Mason Burns is now at Purdue University. Margot Pickering graduated from the University of Kansas Law School.
The Suppression and Justification of Prejudice as a Function of Political Orientation
Article first published online: 14 JAN 2013
Copyright © 2013 European Association of Personality Psychology
European Journal of Personality
Volume 28, Issue 1, pages 44–59, January/February 2014
How to Cite
Webster, R. J., Burns, M. D., Pickering, M. and Saucier, D. A. (2014), The Suppression and Justification of Prejudice as a Function of Political Orientation. Eur. J. Pers., 28: 44–59. doi: 10.1002/per.1896
- Issue published online: 11 FEB 2014
- Article first published online: 14 JAN 2013
- Manuscript Accepted: 1 AUG 2012
- Manuscript Revised: 12 JUL 2012
- Manuscript Received: 23 APR 2012
- political orientation;
- motivation to control prejudice;
- right-wing authoritarianism;
- social dominance orientation
Politically conservative (versus liberal) individuals generally report more prejudice towards various low-status out-groups. Three studies examined whether prejudice suppression factors—specifically, internal and external motivation to suppress (IMS and EMS, respectively) prejudice—can help explain the relationship between political orientation and prejudice. Study 1 showed that IMS and EMS partially mediated the relationship between political orientation and affective prejudice towards Arabs. Study 2 demonstrated that when justification [right-wing authoritarianism (RWA) and social dominance orientation] and suppression (IMS and EMS) factors are simultaneously tested as mediators, only RWA partially mediated the relationship between political orientation and prejudice towards deviant (e.g. gay men) out-groups, whereas RWA and IMS fully mediated the relationship between political orientation and prejudice towards derogated out-groups (e.g. Blacks). Intriguingly, IMS rendered social dominance orientation effects non-significant for derogated out-groups. Study 3 showed that anticipating an out-group interaction (with a Black or lesbian confederate) diminished the mediational contribution of IMS in the political orientation–prejudice relationship because of increased IMS among participants; yet the increases in IMS did not completely eliminate differences in prejudice as a function of political orientation. Ultimately, these three studies demonstrate that suppression (in addition to justification) factors do help explain the relationship between political orientation and prejudice. Copyright © 2013 European Association of Personality Psychology.
Abundant research has shown that politically conservative beliefs or self-identification as conservative is meaningfully correlated with higher levels of self-reported prejudice towards a variety of derogated or disadvantaged groups (e.g. gay men, Blacks, fat people; for reviews, see Duckitt, 1992; Webster, Saucier, & Parks, 2011; Whitley & Kite, 2006). Given this overwhelming convergence of evidence and the associated stereotyping of conservatives as prejudiced (Webster et al., 2011), our objective was to help explain why self-identified conservative (versus liberal) individuals consistently report greater levels of prejudice. In particular, we focused on whether differences in motivation to suppress prejudice (in addition to other justifying ideologies) could help explain such differences.
Political orientation and the justification and suppression of prejudice
According to the justification–suppression model, everyone possesses genuine prejudice, ‘an authentically negative reaction that is usually not directly accessible but that is primary and powerful’ (Crandall & Eshleman, 2003, pp. 416–417). However, individuals do not express ‘genuine’ prejudice; instead, beliefs, values, and social norms ‘restrain’ the expression of unadulterated negativity towards most social groups. Such beliefs, values, and norms are termed suppression factors. Prejudice is ultimately ‘freed’ by justification factors that provide individuals the opportunity to express prejudices without sanction or compunction (e.g. guilt or shame imposed by the self or others).
Jost and colleagues (Jost, 2006; Jost, Federico, & Napier, 2009) postulated that conservatives' and liberals' psychological differences stem from their differential values about social equality and social change (also Webster & Saucier, 2012b). Broadly speaking, because conservatives have not internalized values supporting social equality and social change as much as liberals, conservatives are less concerned about suppressing prejudice and more concerned about justifying their prejudices (i.e. justification and suppression factors should be negatively related), which leads to greater expression of prejudices by conservatives.1
The justification of prejudice as a function of political orientation
First, we argue that more conservative (versus liberal) individuals more greatly endorse ideologies that justify social inequality [social dominance orientation (SDO)] and traditional social conventions [right-wing authoritarianism (RWA)], which help ‘vent’ their prejudices towards particular groups without guilt or sanction.
Right-wing authoritarianism (Altemeyer, 1981, 1998) is comprised of three inter-related beliefs: conventionalism (adherence to traditional norms), submission to authority (adherence to religious bodies or the government), and aggression towards those who deviate from those norms or that which is sanctioned by authority. RWA thus helps justify, broadly, opposition to social change (and also helps justify social inequality to a lesser degree; Webster & Saucier, 2012b), often under the guise of institutional or governmental mandate (Altemeyer, 1981, 1998). Accordingly, RWA best predicts prejudice towards ‘deviant’ groups that challenge or violate traditional social values, such as gay men, feminists, and rock stars (see the dual-process model of ideology and prejudice by Duckitt, 2001, 2006; Duckitt & Sibley, 2007, 2010).
Meanwhile, SDO is characterized by a Darwinian outlook on life: there is a defined power hierarchy for every species, with some groups at the top and some at the bottom (Pratto et al., 1994). Because people higher in SDO perceive that the world's resources are inherently limited, they believe that they must do everything in their power to secure and retain these resources—whether abstract (e.g. ‘power’) or material (e.g. ‘wealth’)—for their in-group. Jost and Thompson (2000) found that SDO was best conceptualized as having two intercorrelated factors: opposition to equality (the tendency to oppose equal conditions for low-status and high-status groups) and group-based dominance (the proclivity to use force to maintain hierarchies). SDO thus helps broadly justify opposition to equality (and inhibits support for social change to a lesser degree; Webster & Saucier, 2012a). Accordingly, SDO best predicts prejudice towards ‘derogated’ groups that are derogated or disadvantaged, such as Blacks, Arabs, or the unemployed (see the dual-process model of ideology and prejudice by Duckitt, 2001, 2006; Duckitt & Sibley, 2007, 2010).
Collectively, the findings on RWA and SDO indicate that not all prejudices are created equal; different types of prejudice may originate from different motivational concerns, with higher SDO helping rationalize prejudice to maintain social inequality and higher RWA helping rationalize prejudice to maintain social conventionalism (also Cohrs, Moschner, Maes, & Kielmann, 2005).
Little research has examined whether RWA or SDO accounts for the relationship between political orientation and prejudice. Sidanius, Pratto, and Bobo (1996) showed that after SDO had been accounted for, the relationship between conservatism and ‘classic’ racism (i.e. endorsement of racist stereotypes and White superiority) was rendered nonsignificant; however, Sidanius et al. did not assess the mediational contribution of RWA in their studies. That is, given the statistical and theoretical relationship between RWA and SDO (e.g. Sibley, Robertson, & Wilson, 2006), RWA might have also predicted classic racism in the Sidanius et al. model. In contrast, Whitley and Lee (2000) conducted a meta-analysis in which SDO, RWA, and conservatism each significantly predicted anti-gay attitudes. Whitley and Lee conducted an additional study in which they found that RWA and SDO only partially explained the relationship between political orientation and sexual prejudice. Nonetheless, Whitley and Lee entered SDO and RWA in separate steps in the regression analysis; thus, we do not know which justification factor explained more variation in prejudice.
We sought to extend these previous studies by simultaneously testing the mediational contributions of RWA and SDO on prejudice towards both deviant (e.g. gay men) and derogated (e.g. Blacks) out-groups, but only after testing the contributions of suppression factors, that is, individual difference variables that facilitate the suppression of prejudice (Sibley and Duckitt, 2010, p. 556).
The suppression of prejudice as a function of political orientation
Plant and Devine (1998; also Butz & Plant, 2009) postulated that people suppress prejudice either because they believe it is inherently wrong to express prejudice (i.e. indicated by higher internal motivation to suppress prejudice, or IMS) or to avoid social conflict or embarrassment from appearing prejudiced (i.e. indicated by higher external motivation to suppress prejudice, or EMS; also Dunton & Fazio, 1997). We posit that more liberal individuals should report higher IMS because they have more greatly internalized values of supporting social change and social equality; conversely, conservatives have internalized such values to a lesser extent (e.g. Greenberg, Simon, Pyszczynski, Solomon & Chatel, 1992; Webster & Saucier, 2012a). We reason that conservatives report greater prejudice because they have lower self-standards for appearing non-prejudicial; that is, IMS should help mediate the relationship between political orientation and prejudice.
