Abstract While much is known about the influence of partisan elites on mass opinion, relatively little is known about peer-to-peer influence within parties. We test the impact of messages signaling political parties’ issue stances on citizens’ own professed policy preferences, comparing the influence of party elites to that of co-partisan peers. Using an online experiment conducted with a quasi-representative sample of Americans, we demonstrate across two policy domains (education and international trade) that the opinions of co-partisan peers are just as influential on citizens’ policy preferences as the opinions of party elites. Further, the mechanisms underlying elite and peer influence appear to differ, with conformity to peers—but not elites—driven almost exclusively by strength of social identification with the party. In Democracy for Realists, Chris Achen and Larry Bartels (2016) posit that, contrary to received folk wisdom, voter preferences are not at the heart of democracy. The opinions of the mass public do not primarily determine or constrain policymaking in the United States. Instead, such notions play “much the same role in contemporary democratic ideology that the divine right of kings played in the monarchical era” (Achen & Bartels, 2016, p. 19). They argue that social identities—and partisan loyalties—are largely responsible for guiding citizens’ vote choices and the very policy preferences believed to be at the root of responsive government. In this article, we test the impact of messages signaling political parties’ issue stances on citizens’ own professed policy preferences. While existing scholarship has emphasized the degree to which the public is influenced by party messages propagated by elites, we argue that perceptions of collective preferences among fellow partisans in the public are at least as influential. Furthermore, we suggest that the reasons for following peer cues may go beyond the information heuristic explanation posited for following elite cues. Partisans who view their affiliation as a social identity are more likely to conform to their co-partisan peers—a dynamic largely absent with respect to partisans’ receptivity to the preferences of party elites. We demonstrate these effects with an experiment embedded in an online survey conducted with a quasi-representative sample of Americans. Preferences for two contemporary issues are measured: Common Core educational standards and the Trans-Pacific Partnership (TPP) international trade agreement. We find that, on average, participants were equally responsive to elite and peer cues across both issues, but that conformity to peers was highly (and uniquely) contingent on the strength of individuals’ partisan social identity. These findings add to growing evidence that the process by which citizens form policy preferences may be less rational than previously assumed; the findings also add needed specificity and nuance with respect to the influence of parties in particular on policy preferences. We conclude that, whereas partisan elite cues may serve heuristic needs, partisan peer cues may serve more social ones. Partisan Influences on Citizens’ Political Preferences An affiliation with a political party is one of the most important factors structuring citizens’ opinions. Partisanship in the United States serves as a powerful lens through which many citizens view and interpret political phenomena (Bartels, 2002; Campbell, Converse, Miller, & Stokes, 1960; Cohen, 2003; Groenendyk, 2015; Kam, 2005; Levendusky, 2009; Rahn, 1993; although for a countervailing view, see, e.g., McGrath, 2016). Political parties are made up of many types of people of course: elected officials, party leaders, activists, and “ordinary citizens.” When political scientists have studied partisan influence, most often they have investigated the influence of party leaders and elected officials on citizens. From Converse (1964) to Zaller (1992) to Cohen et al. (2008), elite influences have loomed large as the primary explanation for how partisanship shapes public opinion. Empirical studies have consistently shown that the views and messages of prominent officials and party leaders powerfully alter public preferences (e.g., Lenz, 2012; Levendusky, 2009; Zaller, 1992;). While the party influence literature has focused nearly exclusively on the persuasive power of opinions expressed by party elites, arguably, the “party cues” to which most American citizens are exposed day-to-day are the opinions of their fellow co-partisans and “out-party” partisans—frequently in the form of public opinion polls. Are the opinions of “rank and file” party members as influential as those of party elites? If so—and given that ordinary peers are not policy experts—why might they be influential? A mostly separate, and older, line of research has examined the impact of peer rather than elite influences on partisans’ political preferences. In parallel to the elite cue literature, such studies have typically investigated the influence of majority opinion on individuals’ own stated views. For example, voting studies in the 1950s emphasized the power of social ties (e.g., family, neighbors, religious affiliations) to shape individuals’ political beliefs, arguing that political preferences were little more than cultural tastes (Berelson, Lazarsfeld, & McPhee, 1954; Katz & Lazarsfeld, 2006 ;). Likewise, studies by social psychologists and communication scholars have also long noted the robust tendency of individuals to conform1 to group norms and behaviors (Asch, 1940, 1951; Noelle-Neuman, 1984; Sherif, 1966 ;). Newcomb’s “Bennington College study” (1963), for instance, found that students who attended the politically liberal school became more liberal in comparison with similar students attending other colleges. While political scientists turned away from studying social influence in general during the 1990s and early 2000s (with some prominent exceptions, see e.g., Huckfeldt & Sprague, 1995), a number of recent studies have revisited this topic. Several field experiments have shown voter turnout to be particularly responsive to “social pressure” surrounding voting norms both offline (Gerber, Green, & Larimer, 2008, 2010; Gerber & Rogers, 2009;) as well as online (Bond et al., 2012; Haenschen, 2016;). Researchers have increasingly demonstrated peer influence on political opinion as well. Group opinion norms—communicated alone, without any persuasive arguments or information—reliably shift individuals’ political views in both face-to-face (Levitan & Verhulst, 2016) and more impersonal (Suhay, 2015) contexts. Recent reassessments of “bandwagon effects” of polls show them to be relatively robust (for a summary of studies, see Hardmeier, 2008). Perceptions of collective preferences—typically altered through randomizing exposure to polling information—can shift opinions by 5–15 percentage points depending on the salience of the