Social conformity and prejudice toward immigrants: the role of political messaging

ABSTRACT Can targeted messages from political elites impact the relationship between valuing conformity and prejudice? We answer this question in a survey experiment on a national probability sample from the Netherlands by exposing respondents to a favorable vs. unfavorable statement about East European immigrants. We find that individuals attaching relatively high importance to the value of conformity express greater agreement with the statement, independently of its content or partisanship. Further, the positive association between valuing conformity and prejudice is significantly decreased when individuals are presented with a favorable statement about East European immigrants. Thus, valuing conformity is not only associated with higher prejudice but can also reduce prejudice because it makes individuals susceptible to elite normative pressure to respond positively.

importance to conformity being more prejudiced toward these groups (Crandall et al., 2002;Duckitt et al., 2002;Thomsen et al., 2008). At the same time, individuals who value conformity are more susceptible to normative influences from peers and authority figures, and therefore more readily accept political messages (Levitan & Verhulst, 2016;Di Palma & McClosky, 1970;Suhay, 2015). Paradoxically, this can mean that the importance attached to conformity does not only predisposes people to respond negatively to minority outgroups but also makes them susceptible to normative pressures to respond positively to these outgroups. As a result, the default positive relationship between attaching higher importance to the value of conformity and higher prejudice against minorities may be substantially decreased when elite messaging is favorable toward the minority group in question. Favorable messaging might lead to lower prejudice, especially among individuals who attach high importance to the value of social conformity and thus are more willing to change their attitudes in order to adhere to expressed social norms.

Social conformity
Historically, obedience to authorities, moral traditionalism, adherence to social conventions and punitiveness toward deviants have been considered among the key aspects of the authoritarian personality (Adorno et al., 1950;Altemeyer, 1981). However, recent research shows that different facets of authoritarianism are theoretically and empirically distinct (Duckitt et al., 2010;Passini, 2017). According to Feldman (2003), the desire for social conformity is the defining feature of the broader authoritarian orientation, with moral traditionalism and hostility toward norm-breakers deriving from this predisposition (see also Stenner, 2005). It has been demonstrated that valuing conformity is the only dimension of authoritarianism that is positively associated with general deference to authority whereas, for instance, moral traditionalism is not (Mallinas et al., 2020). Following these theoretical and empirical insights, we study social conformity as an individual difference variable in its own right, rather than as a manifestation of a broader authoritarian personality. We consider conformity as a social value that emphasizes the importance of uniformity and adhering to social norms that can be independent of obedience as the readiness to submit to authorities. Thus, we assess individual differences in valuing social conformity rather than obedience to authorities.

Conformity, prejudice and political messaging
Views on the relationship between conformity values and prejudice have evolved over time. Initially, conformity was thought to be uniformly and positively associated with prejudice (Allport, 1954), and valuing conformity was seen a predictor of negative attitudes toward dissenting minority groups and immigrants (Sniderman & Hagendoorn, 2007;Sniderman et al., 2000). Recent evidence, however, strongly suggests that this relationship also depends on the content of the social norms. Importantly, when expression of outgroup prejudice is perceived as a violation of positive social norms, individuals who attach high importance to conformity tend to suppress the expression of prejudice (Bilewicz et al., 2017;Oyamot et al., 2017;Plant & Devine, 1998;Roets et al., 2015). This finding suggests that elites' leverage in how likely members of the public are to openly demonstrate prejudices should be particularly high among those who value conformity. Existing research demonstrates that political speech can function as social norm signaling. For example, prejudiced statements coming from political elites are interpreted by citizens as a 'permission' to express prejudice toward ethnic minorities (Crandall et al., 2018;Newman et al., 2021), or women (Georgeac et al., 2018). Exposure to pro-immigrant messages perceived as sanctioned by the system, in turn, can reduce the expression of prejudice (Gaucher et al., 2018). Therefore, individuals who attach higher importance to the value of social conformity should also be more strongly affected by political speech, even if the message is favorable toward 'deviant' minority groups that face prejudice, such as immigrants. As a consequence, exposure to positive outgroup elite messages should reduce the association of valuing conformity with outgroup prejudice.