Moreover, in the past few election cycles, conservatives (from GOP members to Young Republicans) concertedly tried to appear non-prejudiced (especially towards Blacks and women) to woo more minorities into their ranks; however, this push for diversity arguably has been driven by situational constraints (the need to grow conservatives' voting base with younger and more diverse individuals) rather than by the internalization of higher standards to appear non-prejudiced (Webster et al., 2011). Thus, more conservative individuals should not only report lower self-standards for appearing non-prejudiced (i.e. score lower on IMS) but also report greater desire to avoid social conflict or embarrassment from appearing prejudiced (i.e. score higher on EMS). Thus, more conservative individuals would likely be more hesitant to report greater prejudice, but only when social pressure is applied (e.g. interacting with an out-group member). That is, social pressure would increase EMS, thereby decreasing prejudicial responding among more conservative individuals (i.e. the relationship between political orientation and prejudice would be attenuated because of higher EMS). In the absence of social pressure, EMS may positively relate to prejudice for more conservative individuals.
Further, we suggest that external constraints or pressures (i.e. interacting with out-group members) may moderate the relationship between political orientation, EMS, and prejudice expression. People higher in IMS tend to report lower prejudice—at least towards Blacks—across different situational constraints, whereas people higher in EMS are more susceptible to change their responses on the basis of situational constraints (e.g. reporting attitudes in public versus private; Butz & Devine, 2009, p. 1317). However, to our knowledge, researchers have not tested whether or how external pressures may affect EMS or IMS scores and how such changes may then affect the relationship between EMS, IMS, and prejudice. Past researchers have typically had participants complete the IMS and EMS scales in mass screenings early in the semester and then tested the relationships of IMS and EMS with relevant criteria measured at a later time (e.g. Crandall et al., 2002; Plant & Devine, 1998, 2001, 2009; Ratcliff et al., 2006).
Perhaps because IMS scores showed consistent effects across different experimental manipulations and because IMS and EMS showed good test–retest reliability (rs = .77 and .60, respectively; Plant & Devine, 1998), researchers assumed that situational constraints would not affect IMS and EMS scores. However, RWA and SDO scores also exhibit just as high test–retest ability and were also thought to be stable trait-like constructs but are actually susceptible to change on the basis of situational contexts (e.g. Duckitt & Fisher, 2003; Guimond, Dambrun, Michinov, & Duarte, 2003; McFarland, Ageyev, & Djintcharadze, 1996; also Ekehammer, Akrami, Gylje, & Zakrisson, 2004). We suggest that IMS and EMS—like RWA and SDO—may be more rightly considered attitudinal variables (not stable trait-like variables) that can vary depending on the context in which they are measured (indeed, the EMS and IMS items refer to participants' own beliefs and values; e.g. ‘Because of my personal values, I believe that using stereotypes about Black people is wrong’).
Thus, the current studies assessed how IMS and EMS may mediate the relationship between political orientation and prejudice with and without situational constraints. Specifically, given conservatives' effort to attract minority members into their ranks, we were interested in how a potential interaction with an out-group member would affect IMS and EMS scores and how such changes would affect the relationship between political orientation and prejudice. We reasoned that the anticipated out-group contact would help minimize differences in prejudice expression between more liberal and more conservative participants because more conservative participants would report greater desire to avoid social conflict or embarrassment (i.e. because of increased EMS scores). However, with the imminent threat of an out-group interaction, it may be advantageous for participants to (outwardly) report that it is also important to suppress their prejudices because they really believe that it is wrong to express prejudice (i.e. reporting a higher self-standard to appear non-prejudiced). That is, anticipating an out-group interaction may also increase IMS scores, thereby attenuating the relationship between political orientation and prejudice. Our studies also provide the first stringent test of whether IMS is related to decreases in prejudice expression towards multiple out-groups (c.f. Crandall et al., 2002), not just Blacks (Butz & Devine, 2009) or gay men/lesbians (Lemm, 2006; Ratcliff, Lassiter, Markman, & Synder, 2006).
Overview of current studies
In bridging multiple theories [the justification–suppression model of prejudice (Crandall & Eshleman, 2003) and the dual-process model of ideology and prejudice (Duckitt, 2001, 2006)], to our knowledge, Study 1 is the first to examine whether IMS and EMS explain variation in (anti-Arab) prejudice between more liberal and conservative individuals. We then simultaneously tested the mediational contributions of both justification (RWA and SDO) and suppression (IMS and EMS) factors in explaining the relationship between political orientation and prejudice towards deviant, derogated, and criminal groups (Study 2). Finally, in Study 3, we assessed whether the relationships between political orientation, IMS, EMS, and prejudice expression (towards both deviant and derogated groups) changed when participants expected to interact (or not interact) with an out-group member (a lesbian or Black confederate).
Here, we also note that we did not test developmental or causal relationships between political orientation and these suppression/justification factors in our studies. Thus, we are not unequivocally assuming that individuals' political orientation causes RWA/SDO/IMS/EMS or that RWA/SDO/IMS/EMS causes political orientation. We are simply positing that once a general political orientation is adopted (however that developmental process unfolds), more conservative individuals differentially endorse justification (RWA and SDO) and suppression (IMS and EMS) factors to help rationalize the expression of their prejudices (Jost, Federico and Napier, 2009). Thus, even if these factors helped formulate individuals' general political orientation, it does not preclude individuals' endorsement of these factors to justify expressing their prejudices.
The goal of Study 1 was to provide initial evidence that IMS and/or EMS help explain differences in prejudice expression as a function of political orientation for a relevant target group in American society: Arabs (given the increased salience of discrimination against Arabs or individuals perceived to be Arab since the September 11 terrorist attacks; e.g. Rowatt, Franklin, & Cotton, 2005).
In total, 392 (166 men, 224 women; two participants did not report their sex) introductory psychology students from a Midwestern American university participated in the current study in exchange for course credit. Participants were primarily Caucasian (87.5%) with a mean age of 19.01 years (SD = 1.97).
Participants completed the following materials in a classroom setting in groups of 5 to 20 as part of a mass screening. Participants responded to items on a 1 (disagree very strongly) to 9 (agree very strongly) Likert-type scale (unless otherwise noted), and all measures were scored as the average response per item with higher mean values reflecting higher levels of the construct of interest.
We asked participants the following question: ‘Although it is often difficult to summarize one's political, economic, and social views in a single word or phrase, please indicate which of the following positions best represents your viewpoint’. In total, 81 participants self-identified as ‘liberal’, 188 self-identified as ‘middle of the road’, and 120 self-identified as ‘conservative’ (Knight, 1999). We treated political orientation as a continuous variable in our analyses with higher scores reflecting a more conservative political orientation; the scores approached a normal distribution (M = 2.10, SD = 0.71, skewness = −0.15; see Pratto et al., 1994, for an analogous response scale and procedure). We included middle-of-the-road respondents (i.e. moderates) in our analyses to be inclusive; still, the results from our primary analyses were virtually the same whether or not we included them (i.e. the pattern of results remained unchanged if we used a dummy-coded political orientation variable including only ‘liberals’ and ‘conservatives’).
Internal and external motivation to suppress prejudice
To assess levels of IMS and EMS, participants completed a modified version of Plant and Devine's (1998) IMS (five items; M = 6.69, SD = 1.71, alpha = .85) and EMS (five items; M = 5.58, SD = 1.95, alpha = .85) scales. In lieu of referring to a specific race (e.g. ‘Blacks’ or ‘African Americans’) in the IMS and EMS items, we referred to ‘people of other races’ (e.g. ‘Being nonprejudiced toward people of other races is important to my self-concept’ and ‘I try to hide any negative thoughts about people of other races in order to avoid negative reactions from others’, respectively; see Lemm, 2006, and Ratcliff et al., 2006, for similar examples).
Anti-Arab affective prejudice
Participants were asked to rate how they feel towards Arabs on seven positive affective descriptors (e.g. relaxed, good, positive) and seven negative descriptors (e.g. bad, negative, angry) (Whitley, 1999). The scores on the positively worded items were reverse-scored and then averaged together with scores on the negatively worded items to create one affective prejudice mean score for each participant (overall M = 4.86, SD = 1.62; alpha = .92). We chose to assess participants' general affectivity towards groups given that affect/emotion is generally considered the defining characteristic (‘core’) of prejudice (Whitley & Kite, 2006, p. 7).
Results and discussion
Correlations between the variables of interest
As predicted, zero-order correlations indicated that more liberal individuals scored lower on anti-Arab prejudice, r = .27, p < .001. Furthermore, liberal individuals scored higher on IMS, r = −.13, p = .01, and lower on EMS, r = .16, p < .01. Given that correlations can sometimes underestimate the strength of a relationship (Lucas & Schimmack, 2009), we computed effect sizes for the mean differences between liberals and conservatives. Indeed, the effect sizes for the differences between liberals and conservatives on IMS, EMS, and prejudice (ds = .45, .42, and .85, respectively) were medium to large in size (Cohen, 1988). IMS and EMS significantly correlated with anti-Arab prejudice, rs = −.50 and .13, respectively, ps < .01, although it is important to note that the correlation between IMS and prejudice is appreciably larger. Thus, suppression factors correlated with the expression of anti-Arab prejudice, with higher IMS associated with lower prejudice and higher EMS associated with greater prejudice.