issue (Rothschild & Malhotra, 2014). Taken together, these studies offer powerful evidence of how perceptions of collective preferences can shift public opinion. While social scientists may agree that information about peers’ preferences can sway individuals’ political attitudes, the influence of co-partisans specifically remains relatively unexplored. Political scientists researching social conformity have tended to focus on the influence of collective majoritarian preferences of the mass public overall (e.g., Marsh, 1985; Mutz, 1998; Rothschild & Malhotra, 2014;) or of relatively nonpolitical social groups such as neighbors or religious peers (e.g., Sinclair, 2012; Suhay, 2015;). Other than a study by Klar (2014)—who showed that partisans engage in considerably more partisan-motivated reasoning when discussing policies in ideologically homogeneous groups—we are aware of no published scholarship that rigorously tests, e.g., via experimental methods, the extent to which Americans are influenced by the perceived views of their partisan peers. This gap in the literature leads us to formulate our first hypothesis: H1.Participants exposed to information about partisans’ policy preferences will shift their policy preferences in the direction of co-partisans. Further, whereas there have been numerous experimental studies demonstrating the influence of party elite cues on ordinary party members’ attitudes, we are aware of no studies that have compared the influence of elite partisan cues to that of peer partisan cues. The accumulated evidence for peer conformity, however—in general and among a variety of social groups—leads us to suspect that, in the party context, peer cues may be as powerful as elite cues. Thus, we posit a second hypothesis: H2.On average, participants will be at least as influenced by ordinary partisans’ preferences as by party elites’ preferences. Social Identity as a Moderator of Partisan Conformity Why might ordinary citizens be influenced by party cues? Researchers who study elite partisan cues tend to argue that such cues serve a valuable informational purpose. Most citizens are “cognitive misers” and thus relatively uninformed of the details of civic and political affairs. Instead, they rely on cues from trusted sources, especially political parties. In this view, parties are coalitions of individuals with shared ideologies (e.g., Lupia & McCubbins, 1998; Noel, 2014;). Thus, party elites’ political stances serve as cognitive shortcuts or heuristics for rank and file partisans, allowing citizens to align their votes and policy preferences with their values and interests (e.g., Lupia, 1994; Page & Shapiro, 2012; Popkin, 1994; Schaffner & Streb, 2002 ) and avoid being persuaded by arbitrary issue frames (Druckman, 2001). Other scholars are less sanguine about citizens’ use of party cues to determine their political preferences (e.g., Achen & Bartels, 2016; Bartels, 2008; Kuklinski & Hurley, 1994). In contrast to the more cognitive–rational approach described above, many scholars conceive of parties more as social identity groups (e.g., Green, Palmquist, & Schickler, 2002; Huddy, Mason, & Aarøe, 2015). Thus, elites may be influential because they serve as proxies for the party as a social group, not simply because they are a cognitive shortcut linking values or interests with policy. Numerous studies are at least suggestive of the identity–conformity perspective. For example, partisans often take on party leaders’ positions even when they have access to information suggesting they should break with party preferences (Cohen, 2003; Rahn, 1993). Further, opinion surveys suggest citizens misunderstand facts (Bartels, 2002) and alter their interpretation (Gaines et al. 2007) to cast their party in a favorable light. That these biases are reduced when people are incentivized to be accurate (Bullock, Gerber, Hill, & Huber, 2015; Prior, Sood, & Khanna, 2015) does not take away from the fact that such biases appear to spontaneously and consistently arise in the survey context. While there is some evidence that partisanship as a social identity may drive adherence to the views of party elites, decades of social scientific research—mainly in Social Psychology—makes an even stronger case for identity driving adherence to the views of partisan peers. At the heart of self-categorization theory, an important off-shoot of social identity theory (see Turner, 1991; Turner, Hogg, Oakes, Reicher, & Wetherell, 1987) is the notion that group identifiers tend to adopt their social groups’ norms, that is, attitudes and behaviors that are prototypical of the group as a whole, regardless of what group leaders believe.2 Social conformity is considered a peer-to-peer phenomenon, occurring among individuals who perceive themselves to be similarly situated—in the same social group, and of the same status. Scholars have put forward two explanations for why identity drives peer conformity. In-group peers allow us to engage in “social reality testing” (Festinger 1950)—in a complicated and uncertain world, we can rely on people who are similar to help us understand what is “true” (Turner, 1991; Turner, Hogg, Oakes, Reicher, & Wetherell, 1987). Conformity likely also is motivated by the importance of belonging and being accepted by close peers (Gerber, Green, & Larimer, 2008, 2010; Noelle-Neuman, 1984; Suhay, 2015). For both reasons, social identification with a group tends to increase attitudinal conformity.3 There is increasing evidence in Political Science that this general finding from Social Psychology is relevant to partisan dynamics. First, political scientists increasingly point to how closely party identification (PID) resembles other strong social identifications,4 such as racial, religious, or regional identifications (Green, Palmquist, & Schickler, 2002; Greene, 1999; Huddy, Mason, & Aarøe, 2015; Iyengar & Westwood, 2015; Mason, 2015). Second, partisan social identification is clearly contributing to other phenomena long associated with social identification with other groups, including group mobilization, attitude polarization, and prejudice (Huddy, Mason, & Aarøe, 2015; Iyengar & Westwood, 2015; Mason, 2015). In short, we build on these efforts, examining the relationship between partisanship as a social identity and conformity to the party’s modal policy views. As described in the next section, we use a measure developed by Huddy, Mason, and Aarøe (2015) to precisely gauge partisan social identification. We predict that those whose partisan social identity is strongest will be most likely to conform to party views: H3.Partisan social identity will moderate the influence of partisan cues. However, as previously discussed, social identity is more likely to motivate adherence to peers’ views than elites’. Thus, we also posit: H4: Partisan social identity will moderate the influence of partisan cues from peers more so than from elites. Methods and Data To