132, 133–134 design of research on, 5–6, 20–26, 135– 137 disagreements in, 93–95 non-political and political discussions in, 12–13 peers in. See Peers social comparison theory on, 72 social intimacy in, 108. See also Intimacy, social Sprague, John, 12, 18, 74 Student government involvement, 34–35, 67, 151, 151n, 152 Survey evidence in C-SNIP. See Collegiate Social Network Interaction Project (C-SNIP) Panel Survey Tocqueville, Alexis de, 17, 17n Trust in government, 115, 134 in roommate or peer, 8, 91, 93, 96–97, 157 in society, 56 University of Wisconsin, Madison C-SNIP Focus Group Study in, 6, 23n, 23–24. See also Collegiate Social Network Interaction Project (C-SNIP) Focus Group Study C-SNIP Panel Survey in, 6, 21–22. See also Collegiate Social Network Interaction Project (C-SNIP) Panel Survey progressive political culture in, 82, 82n random assignment of roommates in, 21, 21n, 23

183

Validity of research, internal and external, 24–25 Values on importance of civic participation. See Perceived importance of civic participation Verba, Sidney, 3, 3n, 53n, 74, 121 Voice and Equality (Verba, Schlozman, and Brady), 3 Voluntary civic organization participation, 7, 15–16 civic expertise of roommate affecting, 103–106 of college students in fourth year, 116– 117, 152 compared to political activities, 35–38 cost-benefit analysis of, 63–64 data matching procedure on, 43–44, 44n effect of civic talk on, 42–45, 51, 60–61, 75–88, 96–107, 110–113, 116–124 information resources on, 59, 61, 64–65 knowledge about politics and current events affecting, 85–87 lasting effect of civic talk on, 116–124 measurement of, 34, 35, 116 parent civic interest or activity affecting, 77–78 parent education affecting, 87, 88 peer group homogeneity affecting, 100– 103 policy implications of, 15–16, 132 political efficacy affecting, 60–61, 79–81 political ideology affecting, 82–85 prior experience in, 43–45, 75–76 recruitment affecting, 61, 63, 69 self-reinforcement in, 120–123 in service learning opportunities, 40–42, 117, 140 social intimacy affecting, 96–100 survey evidence on, 35–36, 40, 42–45, 51, 60–61, 75–88, 96–107, 110–113, 116–124, 151, 152 types of organizations in, 35, 151, 152 Voter turnout civic engagement affecting, 54, 59, 62, 63, 69, 78–79 civic expertise of roommate affecting, 103–106 of college students in fourth year, 116, 119–120, 123–124, 153 cost-benefit analysis of, 52–56, 63–64 data matching procedure on, 46, 47, 47n

184

Voter turnout (continued) and education level, 4 effect of civic talk on, 46–48, 51, 62–63, 75–88, 96–107, 110–113, 116–124 heterogeneity of community affecting, 95n by income, 133 information resources affecting, 53–54, 62, 63 knowledge about politics and current events affecting, 85–87 long term effect of civic talk on, 119, 123–124 measurement of, 34–35, 116, 153 parent civic interest or activity affecting, 75–77, 78 parent education affecting, 87, 88 peer group homogeneity affecting, 100– 103

Index

political efficacy affecting, 62, 63, 78–81 political ideology affecting, 82–85 prior civic participation experience affecting, 46, 47–48, 75–77, 78, 120– 121 recruitment affecting, 55, 55n, 62, 63, 66 self-reinforcement in, 120–121 social intimacy affecting, 96–100 social networks affecting, 139 social norms affecting, 56, 62, 63 survey evidence on, 34–35, 40, 46–48, 51, 62–63, 75–88, 96–107, 110–113, 116–124, 120, 153, 153 trends in, 34, 116, 132–133, 153 Wolbrecht, Christina, 60n Wolf, Sharon, 56

CASEY A. KLOFSTAD is an Assistant Professor of Political Science

at the University of Miami.