Multiple mediation analysis
Preacher and Hayes' (2008) multiple-mediation analysis estimates the path coefficients in a multiple-mediator model and generates bootstrap confidence intervals (CIs) for total and specific indirect effects of X on Y through one or more mediator variable(s) M. Thus, multiple mediation analysis can simultaneously compute all path coefficients in the mediation model and bootstrap CIs for the total and specific mediational effects of IMS and EMS. We calculated and reported bootstrap CIs because they are preferred over Sobel's test (Preacher & Hayes, 2008; Shrout & Bolger, 2002), especially for smaller samples (which apply more to Studies 2 and 3 in the current paper). Please note the following: (i) all analyses were conducted using the SPSS script found at A. F. Hayes's website (http://www.afhayes.com/spss-sas-and-mplus-macros-and-code.html); (ii) the number of bootstrap resamples was 1000 for all mediation models in the current studies; (iii) we computed 95% CIs for all indirect effects; and (iv) the Preacher and Hayes procedure only provides unstandardized coefficients.
As seen in Figure 1, political orientation significantly predicted both IMS and EMS (B = −0.31 and 0.43, SE = 0.12 and 0.14, p = .01 and .002, respectively). IMS and EMS then significantly predicted anti-Arab prejudice (B = −0.49 and 0.16, SE = 0.04, respectively, and 0.04, ps < .001), such that those higher on IMS scored lower on prejudice, whereas those higher on EMS scored higher on prejudice. Most importantly, when IMS and EMS are controlled for, the relationship between political orientation and prejudice was attenuated but still significant (c path B = 0.63, SE = 0.11, p < .001; c-prime B path = 0.41, SE = 0.10, p < .001). However, bootstrap CIs for the IMS (0.04, 0.28) and EMS (0.02, 0.13) indirect effects were both significant, that is, they did not include 0. Thus, we have evidence that IMS and EMS partially mediated the relationship between political orientation and prejudice. That is, as people report that they are more liberal, they report lower levels of prejudice partially because of both higher internal and lower external motivations2 to control the expressions of prejudicial affect, at least in this case, towards Arabs.
Study 1 evidenced that suppression factors helped explain why conservatives reported more prejudice towards Arabs. However, as previously discussed, it was important to consider whether both suppression and justification factors would more wholly explain the relationship between political orientation and prejudice. Accordingly, in our second study, we assessed whether IMS, EMS,3 and the aforementioned justification factors—RWA and SDO—could completely account for the relationship between political orientation and prejudice—in this case, towards three different categories of out-groups: derogated, deviant, and criminal.
In total, 135 (60 men, 69 women; six participants did not report their sex) introductory psychology students from a Midwestern American university participated in the current study in exchange for course credit. Participants were primarily Caucasian (76.3%) with a mean age of 19.45 years (SD = 2.44).
Participants completed the following materials in a classroom setting in groups of 5 to 20. Participants responded to items on a 1 (disagree very strongly) to 9 (agree very strongly) Likert-type scale (unless otherwise noted), and all measures were scored as the average response per item with higher mean values reflecting higher levels of the construct of interest. Half of the participants were randomly assigned to complete the prejudice measures first and then the suppression and justification scales; the other half of the participants were randomly assigned to complete the justification and suppression scales first and then the prejudice measures. The order of materials did not significantly correlate with any scale means (ps > .14).
We used the same single-item political orientation measure as in Study 1; again, scores approximated a normal distribution (M = 1.89, SD = 0.65, skewness = 0.12). As in Study 1, we included moderates in our analyses to be inclusive; still, the results from our primary analyses were virtually the same whether or not we included them (i.e. the pattern of results remained unchanged whether we used political orientation as a continuous or categorical variable). In addition, for converging evidence, participants provided their political views on social and economic issues on a 1 (very conservative) to 9 (very liberal) Likert-type scale (Knight, 1999); responses to these two items were averaged together to form a composite variable of political orientation (alpha = .85).
Internal and external motivation to suppress prejudice
As in Study 1, participants completed our slightly modified versions of Plant and Devine's IMS (alpha = .89) and EMS (alpha = .86) scales.
Participants completed Funke's (2005) 12-item RWA scale (alpha = .75). Examples of aggression, conventionalism, and submission items are, respectively, ‘What our country really needs is a strong determined leader who will crush evil, and take us back to our true path’, ‘The withdrawal from tradition will turn out to be a fatal fault one day’, and ‘Obedience and respect for authority are the most important virtues children should learn’. Please note that one of the items on Funke's aggression component referred to the German ‘Chancellor’, which was replaced with ‘President’ for the current American sample.
Social dominance orientation
Participants completed the Jost et al. revised (‘balanced’) version of the social dominance orientation scale of Pratto et al.’s (1994) (16 items; alpha = .89). Examples of opposition to equality and group-based dominance items are ‘It would be good if all groups could be equal’ (reverse scored) and ‘If certain groups stayed in their place, we would have fewer problems’, respectively.
In a slight deviation from Study 1, participants completed 10 items (five positively worded and five negatively worded) randomly selected from Whitley's (1999) 14-item affective prejudice scale for each of the following social and ethnic minority groups: gay men, lesbians, feminists, Arabs, Blacks, Belgians, prostitutes, drug dealers, and terrorists (all nine scale alphas > .81). We included groups that are likely to be perceived as threatening conventional values (e.g. gay men), groups likely to be perceived as disadvantaged or competitively challenging in the social, economic, or political realm (e.g. Blacks), as well as groups likely to be perceived as blatantly violating legal norms towards which liberals and conservatives should hold similar (negative) views (e.g. drug dealers; Duckitt & Sibley, 2007).
Correlations among the variables of interest
Correlations among the variables of interest can be found in Table 1, along with means and standard deviations. Given that correlational and mediational results are virtually identical when using either political orientation measure, we will describe results using the single-item political orientation measure as in Study 1.
|1. One item||2.09||0.76|
|2. Two items||5.21||2.14||0.74|
|7. Gay men||4.24||2.18||0.44||0.44||−0.04||−0.30||0.53||0.30|
|14. Drug dealers||6.93||1.71||0.12||0.15||0.36||−0.05||0.16||0.01||0.02||0.12||0.00||0.07||0.02||0.07||0.37|
Further, as with Study 1, more liberal participants reported less prejudice towards many of the out-groups (except, as hypothesized, drug dealers, prostitutes, and terrorists) and scored higher on IMS; however, whereas liberals scored lower on EMS, this correlation was not significant (likely because of the smaller sample size in Study 2) but was of the same magnitude as the correlation found in Study 1. Additionally, as expected, more liberal individuals reported lower scores on the RWA and SDO scales. Overall, more liberal individuals reported again that they were more internally motivated to control their prejudices, reported less justification for expressing prejudices, and reported less prejudice towards a variety of minority groups.
Further, IMS was negatively correlated with RWA and especially with SDO; that is, as participants' level of justification increased (via SDO and RWA), their level of suppression decreased (via lower reported self-standards of appearing non-prejudicial; also Lowery, Hardin, & Sinclair, 2001). Meanwhile, EMS was not correlated with either SDO or RWA; this is not unsurprising given that EMS positively correlated with prejudice towards criminal groups only. Conversely, IMS was negatively correlated with prejudice towards every group, except criminal groups. IMS and EMS were again positively but not significantly correlated. Lastly, RWA and SDO were predictably correlated with prejudice towards a variety of out-groups.
Factor analysis of prejudice group scores
Following Duckitt and Sibley's (2007) example, we conducted a factor analysis on group prejudice mean scores to assess whether certain target groups could be categorized together. A three-factor principal axis analysis with Promax rotation (kappa = 1) revealed a clean (i.e. all items loaded onto their primary factor >0.59 and <0.40 onto all other factors) solution (with eigenvalues of 2.64, 1.78, and 1.27) to explain 47.15% of the variance (after rotation). Lesbians, gay men, and feminists loaded on the first factor; Arabs, Blacks, and Belgians loaded onto the second factor; and drug dealers, terrorists, and prostitutes loaded onto the third factor. Thus, it appeared that the first factor represented more or less socially deviant groups; the second factor represented derogated (socioeconomically competitive or disadvantaged) groups; and the third factor represented criminal groups (c.f. Duckitt & Sibley, 20074). We aggregated the group means to create three composite prejudice variables (alphas = .73, .66, and .58, respectively).
As in Study 1, the primary multiple-mediation results were virtually the same whether or not we included the middle-of-the-road participants; thus, we included all participants for our analyses. We also note that there was no evidence of multicollinearity given that all variance inflation factor values were <1.92.