test the effects of information about the preferences of peers and elites on partisans’ support for policy issues, we designed a novel experiment and embedded it in a survey of U.S. citizens. In this section, we provide information about the survey, treatment stimuli, and variables used in the analyses that follow. Note that the design of this study was preregistered under the authors’ names at the Open Science Framework. Description of the Sample An online survey was conducted August 16–23, 2016, using a sample (N = 1,024) recruited by Social Sampling International (SSI) from their panel of respondents.5 The sample was balanced using quotas corresponding to age, gender, race, income, geographic region, and partisan affiliation to reflect the overall U.S. public (see Table I below). Table I Overview of Study Sample Study sample U.S. population overall Gender (% female) 55.1 50.8 Age (% over 65 years) 15.3 14.9 Race (% Black) 13.5 13.3 Ethnicity (% Hispanic) 10.6 17.6 Household income (Median) $50,000–$59,999 $53,482 Education (% Bachelor’s or higher) 44.0 29.3 PID (% Democrat) 37.8 32.0 PID (% Republican) 28.2 27.0 Study sample U.S. population overall Gender (% female) 55.1 50.8 Age (% over 65 years) 15.3 14.9 Race (% Black) 13.5 13.3 Ethnicity (% Hispanic) 10.6 17.6 Household income (Median) $50,000–$59,999 $53,482 Education (% Bachelor’s or higher) 44.0 29.3 PID (% Democrat) 37.8 32.0 PID (% Republican) 28.2 27.0 Note. U.S. population estimates are from the 2015 Current Population Survey. For reference, PID was drawn from a Pew Research Center survey conducted August 9–16, 2016. Data are unweighted; recruitment was balanced using quotas. PID = party identification. Table I Overview of Study Sample Study sample U.S. population overall Gender (% female) 55.1 50.8 Age (% over 65 years) 15.3 14.9 Race (% Black) 13.5 13.3 Ethnicity (% Hispanic) 10.6 17.6 Household income (Median) $50,000–$59,999 $53,482 Education (% Bachelor’s or higher) 44.0 29.3 PID (% Democrat) 37.8 32.0 PID (% Republican) 28.2 27.0 Study sample U.S. population overall Gender (% female) 55.1 50.8 Age (% over 65 years) 15.3 14.9 Race (% Black) 13.5 13.3 Ethnicity (% Hispanic) 10.6 17.6 Household income (Median) $50,000–$59,999 $53,482 Education (% Bachelor’s or higher) 44.0 29.3 PID (% Democrat) 37.8 32.0 PID (% Republican) 28.2 27.0 Note. U.S. population estimates are from the 2015 Current Population Survey. For reference, PID was drawn from a Pew Research Center survey conducted August 9–16, 2016. Data are unweighted; recruitment was balanced using quotas. PID = party identification. Respondents were asked to complete a 10-min questionnaire, which measured these and other relevant demographic characteristics and political attitudes.6 PID was measured conventionally (“Generally speaking, do you think of yourself as a Democrat, Republican, Independent, or Other”), excluding respondents who did not identify with either of the major parties. To better assess variation in social identity among Democratic and Republican identifiers, we relied on a three-item battery adapted from Huddy, Mason, and Aarøe (2015). The items were reliable (α = 0.82) and combined by averaging across questions. The resulting measure ranged from 0 to 4 (M = 2.16; SD = 0.96).7 In addition, to gauge partisan affinities more closely tied to ideological differences between the parties, respondents were asked how close they felt to each of the political parties “in terms of ideas and interests” on a scale ranging from “extremely distant” to “extremely close,” an item adapted from the American National Election Studies (ANES). These positive and negative assessments (Abramowitz & Webster, 2016; Medeiros & Noel, 2014) were combined in a composite measure reflecting net attitudes toward the parties and ranged from −3 to +6 (M = 3.60, SD = 2.00). While the conventional PID scale frequently used by political scientists distinguishes only between strong and weak identifiers, these separate measures tap into additional gradations of partisanship: one that assesses partisan affinity as a social identity and the other how partisans view themselves ideologically.8 Experimental Design The last portion of the survey included a randomized experiment in which respondents were presented with information about two policy debates and, afterward, asked to state their own view. The issues were (1) “Common Core” curricular standards, and (2) the TPP trade deal. These issues were selected for two reasons. First, rather than using fictionalized or obscure issues, by measuring conformity on real, contemporary policy debates, we hoped to maintain realism and preserve some external validity in the study. Second, these particular issues were selected because both policies have attracted vocal support and opposition among partisans on each side (see Bradner & Walsh, 2015; Clement, 2015; Whitman, 2015; Williams, 2014). By selecting issues such as these, where partisan cleavages are less obvious than on other matters (e.g., Obamacare), we were able to vary the level of partisan support/opposition plausibly across treatments as well as examine issues where opinions were still in the process of being formed. Finally, note that the order in which the issues were presented was randomized and did not impact results. Past studies have shown that the effects of party cues are reduced when policy information is provided (Boudreau & MacKenzie, 2014; Bullock, 2011), prompting individuals to think more independently about their own preferences. Therefore, we conservatively designed this study, providing respondents with explanations of the policy debates in the form of excerpted news stories with arguments on either side. For each issue, this information remains constant across the experimental and control conditions. Two examples of these excerpts are provided in Figure 1 (see the Supplementary Appendix for the complete set of stimuli). Figure 1 View largeDownload slide Examples of experimental stimuli Figure 1 View largeDownload slide Examples of experimental stimuli For each issue, respondents were randomly assigned to a control group or one of four treatment groups, corresponding to a 2 × 2 design: varying which political party supported the policy (with percentage values held consistent for each issue) and the group referenced in the stimulus—either registered voters (peers) or members of Congress (elites). Both manipulations occurred through phrasing in the final paragraph of the stimulus and an accompanying infographic. Respondents assigned to control groups received similar news excerpts that omitted the final paragraph and, for parallelism, displayed a similar infographic without any survey results. Randomization was checked for both the Common Core experiment ( X2=42.73,p=0.69) and TPP experiment ( X2=42.67,p=0.31) by regressing assigned groups for each issue on demographics in multinomial logit models and testing for joint significance of