Social norms versus social motives: the effects of social influence and motivation to control prejudiced reactions on the expression of prejudice

Abstract

This study examined how individual differences in motivation to control prejudiced reactions (MCPR) affected one's sensitivity to social norms regarding the expression of gay rights attitudes. After measuring their political beliefs and MCPR, pro-gay rights and anti-gay rights participants took part in a discussion where they faced discussion groups that either opposed their position unanimously or nonunanimously (i.e., four opposing confederates vs. three opposing confederates and one undecided confederate). Anti-gay rights individuals showed more conformity overall, especially when high in MCPR. Anti-gay rights individuals also showed norm-consistent attitude change, regardless of MCPR, whereas attitude change among pro-gay rights individuals depended upon their level of MCPR.

Keywords::

• attitude change
• conformity
• political psychology
• prejudice
• social norms

When Allport (1954) wrote his seminal text on The Nature of Prejudice he argued that the majority of prejudiced attitudes arose from conformity and social norms. More than 50 years later, a small but strong body of literature supports many of Allport's original hypotheses regarding the role of social influence in the perpetuation of prejudice (Crandall, Eshleman, & O'Brien, 2002; Crandall & Stangor, 2005; Stangor, Sechrist, & Jost, 2001). For instance, correlational studies have shown that people's willingness to express their prejudices is strongly related to their perception of what society deems acceptable (Crandall et al., 2002). Experimental studies have further demonstrated that manipulating social norms can reduce prejudice and increase the endorsement of policies aimed at reducing prejudice (Clark & Maass, 1990; Monteith, Deneen, & Tooman, 1996).

Yet, people are not universally susceptible to norms regarding prejudice. Suppression scales have been identified as potential predictors of how people respond to normative rules about the expression of prejudice (Crandall et al., 2002), but few tests connecting these scales to behavior have been conducted. Of interest to the present study is the motivation to control prejudiced reactions (MCPR) scale, which aims to assess how motivated people are to conform to nonprejudiced standards (Dunton & Fazio, 1997). In this study we examined whether MCPR moderates the amount of influence that social norms—whether tolerant or intolerant—have on people's attitudes toward policies directed at reducing prejudice.

Social norms and prejudice

In the present study, we were interested in the current debate surrounding legislation that aims to reduce sexual prejudice. Sexual prejudice is defined here as a negative attitude toward members of a group based on sexual orientation (Stangor, 2009). The expression of sexual prejudice has traditionally taken on overt forms (Sniderman & Carmines, 1997). However, just as overt expressions of racism have given way to more subtle forms (i.e., symbolic prejudice; see Fiske, 1998, for a review), prejudice directed at homosexuals may be moving toward more subtle forms as well (Cowan, Heiple, Marquez, Khatchadourian, & McNevin, 2005). One way that prejudiced beliefs are voiced in a culture that is becoming increasingly intolerant of blatant expressions of prejudice is via resistance to distributive social justice policies that serve to reduce prejudice and discrimination (e.g., same-sex marriage, hate crime legislation).

A main thrust of prejudice research has centered on explaining the process of prejudice, as well as the content and structure of prejudiced attitudes. However, an equally important endeavor in the study of prejudice remains identifying strategies to reduce prejudice (Swim, Gervais, Pearson, & Stangor, 2009). One important way to reduce prejudice is through salient social norms (see Crandall & Stangor, 2005, for review). For example, Crandall et al. (2002) found that the extent to which an individual expressed a prejudicial or tolerant view of a particular group (e.g., African Americans, homosexuals) was closely linked to their perceptions of what others felt was appropriate.

Just as individuals' policy attitudes can serve as a medium in which they may covertly express prejudice (Federico & Sidanius, 2002), one's endorsement of social justice policies can be influenced by social norms (Clark & Maass, 1990; Monteith et al., 1996). A handful of studies have examined the impact of social norms on the endorsement of gay rights. One of the first was conducted by Clark and Maass (1990). They recruited individuals who were opposed or neutral on the issue of gay rights and subjected them to varying levels of normative pressure (where they were told that the majority was either for or against gay rights). When assigned to a condition where the majority was for, participants expressed more positive attitudes toward homosexuals when they anticipated having to discuss their positions with the majority. When the majority against gay rights increased in number, the individual's gay rights attitudes became more reflective of the majority position (i.e., against gay rights). Therefore, the ability of a social norm to influence individuals' attitudes toward granting rights to gays and lesbians may be moderated by the solidarity of the group conveying the message (i.e., numerical strength of the majority).

Although it is clear that people follow norms when expressing prejudice, it may not necessarily be true that people follow them unequivocally. In a society with strong overarching norms of equality and tolerance, it seems unlikely that a prejudiced norm could win out as easily as a tolerant norm. Monteith et al. (1996) found that when participants were presented with a nonprejudiced norm—in this case it was a confederate participant publicly voicing support for several campus level gay rights initiatives—they showed more support for local gay rights policies. Yet the presentation of a prejudiced norm failed to decrease support.

This tolerant norm effect—whereby individuals' attitudes change to match the presentation of tolerant social norms more so than intolerant norms—has been found to be even more prevalent when the position being advocated was either consistent with the larger population's norm or was perceived as potentially becoming the more popular opinion (Clark & Maass, 1990; Glynn, Hayes, & Shanahan, 1997; Shamir, 1997). Polling clearly demonstrates a national trend of Americans increasingly supporting gay rights (Kohut, 2013). Therefore, when it comes to gay rights, one might expect that the nonprejudicial norm would exert more social influence than the prejudicial norm.

Individual differences in susceptibility to social norms

Although the role of social norms in the expression of prejudice has been established, it is important to examine how certain aspects of the person might affect how people respond to such norms. To further understand how social norms and preexisting beliefs predict the expression or suppression of prejudice, an individual difference variable—motivation to control prejudice responding—was included to examine whether individual differences moderated willingness to express or suppress one's preexisting political beliefs.

Certain individuals may be more motivated to avoid negative evaluation and—especially in a discussion of intergroup attitudes—may be afraid of being labeled as prejudiced. Individuals who suppress the expression of prejudice to avoid negative reactions score high on Dunton and Fazio's (1997) MCPR scale. Furthermore, there is some evidence that MCPR affects conformity when it comes to social policy. In one study, Gabriel, Banse, and Hug (2007) found that people high in MCPR were more likely to donate money and sign a petition for a local gay, lesbian, bisexual, and transgender (GLBT) organization whenever in public as opposed to being in private, indicating that individuals may be apprehensive of the perceptions of others regarding norms about sexual prejudice. However, to our knowledge, studies of this kind which examine the interplay between MCPR, the social environment, and actual behavior are few.

Previous experimental research on MCPR has often focused on how MCPR affects public expressions of prejudice (Gabriel et al., 2007) or automatically activated attitudes (Fazio & Hilden, 2001; Glaser & Knowles, 2008; Olsen & Fazio, 2004). MCPR is alleged to assess people's inclination to follow social norms regarding prejudice. Yet, little systematic research has been undertaken on whether individual differences in MCPR predict how people's expression or suppression of prejudice are influenced by situationally-salient social norms. For example, in the Gabriel et al. study they manipulated only the public nature of the context, not the social norm. Thus, validation of whether MCPR differences predict behavioral differences in the receptivity to social norms remains to be seen. This study addressed this gap in the literature by examining whether people's level of MCPR moderates their support for social justice policies when under the pressure of an egalitarian or prejudiced social norm.

Hypotheses

The purpose of our study was to examine how individuals' political position (pro- vs. anti-gay rights) and MCPR would affect their likelihood to express their opinions when confronted by a majority opposing their views. To test this, participants first completed an initial survey that assessed their policy attitudes and MCPR. At a later date, participants were placed into discussion groups where a group either unanimously or partially opposed their initial policy attitudes. During the discussion, participants were asked to participate in a series of public votes on the issue of gay rights. Later, in a private room, participants were asked to provide their final opinion on the gay rights issues raised in the voting session.

In our study we chose to operationalize the effects of social influence in two ways. First, conformity reflected the extent to which participants publicly expressed a vote on gay rights that was contrary to their own previously expressed private attitudes on gay rights. Second, private attitude change reflected the extent to which individuals changed reports of their privately held attitudes on gay rights after being exposed to a social norm that contradicts them. Our hypotheses were as follows.

Hypothsis 1

=

Participants exposed to the unanimous opposition condition will show more public conformity to the group as well as evince more attitude change in the direction of the group norm than those in the partial opposition condition.

Hypothsis 2

=

Participants with initially more anti-gay rights attitudes will show greater public conformity and attitude change in the direction of the group norm than those with initially more pro-gay rights attitudes.

Hypothsis 3a

=

Participants with high MCPR will show greater public conformity and attitude change in the direction of the group norm than those with low MCPR.

Hypothsis 3b

=

MCPR will moderate the effects of normative pressure on conformity and attitude change such that the effects of high normative pressure will be stronger in those who are high in MCPR than those who are low in MCPR.

Political position and MCPR

We also were interested how one's MCPR might affect their susceptibility to social norms about the expression of prejudice, in particular, norms of tolerance. A lack of preexisting research prevented us from making exact predictions about the pattern of interaction between one's position on gay rights and their level of MCPR. As such, our examination of interactions of MCPR and political position are exploratory.

Method

Participants

Students at a large university in the Southeastern United States (n = 457) participated in the study for course credit. Of the initial 457 participants, 403 completed both Part 1 and Part 2 of the study. There were no significant differences between those attending and those dropping on attitude valence or extremity. Most participants were female (60%) and either Caucasian (64.6%) or African American (27.8%). The average age was 19.35 years (SD = 3.37). Republicans (46.9%) and Democrats (32.2%) were the most common political affiliations.

Materials and procedure

This study employed a two-part methodology. Part 1 was a screening survey gauging MCPR and gay rights attitudes. Part 2 consisted of an initial public practice vote about local campus policies, followed by a public voting session about six gay rights issues, a 20-minute group discussion about gay rights, a revote on the gay rights voting topics, and a private postdiscussion attitude change measure.

Subject variables

Political position

In Part 1, participants were given the Political Positions Inventory (PPI), a questionnaire that assessed their political attitudes on, among other topics, gay rights. Using a seven-point Likert scale (1 = strongly disagree, 7 = strongly agree), participants responded to 18 items that assessed participants' attitudes toward homosexuals (12 items) and gay rights policies (6 items). Examples of policy items are “Homosexuals should be able to adopt children” and “Homosexuals should be allowed to marry, legally.” Participants were determined to be “for” or “against” gay rights based on their average score on the 18 items. Participants who scored below “4” were “for” gay rights (n = 182), whereas those who scored above “4” were “anti” gay rights (n = 221). The reliability of the 18-item gay rights scale was α = .94.

Motivation to control prejudiced reactions

Motivation to suppress prejudice was assessed using Dunton and Fazio's (1997) MCPR scale. The MCPR scale was adapted to reflect prejudice toward minorities in general rather than just racial minorities. The scale had 18 items (6 reversed), such as “In today's society it is important to not appear prejudiced in any way.” Participants responded on a six-point Likert scale from 1 (strongly disagree) to 6 (strongly agree). Reliability was α = .71.

Independent variable

Normative pressure

To manipulate normative pressure, participants were randomly assigned an experimental or control condition. In the unanimous normative pressure (experimental) condition participants encountered four confederates who opposed their initial position (as determined by his/her responses to the screening survey). In the nonunanimous normative pressure (control) condition, participants encountered three confederates with opposing beliefs and one undecided confederate who took a neutral stance on the issue.

Procedure

Following completion of the PPI in Part 1, participants were told to come back to the lab at a later date to participate in Part 2 of the experiment wherein they would discuss a political topic with peers. Upon arrival, discussants were led from the waiting room to the lab conference table individually. Once seated, the experimenter provided the group with discussion guidelines and information about the format of the discussion. The group was told that there would be two initial voting sets followed by an open-ended discussion on a “randomly selected” topic. For each question that was read to the group, participants were instructed to vote in the order of their seat (1–5). Participants were instructed respond “yes,” “no,” or “undecided” to each vote.

The first public voting set, “University policies,” was announced. Consistent with the methodology employed by Asch (1956) wherein confederates gave the right answers before the wrong answers, the “University policies” policies voting set was designed such that all of members of the group should agree and would serve as a practice vote. Next, a public vote (Vote 1) was taken on six gay rights issues to see if participants would conform by voting against the position they reported in the screening survey. Although worded differently, these voting issues mirrored the six policy items assessed in the PPI. The voting set included the following items: “Sexual orientation should be included in anti-discrimination laws,” “Sexual orientation should be included as a protected class in hate crime legislation,” “There should be a law prohibiting same-sex marriage,” “There should be laws restricting homosexuals from adopting,” “Homosexuals should be prohibited from serving openly in the military,” and “Same-sex couples should be afforded the same benefits as heterosexual couples (such as hospital visitation rights).”

For each policy vote, the research assistants serving as confederates took the positions opposite of the participant (e.g., if the participant was “for,” confederates were “against”). Depending on the condition, either four opposing confederates (unanimous) or three opposing and one undecided confederate (nonunanimous) were present during voting. When taking a neutral stance, the confederate voted “undecided” across all gay rights policies and in the discussion offered mixed opinions, acknowledging both sides of the issues. The undecided participant always voted third, breaking the solidarity of the opposing group norm right before the participant would vote. The participant was always assigned to the fourth seat and thus had three confederates speak before him/her and one after.

The confederates were trained with an extensive list of arguments for and against gay rights piloted and balanced for quality and strength. Research assistants were unaware of the hypotheses. Furthermore, they were unaware whether the participant was for or against gay rights until the day of the discussion and were never aware of the participant's MCPR score.

A 20-minute discussion on gay rights followed the first voting set. During the discussion confederates argued consistent with their votes. After the discussion, participants and the confederates had a second public vote (Vote 2) on the same six gay rights issues in Vote 1. Confederate votes remained consistent across voting sets. Finally, participants were escorted to a private room to report postdiscussion attitudes by responding to the six gay rights policy items from the screening survey.

After completing the postdiscussion measures, the experimenter debriefed the participant using the Aronson, Ellsworth, Carlsmith, and Gonzales (1990) post-experimental interview procedure. Following debriefing, techniques (i.e., participant knowledge check via guessing game, confidentiality agreements) established by Edlund, Sagarin, Skowronski, Johnson, and Kutter (2009) were used to detect and reduce cross talk.

Dependent variables

Public conformity

Public conformity was determined by how often the participant voted with the group on Vote 1 and Vote 2. A vote contrary to the participant's reported position on their survey in Part 1 was coded as active conformity. Thus, for every vote, of which there were 12 (6 in Vote 1 and 6 in Vote 2), if participants voted with the group they were assigned a score of “1.” If participants voted against the group, they were assigned a score of “0.” If participants indicated they were undecided, they were assigned a score of “0.5.” These scores were then summed across Votes 1 and 2 for an average of conforming votes. Votes across Votes 1 and 2 were highly correlated, r = .80, thus the two indices were combined into a single measure of conformity.1

Private attitude change

Attitude change was determined by subtracting participants' scores on six gay rights policy items completed privately postdiscussion (Part 2) from those same items completed during screening (Part 1). Negative scores indicated that attitudes became more anti-gay rights and positive scores indicated attitudes became more pro-gay rights.

Results

Descriptive statistics were generated to provide an overall assessment of where participants stood on the subject variables. Although PPI scores were used to determine and thus categorize participants into political positions (pro vs. anti) for the experiment, it was used as a continuous variable (i.e., participant scores on the screening survey gay rights items) in analyses. High PPI scores indicated resistance to gay rights policies and low scores indicated being in favor of these policies. The average PPI score was 3.87 (SD = 1.50) and the average MCPR score was 3.56 (SD = 0.57). Skewness and kurtosis statistics indicated that scores were within acceptable ranges. Hierarchical regression analyses were used to test how well participants' preexisting political position, normative pressure, and MCPR predicted conformity and attitude change.2 Separate regressions were run for each dependent variable. Political position and MCPR scores were centered and normative pressure was dummy coded (0 = nonunanimous opposition, 1 = unanimous opposition). As recommended by Aiken and West (1991), we entered direct effects (political position, normative pressure, and MCPR) at Step 1, and two-way interactions (political position × MCPR, political position × normative pressure, and MCPR × normative pressure) at Step 2. The three-way interaction was nonsignificant and was therefore omitted from this report.

Conformity

Our first dependent variable, conformity, was measured by how often participants voted with the group. The final regression model explained 26% of the variance in conformity scores, F(6, 402) = 23.21, p < .0005, with significant main effects of political position, normative pressure, and MCPR, and a significant two-way political position × MCPR interaction (see Table 1).

Table 1 Summary of hierarchical regression analysis for variables predicting conformity.

	Step 1			Step 2
Variable	B	SE	β	B	SE	β
Political position (PP)	0.100	0.010	0.471^*^*^*	0.090	0.014	0.424^*^*^*
Normative pressure (NP)	0.062	0.028	0.097^*	0.062	0.028	0.097^*
MCPR	0.113	0.026	0.199^*^*^*	0.124	0.036	0.218^*^*
PP × NP				0.027	0.019	0.089
PP × MCPR				0.047	0.014	0.147^*^*
NP × MCPR				0.025	0.051	0.032

^*p < .05; ^**p < .005;^***p < .001.

Note: NP was coded 0 = nonunanimous (one confederate dissenter out of four) vs. 1 = unanimous (all four confederates opposing participant). PP was scored such that higher scores indicated opposition to gay rights. R² = .23 for Step 1; ρR² = .03 for Step 2; [F(3, 396) = 5.148, p = .002, R² = .26].

Consistent with our first hypothesis, those who were placed into groups where they faced unanimous (M = 0.59, SD = 0.30) as opposed to nonunanimous opposition (M = 0.52, SD = 0.34) toward their position on gay rights showed more public conformity. Also, consistent with our second hypothesis, those who started the study with more anti-gay rights (M = 0.72, SD = .024) attitudes tended to show more public conformity than those who were more pro-gay rights (M = 0.36, SD = 0.30).

Consistent with hypothesis 3a, higher MCPR scores were associated with increased conformity. Contrary to hypothesis 3b, the interaction of normative pressure and MCPR was nonsignificant. However, results of the exploratory analysis revealed a two-way political position × MCPR interaction. To investigate the significant interaction further we decomposed it for interpretation and performed simple slopes tests (see Aiken & West, 1991). Values of the interaction term were plotted at ± 1 SD and the mean were substituted in a simple linear regression equation. These were then plotted to display the interaction (refer to Figure 1). Simple slopes tests revealed that individuals who were more anti-gay rights (+1 SD) showed greater conformity when high in MCPR than when low (b = .195, t = 6.76, p < .01) whereas those who were more pro-gay rights showed no difference in conformity whether high or low in MCPR (b = .054, t(396) = 1.56, p = .120).

Figure 1 Interaction of political position and MCPR on Conformity.Note: On the Y axis, scores represent an average proportion of conformity. Higher scores represent greater proportions of conformity.

Figure 2 Interaction of political position and MCPR on attitude change.Note: On the Y axis, scores could range from − 6 to 6. A negative Direction of Attitude Change score represents an increase in anti-gay rights attitudes from the PPI to the postdiscussion. A positive Direction of Attitude Change score represented change from being anti-gay rights on the PPI to being pro-gay rights postdiscussion. A Direction of Attitude Change score of 0 represents no change in gay rights attitudes from PPI to postdiscussion.

Attitude change

Attitude change was assessed by computing a difference score between participants' screening survey responses to gay rights policies and their response to the same items completed privately postdiscussion. For the analysis of attitude change, the effects of any variable besides participants' initial position were expected to take the form of interactions because pro- and anti-gay rights participants were expected to change their attitudes in different directions.

The final regression model for attitude change explained 28% of the variance in attitude change scores, F(6, 402) = 26.50, p < .0005, with a significant main effect of political position and a significant political position × MCPR interaction (refer to Table 2).

Table 2 Hierarchical regression analysis for variables predicting attitude change.

	Step 1			Step 2
Variable	B	SE	β	B	SE	β
Political position (PP)	0.528	0.045	0.513^*^*^*	0.447	0.065	0.435^*^*^*
Normative pressure (NP)	0.163	0.133	0.053	0.168	0.132	0.054
MCPR	− 0.044	0.121	− 0.016	− 0.189	0.170	− 0.069
PP × NP				0.176	0.090	0.121^†
PP × MCPR				0.148	0.068	0.095^*
NP × MCPR				0.278	0.244	0.070

^†p < .07; ^*p < .05; ^***p < .001.

Note: NP was coded 0 = nonunanimous (one confederate dissenter out of four) vs. 1 = unanimous (all four confederates opposing participant). PP was scored such that higher scores indicated opposition to gay rights. R² = .27 for Step 1; ρR² = .02 for Step 2; [F(3, 392) = 26.496, p = .004, R² = .29]; Step 3 was not a significant addition to the model, and was thus deleted [F(1, 391) = 0.38 p = .538].

Contrary to our first hypothesis, there was no significant interaction of normative pressure and political position in predicting attitude change. Consistent with our second hypothesis, people who were initially against gay rights (M = 1.41, SD = 0.24) showed a larger norm-consistent attitude change than those who were initially pro-gay rights (M = − 0.48, SD = 1.28) after exposure to group norms. However, because those who were pro- and anti-gay rights were expected to change their attitudes in opposite directions, it was necessary to examine differences in absolute attitude change to determine whether the magnitude of change was different for those who were pro- and anti-gay rights. A t-test was conducted to follow-up on the significant effect of initial position to test whether the absolute value of attitude change scores were significantly different for those who were pro- and anti-gay rights. Those who were initially anti-gay rights (M = 1.56, SD = 0.98) showed significantly greater attitude change than those who were initially pro-gay rights (M = 0.99, SD = 0.94), t(397) = 5.90, p < .001.

Results revealed a significant interaction of position and MCPR. We thus plotted values consistent with the recommendation of Aiken and West (1991) and followed up with simple slopes tests. Simple slopes tests revealed that MCPR only moderated attitude change in those who were more pro-gay rights ( − 1 SD) with those both pro-gay rights and high in MCPR becoming slightly more anti-gay rights postdiscussion than their low-MCPR counterparts (b = − .412, t = − 1.978, p < .05). However, for those who were initially more anti-gay rights, their attitude change—while greater overall—was not contingent upon MCPR (b = .034, t = 175, p < .861). The observed interaction pattern partially supports hypothesis 3a in that MCPR led to increased attitude change, but only for those who were pro-gay rights. Contrary to hypothesis 3b, the interaction of normative pressure and MCPR was nonsignificant.

Discussion

Our study demonstrated that an individual's political position, the solidarity of the opposing group norm, and the level of a person's MCPR affects attitude change and conformity to an opposing group norm. People who were initially anti-gay rights conformed more to the group and showed more norm-consistent attitude change than participants who were initially pro-gay rights. Also, participants facing a group with complete solidarity, as opposed a group with a dissenter, showed more conformity. However, the solidarity of the group norm failed to predict whether individuals changed their private attitudes. Finally, participants high in MCPR were more likely to publicly conform than participants low in MCPR if they were anti-gay rights. Yet MCPR only moderated private attitude change in those who were pro-gay rights. Anti-gay rights participants tended to show higher levels of attitude change than pro-gay rights participants regardless of MCPR.

Social norms and intergroup policy attitudes

It has been argued that the effects of social influence on the expression of prejudice are more powerful when the position being advocated is perceived as being consistent with the larger societal norm or consistent with the view becoming the more popular opinion (Clark & Maass, 1990; Glynn et al., 1997; Shamir, 1997). The pro-gay rights position fits these conditions. National polling shows that attitudes toward homosexuals are becoming less negative and support for gay rights initiatives has increased (e.g., 54% support benefits for Federal worker same-sex spouses, 61% support adoption rights; see Gallup.com). However, a 2013 article on Gallup.com noted that the majority (63%) feels that the nation is opposed to gay rights initiatives (Jones, 2013). Meaning, there isn't much evidence to support that our participants' perception of national trends enabled them to stand up to intolerance. Still the possibility that our findings are a matter of perceived trends remains to be seen.

Although the administration of this study occurred before Section 3 of the “Defense of Marriage Act” was declared unconstitutional by the United States Supreme Court, such significant changes in institutional norms might cause a shift in individual beliefs. As was observed in Mexico City, once a comprehensive antismoking law was set in place—thus shifting the institutional norm—individual attitudes toward smoking followed suit (Thrasher, Pérez-Hernàndez, Swayampakala, Arillo-Santillàn, & Bottai, 2010). However, the extent to which significant policy changes or trends in national opinions might promote resistance of intolerance, or vice versa, has yet to be determined.

It may also be the case that there is a tolerant social norm effect—in which the tolerant norm exerts more influence than the intolerant norm. The existence of a tolerant norm effect may help explain why US trends have moved toward tolerance over the last several decades, as those endorsing egalitarian values largely hold their ground while those opposed concede. It would also explain why Monteith et al. (1996) similarly found tolerant norms to be more persuasive than intolerant norms even at a time where there was not national endorsement of gay rights. It is worth exploring whether and why a tolerant norm effect might exist. If it exists, one would expect that this pattern—of those egalitarian resisting prejudicial norms and those higher in prejudice conforming to tolerant norms—would extend across many intergroup issues (e.g., immigrant rights, women's rights).

MCPR and the expression of prejudice

In terms of public expressions of policy attitudes, individuals who were high in MCPR showed more conformity than those who were low in MCPR, but only those who were anti-gay rights. When discussing gay rights in a group that is largely pro-gay rights, people seem to anticipate that opposing gay rights will be viewed as prejudiced. Those high in MCPR adjust their public responses accordingly. This finding lends predictive validity to the MCPR measure, which proved sensitive enough to predict how certain people behaved in a social situation where an intergroup issue was being discussed. Furthermore, it indicates that the discussion of gay rights is a discussion of prejudice—or is at least perceived as such by those sensitive to appearing biased—despite arguments that opposition to gay rights is not about bias (Ball, 2010).

When it came to private attitudes, MCPR only mattered for those who were initially pro-gay rights. For those who were anti-gay rights, the power of the tolerant social norm alone—regardless of MCPR—triggered attitude change. In contrast, those who were pro-gay rights were less likely to change their attitudes overall. It was only those who were pro-gay rights and high in MCPR went along with prejudicial norms. True to their self-reports of only suppressing prejudice when norms dictate that is the appropriate response, they changed their private attitudes to match the local intolerant norm (i.e., the discussant group). Those high in MCPR may have felt inclined to report egalitarian views in the initial survey but less so after discovering that the group was against gay rights. It is as if they are the “fair weather friends” of gay rights. As such, MCPR may entail a certain degree of vigilance toward social norms about prejudice. This type of flexibility would undoubtedly be useful in avoiding violations of norms in social situations that condemn prejudice.

The results of our study suggest that when it comes to norms that dictate the acceptability of prejudice, people do not always blindly follow. Rather, the tendency to yield to the majority on these issues is partly guided by how preoccupied people are with abiding by what social norms deem appropriate. Individuals who report egalitarian beliefs, but who are not necessarily concerned about what society views as appropriate, have attitudes that hold up best to opposition.

Limitations and future directions

One limitation of the present research is that our study was conducted at Mississippi State University, which is perceived to have a state-level norm that is largely anti-gay rights. This may affect the generalizability of our results. Although the anti-gay rights position was only a small majority (54.8%) in our sample, it would be interesting to conduct this experiment in a political climate where those who were pro-gay rights were in the overwhelming majority. In our study, pro-gay rights individuals may have more easily adhered to their beliefs because they are accustomed to perceiving themselves as the minority voice, and have thus developed counter-arguments to inoculate themselves against persuasion attempts (see McGuire, 1964).

Conclusion

The findings of this study expand upon previous knowledge in several ways. First, they provide insight about the ways in the public expressions of prejudice and privately held prejudicial beliefs can be altered through social influence. Specifically, we found evidence that not all norms are created equal; pro-gay rights norms appear to initiate more norm-consistent behavior and attitude change in people than anti-gay rights norms. Second, studies on MCPR have yet to establish that it is predictive of both attitudes and behaviors. Our study is, to our knowledge, the first to show that MCPR moderates people's actual behavior in a situation that dictates the acceptability of prejudice. We found that not all individuals respond to social norm pressures equally. Individuals highly concerned with not appearing prejudiced were particularly susceptible to conforming to normative information about gay rights, and in certain circumstances, changing their privately held attitudes. In contrast, those less concerned with appearing prejudiced were more willing to openly express and resist changing their opinions about gay rights.

Notes

^1. Separate regression analyses were run for voting sets 1 and 2, as voting set 1 could be considered a measure of pure conformity whereas voting set 2 could be said to be an index of persuasion due to the arguments voiced during the discussion. However, the regression analyses were identical in terms of the significant effects that emerged therefore conformity total was used as the final dependent variable.

^2. Because any differences in conformity or attitude change between those who were pro- and anti-gay rights could be potentially attributed to differences in attitude extremity, a t-test was conducted to test for any differences in attitude extremity between those who were pro- and anti-gay rights. Those who were initially pro-gay rights (M = 5.58, SD = 0.59) were not significantly different in attitude extremity than those who were initially anti-gay rights (M = 5.57, SD = 0.63), t(415) = − 0.193, p = .479.