The paths from political orientation to the proposed mediators were, of course, the same regardless of the criterion. Political orientation negatively predicted IMS (B = −0.54, SE = 0.22) and positively predicted SDO (B = 0.44, SE = 0.15) and RWA (B = 1.09, SE = 0.12); a more conservative orientation was related to lower IMS and higher RWA and SDO.
First, we assessed whether suppression (IMS and EMS) and/or justification (RWA and SDO) factors mediated the relationship between political orientation and prejudice towards deviant groups using Preacher and Hayes' (2008) multiple-mediation analysis. As seen in Figure 2, political orientation predicted prejudice, c-path B = 1.20, SE = 0.17. After the shared variance between the suppression and justification factors is accounted for, RWA was the only significant predictor of prejudice (B = 0.50, SE = 0.12); thus, IMS (B = −0.08, SE = 0.08), EMS (B = −0.07, SE = 0.06), and SDO (B = 0.05, SE = 0.12) did not uniquely predict prejudice. Further, RWA was the only potential mediator to have a significant indirect effect (bootstrap CI = 0.28, 0.85); however, RWA only partially mediated the relationship between political orientation and prejudice given that political orientation still predicted prejudice after controlling for RWA (c-prime path B = 0.62, SE = 0.20). Thus, more conservative individuals reported more prejudice towards deviant groups partially because of higher levels of RWA. The model explained an impressive 42.1% of the variance in prejudice as well.
As seen in Figure 3, political orientation predicted prejudice towards these groups (B = 0.44, SE = 0.13). As for the potential mediating variables, RWA and IMS were the only significant unique predictors of prejudice (Bs = 0.27 and −0.21, SEs = 0.08 and 0.06, respectively); thus, neither EMS (B = −0.03, SE = 0.04) nor SDO (B = 0.13, SE = 0.09) uniquely predicted prejudice scores. Further, both RWA (bootstrap CI = 0.11, 0.49) and IMS (bootstrap CI = 0.02, 0.29) exhibited significant indirect effects on prejudice towards derogated groups and rendered the path between political orientation and prejudice (c-prime path) nonsignificant (B ≤ 0.01, SE = 0.14), indicating full mediation. Thus, more politically conservative individuals reported more prejudice towards derogated groups because of lower scores on IMS and higher scores on RWA. The model explained an impressive 38.7% of the variance in prejudice as well.
Why did SDO not uniquely predict prejudice towards derogated out-groups given that past research has shown SDO to be a robust predictor of prejudice towards such groups (e.g. Duckitt, 2006)? We examined whether SDO would have been a significant predictor without the primary suppression variable—IMS—in the mediation model. Indeed, without IMS in the model, SDO was a significant predictor of prejudice towards the disadvantaged out-groups (B = 0.32, SE = 0.07, p < .001) and was a significant mediator (bootstrap CI for indirect effect = 0.05, 0.25). The addition of IMS appreciably increased the fit of the model by adding 6% of explained variance (ΔR2 = 0.058) and rendered the SDO predictive and the mediating paths nonsignificant, which clearly demonstrates the value of adding suppression factors in the mediation model.
Political orientation did not predict prejudice towards criminal groups (B = 0.16, SE = 0.13, p = .20); however, this does not completely eliminate the possibility of indirect effects through IMS or EMS. EMS only uniquely and significantly predicted prejudice towards criminal groups (B = 0.20, SE = 0.05, p < .001; all other Bs < |0.14|, SEs > .07). Regardless, none of the potential mediators exhibited significant indirect effects (all bootstrap CIs included 0).
Overall, then, results from Study 2 showed that after accounting for the shared variance between our justification (RWA and SDO) and suppression (IMS and EMS) factors, RWA partially mediated the relationship between political orientation and prejudice towards deviant groups, whereas RWA and IMS fully mediated the relationship between political orientation and prejudice towards derogated groups. The prediction of RWA is not entirely surprisingly given that such groups could pose both economic and cultural threats, thereby drawing the ire of both highly dominating and highly authoritarian individuals (e.g. Whitley, 1999), although the addition of the suppression factors rendered the effects of SDO on socioeconomically disadvantaged groups nonsignificant.
Only EMS uniquely and positively predicted prejudice towards criminal groups (which we discuss in the General Discussion section within the broader context of all studies), which is surprising given that RWA and SDO both have shown to significantly predict prejudice towards such groups (Duckitt & Sibley, 2007). Given the appreciably high prejudice scores against these groups, motivation to avoid confrontation with such criminal groups—but not sociopolitical ideology—may have driven prejudice against such groups, especially given the gravity of the crimes (prostitution, drug dealing, and terrorism). In support of this explanation, even political orientation did not relate to prejudice towards such groups in Study 2. It is also possible that people overall felt very negative towards such groups but would have disagreed on how to punish such individuals (e.g. support for the death penalty); that is, we would find more covariation between criminal sentencing (versus anti-sentiment) and political orientation/SDO/RWA.
As discussed in the introduction, we reasoned that situational constraints would help minimize differences in prejudice expression between more liberal and more conservative participants because more conservative participants would report greater desire to avoid social conflict or embarrassment (i.e. because of increased EMS scores). However, with the imminent threat of an out-group interaction, it may be advantageous for participants to report that it is more important to suppress their prejudices because they really believe that it is wrong to express prejudice (i.e. reporting a higher self-standard to appear non-prejudiced) rather than because they want to avoid social embarrassment or conflict with the out-group member. That is, anticipating an out-group interaction may also increase IMS scores.
Thus, Study 3 assessed how IMS and EMS may mediate the relationship between political orientation and prejudice with and without situational constraints—in this case, anticipating (versus not anticipating) an out-group interaction with either a Black or lesbian confederate. We varied both the expectation of interaction (anticipated contact versus no contact) and the target group (Black versus Lesbian) to assess whether effects would be equivalent across different out-group members.
In total, 153 introductory psychology students from a Midwestern American university participated in exchange for course credit. All non-White participants (n = 27) were removed before conducting any analyses. The final sample of 126 consisted of 44 men and 82 women (M age = 18.90, SD = 2.81).
Materials and procedure
Participants completed the experiment and all materials in a classroom setting in groups (with a maximum of 12 participants per group). Participants responded to items on a 1 (disagree very strongly) to 9 (agree very strongly) Likert-type scale (unless otherwise noted), and all measures were scored as the average response per item with higher mean values reflecting higher levels of the construct of interest. Students were randomly assigned in a 2 (experimenter: Black versus Lesbian) × 2 (experimenter contact: anticipated contact with experimenter versus no anticipated contact) between-groups design (ns for each cell ranged from 26 to 39).
Specifically, each group of students was randomly assigned to have either a Black male or lesbian undergraduate confederate experimenter. The confederates were trained to conduct each session in a consistent and professional manner using an identical script. The lesbian confederate wore a ‘pride’ tee shirt that had a rainbow flag displayed at the top of the shirt; below the flag, the shirt read ‘Check all that apply:’. Under ‘Check all that apply:’, two sets of two descriptors were consecutively placed beneath each other: ‘Straight’ and ‘Lesbian’ and ‘Proud’ and ‘Ashamed’. Next to each descriptor was a check box; accordingly, the boxes next to ‘Lesbian’ and ‘Proud’ contained a checkmark. Nothing else was on the tee shirt.
Each participant was randomly given a survey packet containing the other experimental manipulation. Experimenters instructed participants to complete the first page and then to wait for further instructions. The first page assessed demographic information (gender, age, and ethnicity), including one item assessing political orientation. For Study 3, participants rated their political orientation on a 1 (very liberal) to 9 (very conservative) Likert-type scale (M = 5.31, SD = 1.76, skewness = −0.85); we used this scaling to maximize the variance of political orientation within each experimental cell.
After participants completed the first page, experimenters then explained:
This will be a two-part experiment. First, you will all complete the survey packet in front of you. For the second part, you will do one of two things. Either you will come with me to another room individually and I will ask you some questions about your survey responses; or you will turn in your survey packet and complete a second, briefer survey. Because we cannot talk to you all individually, you were randomly assigned to either talk to me or do the second survey. If on page 3 there is a square, you have been selected to briefly talk to me. If you have a triangle, you will complete the second survey.
The experimenter then ensured that all participants knew to which condition they had been assigned.
Participants then completed items [modified from Plant & Devine's (2003) interracial anxiety measures] assessing their mood (Positive and Negative Affect Schedule, expanded form; Watson & Clark, 1994) and feelings about interacting with the experimenter. These latter items were worded so that participants could appropriately provide responses regardless of whether or not they were anticipating experimenter contact. Specifically, we assessed participants' feelings of contact avoidance (four items; e.g. ‘If I could avoid interacting with the experimenter, I would’), contact hostility (four items; e.g. ‘I would feel hostile when interacting with the experimenter’), and outcome expectancies (two items; ‘I would expect the experimenter to like me after interacting with me’ and ‘I do not think the experimenter would find me friendly after interacting with me’); please note that the former item was reverse-coded so that higher scores reflected greater negative expectations. The items for contact hostility and contact avoidance were aggregated to form a very reliable composite (alpha = .89), referred to as contact negativity. The two outcome expectancy items correlated too weakly (r = .11) to warrant aggregation. Thus, these items were treated as separate variables.
Next, participants completed Plant and Devine's (1998) IMS and EMS scales, with one change. In Studies 1 and 2, we revised the IMS and EMS scales to more broadly assess motivation to suppress prejudice towards ‘people of different races’; in Study 3, we revised the wording to ‘social/racial groups’ to more fittingly reflect our target groups (e.g. gays and Blacks, respectively).
Participants then completed six affective prejudice items (Whitley, 1999; three positively worded and three negatively worded) towards 10 groups: gays, lesbians, feminists, people who support women's choice regarding abortion (‘pro-choice supporters’), people who believe in evolution, Blacks, Arabs, Belgians, immigrants, and the unemployed. All derogated and deviant groups from Study 2 were included (i.e. criminal groups were not included because our focus was on groups towards which more conservative individuals reported greater prejudice); however, we added two additional deviant (pro-choice supporters and people who believe in evolution) and two additional derogated (immigrants and the unemployed) groups to help increase reliability (Duckitt & Sibley, 2007). As in Studies 1 and 2, the positively worded items were reverse coded and then averaged together with the negatively worded items to create composite prejudice scores for each group (alphas ranged from .72 to .92).
After all participants finished the first survey packet, experimenters ended the research sessions (i.e. without the participants having the interactions or completing additional surveys) and debriefed participants about the true purpose of the study.
Anticipating contact with an out-group member in Study 3 appeared to be threatening to participants. Regardless of political orientation, participants reported more fear (via the fear scale on the Positive and Negative Affect Schedule, expanded form) when anticipating contact with an out-group member (M = 2.70, SD = 1.55) than not (M = 2.12, SD = 1.09), F(1, 130) = 6.34, p = .013, d = 0.44. Fear scores were unaffected by which experimenter (Black versus lesbian) was present, F < 1.0. Although fear scores negatively related to contact negativity (r = .50, p < .001), fear did not correlate with IMS, EMS, or prejudice, rs < |.14|, ns.
Factor analysis of prejudice group scores
Following the approach in Study 2, a two-factor principal axis analysis with Promax rotation (kappa = 1) was conducted on the aggregated prejudice scores that explained 43.25% of the variance after rotation (eigenvalues = 3.85 and 1.53). Four groups loaded cleanly (i.e. loaded >0.48 on their primary factor and <0.40 on the other factor) on each of the two factors. Deviant groups (evolution supporters, gays, lesbians, and pro-choice supporters) loaded on Factor 1; derogated groups (Arabs, immigrants, Belgians, and Blacks) loaded onto Factor 2. The prejudice scores for the unemployed and feminist targets did not load >0.40 onto either group and thus were discarded from the primary analyses. The prejudice scores for the groups on Factors 1 and 2 were aggregated to create reliable composite scores (alphas = .83 and .75, respectively).
Correlations among the variables of interest
As Table 2 shows, people with a more conservative political orientation felt greater negative feelings about an expected interaction with the experimenter, reported lower IMS, and greater rates of prejudice towards all groups (except not significantly towards Blacks). In turn, greater IMS was also negatively related to most prejudices as well. However, EMS was not related to political orientation or prejudice. Meanwhile, greater contact negativity related to higher prejudice towards all deviant groups. This pattern was not consistent across derogated groups; greater contact negativity only related to higher anti-immigrant prejudice. Because the outcome expectancy criteria were single items and demonstrated uninteresting pattern of correlations with the other variables, we did not consider them further.
|1. Political orientation||4.75||1.57||—|
|Interaction with experimenter|
|2. Contact negativity||3.13||1.39||0.31||—|
|3. Will not like me||5.32||2.02||−0.02||−0.05||—|
|4. Will not find me friendly||2.17||1.57||0.13||0.27||0.11||—|
|14. Evolution supporters||4.69||2.00||0.36||0.33||−0.04||0.01||0.11||−0.14||0.26||0.23||0.29||0.24||0.60||0.61||0.56|
Predicting internal and external motivation to suppress prejudice and contact negativity: the effects of political orientation
We conducted a series of hierarchical regression analyses to assess whether political orientation, experimenter (Black versus lesbian), and experimenter contact (anticipating contact versus no contact) predicted EMS, IMS, or contact negativity. In Step 1, we entered scores for political orientation and dummy-coded variables for experimenter (0 = ‘Black’ and 1 = ‘lesbian’) and experimenter contact (0 = ‘no contact’ and 1 = ‘anticipated contact with experimenter’). In Step 2, we entered all two-way interaction terms between political orientation (after standardization), experimenter, and experimenter contact. In Step 3, we entered the three-way interaction between political orientation, experimenter, and experimenter contact.
Results of these regressions are summarized in Table 3. First, a more conservative political orientation related to lower IMS, greater contact negativity, and higher prejudice towards both deviant and derogated groups. However, the relationship between political orientation and IMS was moderated by condition: political orientation and IMS were only negatively related in the no-contact (versus contact) condition. When this effect is further probed, it appeared that IMS scores for people scoring in the top (more conservative orientation) and middle (more moderate orientation) third increased when anticipating contact with the experimenter (M = 6.93 and 7.48, SD = 1.40 and 1.29, respectively) versus anticipating no contact (M = 5.75 and 6.84, SD = 1.92 and 1.67), d = 0.68 and 0.44, respectively. Meanwhile, IMS scores for people scoring in the lower third (more liberal orientation) appeared not to change much when anticipating contact with the experimenter (M = 7.75, SD = 1.06) versus anticipating no contact (M = 7.83, SD = 1.33), d = −0.07. Thus, the effect of political orientation on IMS was attenuated in the experimenter contact condition because of increased IMS scores among more moderate and conservative participants, not because more liberal participants' scores decreased.
|Step 1||Step 2|
|Political orientation||Contact (versus no contact)||Experimenter (lesbian versus Black)||Political orientation × Contact||Political orientation × Experimenter||Contact × experimenter|
|EMS||β = −0.31†||β = 0.34‡|
|IMS||β = −0.34||β = 0.18||β = −0.31§|
|Contact negativity||β = 0.29|
|Deviant groups||β = 0.44||β = −0.16|
|Derogated groups||β = 0.23||β = −0.18|
Moreover, anticipating experimenter contact (versus no contact) increased IMS and decreased prejudice towards both deviant and derogated groups. Also, EMS scores increased more liberal participants in the no-contact condition only, whereas EMS scores increased when anticipating contact with the lesbian confederate, but not with the Black confederate.
Predicting prejudice scores: the effects of internal and external motivation to suppress prejudice and contact negativity
We next assessed whether experimenter contact or experimenter moderated the prediction of the potential mediators—IMS, EMS, and contact negativity—on prejudice. Contact negativity scores were entered in the very first step given that, chronologically, participants completed this measure before the IMS and EMS scales. The main effects of IMS, EMS, experimenter, and experimenter contact were entered in Step 2. (Please note that IMS, EMS, or contact negativity interacted with the experimental manipulations to predict prejudice, all ΔR2 < 0.01, Fs < 1.00.)
Results of these regressions are summarized in Table 4. In sum, only higher contact negativity predicted greater prejudice towards deviant groups; meanwhile, both higher contact negativity and lower IMS significantly predicted greater prejudice towards derogated groups. Further, experimenter contact remained as a significant predictor of prejudice towards deviant groups (with experimenter contact lowering prejudice), as seen in the regression models in Table 3; however, experimenter contact was no longer a significant predictor of prejudice towards derogated groups as it was in the model that only included political orientation as a predictor of prejudice (Table 3). This indicated that the addition of IMS and contact negativity rendered the effect of experimenter contact nonsignificant and quite possibly mediated the effect of experimenter contact on prejudice. We investigate this possibility in the following.
|Step 1||Step 2|
|Contact negativity||Contact (versus no contact)||Experimenter (lesbian vs. Black)||IMS||EMS|
|Deviant groups||β = 0.36||β = −0.17||β = −0.19*|
|Derogated groups||β = 0.22||β = −0.28|
Given these results, we conducted a series of mediational analyses using the Preacher and Hayes (2008) mediational script to test potential mediators of both the political orientation–prejudice and experimenter contact–prejudice relationships as a function of experimenter contact (versus experimenter) given that anticipated experimenter contact (versus no contact)—but not experimenter (lesbian versus Black)—has yielded more consistent and meaningful differences on our criteria.
Political orientation prejudice towards deviant groups
We tested whether contact negativity mediated the relationship between political orientation and prejudice towards deviant groups in the contact and no-contact conditions. In the no-contact condition, contact negativity exhibited a significant indirect effect (bootstrap CIs = −0.48, −0.03); however, political orientation remained a significant predictor of prejudice even after controlling for contact negativity: c-path B = 0.86, SE = 0.22, p < .001 vs. c-prime path B = 0.68, SE = 0.21, p = .002. However, when experimenter contact was anticipated, contact negativity no longer exhibited a significant indirect effect (bootstrap CI = −0.25, 0.13).
Contact prejudice towards deviant groups
Nonetheless, recall that overall anticipating experimenter contact both increased IMS scores and lowered prejudice. Thus, did the increase in IMS scores explain why anticipating experimenter contact decreased prejudice towards social deviant groups? Indeed, IMS exhibited a significant indirect effect (bootstrap CI = −0.38, −0.02); however, experimenter contact remained a significant predictor—barely—of prejudice after controlling for IMS scores: c-path B = −0.76, SE = 0.30, p = .01 vs. c-prime path B = −0.61, SE = 0.30, p = .05.
Political orientation prejudice towards derogated groups
We tested whether IMS and contact negativity mediated the relationship between political orientation and prejudice towards such groups when participants anticipated and did not anticipate contact with the experimenter. In the no-contact condition, IMS (bootstrap CI = −0.35, −0.03)—but not contact negativity (bootstrap CI = −0.14, 0.01)—exhibited significant indirect effects. Further, IMS scores rendered the effect of political orientation on prejudice nonsignificant: c-path B = 0.26, SE = 0.13, p = .04 vs. c-prime path B = 0.08, SE = 0.14, p = .56. When experimenter contact was anticipated, neither IMS (bootstrap CI = −0.18, 0.03) nor contact negativity (bootstrap CI = −0.24, 0.05) scores exhibited significant indirect effects. This may be because the direct effect of political orientation on prejudice was already attenuated when expecting experimenter contact (B = 0.28, SE = 0.14, p = .058).5
Contact prejudice towards derogated groups
Recall that overall anticipating experimenter contact both increased IMS scores and lowered prejudice. We thus tested whether IMS scores significantly mediated the effect of anticipated experimenter contact (versus no contact) on prejudice towards derogated groups. Indeed, IMS exhibited a significant indirect effect (bootstrap CI = −0.32, −0.03); moreover, IMS scores rendered the effect of experimenter contact on prejudice nonsignificant: c-path B = −0.44, SE = 0.18, p = .02 vs. c-prime path B = −0.30, SE = 0.18, p = .11.
Overall, then, Study 3 showed that situational constraints (in this case, anticipating out-group member contact) decreased prejudice expression—especially for derogated groups—among all participants (i.e. across all levels of political orientation) because of increased reported self-standards for appearing non-prejudicial (primarily for moderate and conservative individuals); yet increases in self-standards could not completely attenuate differences in prejudice as a function of political orientation, especially for prejudice towards deviant groups.
The three current studies provide a clearer understanding of the suppression (via IMS and EMS) and justification (via RWA, SDO) of prejudice as a function of political orientation (c.f. Crandall & Eshleman, 2003). In Study 1, we showed that IMS and EMS partially mediated the relationship between political orientation and anti-Arab affective prejudice. In a more comprehensive test, Study 2 showed that after the mediational contributions of justification (RWA and SDO) and suppression (IMS and EMS) factors are simultaneously tested, only RWA (partially) mediated the relationship between political orientation and prejudice towards deviant groups, whereas IMS and RWA fully mediated the relationship between political orientation and prejudice towards derogated groups. Moreover, the addition of IMS into the mediation model predicting prejudice towards disadvantaged out-groups actually rendered SDO nonsignificant as a predictor and mediator, thus clearly demonstrating the value of suppression factors in our theoretical framework.
Results also showed that people higher in SDO were less motivated to suppress their prejudices, which makes sense given that people higher in SDO use a variety of legitimating myths to justify their prejudices and discriminatory behaviours (Sidanius & Pratto, 1999). The finding that IMS overpowered SDO in Study 2 warrants future research clarifying their relationship to prejudice (e.g. does priming social dominance lower IMS thereby leading to increased prejudice?). Lastly, in Study 2, although political orientation was unrelated to prejudice towards criminal groups, higher EMS was.
Study 3 assessed whether anticipating experimenter contact (vs. no contact) with a Black or lesbian confederate moderated the relationship between political orientation, IMS/EMS, and prejudice. We reasoned that the ‘threat’ of out-group contact (versus no contact) may help minimize differences in prejudice as a function of political orientation most likely because more conservative individuals would seek to avoid conflict/social embarrassment (i.e. because of higher EMS). In fact, though, Study 3 is the first to show that anticipating out-group contact increased participants' reported levels of IMS, which fully explained the decrease in prejudice towards derogated groups and predominantly explained the decrease in prejudice towards deviant groups. The mere presence of an out-group member appeared to raise liberals' concerns about social embarrassment and conflict (i.e. increased EMS scores) in the no-contact condition, but increased EMS did not increase prejudice in this case.
More conservative participants did score higher in prejudice towards deviant and derogated groups regardless of whether they were expecting out-group contact. More conservative participants also scored lower in IMS, but only when they were not expecting out-group contact; anticipating contact increased reported self-standards about appearing non-prejudicial primarily among more moderate and conservative individuals, thereby diminishing the relationship between political orientation and IMS. Accordingly, IMS could only have mediated the political orientation–prejudice relationship in the no-contact condition; indeed, IMS fully explained why conservatives scored higher in prejudice towards derogated groups, which replicated our findings from Study 2.
However, IMS did not exhibit any mediational contribution for the relationship between political orientation and prejudice towards social deviants, which also replicated Study 2 results. More conservative participants, though, reported more negative feelings about interacting with the out-group experimenter regardless of whether they were actually assigned to meet with the experimenter. These negative feelings partly explained why more conservative individuals reported more prejudice towards social deviants, but only in the no-contact condition. Ultimately, no variables we measured significantly mediated the relationship between political orientation and prejudice towards social deviants when participants expected to interact with the experimenter.
Overall, Study 3 showed that although anticipating out-group contact lowered prejudice expressions and increased reported self-standards for appearing nonprejudicial, these shifts did not sufficiently attenuate differences in prejudice expression between more liberal and more conservative individuals. Again, these results do not mean that conservative individuals did not inhibit prejudice expression when anticipating out-group contact. That is, given the main effects of—but no interaction between—political orientation and experimenter contact on prejudice, both more liberal and conservative participants showed relatively equal decreases in prejudice when anticipating experimenter contact. Nevertheless, anticipating an out-group interaction did minimize the difference in IMS scores between more liberal and conservative participants. Thus, conservatives likely rely more on justification factors to rationalize prejudice expression, especially for deviant groups, when anticipating contact with out-group members. Anecdotally, this may help explain why people assert that ostensibly veiled discriminatory statements (often uttered in the presence of out-groups or their allies) are not based on prejudice (e.g. ‘I am not prejudiced/racist/sexist, but…’) but based on other functional reasons (e.g. ‘…We need to preserve traditional moral values’).
Overall, these three studies converge to further understanding of the suppression of prejudice, particularly as a function of political orientation. It appears IMS (versus EMS) is the more important factor in facilitating prejudice suppression in general. First, IMS—but not EMS—was consistently and negatively correlated with prejudice towards multiple target groups across all three studies. Second, as predicted, liberals consistently scored higher on IMS in each study, which in turn fully explained why liberals reported lower levels of prejudice against derogated (but not deviant) groups, at least when not expecting out-group contact. In sum, regarding political orientation, both justification and suppression factors helped explain differences in prejudice towards derogated out-groups, whereas only justification factors helped explain differences in the expression of prejudice towards deviant groups.
In the USA, it is much less socially acceptable to express prejudice or exhibit discrimination against derogated groups (which are typically based on race/ethnicity; Crandall et al., 2002). Ultimately, we suggest the possibility that the correlation between IMS and prejudice expression may vary according to how ‘acceptable’ it is in general for people to express prejudice against a particular target group (Crandall et al., 2002). Indeed, when we look across our studies, IMS more highly correlated with prejudice towards derogated groups (rs = −.53 and −.37 for Studies 2 and 3, respectively) versus deviant groups (rs = −.30 and −.25 for Studies 2 and 3, respectively). This is consistent with the research of Crandall et al. (2002) that shows the following: (i) people tended to report that it is less acceptable to express prejudice towards derogated groups based on race/ethnicity; and (b) scores from their own suppression of prejudice scale (which does not discriminate between internal and external motivation to control prejudice) only negatively correlated with higher prejudice among more normatively unacceptable prejudices. This is also consistent with recent experimental work showing that motivating people to suppress prejudice using autonomous or internal controls (emphasizing choice and why prejudice reduction is important and worthwhile to oneself) lowered implicit and explicit prejudice expression (at least, towards Blacks), whereas motivating people to suppress prejudice expression using external controls (complying to social norms) increased implicit and explicit racism even when compared with a no-control intervention (Legault, Gutsell, & Inzlicht, 2011).
Further, the relationship between generalized levels of EMS and prejudice may also vary as a function of perceived dangerousness of expressing prejudice towards particular groups. EMS did not consistently correlate with prejudice against derogated or deviant groups, even under the threat of out-group interaction. However, EMS did correlate positively with prejudice towards criminal groups in Study 2 and Arabs in Study 1. EMS may then only correlate with prejudice expression towards groups towards which people deem that it is generally more dangerous to express prejudice (Crandall et al., 2002). It is also possible that the correlation between EMS and prejudice depends on the target groups being rated. We did not specify target groups or prejudices in our modified EMS and IMS scales, and to our knowledge, we are the first to do so. Thus, EMS may have positively correlated with anti-Arab prejudice in Study 1 because it was the only group being rated after completing the EMS scale, whereas EMS have positively related to criminal groups in Study 2 because such groups stood out against the other targets (indeed, the mean prejudice scores for the criminal groups were appreciably higher compared with the other groups). Future research can probe how the specificity of the IMS and EMS scales and the type of target groups being rated affects the relationship between IMS, EMS, and prejudice.
Ultimately, we suggest then that some people (e.g. those with more liberal political orientations) have stronger internalized social norms about appearing nonprejudiced (Myrdal, 1944) and apply these internal norms more often, particularly for derogated groups; meanwhile, others have not yet internalized such norms, but nevertheless are able to change their reported self-standards about appearing nonprejudiced in the face of out-group interactions (e.g. more conservative individuals). When it is moderately okay to express prejudice against groups (e.g. gay men), social norms—and thus, IMS—do not factor as much into whether prejudice is expressed against such groups.
One may reason though that IMS scores increased in Study 3 because of social desirability concerns, not because of increased ‘genuine’ concern about (i.e. internalization of), self-standards for appearing non-prejudicial. The contact manipulation in Study 3 did aim to induce social pressure to decrease prejudicial responding, but what matters most is that the manipulation was successful: conservatives decreased prejudice expression in Study 3 because of explicit, intrinsic concerns about appearing non-prejudicial (at least towards derogated groups). Whether or not such standards were internalized on a ‘deeper’ level is a secondary concern, particularly because even superficial increases in internal (versus external) standards can ultimately lead to greater internalization and application of such standards in future contexts because of increased self-regulation (Monteith, Lybarger, & Woodcock, 2009). It will, of course, be valuable then to assess the extent to which participants internalized such standards (e.g. on an implicit level; Glaser & Knowles, 2008) in future studies and how long such internalization endures. Regardless, demonstrating that social pressure can increase IMS thereby decreasing prejudice—even among more conservative individuals—is an important and novel finding. We now know that IMS has some socially normative component despite not correlating with self-reports of social desirability or social monitoring in previous studies (Plant & Devine, 1998).
In any case, we acknowledge that most of our data were correlational and did not adequately test for causal or developmental relationships; yet we hope that our studies will stimulate more experimental research in which key variables are manipulated to test our underlying causal hypothesis that because conservatives have not internalized values supporting social equality and social change as much as liberals, conservatives are less concerned about suppressing prejudice and more concerned about justifying their prejudices, which leads to greater expression of prejudices by conservatives. For example, researchers can assess whether manipulating perceptions of target groups (e.g. as derogated or deviant; Duckitt & Sibley, 2010) or manipulating self-standards and societal standards (Legault et al., 2011) to appear nonprejudicial may moderate the relationship between political orientation and prejudice expression. It may be more difficult to manipulate people's actual self-identified political orientation; however, there may be proxies for such manipulations, such as manipulating cognitive resource availability given that cognitive load produces more conservative attitudes or behaviour (Eidelman & Crandall, 2009; Eidelman, Crandall, Goodman, & Blanchar, in press).
Lastly, as with previous research, the three current studies focused on those groups towards which more conservative individuals would report more negative attitudes. However, Gordon Allport (1954)—who wrote the seminal book on prejudice (The Nature of Prejudice)—and even prejudice theorists before him (Webster, Saucier, & Harris, 2010) acknowledged that everyone hold prejudices. In line with the assertion of Jost et al. (2009) that differing attitudes about social change and social equality underlie differences between liberals and conservatives, Webster and Saucier (2012b) have found that conservatives responded more negatively towards groups perceived as supporting social progress and equality (e.g., liberals, environmentalists, gay men) because conservatives' direct opposition to social equality and social change made it more likely to see such groups as violating their worldviews (i.e. most cherished beliefs and values); meanwhile, liberals responded more negatively towards groups opposing social progress and equality (e.g., conservatives, pro-life advocates, Tea Party Members) because liberals' explicit support for social equality and social change made it more likely to view such groups as violating their worldviews. It appears then that perceived violations of worldviews relating to social change and social equality likely underlie the development and maintenance of prejudices among both liberals and conservatives.
Taken together, our results indicate that justification and suppression factors are both useful in explaining the relationship between political orientation and prejudice. Accordingly, these results support the justification–suppression of prejudice model (Crandall & Eshleman, 2003) in that individuals tend to express prejudice when suppression is lower and/or justification is higher, while concurrently affirming Duckitt's (2001, 2006) dual-process model of prejudice in that not all prejudices are created equal—that is, it is valuable to distinguish between deviant, derogated, and criminal target groups.
Moreover, the current research demonstrates that when confronted with out-group contact letting people choose (or perceive to choose) to change their own self-standards about appearing non-prejudicial can help suppress prejudice expression; however, our results show that this may be more difficult to do with individuals who more strongly endorse factors that help justify the expression of prejudice and discrimination (e.g. those with a more conservative political orientation) and for target groups that violate traditional social values (e.g. gay men).
Past research has documented well the differences in the expression of prejudice between liberals and conservatives. Ultimately, our studies extend this literature to demonstrate that more conservative (versus liberal) individuals are less concerned about suppressing prejudice and more concerned about justifying their prejudices, which leads to greater expression of prejudices by conservatives. By doing so, we offer evidence of a theoretical explanation for these well-documented differences.
The authors would like to thank the Saucier lab for helping collect and enter data. In particular, we would like to thank our confederates for their convincing performances in Study 3: Evan Jones, Chantalle Hanschu, and Jennifer Cooper (who also helped design the t-shirt).
Some researchers have listed political conservatism as a justification factor in itself (Crandall, Eshleman, & O'Brien, 2002). We distinguish individuals' general political orientation (as liberal or conservative) from self-reported agreement with specific political causes or beliefs (e.g. affirmative action, military spending, same-sex marriage; e.g. Pratto, Sidanius, Stallworth, & Malle, 1994). That is, we would expect someone who identifies as (more or less) liberal or conservative to be more likely to report correspondingly (more or less) liberal or conservative values or beliefs, respectively, including the justification and suppression factors in the current studies (Knight, 1999). We also posit that conceptualization of political orientation (as liberal or conservative) is closer to the general public's colloquial definition and thus more ecologically valid.
We note that in past research IMS and EMS sometimes—but not always (e.g. Lowery et al., 2001; Ratcliff et al., 2006)—interact to predict prejudice; when this occurs, typically, people higher in IMS and lower in EMS tend to report the lowest prejudice expression and be less susceptible to situational constraints designed to manipulate prejudicial responding (Butz & Devine, 2009). In our studies, though, we did not find any evidence that IMS and EMS interacted together or interacted with political orientation, RWA, or SDO to predict prejudice in any of our three studies. Political orientation did not interact with RWA or SDO to produce prejudice either.
Dispositional empathy (Davis, 1983) tends to be a robust predictor of prejudice (Bäckström & Björklund, 2007; McFarland, 2010); however, empathy neither meaningfully added any variance to the mediation models in Study 2 nor did it predict prejudice after accounting for the variance contributed by IMS, EMS, RWA, and SDO. This is not surprising given the attitudinal specificity (Ajzen & Fishbein, 1980) of the other predictors, especially of IMS.
These groupings do not entirely parallel those found by Duckitt and Sibley (2007). In their study, prejudice scores for gay men and Arabs loaded together with other ‘derogated’ groups (towards which SDO best predicted prejudice), feminists loaded with the ‘dissident’ groups (of which RWA and SDO best predicted prejudice), and those for criminal offenders loaded with both ‘dangerous’ (terrorists and drug dealers; towards which RWA best predicted prejudice) and ‘dissident’ (prostitutes) groups. We do not expect perfect unity with Duckitt and Sibley's groupings, especially given the following: (i) we measured prejudice only towards nine (versus 24) groups; and (ii) we sampled from different cultures (USA versus New Zealand). Thus, the differences in groupings may be due to the other referent groups included in the factor analysis or cultural differences. However, given that we achieve groupings in Study 3 that are relatively similar to those in Study 2, we do not think the groupings we found in our studies are just due to random error.
The small difference between the effects of political orientation on prejudice in the experimenter contact and no-contact conditions is small enough to not expect—and, accordingly, we did not find—a significant interaction between political orientation and experimenter contact to predict prejudice towards derogated groups.
- 1980). Understanding attitudes and predicting social behavior. Englewood Cliffs, NJ: Prentice-Hall. , & (
- 1954). The nature of prejudice. New York: Perseus. (
- 1981). Right-wing authoritarianism. Winnipeg: University of Manitoba Press. (
- 1998). The other ‘authoritarian personality’. Advances in Experimental Social Psychology, 30, 47–92. (
- 2007). Structural modeling of generalized prejudice: The role of social dominance, authoritarianism, and empathy. Journal of Individual Differences, 28, 10–17. , & (
- 2009). Prejudice control and interracial relations: The role of motivation to respond without prejudice. Journal of Personality, 77, 1311–1342. , & (
- 1988). Statistical power analysis for the behavioral sciences (2nd ed.) Mahwah, NJ: Lawrence Erlbaum Associates. (
- 2005). The motivational bases of right-wing authoritarianism and social dominance orientation: Relations to values and attitudes in the aftermath of September 11, 2001. Personality and Social Psychology Bulletin, 31, 1425–1434. , , , & (
- 2003). A justification–suppression model of the expression and experience of prejudice. Psychological Bulletin, 129, 414–446. , & (
- 2002). Social norms and the expression and suppression of prejudice: The struggle for internalization. Journal of Personality and Social Psychology, 82, 359–378. , , & (
- 1983). Measuring individual differences in empathy: Evidence for a multidimensional approach. Journal of Personality and Social Psychology, 44, 113–126. (
- 1992). The social psychology of prejudice. New York: Praeger Publishers. (
- 2001). A cognitive-motivational theory of ideology and prejudice. Advances in Experimental Social Psychology, 33, 41–113. (
- 2006). Differential effects of right wing authoritarianism and social dominance orientation on outgroup attitudes and their mediation by threat from and competitiveness to outgroups. Personality and Social Psychology Bulletin, 32, 684–696. (
- 2003). The impact of social threat on worldview and ideological attitudes. Political Psychology, 24, 199–222. , & (
- 2007). Right-wing authoritarianism, social dominance orientation, and the dimensions of generalized prejudice. European Journal of Personality, 21, 113–130. , & (
- 2010). Right-wing authoritarianism and social dominance orientation differentially moderate intergroup effects on prejudice. European Journal of Personality, 24, 583–601. , & (
- 1997). An individual difference measure of motivation to control prejudice. Personality and Social Psychology Bulletin, 23, 316–326. , & (
- 2009). On the psychological advantage of the status quo. In J. T. Jost, A. C. Kay, & H. Thorisdottir (Eds.), Social and psychological bases of ideology and system justification (Ch. 4). New York, NY: Oxford University Press. , & (
- Low-effort thought promotes political conservatism. Personality and Social Psychology Bulletin. , , , & (in press).
- 2004). What matters most to prejudice: Big Five personality, social dominance orientation, or right-wing authoritarianism? European Journal of Personality, 18, 463–482. , , , & (
- 2005). The dimensionality of right-wing authoritarianism: Lessons from the dilemma between theory and measurement. Political Psychology, 26, 195–218. (
- 2008). Implicit motivation to control prejudice. Journal of Experimental Social Psychology, 44, 164–172. , & (
- 1992). Terror management and tolerance: Does mortality salience always intensify negative reactions to others who threaten one's worldview? Journal of Personality and Social Psychology, 63, 212–220. , , , , & (
- 2003). Does social dominance generate prejudice? Integrating individual and contextual determinants of intergroup cognitions. Journal of Personality and Social Psychology, 84, 697–721. , , , & (
- 2006). The end of ideology. American Psychologist, 61, 651–670. (
- 2000). Group-based dominance and opposition to equality as independent predictors of self-esteem, ethnocentrism, and social policy attitudes among African Americans and European Americans. Journal of Experimental Social Psychology, 36, 209–232. , & (
- 2009). Political ideology: Its structure, function, and elective affinities. Annual Review of Psychology, 60, 307–337. , , & (
- 1999). Liberalism and conservatism. In J. P. Robinson, P. R. Shaver, & L. S. Wrightsman (Eds.), Measures of social psychological attitudes (pp. 59–158). San Diego, CA: Academic. (
- 2011). Ironic effects of anti-prejudice messages: How motivational interventions can reduce (but also increase) prejudice. Psychological Science, 22, 1472–1477. , , & (
- 2006). Positive associations among interpersonal contact, motivation, implicit and explicit attitudes toward gay men. Journal of Homosexuality, 51 (2), 79–99. (
- 2001). Social influence on automatic racial prejudice. Journal of Personality and Social Psychology, 81, 842–855. , , & (
- 2009). Income and well-being: How big is the gap between the rich and the poor? Journal of Research in Personality, 43, 75–78. , & (
- 2010). Authoritarianism, social dominance, and other roots of generalized prejudice. Political Psychology, 31, 453–477. (
- 1996). Russian authoritarianism two years after communism. Personality and Social Psychology Bulletin, 22, 210–217. , , & (
- 2009). Schooling the cognitive monster: The role of motivation in the regulation and control of prejudice. Social and Personality Compass, 3, 211–226. , , & (
- 1944). An American dilemma: The negro problem and modern democracy. New York: Harper & Bros. (
- 1998). Internal and external motivation to respond without prejudice. Journal of Personality and Social Psychology, 75, 811–832. , & (
- 2001). Responses to other-imposed pro-Black pressure: Acceptance or backlash? Journal of Experimental Social Psychology, 37, 486–501. , & (
- 2003). The antecedents and implications of interracial anxiety. Personality and Social Psychology Bulletin, 29, 790–801. , & (
- 2009). The active control of prejudice: Unpacking the intentions guiding control efforts. Journal of Personality and Social Psychology, 96, 640–652. , & (
- 1994). Social dominance orientation: A personality variable predicting social and political attitudes. Journal of Personality and Social Psychology, 67, 741–763. , , , & (
- 2008). Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behavior Research Methods, Instruments, & Computers, 40, 879–891. , & (
- 2006). Gender differences in attitudes toward gay men and lesbians: The role of motivation to respond without prejudice. Personality and Social Psychology Bulletin, 32, 1325–1338. , , , & (
- 2005). Patterns and personality correlates of implicit and explicit attitudes toward Christians and Muslims. Journal for the Scientific Study of Religion, 44, 29–43. , , & (
- 2002). Mediation in experimental and nonexperimental studies: New procedures and recommendations. Psychological Methods, 7, 422–445. & (
- 2010). The ideological legitimation of the status quo: Longitudinal tests of a social dominance model. Political Psychology, 31, 109–137. , & (
- 2006). Social dominance orientation and right-wing authoritarianism: Additive and interactive effects. Political Psychology, 27, 755–768. , , & (
- 1999). Social dominance: An intergroup theory of social hierarchy and oppression. New York: Cambridge University Press. , & (
- 1996). Racism, conservatism, affirmative action and intellectual sophistication: A matter of principled conservatism or group dominance? Journal of Personality and Social Psychology, 70, 476–490. & (
- 1994). The PANAS-X: Manual for the Positive and Negative Affect Schedule—Expanded form (University of Iowa). Retrieved April 11, 2012, from http://www.psychology.uiowa.edu/Faculty/Clark/PANAS-X.pdf. , & (
- 2012a). Right-wing authoritarianism and social dominance orientation subscales differentially predict prejudice. Manuscript in preparation, Kansas State University. , & (
- 2012b). Spanning the political divide: Why are liberals and conservatives explicitly prejudiced toward different groups? Manuscript in preparation, Kansas State University. , & (
- 2010). Before the measurement of prejudice: Early psychological and sociological papers on prejudice. Journal of the History of the Behavioral Sciences, 46, 300–313. , , & (
- 2011). New bottle, same old wine: The GOP and race in the age of Obama. In G. S. Parks & M. Hughey (Eds.), The Obamas and a (Post) Racial America (pp. 266–284). New York: Oxford University Press. , , & (
- 1999). Right-wing authoritarianism, social dominance orientation, and prejudice. Journal of Personality and Social Psychology, 77, 126–134. (
- 2006). The psychology of prejudice and discrimination. Belmont, CA: Thomson-Wadsworth. & (
- 2000). Relationships of authoritarianism and related constructs to attitudes toward homosexuality. Journal of Applied Social Psychology, 30, 144–170. , & (