the covariates (see Supplementary Appendices for full models). After each treatment, respondents were asked, “Do you SUPPORT or OPPOSE the Trans-Pacific Partnership trade agreement [Common Core state standards]?” and asked to respond on a five-point scale from “strongly oppose” to “strongly support.” This item served as the dependent variable in the study. To reduce suspicions about the study’s objectives, respondents were also asked to rate how informative they found the articles. Results Across both issues tested (Common Core and TPP) and across both partisan subgroups, we find evidence of partisan conformity. The analyses below offer support for our hypotheses.9 Treatment Effects Conditional on Partisan Identification Results reveal clear systematic differences in treatment effects in line with our expectation that information about party preferences would induce partisan conformity. Differences were shown to be statistically significant and in the expected direction in most cases regardless of the nature of the stimulus or issue. These findings are emphasized in Figure 2, which plots the first differences between treatment and control groups, pooling across the two issues to simplify reporting. In Supplementary Appendices (Table C-3), we separately report the mean level of support for each issue segmented by treatment group and party. Figure 2 View largeDownload slide Conditional average treatment effects are plotted for each partisan subgroup with 90% confidence intervals, pooling across the two issues Figure 2 View largeDownload slide Conditional average treatment effects are plotted for each partisan subgroup with 90% confidence intervals, pooling across the two issues We hypothesized in H1 that partisans would conform to messages about their party’s positions, and we find that treatment effects were indeed significant in the expected directions for both Democrats and Republicans.10 The lone exceptions (of eight separate tests) occurred among Republicans who were told that fellow partisans were opposed to the proposals.11 The results also lend support to H2, which hypothesized that individuals would be at least as influenced by ordinary partisans’ preferences as by party elites’ preferences. In three of the four cases above, the magnitude of treatment effects on participants was statistically indistinguishable regardless of whether the stimulus referenced members of Congress or registered voters, and, in most cases, slightly larger in the “peers” condition than the “elites” condition.12 Partisan Social Identity as a Moderator of Effects Our third and fourth hypotheses stated that partisan social identity would moderate the effects of partisan cues (H3) and that it would moderate the effects more so when the stimulus referenced peers compared with references to elites (H4). We find evidence for both hypotheses. To examine how partisan social identity influenced conformity, we estimated multivariate linear models of support for the two issues as a function of respondents’ experimental group,13 strength of social identification with their party, and several potentially relevant control variables, including congressional approval ratings.14 The inclusion of control variables improves precision by reducing disturbance variability; however, findings were not substantively different when models were estimated without controls (see Supplementary Appendices). To enable generalization across partisan groups, we grouped participants according to whether their partisan “in-group” supported—or opposed—the issue at hand; for example, Democrats who received a stimulus indicating that fellow Democratic voters supported the issues were grouped with Republicans who received a stimulus indicating fellow Republicans supported the issues. In short, this coding reflects co-partisan support (or opposition) from the perspective of the participant. While results were broadly similar for the two issues when analyzed separately, to improve statistical power, we also pooled across the issues in a multilevel model with random effects for the policy issue. Effects of partisan social identity were specifically tested by examining an interaction between partisan social identity and indicators for the four treatment groups, using the control group as the omitted category (see Table II). We find that the coefficient on the partisan social identity interaction term is statistically significant in two of the four interactions—specifically, only those stimuli in which other partisan voters were referenced in the treatment. Table II Partisan Social Identity as Moderator of Treatment Effects TPP trade deal Common core Pooled issues Partisan social identity 0.09 0.04 0.07 (0.09) (0.10) (0.07) Group 1: Co-Partisan Voters Support −0.13 0.09 −0.02 (0.29) (0.36) (0.23) Group 2: Co-Partisan Voters Oppose 0.41* 0.14 0.27 (0.31) (0.36) (0.24) Group 3: Co-Partisan Elites Support 0.09 0.28 0.16 (0.29) (0.35) (0.23) Group 4: Co-Partisan Elites Oppose −0.16 −0.70** −0.40** (0.28) (0.35) (0.23) Social identity × Group 1 0.19** 0.17 0.17** (0.12) (0.14) (0.09) Social identity × Group 2 −0.27*** −0.15 −0.22*** (0.12) (0.15) (0.10) Social identity × Group 3 0.09 -0.01 0.05 (0.12) (0.14) (0.09) Social identity × Group 4 −0.03 0.21* 0.08 (0.12) (0.14) (0.09) Random effects? No No Yes Observations (N) 667 668 1,335 R2 0.16 0.18 Adjusted R2 0.13 0.16 AIC 1,869 2,146 4,129 TPP trade deal Common core Pooled issues Partisan social identity 0.09 0.04 0.07 (0.09) (0.10) (0.07) Group 1: Co-Partisan Voters Support −0.13 0.09 −0.02 (0.29) (0.36) (0.23) Group 2: Co-Partisan Voters Oppose 0.41* 0.14 0.27 (0.31) (0.36) (0.24) Group 3: Co-Partisan Elites Support 0.09 0.28 0.16 (0.29) (0.35) (0.23) Group 4: Co-Partisan Elites Oppose −0.16 −0.70** −0.40** (0.28) (0.35) (0.23) Social identity × Group 1 0.19** 0.17 0.17** (0.12) (0.14) (0.09) Social identity × Group 2 −0.27*** −0.15 −0.22*** (0.12) (0.15) (0.10) Social identity × Group 3 0.09 -0.01 0.05 (0.12) (0.14) (0.09) Social identity × Group 4 −0.03 0.21* 0.08 (0.12) (0.14) (0.09) Random effects? No No Yes Observations (N) 667 668 1,335 R2 0.16 0.18 Adjusted R2 0.13 0.16 AIC 1,869 2,146 4,129 Note. Linear models with standard errors in parentheses. Random effect added for the issue in the pooled model. Models include controls for age, gender, race, church attendance, education, income, congressional approval, political interest, and an indicator for which party the respondent identified with (nonidentifiers were excluded). Full regression output and alternative specifications are provided in Supplementary Appendices. * p < .10; ** p < .05; *** p < .01 (one-tailed tests). None of the estimates rise to p < .001 significance (****). Table II Partisan Social Identity as Moderator of Treatment Effects TPP trade deal Common core Pooled issues Partisan social identity 0.09 0.04 0.07 (0.09) (0.10) (0.07) Group 1: Co-Partisan Voters Support −0.13 0.09 −0.02 (0.29) (0.36) (0.23) Group 2: Co-Partisan Voters Oppose 0.41* 0.14 0.27 (0.31) (0.36) (0.24) Group 3: Co-Partisan Elites Support 0.09 0.28 0.16 (0.29) (0.35) (0.23) Group 4: Co-Partisan Elites Oppose −0.16 −0.70** −0.40** (0.28) (0.35) (0.23) Social identity × Group 1 0.19** 0.17 0.17** (0.12) (0.14) (0.09) Social identity × Group 2 −0.27*** −0.15 −0.22*** (0.12) (0.15) (0.10) Social identity × Group 3 0.09 -0.01 0.05 (0.12) (0.14) (0.09) Social identity × Group 4 −0.03 0.21* 0.08 (0.12) (0.14) (0.09) Random effects? No No Yes Observations (N) 667 668 1,335 R2 0.16 0.18 Adjusted R2 0.13 0.16 AIC 1,869 2,146 4,129 TPP trade deal Common core Pooled issues Partisan social identity 0.09 0.04 0.07 (0.09) (0.10) (0.07) Group 1: Co-Partisan Voters Support −0.13 0.09 −0.02 (0.29) (0.36) (0.23) Group 2: Co-Partisan Voters Oppose 0.41* 0.14 0.27 (0.31) (0.36) (0.24) Group 3: Co-Partisan Elites Support 0.09 0.28 0.16 (0.29) (0.35) (0.23) Group 4: Co-Partisan Elites Oppose −0.16 −0.70** −0.40** (0.28) (0.35) (0.23) Social identity × Group 1 0.19** 0.17 0.17** (0.12) (0.14) (0.09) Social identity × Group 2 −0.27*** −0.15 −0.22*** (0.12) (0.15) (0.10) Social identity × Group 3 0.09 -0.01 0.05 (0.12) (0.14) (0.09) Social identity × Group 4 −0.03 0.21* 0.08 (0.12) (0.14) (0.09) Random effects? No No Yes Observations (N) 667 668 1,335 R2 0.16 0.18 Adjusted R2 0.13 0.16 AIC 1,869 2,146 4,129 Note. Linear models with standard errors in parentheses. Random effect added for the issue in the pooled model. Models include controls for age, gender, race, church attendance, education, income, congressional approval, political interest, and an indicator for which party the respondent identified with (nonidentifiers were excluded). Full regression output and alternative specifications are provided in Supplementary Appendices. * p < .10; ** p < .05; *** p < .01 (one-tailed tests). None of the estimates rise to p < .001 significance (****). In Figure 3, we illustrate these results, depicting changes in the treatment effect (or first differences when compared with the control group) on the y-axis and levels of partisan social identity on the x-axis, holding all other variables at their mean values.15 The results show that, as predicted by H3, partisan social identity does moderate treatment effects such that those who are higher on the identity scale are considerably more likely to conform to their party’s prevailing preferences. However, in line with H4, the moderating effects of partisan social identity were confined only to the two treatment groups involving the preferences of peer groups. Where respondents were provided with information about elite preferences instead, conformity effects were unrelated to levels of partisan social identity.16 Figure 3 View largeDownload slide Partisan social identity moderates conformity, but only when messages refer to voter preferences. First differences are plotted for pooled linear mixed model, varying treatment/control group indicators and respondents' partisan social identity and holding other variables at their means. Shaded areas depict 90% confidence intervals derived through simulations. Full results are detailed in supplementary appendices Figure 3 View largeDownload slide Partisan social identity moderates conformity, but only when messages refer to voter preferences. First differences are plotted for pooled linear mixed model, varying treatment/control group indicators and respondents' partisan social identity and holding other variables at their means. Shaded areas depict 90% confidence intervals derived through simulations. Full results are detailed in supplementary appendices Anticipating a counterargument to the conclusions we draw from the analyses above, we conduct a final test, examining whether the same dynamics are evident when variation in partisanship is measured using net closeness toward both parties “in terms of ideas and interests.” If such a measure—which gauges ideological closeness to the parties—reveals the same pattern of moderation, then we cannot conclude that partisan social identity uniquely drives conformity to partisan peers. In contrast to Figure 3, we find that this alternative measure does not moderate the effect of peer cues (see Supplementary Appendices, Figure D-3). In the first instance (co-partisans support), the effects of peer cues are constant across the range of the measure; in the second instance (co-partisans oppose), effects trend in the “right” direction, but the difference is not statistically significant. This alternate specification provides further evidence in support of our general framework that the persuasive power of partisan peers (as opposed to elites) rests uniquely in citizens’ social identification with a political party. Discussion and Conclusion This article examines the nature of partisanship and how it shapes citizens’ expressed views on contemporary policy debates. We demonstrated using a randomized experiment embedded in a survey of Americans how susceptible those who identify with a party may be to impulses to align their own positions, so that they match the preferences of their partisan group. Our findings also show something previously unknown about how partisans in the United States respond to messages about party preferences: while information about the preferences of party elites can induce considerable opinion change, polls about the preferences of Democrats and Republicans in the mass public are at least as likely to influence the positions people take on policies. The shifts are also considerable in size, as much as half a point on a scale from −2 to 2. Further, we demonstrate that such effects are strongest among those whose association with their party is most deeply felt as a social identity. These additional findings bolster our argument that partisanship may convey more than informational heuristics. We find that (1) these effects are not replicated when we replace the identity measure with a measure of ideological closeness to the parties, and (2) moderation by social identity occurs in response to mass but not elite partisan cues. These findings align with research on identity-driven conformity generally, which argues that individuals will alter their views in response to the majority of relatively co-equal group members. It is (perceived) co-equals who make up the “group” to which the individual belongs, and it is co-equals who form the affiliative bonds that are theorized as motivating social conformity. This body of theory stands in contrast to frameworks discussing the power of elites. Citizens tend to adhere to elites’ views for other reasons, primarily elites’ greater access to information, as well as power. In our study, the irrelevance of social identity to participants’ responsiveness to elite cues lends support to this perspective as well. These findings have several implications for the empirical study of politics. At the aggregate level, conformity