Message content
What kinds of messages are particularly effective in changing the expressed attitudes of those who put high value on conformity? We suggest that messages describing the target outgroup as beneficial for society rather than threatening is particularly relevant for moderating the relationship between valuing conformity and group-specific prejudice. Immigrant groups that are seen as contributing to society are evaluated more positively (e.g., Thravalou et al., 2021), and it has been shown that normative threats, either real or perceived, play a central role in translating the individual predisposition toward social conformity into outgroup negativity (Stenner, 2005). Therefore, messages describing the target outgroup as benign (vs. threatening) should be particularly efficient in influencing the relationship between valuing conformity and group-specific prejudice.
The model presented in Figure 1 starts from two assumptions. First, valuing conformity with its emphasis on uniformity and consensus is positively related to antiimmigrant prejudice. Second, valuing conformity is also positively associated with acceptance of elites' messages as the critical mediating process is higher susceptibility to social influence. The novelty of our model concerns the relation between acceptance of elite messages and prejudice with the corresponding indirect effect of valuing conformity on prejudice. Specifically, acceptance of a positive elite message about immigrants is expected to be associated with lower prejudice, whereas acceptance of a negative message should be associated with higher prejudice. As a result, and in addition to a positive direct effect, valuing conformity can produce either a positive or negative indirect effect on prejudice depending on the message content. Thus, we expect that political elites have an impact on the expressed prejudice toward a specific immigrant group among individuals who attach high importance to the value of social conformity by portraying that group in either positive (benign) or negative (threatening) light. We test the proposed model using a survey experiment among a national sample in the Netherlands in which we manipulate the content of the political message about a specific immigrant group. In addition, we also manipulate the source of the political message to be able to account for the potential impact of partisanship, which we explain next.

Normative influence and partisanship
The psychological processes linking valuing conformity, outgroup prejudice, and political messaging described above are distinct from the 'party over policy' effect (Cohen, 2003). Partisan identity is an important variable in political communication as voters are more likely to adopt policy positions supported by their parties, not only in two-party democracies such as the United States and the United Kingdom but also in multiparty democracies such as the Netherlands (Verkuyten & Maliepaard, 2013). Our argument, however, follows a more general approach to effectiveness of political communication that emphasizes overall credibility of elite messaging (Druckman, 2001), with co-partisanship being one possible way to achieve such credibility. We suggest that respondents who attach high importance to the value of social conformity are influenced by normative messages from figures of political authority, relatively independently of party affiliation. Thus, the proposed underlying mechanism is susceptibility to normative influence rather than partisanship, although the latter is likely to strengthen the effect. Therefore, we also manipulated the ingroup versus outgroup source of the political message to account for the potential impact of partisanship.

Hypotheses
The theoretical discussion above leads us to formulate several testable hypotheses (also see Figure 1). First, overall levels of prejudice should be lower in the favorable message condition than in the unfavorable message condition (H1). Second, there should be a positive direct effect of conformity values on prejudice, independently of condition (H2). Third, there should be a positive effect of conformity values on message acceptance, independently of condition (H3). Fourth, there should be a negative effect of message acceptance on prejudice in the favorable message condition and a positive effect of message acceptance on prejudice in the unfavorable message condition (H4). Finally, there should be a negative indirect effect of conformity values on prejudice in the favorable message condition and a positive indirect effect of conformity values of prejudice in the unfavorable message condition (H5).