to partisan peers adds another explanation for the correlation between partisanship and policy views, despite citizens’ relative inattentiveness to policy debates. In terms of change over time, partisan conformity can help to explain campaign “momentum”—with support within a party increasing when a preference reaches majority support—as well as the relatively slow rate at which mass opinion within partisan (and other) groups tends to change. As Huddy and colleagues (Huddy, Mason, & Aarøe, 2015) show, a social identity orientation toward partisanship especially predicts “expressive” participation in political life, which may further contribute to a cycle of partisan conformity: the same individuals most susceptible to conformity effects may also be the most likely to express these views publicly, reinforcing perceptions of consensus among partisans (Noelle-Neuman, 1984). Partisan conformity has important normative implications as well. First and foremost, the tendency to adhere to group norms may prevent individual citizens from considering whether specific policy proposals offered by their party align with their values and interests. In addition, by increasing ideological coherence within the parties, partisan conformity also contributes to differences between the parties (i.e., partisan polarization) and all the difficulties of democratic governance that follow. The normative consequences of partisan conformity are not entirely negative, however. For example, conformity effects make it more likely that party members will agree on an issue agenda that, realistically, must (at least in a two-party system) integrate a variety of positions that are only loosely connected, and sometimes even contradict one another (see Converse, 1964; Karol, 2009). Finally, conformity also brings a certain amount of stability to mass politics—stability which is often taken for granted in developed democracies. Although the effects observed in our study appear robust, this line of research remains relatively new. We hope future studies will seek to replicate our results and pursue related questions our study could not answer. For example, given increasing affective and social polarization between Democrats and Republicans in the United States (Iyengar, Sood, & Lelkes, 2012; Mason, 2015), could it be that the kind of peer-to-peer conformity effects we observed are larger or more consistent today than in the past? Even if strong conformity effects are observed among partisans, how quickly do they decay if not reinforced (see Chong & Druckman, 2013)? Finally, we investigated issues where we expected opinions to be less crystallized compared to issues where divides were well worn (e.g., Obamacare). To what extent do effects vary according to issue-specific ambivalence? In sum, while most scholarship suggests that partisan conformity occurs top-down as an elite-driven process with individuals relying on leaders’ positions as an informational heuristic, our findings indicate that this explanation is incomplete. Partisans are just as likely to be influenced horizontally by polls about where their co-partisan peers stand, and an orientation to partisanship as a social identity is positively and uniquely related to these group conformity effects. These findings may bring comfort to those who worry that conventional theories of party cueing place too much emphasis on the influence of elites, but they may also raise new concerns about the coercive influence of group identity in structuring political attitudes. Supplementary Data Supplementary Data are available at IJPOR online. Conflicts of interest: None declared Benjamin Toff is an Assistant Professor in the Hubbard School of Journalism and Mass Communication at the University of Minnesota. School of Journalism and Mass Communications at the University of Minnesota. He holds a PhD in Political Science from the University of Wisconsin-Madison. Elizabeth Suhay is an Assistant Professor in the Department of Government at American University. She holds a PhD in Political Science from the University of Michigan. Footnotes 1Psychologists tend to use the term “social conformity” to describe the phenomenon whereby individuals adopt group opinion norms as their own simply because they are the norm. Common synonyms include “peer pressure,” “group pressure,” and “majority influence.” 2Psychologists tend to classify adherence to authority figures as a distinct phenomenon: trust-driven obedience, not identity-driven conformity (see Turner, 1991). 3We follow Turner (1991) in making a distinction between “conformity” and “compliance.” Whereas the former represents authentic, internal belief, or attitude change, the latter represents mere surface change, usually to protect one’s reputation or otherwise advance one’s interests. 4We define social (or group) identification as most contemporary scholars of political behavior do. Social identity reflects both a self-categorization (incorporating the group into one’s personal identity) and a feeling of emotional closeness—an emotional bond—with a group (Green, Palmquist, and Schickler, 2002; Huddy, Mason, and Aarøe, 2015) 5The full sample (N = 1,024) omits respondents who did not consent to participate or wished to withdraw their responses following a debriefing (N = 252). Randomization tests showed these exclusions did not affect the balance between treatment groups. 6To avoid posttreatment bias with respect to potential moderator variables, these items were placed in the first half of the survey, before the embedded experiment. 7The questions were: (1) How important is being a Republican [Democrat] to you?; (2) How well does the term Republican [Democrat] describe you?; (3) When talking about Republicans [Democrats], how often do you use “we” instead of “they?” 8These new measures were correlated (r = .46), but they capture distinct dimensions of partisan affinity. A considerable number of Democrats (14%) and Republicans (11%) expressed neutral or net negative feelings toward their own party in terms of “ideas and interests”; yet, these individuals’ partisan social identity levels were not significantly different on average from the rest of the respondents in the sample. 9Owing to concerns about posttreatment biases arising from excluding respondents, we do not omit respondents who failed an attention check or appeared to skip reading the news excerpts (N = 155 for Common Core; N = 162 for TPP). However, supplementary analyses excluding these individuals do not substantively change results. 10The issues are broken out separately in the Supplementary Appendix. 11We suspect this null finding may be because of a failure of the treatment to shift perceptions in this case, as Republicans are in fact somewhat more opposed than Democrats to both issues. 12Democrats in the co-partisan support and oppose conditions were 0.08 more supportive ( t=0.65,p=0.52) and 0.18 more opposed ( t=1.45,p=0.15) when the treatment group referenced co-partisan peers instead of elites. For Republicans, this dynamic held only with respect to treatments indicating co-partisan support for the issues: a 0.18 relative increase in support ( t=0.34,p=0.74) for peers rather than elites. Republicans in the “co-partisans oppose” conditions were 0.27 more responsive to elite cues than peer cues (t = 1.75, p = .08); however, neither of these conditional treatment groups were significantly different from Republicans in the control group. 13Because respondents were randomly assigned to one of the five groups, observations are independent from one another, and it is appropriate to combine them in the same model. 14In alternate specifications, we estimated a three-way interaction between this variable and the partisan social identity variable, but it did not substantively change the findings. 15Figures were produced in R using estimates obtained using the “merTools” package. 16Owing to concerns that linearity assumptions may be violated in multiplicative interactions (see Hainmueller, Mummolo, and Xu, 2017), we performed an additional diagnostic test by binning the social identity variable in three quantiles and separately estimating the model on subsets of the data. These results are provided in Figure D-2 in the Supplementary Appendices. References Abramowitz A. I. , Webster S. ( 2016 ). The rise of negative partisanship and the nationalization of U.S. elections in the 21st century . Electoral Studies , 41 , 12 – 22 . Google Scholar CrossRef Search ADS Achen C. H. , Bartels L. M. ( 2016 ). Democracy for realists: why elections do not produce responsive government . Princeton, NJ : Princeton University Press . Google Scholar CrossRef Search ADS Asch S. E. ( 1940 ). Studies in the principles of judgements and attitudes: II. Determination of judgements by group and by ego standards . Journal of Social Psychology , 12 , 433 – 465 . Google Scholar CrossRef Search ADS Asch S. E. ( 1951 ). Effects of group pressure upon the modification and distortion of judgments. In Guetzkow H. (Ed.), Groups, leadership, and men: Research in human relations (pp. 177 – 190 ). Pittsburgh, PA : Carnegie Press . Bartels L. M. ( 2002 ). Beyond the running tally: Partisan bias in political perceptions . Political Behavior , 24 , 117 – 150 . Google Scholar CrossRef Search ADS Bartels L. M. ( 2008 ). Unequal democracy: The political economy of the new gilded age . New York, NY : Russell Sage Foundation . Berelson B. R. , Lazarsfeld P. F. , McPhee W. N. ( 1954 ). Voting: A study of opinion formation in a presidential campaign . Chicago : University of Chicago Press . Bond R. M. , Fariss C. J. , Jones J. J. , Kramer A. D. , Marlow C. , Settle J. E. , Fowler J. H. ( 2012 ). A 61-million-person experiment in social influence and political mobilization . Nature , 489 , 295 – 298 . Google Scholar CrossRef Search ADS PubMed Boudreau C. , MacKenzie S. A. ( 2014 ). Informing the electorate: How party cues and policy information affect public opinion about initiatives . American Journal of Political Science , 58 , 48 – 62 . Google Scholar CrossRef Search ADS Bradner E. , Walsh D. ( 2015, June 23 ). Democrats reject Obama on trade. CNN.com. Retrieved from http://edition.cnn.com/2015/06/12/politics/white-house-tpp-trade-deal-congress/ Bullock J. G. ( 2011 ). Elite influence on public opinion in an informed electorate . American Political Science Review , 105 , 496 – 515 . Google Scholar CrossRef Search ADS Bullock J. G. , Gerber A. S. , Hill S. J. , Huber G. A. ( 2015 ). Partisan bias in factual beliefs about politics . Quarterly Journal of Political Science , 10 , 519 – 578 . Google Scholar CrossRef Search ADS Campbell A. , Converse P. E. , Miller W. E. , Stokes D. E. ( 1960 ). The American voter . New York, NY : John Wiley & Sons . Chong D. , Druckman J. N. ( 2013 ). Counterframing effects . Journal of Politics , 75 ( 1 ), 1 – 16 . Google Scholar CrossRef Search ADS Clement S. ( 2015, January 28 ). Conservatives hate common core. The rest of America? Who knows. Washington Post. Retrieved from http://www.washingtonpost.com/news/the-fix/wp/2015/01/28/conservatives-hate-common-core-the-rest-of-america-who-knows/ Cohen G. ( 2003 ). Party over policy: The dominating impact of group influence on political beliefs . Journal of Personality and Social Psychology , 85 , 808 – 822 . Google Scholar CrossRef Search ADS PubMed Cohen M. , Karol D. , Noel H. , Zaller J. ( 2008 ). The party decides: Presidential nominations before and after reform . Chicago : University of Chicago Press . Google Scholar CrossRef Search ADS Converse P. E. ( 1964 ). The nature of belief systems in mass publics. In Apter D. E. (Ed.), Ideology and discontent (pp. 206 – 261 ). New York, NY : Free Press . Druckman J. N. ( 2001 ). Using credible advice to overcome framing effects . Journal of Law, Economics, and Organization , 17 , 62 – 82 . Google Scholar CrossRef Search ADS Festinger L. ( 1950 ). Informal social communication . Psychological Review , 57 , 271 – 282 . Google Scholar CrossRef Search ADS PubMed Gaines B. J. , Kuklinski J. H. , Quirk P. J. , Peyton B. , Verkuilen J. ( 2007 ). Same facts, different interpretations: Partisan motivation and opinion on Iraq . Journal of Politics , 69 , 957 – 974 . Google Scholar CrossRef Search ADS Gerber A. S. , Green D. P. , Larimer C. W. ( 2008 ). Social pressure and vote turnout: Evidence from a large-scale field experiment . American Political Science Review , 102 , 19 – 31 . Google Scholar CrossRef Search ADS Gerber A. S. , Green D. P. , Larimer C. W. ( 2010 ). An experiment testing the relative effectiveness of encouraging voter participation by inducing feelings of pride or shame . Political Behavior , 32 , 409 – 422 . Google Scholar CrossRef Search ADS Gerber A. S. , Rogers T. ( 2009 ). Descriptive social norms and motivation to vote: Everybody’s voting and so should you . The Journal of Politics , 71 , 178 – 91 . Google Scholar CrossRef Search ADS Green D. , Palmquist B. , Schickler E. ( 2002 ). Partisan hearts and minds: Political parties and the social identities of voters . New Haven : Yale University Press . Greene S. ( 1999 ). Understanding party identification: A social identity approach . Political Psychology , 20 , 393 – 403 . Google Scholar CrossRef Search ADS Groenendyk E. ( 2015 ). Competing motives in the partisan mind: How loyalty and responsiveness shape party identification and democracy . New York, NY : Oxford University . Haenschen K. ( 2016 ). Social pressure on social media: Using Facebook status updates to increase voter turnout . Journal of Communication , 66 , 542 – 563 . Google Scholar CrossRef Search ADS Hainmueller J. , Mummolo J. , Xu Y. ( 2017 ). How much should we trust estimates from multiplicative interaction models? Simple tools to improve empirical practice. (Working Paper). Retrieved from https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2739221 Hardmeier