Data and method
A national sample of ethnic Dutch adults was recruited using a probability sample from GfK, a market-research company that maintains a large respondent panel representative of the Dutch population. Panel participants have previously agreed to participate in a panel. They regularly receive invitations from the company to participate anonymously in a survey and receive credits from the company for doing so. The total of 1,675 panel members were contacted and 867 completed the questionnaire, yielding the response rate of 52% that is common for online surveys in the Netherlands (Stoop, 2005). Our experiment was embedded into a larger survey of the Dutch public on the topics of immigration and diversity. 1 National surveys are rather costly and typically involve multiple researchers with their own interests and questions. Thus, each researcher is limited in the number of questions that can be asked. In the current research we were able to administer our survey experiment randomly to half of the sample, so our analysis included 435 respondents which provides adequate statistical power. Sample age ranged between 18 and 89 years and was similar to the national average of the adult population with the median respondent being around 50 years of age. The gender ratio was 50.1% male to 49.9% female. Finally, 28.7% of respondents had low levels of education, 41.8% were moderately educated, and 29.4% were highly educated.
All measures that were developed for our experiment are included and used. 2 Respondents were first presented with a set of four items (7-point Likert scales) intended to measure the importance attached to the value of conformity. These items based on previous research in the Netherlands (Sniderman & Hagendoorn, 2007), and Italy (Sniderman et al., 2000), and amended so that none of the items explicitly mentioned immigrants. Two sample items are 'People must abide by the usual norms and rules,' and 'Customs and traditions must be preserved and cherished rather than amended and changed' (α = .77).
Then, respondents were asked about their partisanship and this information was used to define the in-party and the out-party for the experimental conditions. The in-party was the one that the respondent self-reported as being closest to. The out-party was defined as the mainstream party most disliked by the people who shared the respondent's partisan identity according to the 2010 Dutch Parliamentary Election Study.
In the experimental part of the study that used a 2-by-2 factorial design, respondents were randomly exposed to a statement that came from either an in-party or an out-party leader, and contained either a favorable or an unfavorable message about East Europeans. We focused on East Europeans an important immigrant group in the Netherlands because it was possible to create both favorable and unfavorable statements about this group, plausibly coming from both left-wing and right-wing parties. In choosing the statements' content, we followed the actual rhetoric that Dutch and other West European politicians employed to describe East European immigrants. However, the statements were not literally taken from political speeches and the participants were fully debriefed about the goal and design of the experiment.
In all conditions, respondents were asked a single direct question (7-point scale) about their degree of agreement with the statement to measure message acceptance as the mediating variable. On the screen immediately following the one with the leader's statement, respondents were asked about their general feelings (−4 = Very cold/ Negative, +4 = Very warm/Positive) toward immigrants from Eastern Europe and toward non-Western immigrants (a term used in the Netherlands to describe people from outside the European Union). The latter attitude was measured to check whether the message impacted only prejudice against the mentioned immigrant group rather than against immigrants more broadly. For the wordings of the experimental manipulation and all survey items included in the analysis, the list of parties, and descriptive statistics see respectively Tables A1, A2, and A3 in Online Appendix.

Results
We first run one-way ANOVAs to compare the mean scores for message acceptance and prejudice toward East Europeans across favorable vs. unfavorable message content. Next, we examine whether message acceptance mediates the relationship between valuing conformity and prejudice toward East Europeans. We use multigroup structural equation modeling (SEM) in which the associations are estimated across the favorable vs. unfavorable message conditions. Valuing conformity is included in the model as a latent construct with its four indicators (items). Multigroup SEM estimates all effects of interest simultaneously, constrains the measurement model to be equivalent for the two experimental conditions and provides information on whether the model fits the data. This analytic strategy allows us to estimate the expected effects presented in Figure 1 while taking measurement error of the specific indicators of conformity values into account (Kline, 2015).
The core findings are presented in Figure 2 using standardized effects. Overall, the model demonstrates an adequate fit according to multiple indices, χ 2 (22) = 51.8, p < .001. RMSEA = 0.079, CFI = 0.951, SRMR = 0.053. As expected, attaching more importance to the value of conformity is positively associated with prejudice toward East Europeans and with message acceptance (agreement) in both the favorable and unfavorable message conditions. Importantly, the estimated relationships between conformity and message acceptance do not significantly differ between the two message content conditions (Δβ = 0.04, p = .674). 3 In order to check whether the effect of conformity on message acceptance depends on the message source, we compared these effects by in-party vs. outparty condition and did not find a significant difference (Δβ = 0.03, p = .683).
The relationships between message acceptance and prejudice, in turn, are significantly different across the two message content conditions (Δβ = 0.38, p < .001). Agreement with a favorable message is associated with lower prejudice toward East European immigrants, whereas agreement with an unfavorable message is associated with higher prejudice toward this group. This difference leads to divergent indirect effects of the importance attached to conformity on prejudice across the two conditions. Estimated indirect and net (total) effects from the mediation model are presented in Table 1. 4 These indicate significant indirect effects of conformity on prejudice toward East Europeans in both conditions: a negative indirect effect in the favorable message condition and a positive indirect effect in the unfavorable message condition. 5 The indirect effects are smaller in magnitude than the corresponding direct effects, and the resulting net effects are positive in both experimental conditions. However, the estimated net effects of conformity on prejudice toward East Europeans across conditions differ in size by the  Table A4 in Online Appendix for the full results (unstandardized).
factor of almost two and the estimated difference between them is significant at the 95% confidence level. In other words, even though elite messaging does not change the direction of the positive relation between the importance that people attach to the value of conformity and anti-immigrant prejudice, it significantly affects its size.

Discussion
The goal of this research has been to understand whether political elites have influence over the degree to which the importance that people attach to the value of conformity is translated into prejudice toward immigrant minorities. Using the top-down model of political communication (Zaller, 1992), we have hypothesized that people who value conformity are susceptible to elite normative influences that can produce indirect effects on prejudice depending on message content. Specifically, we have tested the prediction that a favorable message about a specific immigrant group from a figure of political prominence reduces the net association between the importance attached to conformity and higher prejudice toward this specific target group. Altogether, our findings support this expectation. Specifically, participants who attached high importance to conformity as a social value were more likely to accept the favorable message about immigrants from Eastern Europe that reduced their expressed prejudice toward that target group but not toward other immigrant groups. This latter finding indicates that politicians can be influential by singling out specific immigrant or minority groups in either positive or negative ways. People often have differential attitudes toward different outgroups (e.g., Axt et al., 2014), and politicians can play a role in signaling which group differences should be considered meaningful. The indirect effects of conformity on prejudice were relatively small but consequential with the corresponding net effect being significantly smaller in the favorable message condition.
This effect did not depend on the source of the message. Rather it was found independently of whether the message came from an in-party or an out-party sourcein contrast with the 'party-over-policy' effect that was found in the United States (Cohen, 2003;Georgeac et al., 2018), and also in the Netherlands (Verkuyten & Maliepaard, Estimates based on the mediation model presented in Figure 2. Standardized effects presented. Tests based on unstandardized effects.

2013
). This suggests that the psychological mechanism behind our finding is one of being susceptible to elite normative influence rather than partisan identity. However, it remains to be seen in how far this finding is specific to the fractionalized and relatively volatile Dutch party system (Aarts & Thomassen, 2008;Mair, 2008;Van der Meer et al., 2012). The presence of multiple parties with often blurred distinctions makes it relatively easy for Dutch voters to switch their votes across elections. The role of partisanship may be more important as an identity marker in polarized political contexts structured around two main parties, such as the United States and the United Kingdom (Finkel et al., 2020). In these countries political influence can depend more on party identity and less on normative influence. It is also important to note that our experimental manipulations were relatively weak: Participants simply read a short text in an online questionnaire. Nevertheless, the fact that a weak manipulation still affects the relationship between conformity and prejudice suggests that even relatively mild communications from the political elites can influence the degree to which individuals who attach high importance to conformity as a social value express prejudice. It is possible that more extensive (newspaper articles, political speeches) and vivid (visual, auditive) manipulations may yield stronger effects. Also, repeated exposure to statements with similar content over extended periods of time may have a more pronounced impact due to accumulation of smaller effects.
Furthermore, the specific message contents that we use in the experimental manipulations may be important. It is not fully clear how well the reported effects can be generalized beyond this particular example. Since we were interested in the role of valuing conformity, we framed the positive message in terms of immigrants' cultural similarity but the negative message emphasized economic competition. It may be possible that respondents who value conformity are particularly affected by this cultural framing and will be influenced less by a positive economic message. However, we adopt the actual rhetoric of some West European politicians from across the political spectrum who tend to portray immigrants from Eastern Europe as an economic threat but, occasionally, also highlight their cultural proximity. Therefore, our findings suggest that these real-world political messaging strategies may be effective.
The findings demonstrate that the indirect effect is smaller in magnitude compared with the direct positive association between conformity values and prejudice, for both favorable and unfavorable messages. Thus, even though a pro-immigrant political message reduces the association between the importance attached to the value of conformity and prejudice, this effect is not strong enough to suppress, let alone reverse it. However, the relative importance of the indirect effect of conformity on prejudice likely depends on the magnitude of the corresponding direct effect. Prejudice toward some groups can be less (or more) common making the indirect social influence effect more (or less) strong and consequential.
We have focused on individual differences in conformity values and it can be argued that these reflect differences in authoritarian predisposition that drives the effect. The conformity measure may function as a proxy for authoritarianism that does not only involves valuing conformity but also aggressiveness and obeying established authorities (Altemeyer, 1981;Funke, 2005). We cannot rule out the possibility that obedience in particular underlies the effect that we found. However, the three aspects of authoritarianism tend to be differentially associated with various social and political outcomes such as policy opinions, party preferences, values, and generalized prejudice (Duckitt et al., 2010;Passini, 2017). These and other findings (see Stenner, 2009) indicate that the effects of conformity values on social and political attitudes can be analyzed separately rather than under the umbrella of 'authoritarianism syndrome'. In addition, some scholars have suggested that the desire for social conformity as opposed to personal autonomy is the defining characteristic of authoritarianism (Feldman, 2003;Stenner, 2005).
Even taking these limitations into account, our results confirm that political elites have relatively high leverage in influencing mass opinions among those who value social conformity, even in areas as sensitive as attitudes toward immigrant groups. Thus, paradoxically, the importance attached to the value of conformity does not only predispose people to be prejudiced toward cultural others but can also reduce prejudice by rendering them susceptible to normative influences to respond positively. Thus, our findings are in line with the proposition that the association between conformity and prejudice depends on the normative context (e.g., Plant & Devine, 1998), and that political messages are an important aspect of this context. Furthermore, our findings have potential implications beyond the realm of political messaging. For instance, intergroup contact theory postulates that authority support is an important condition for contact to reduce prejudice (Pettigrew, 1998). It may be of interest to examine whether individuals who value conformity are more susceptible to authority support for intergroup contact. Notes 1. The survey was fielded in Dutch. In the text and in Online Appendix we present the English translations of the survey questions and response options. 2. The study was ethically approved by the Ethics Committee of the Faculty of Social and Behavioral Sciences, Utrecht University. Number FETC18-063. 3. Δβ is the estimated difference in standardized coefficients across the two conditions. Since multigroup SEM estimates the effects by conditions, we calculate the differences and test them for significance using the procedure proposed by Clogg et al. (1995). 4. We refer to total effects (standard SEM term) as net effects because, according to our findings, positive direct effects and negative indirect effects of conformity on prejudice in the positive message condition partly cancel each other out. 5. When controlled for demographics (age, gender, and education), the indirect effect of conformity on prejudice in the unfavorable condition remains positive but significant only at the 90% confidence level. The differences between indirect and net effects across the two conditions remain significant at the 99% and 95% confidence levels respectively.

Disclosure statement
No potential conflict of interest was reported by the author(s).