S. ( 2008 ). The effects of published polls on citizens. In Donsbach W. , Traugott M. W. (Eds.), The SAGE Handbook of Public Opinion Research (pp. 504 – 515 ). London : SAGE Publications . Google Scholar CrossRef Search ADS Huckfeldt R. , Sprague J. ( 1995 ). Citizens, politics and social communication: Information and influence in an election campaign . New York, NY : Cambridge University Press . Google Scholar CrossRef Search ADS Huddy L , Mason L. , Aarøe L. ( 2015 ). Expressive partisanship: Campaign involvement, political emotion, and partisan identity . American Political Science Review , 109 ( 1 ), 1 – 17 . Google Scholar CrossRef Search ADS Iyengar S. , Sood G. , Lelkes Y. ( 2012 ). Affect, not ideology: A social identity perspective on polarization . Public Opinion Quarterly , 76 , 405 – 431 . Google Scholar CrossRef Search ADS Iyengar S. , Westwood S. J. ( 2015 ). Fear and loathing across party lines: New evidence of group polarization . American Journal of Political Science , 59 , 690 – 707 . Google Scholar CrossRef Search ADS Kam C. D. ( 2005 ). Who toes the party line? Cues, values, and individual differences . Political Behavior , 27 , 163 – 182 . Google Scholar CrossRef Search ADS Karol D. ( 2009 ). Party position change in American politics: Coalition management . Cambridge, UK : Cambridge University Press . Google Scholar CrossRef Search ADS Katz E. , Lazarsfeld P. F. ( 2006  ). Personal influence: The part played by people in the flow of mass communications . New Brunswick, NJ : Transaction Publishers . Klar S. ( 2014 ). Partisanship in a social setting . American Journal of Political Science , 58 , 687 – 704 . Google Scholar CrossRef Search ADS Kuklinski J. H. , Hurley N. L. ( 1994 ). On hearing and interpreting political messages: A cautionary tale of citizen cue-taking . The Journal of Politics , 56 , 729 – 751 . Google Scholar CrossRef Search ADS Lenz G. S. ( 2012 ). Follow the leader? How voters respond to politicians’ policies and performance . Chicago : University of Chicago Press . Google Scholar CrossRef Search ADS Levendusky M. ( 2009 ). The Partisan Sort: How Liberals Became Democrats and Conservatives Became Republicans . Chicago : University of Chicago Press . Google Scholar CrossRef Search ADS Levitan L. C. , Verhulst B. ( 2016 ). Conformity in groups: The effects of others’ views on expressed attitudes and attitude change . Political Behavior , 38 , 277 – 315 . Google Scholar CrossRef Search ADS Lupia A. ( 1994 ). Shortcuts versus encyclopedias: Information and voting behavior in California insurance reform elections . The American Political Science Review , 88 , 63 – 76 . Google Scholar CrossRef Search ADS Lupia A. , McCubbins M. D. ( 1998 ). The democratic dilemma . New York, NY : Cambridge University Press . Marsh C. ( 1985 ). Back on the bandwagon: The effect of opinion polls on public opinion . British Journal of Political Science , 15 , 51 – 74 . Google Scholar CrossRef Search ADS Mason L. ( 2015 ). ‘ I disrespectfully agree’: The differential effects of partisan sorting on social and issue polarization . American Journal of Political Science , 59 , 128 – 145 . Google Scholar CrossRef Search ADS McGrath M. C. ( 2016 ). Economic behavior and the partisan perceptual screen . Quarterly Journal of Political Science , 11 , 363 – 383 . Google Scholar CrossRef Search ADS Medeiros M. , Noel A. ( 2014 ). The forgotten side of partisanship: Negative party identification in four Anglo-American democracies . Comparative Political Studies , 47 , 1022 – 1046 . Google Scholar CrossRef Search ADS Mutz D. C. ( 1998 ). Impersonal influence: How perceptions of mass collectives affect political attitudes . Cambridge, UK : Cambridge University Press . Google Scholar CrossRef Search ADS Newcomb T. ( 1963 ). Persistence and regression of changed attitudes: Long-range studies . Journal of Social Issues , 19 , 3 – 14 . Google Scholar CrossRef Search ADS Noel H. ( 2014 ). Political ideologies and political parties in America . New York, NY : Cambridge . Noelle-Neuman E. ( 1984 ). The spiral of silence . Chicago : University of Chicago Press . Page B. I. , Shapiro R. Y. ( 2012 ). The rational public: Fifty years of trends in Americans’ policy preferences . Chicago : University of Chicago Press . Popkin S. L. ( 1994 ). The reasoning voter: Communication and persuasion in presidential campaigns . Chicago : University of Chicago Press . Prior M , Sood G. , Khanna K. ( 2015 ). You cannot be serious: The impact of accuracy incentives on partisan bias in reports of economic perceptions . Quarterly Journal of Political Science , 10 , 489 – 518 . Google Scholar CrossRef Search ADS Rahn W. M. ( 1993 ). The role of partisan stereotypes in information processing about political candidates. American Journal of Political Science , 37 , 472 – 496 . Google Scholar CrossRef Search ADS Rothschild D. , Malhotra N. ( 2014 ). Are public opinion polls self-fulfilling prophecies? Research and Politics , 1 ( 2 ): 1 – 10 . Google Scholar CrossRef Search ADS Schaffner B. F. , Streb M. J. ( 2002 ). The partisan heuristic in low-information elections . Public Opinion Quarterly , 66 , 559 – 581 . Google Scholar CrossRef Search ADS Sherif M. ( 1966 ). The psychology of social norms . New York, NY : Harper & Row . Sinclair B. ( 2012 ). The social citizen: Peer networks and political behavior . Chicago : University of Chicago Press . Google Scholar CrossRef Search ADS Suhay E. ( 2015 ). Explaining group influence: The role of identity and emotion in political conformity and polarization . Political Behavior , 37 , 221 – 251 . Google Scholar CrossRef Search ADS Turner J. C. ( 1991 ). Social influence . Pacific Grove, CA : Brooks/Cole . Turner J. C. , Hogg M. A. , Oakes P. J. , Reicher S. D. , Wetherell M. S. ( 1987 ). Rediscovering the social group: A self-categorization theory . Oxford : Basil Blackwell . Whitman D. ( 2015 ). The surprising roots of the common core: How conservatives gave rise to ‘Obamacore’. The Brookings Institution: The Brown Center on Education Policy. Retrieved from https://www.brookings.edu/wp-content/uploads/2016/06/Surprising-Conservative-Roots-of-the-Common-Core_FINAL.pdf Williams J. P. ( 2014, February 27 ). Who is fighting against common core? The push against Common Core is coming from both sides of the political aisle. U.S. News. Retrieved from http://www.usnews.com/news/special-reports/a-guide-to-common-core/articles/2014/02/27/who-is-fighting-against-common-core Zaller J. R. ( 1992 ). The nature and origins of mass opinion . New York, NY : Cambridge . Google Scholar CrossRef Search ADS © The Author(s) 2018. Published by Oxford University Press on behalf of The World Association for Public Opinion Research. All rights reserved. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)
International Journal of Public Opinion Research – Oxford University Press
Published: May 29, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera