Confronting prejudice: The who, what, and why of confrontation effectiveness

Abstract

Following a task that ostensibly revealed implicit racism or sexism, participants were confronted about their bias. The confronter's race and gender were manipulated to test whether group membership (whether one is part of the group experiencing prejudice) determined confrontation effectiveness. Results showed that the confrontation of racism met with greater acceptance when performed by a White than a Black confronter, indicating that group membership drives confrontation impact. Further, tests of two possible mediational accounts showed mediation by perceptions of the Black confronter as a complainer. Participants trivialized the confrontation concerning sexism, underscoring the need to strengthen social norms against sexism. Implications for engaging in confrontation are discussed.

Keywords:

• Prejudice reduction
• Confrontation
• Racism
• Sexism

Confrontation of prejudice is often envisaged as a way to wield influence over injustices otherwise resulting from unchecked prejudice. Especially in today's society, where even egalitarian-minded people may engage in subtle discriminatory responses (e.g., Dovidio, Kawakami, Johnson, Johnson, & Howard, 1997), and many people are externally motivated to control expressions of prejudice (Plant & Devine, 1998), confrontation may be a useful strategy for curbing discriminatory behaviors.

Members of stereotyped groups experience everyday discrimination and have ample opportunity for confrontations (Swim, Hyers, Cohen, & Ferguson, 2001; Swim, Hyers, Cohen, Fitzgerald, & Bylsma, 2003). Some person-centered factors increase the likelihood that targets of bias will confront, such as optimism (Kaiser & Miller, 2004; Wellman, Czopp, & Geers, 2009) and the general belief that others can change (Rattan & Dweck, 2010). Also, targets are more likely to confront in certain situations, such as private contexts (Stangor, Swim, Van Allen, & Sechrist, 2002), and with low costs of confronting (Shelton & Stewart, 2004). However, by and large, targets do not confront people about their biased responses, often fearing possible negative consequences (Shelton, Richeson, Salvatore, & Hill, 2006; Swim & Hyers, 1999; Woodzicka & LaFrance, 2001).

Opportunities for confrontation also arise for “non-targets,” or people who observe instances of bias but are not members of the targeted group. However, non-targets overestimate how much distress they will experience when observing racist behavior, instead failing to confront and even showing social acceptance of the perpetrator of bias (Kawakami, Dunn, Karmali, & Dovidio, 2009). Like targets, non-targets may be reluctant to confront others’ bias because of possible negative consequences, such as being disliked (Czopp, Monteith, & Mark, 2006; but see Mallett & Wagner, 2011).

Are targets’ and non-targets’ concerns about negative outcomes of confrontation equally valid, or are confrontations more likely to be received positively when they are performed by certain confronters? The present research focuses on understanding reactions to confrontations of racism and sexism depending on who does the confronting. We were interested in confronter group membership; specifically whether targets (e.g., a Black person confronting racism) or non-targets (e.g., a White person confronting racism) are associated with more favorable reactions, and if so, why. Unique from past work, our research used non-interpersonal confrontation, in which participants were confronted through written materials. Everyday examples of such contexts for confrontation abound (e.g., newspaper editorials and online communications). Also unique from past work, participants were confronted about their implicit biases and associated discriminatory responses. Such confrontations are evident, for example, in educational exercises where students may take an implicit association test (IAT) and learn about their subtle, unconscious bias, along with the likelihood that their biases are linked to discriminatory treatment of others (see Morris & Ashburn-Nardo, 2010). Are people who are confronted with such messages convinced that they hold subtle biases and that they should try to change them?

The impact of group membership and type of prejudice

Only a few studies have compared the effectiveness of non-target versus target confrontations. Czopp and Monteith (2003) had White participants imagine scenarios in which they engaged in prejudiced responses and were confronted by either targets or non-targets. Participants were more responsive to non-target (i.e., White) than target (i.e., Black) confrontations, reporting more guilt and self-disappointment. Rasinski and Czopp (2010) had White participants watch a video debate between two students, and manipulated whether a White or Black speaker confronted a White speaker about a racially biased remark. Participants were more persuaded by the video with a White confronter than a Black confronter, and they perceived the Black confronter as relatively rude.

Previous research has not addressed why confronter group membership may be an important determinant of confrontation effectiveness. We tested two possible reasons why non-targets may be more influential confronters than targets. First, past research has shown that targets are readily pegged as complainers and hypersensitive when they point out discrimination (Crosby, 1993; Feagin & Sikes, 1994; Kaiser & Miller, 2001; Swim & Hyers, 1999). Thus the tendency to regard targets that make accusations of bias as complainers may undermine their ability to be effective confronters of prejudice, relative to non-target confronters. Alternatively, differential expectancies about who will speak out against prejudice may play a role. Classic persuasion research showed that communicators promoting a message that does not advance their own interests are surprising and their messages are viewed as more valid, compared to those who argue in favor of their own interests (Eagly, Wood, & Chaiken, 1978; Petty, Fleming, Priester, & Feinstein, 2001; Walster, Aronson, & Abrahmson, 1966). Thus non-targets may be more effective confronters than targets because they are not perceived as having a vested interest in curbing prejudice, so people do not expect them to confront, and their messages are taken more seriously.

Additionally, the present research examined confrontations of both racism and sexism. Contemporary norms surrounding racism suggest that it is a more serious problem than sexism (see Glick & Fiske, 1996, 2001; Rodin, Price, Bryson, & Sanchez, 1990). Indeed, in Czopp and Monteith's (2003) research participants were more amused when they imagined being confronted about sexism, relative to being confronted about racism. Simply put, people seem to care less about possible sexism, especially more subtle forms of sexism. For example, Saunders and Senn (2009) found that men who imagined they were confronted about subtle gender-biased behavior (references to women doing the cooking and needing to keep in shape for men) felt amused and that the confrontation was “laughable.” In the present experiment we examined an actual confrontation of implicit sexism, and we suspected that participants would be relatively unresponsive to the confrontation, regardless of characteristics of the confronter.

Experiment overview

Participants completed a study supposedly sponsored by a researcher at another college, and the gender and race of this researcher were manipulated. Participants performed a computer task that supposedly revealed their implicit racism or sexism, and then they were confronted about their bias with an essay written by the researcher. The essay underscored that “you” (i.e., directed at participants) are responsible for perpetuating subtle racism/sexism, and “you” should try to reduce “your” prejudice. Effectiveness of the confrontation was assessed in terms of acceptance of the confrontational message (e.g., as being valid and applying to the self).

Our design enabled us to test reactions to confrontations made by confronters who varied in group membership. For instance, in the case of confronting racism, the confrontation was made by a Black researcher (either male or female), or a White researcher (either male or female). If group membership is the important factor in driving reactions to confrontations, Black confronters should be less effective than White confronters.

We expected to find support for this group membership hypothesis in the case of racism. Specifically, we expected that the White researcher's confrontation would engender more acceptance than the Black researcher's confrontation. We also expected that the Black confronter would be viewed as more of a complainer than the White confronter, and that the Black researcher's confrontation would be perceived as more consistent with participants’ expectations than the White researcher's confrontation. Importantly we were able to test whether impressions of the confronter as a complainer or consistency with expectations played a mediating role in the effect of confronter race on acceptance of the confrontation and, if so, which was a stronger mediator.

In contrast to racism, we anticipated that participants would not care much about confrontations of implicit sexism and associated sexist responses, regardless of whether a target (female) or non-target (male) performed the confrontation. This result would be consistent with prior work suggesting that sexism elicits less concern than racism (e.g., Glick & Fiske, 1996, 2001).

Method

Participants

Introduction to Psychology students received course credit in return for their participation. Two participants disallowed the use of their data. Data from three Black participants who ended up in the racism condition were also deleted because Blacks may well react differently to confrontations of racism than non-Blacks, but we had too few Black participants to examine this possibility. This left 193 participants (56% male; 87% White; 6.7% Asian/Pacific Islander; 1.6% Latino; 3.6% other).1

Design

The experiment used a 2 (type of prejudice: racism or sexism) × 2 (confronter race: White or Black) × 2 (confronter gender: female or male) × 2 (participant gender) between-participants design.

Procedure

Participants in groups of up to four were told that the study was being conducted for another university that did not have a participant pool. They learned that a researcher from Casper College (a college unknown to participants) had developed an “association judgment” task and had written an article based on the preliminary findings. Participants were informed that they would complete this task, read the article, and then provide feedback to be passed along to the researcher.

The first part of the confrontation was the association judgment task. This was a mock lexical decision task in which primes (presented for 150 ms) were followed by a string of letters and participants indicated whether each string of letters was a word or non-word with a keystroke. Primes were photos of males or females (face only) who were either White or Black. Words were presented on half of the 96 trials, and these words were related to intelligence on 24 trials. Interest centered not on participants’ performance but rather on confronting them with their supposedly subtle stereotypic biases. Specifically, after the last trial of the lexical decision task, a screen appeared that explained the general logic behind the task.

Participants read that potentially unconscious stereotypic associations were measured in the task. Participants were informed that stereotypes about Blacks (racism condition)/females (sexism condition) being less intelligent than Whites/males were measured. Then participants saw (bogus) summary statistics showing that they took 409 ms on average to identify intelligence-related words when they followed Black faces (racism condition)/female faces (sexism condition), and 205 ms on average to identify such words when they followed White/male faces. Participants also read that this meant they unconsciously viewed Whites/males as more intelligent than Blacks/females.

The second component of the confrontation was the article that participants were then given, which was ostensibly written by the researcher. The article was accompanied by a color photo to manipulate race and gender of the confronter. The one-page, single-spaced article entitled “Are You Part of the Problem?” emphasized that (1) racial/sexist biases assessed in the task can exist even without conscious stereotype endorsement because of common portrayals of Blacks/females and social learning, (2) these biases can lead people “like you” to discriminate in often subtle but pervasive and damaging ways, and (3) if “you are part of this problem,” you should create positive change by trying to change your own biases.

Participants were then asked to provide reactions to be passed on to the professor, providing a context for the completion of the dependent measures. Participants were then probed for suspicion (no one voiced suspicion), debriefed, and dismissed.

Dependent measures

Acceptance

A total of 11 items tapped into the extent to which participants accepted the confrontation as accurate, true of them, and something to which they needed to attend. Example items are: To what extent do you agree with the content of the article? To what extent do you think the article pertains to your own behavior and reactions? How much thought will you give to the topic of the article in the future? Do you believe that you associate, perhaps unconsciously, certain groups of people with intelligence more than you associate other groups of people with intelligence? To what extent do you think you need to watch yourself in the future so that you won’t be biased by stereotypes? Ratings were made on 1 (not at all) to 7 (very much) scales. Factor analysis of the 11 items yielded two factors, one with items focused more on the article (e.g., agreement) and one with items focused more on the self (e.g., To what extent do you think the article pertained to your own behavior and reactions?). However, analyses of the two factors yielded the same pattern of results. Thus, a single acceptance index was formed (α = .90).

Perceptions of the confronter

Five items (among fillers) assessed the extent to which participants perceived the confronter as a complainer (e.g., whiner, complainer), α = .82, with 7-point rating scales.

Expectations

The items measuring consistency with expectations were “How consistent with your expectations were the researcher's arguments in the article?” and “How surprised were you that the researcher chose to research this particular topic?” (reverse-scored) on 1 (not at all) to 7 (very much). Although these items did not correlate strongly (r = .15, p < .04), analyses of each yielded the same pattern of results and so results averaging across the items are reported.

Results

The dependent measures were initially analyzed in a series of 2 (type of prejudice) × 2 (confronter race) × 2 (confronter gender) × 2 (participant gender) between-participant ANOVAs. We strove to have equal numbers of participants per condition when assigning participants, so that we had approximately 12 participants per cell. We found that gender was not systematically associated with patterns of results. Thus, we collapse across gender in analyses reported here.

Acceptance2

The analysis of the acceptance index revealed only one significant effect in the form of an interaction between type of prejudice and confronter race, F(1, 184) = 5.35, p < .03, = .03. As shown in Table 1, participants in the racism condition showed greater acceptance of the confrontation message when the ostensible researcher claiming racial bias was White than when he was Black, t(90) = 2.04, p < .05, r = .20. In contrast, confronter gender was not associated with any significant effects, Fs < 1.50, ps > .23. Participants in the sexism condition had acceptance scores that were similar to participants in the Black confronter–racism condition, t(143) = .41, p = .68, r = .04, and significantly lower than participants in the White confronter–racism condition, t(145) = 1.95, p = .05, r = .17. This pattern suggests that participants were relatively unmoved by the confrontation of sexism, a finding that is in keeping with the idea that people are rather unbothered by sexism (Czopp & Monteith, 2003; Fiske & Stevens, 1993).

Table 1. Mean scores on dependent variables as a function of type of prejudice and confronter race

	Type of prejudice
	Racism		Sexism
	Confronter race
Measure	White	Black	White	Black
Acceptance	4.45a (1.10)	3.99b (1.09)	3.92 (1.15)	4.19 (1.09)
Complainer	3.42a (1.12)	4.22b (1.20)	3.70 (1.33)	3.56 (1.41)
Expectations	4.51a (1.17)	5.19b (1.10)	4.42 (1.27)	4 .41 (1.06)
Within each type of prejudice condition, confronter race means differ significantly from each other p < .05 at least if they do not share a common subscript.

Impressions of confronter

Analysis of the extent to which participants perceived the researcher as a complainer revealed a significant interaction between type of prejudice and confronter race, F(1, 185) = 7.59, p < .01, = .04. As shown in Table 1, participants in the racism condition reported a significantly stronger impression of the researcher as a complainer when the researcher was Black rather than White, t(91) = 3.32, p < .001, r = .33. This result again supports the idea that confronter ethnic group membership is the important determinant of reactions to confrontations of racism. In contrast, the effect of gender did not approach significance in the sexism condition (p > .56). Indeed, participants in the sexism condition did not have particularly strong impressions of the confronter as a complainer. Complainer scores in the sexism condition were comparable to scores in the White confronter racism condition, t(146) = .97, p = .33, r = .08, and they were significantly lower compared to the Black confronter, racism condition, t(143) = 2.70, p < .01, r = .24. Together with the acceptance results, this pattern suggests that participants were rather unaffected by confrontations of sexism.

Expectations

A significant main effect for type of prejudice was found, F(1, 185) = 6.60, p < .02, = .03, such that participants in the racism condition reported that the researcher's arguments were significantly more consistent with their expectations (M = 4.84. SD = 1.18) than participants in the sexism condition (M = 4.42, SD = 1.16). A significant main effect for confronter race was also obtained, with higher expectation scores in the Black (M = 4.77, SD = 1.22) than the White (M = 4.47, SD = 1.14) confronter condition, F(1, 185) = 3.99, p < .05, = .02. These main effects were qualified by a significant interaction between type of prejudice and confronter race, F(1, 185) = 4.34, p < .04, = .02. As shown in Table 1, participants in the racism condition reported that the researchers’ arguments about racial bias were significantly more consistent with their expectations in the Black confronter condition than in the White confronter condition, t(91) = 2.87, p < .01, r = .29. The overall pattern in the sexism condition indicated that being confronted about sexism was not perceived as strongly consistent with participants’ expectations. When participants in the sexism condition were compared to participants in the White confronter–racism condition, expectancy scores were comparable, t(146) = .44, p = .66, r = .04. In contrast, being confronted about racism by a Black confronter was much more consistent with participants’ expectations than being confronted about sexism, t(143) = 3.76, p < .001, r = .32.

Tests of mediation in the racism condition

The consistent pattern across the acceptance, complainer, and expectancy variables was that confronter race influenced acceptance but only in the racism condition. Thus we performed a multiple mediational test in the racism condition only to determine whether complaining or expectancy mediated the effect of confronter race on acceptance. The bootstrapping approach (Preacher & Hayes, 2008) was used with a total of 5000 resamples and 95% confidence intervals (i.e., establishing that the indirect effects were significant at p < .05, two-tailed). Using complainer and expectancy as the mediators, we found that the indirect effects were significant, complainer_ab = −.45, SE = .144, 95% CI [−.735, −.221], expectancy_ab = −.16, SE = .078, 95% CI [.039, .370], and the direct effect was reduced to non-significance, b = −.18, SE = .197, t(91) = −.93, p > .35 (see Figure 1). Examination of the pairwise contrasts of the indirect effects demonstrated that the specific indirect effect through complainer was larger than the specific indirect effect through expectancy, with a CI of −.950 to −.332 (this is the Bias-Corrected and Accelerated confidence interval, as recommended by Preacher & Hayes, 2008). Altogether, the mediational tests suggest that greater perceptions of complaining, above that of expectancy, mediate the effect of race on acceptance.

Figure 1. The relation between type of confronter and acceptance mediated by complainer and expectancy in the racism condition. All coefficients are significant at p < .05 or less except the direct effect (c′) of type of confronter on persuasion when the mediators are included in the model.

Discussion

The present research moves beyond past confrontation findings to address who is likely to be an effective confronter of what form of prejudice and why. Our results showed that people are more likely to accept confrontations of racism as convincing and suggestive of their personal need to work on bias reduction when the confrontations are performed by Whites rather than Blacks. Further, consistent with findings that targets who make claims of discrimination are perceived as complainers, even when such claims are perfectly legitimate (Garcia, Reser, Amo, Redersdorff, & Branscombe 2005; Kaiser & Miller, 2001), we found that participants’ viewed Black confronters more as complainers than White confronters. Finally, our mediational tests indicated that Black confronters produced less acceptance of the confrontation primarily because of participants’ impressions of them as complainers.

Participants appeared to trivialize the confrontation of sexism and were not very persuaded, even by the male confronter. Participants appeared to be dismissive of the feedback from the implicit sexism task and the associated confrontational message, and relatively unconcerned about the possibility that they were prone to subtle forms of sexism. These findings add to the persistent finding that people are generally less concerned about sexism than racism (Glick & Fiske, 1996; Rodin et al., 1990; Saunders & Senn, 2009). Although a focus on more blatant sexism rather than the subtle form we examined may yield stronger reactions, we believe that our findings generally speak to the need to strengthen social norms against sexism.

That the patterns of findings in the racism condition were obtained in the context of a non-interpersonal confrontation about subtle prejudice is notable. Non-interpersonal confrontations may occur frequently in day-to-day life, such as in editorials, television talk shows, or online exposures. Also, given growing knowledge of the pervasive presence and impact of implicit prejudices (see Monteith, Woodcock & Lybarger, in press), confrontations of subtle bias are on the rise (e.g., see Project Implicit, https://implicit.harvard.edu/implicit/). Would our findings have been obtained with an actual person-to-person confrontation? We suspect that different processes work in such confrontations, such as feeling especially contrite when the confronter is (or is believed to be) physically present (Czopp et al., 2006), and the operation of social constraints such as strong norms to be liked (Mallett & Wagner, 2011). Future research will be needed to address this issue empirically by comparing non-interpersonal and interpersonal confrontations directly.

Who should confront prejudice?

The confronter group membership results of our investigation can be discussed in the context of whether certain people should be encouraged to confront bias more than others. Previous research has indicated that targets are hesitant to confront discrimination (Shelton & Stewart, 2004; Stangor et al., 2002; Swim & Hyers, 1999). The anxieties that fuel targets’ caution include concerns about social pressures to be polite, retaliation, and the possibility of being disbelieved (e.g., Swim & Hyers, 1999). Clearly situations that present the opportunity for targets to confront prejudice and discrimination can create much angst. The present research provides justification for targets’ concerns, indicating that Blacks who confront prejudice are more likely to be perceived as complainers than Whites, which leads to less acceptance of the confrontation.

We thus might conclude that the business of confronting prejudice is best left to willing members of non-stigmatized groups, but this solution poses obvious problems. For example, non-stigmatized group members often will not be aware of the everyday discrimination experienced by members of stigmatized groups (e.g., Kessler, Mickelson, & Williams, 1999), and failure to confront prejudice will only reinforce the status quo. Rather than discouraging targets from confronting bias, research is needed to determine how best to minimize the likelihood that they will be perceived as complainers. Also important, targets may well be effective confronters in some circumstances.

The current findings do suggest that non-target confronters of racism will have advantages over target group members when the confrontation is not taking place in a one-on-one interpersonal context. The findings thus have implications for the strategic use of non-targets, as when crafting public messages that encourage people to recognize and curb their prejudice. Non-targets may fear negative costs of confronting (e.g., Ratner & Miller, 2001), but they may be willing to incur such costs if they know that their confrontations will, in fact, help to curb bias.

How should prejudice be confronted?

Our finding that perceiving the confronter as a complainer was a critical mediator suggests that targets should try to avoid being perceived as complainers. How might this be accomplished? Because group membership apparently signals that the confronter may be “merely complaining,” confrontational messages that downplay target group membership are advisable. Also, confrontation may be more effective to the extent that a target confronter emphasizes a larger superordinate group that includes targets and non-targets (e.g., Americans, college students) (see Gaertner & Dovidio, 2005).

Complaints that are not threatening or infused with negative affect should be met with more constructive and supportive reactions, and complaints will be reacted to more favorably if they are directed at another person or event rather than at the individual (Kowalski, 1996). The individual certainly must be implicated to some extent in confrontations of prejudice, but perhaps focusing on the person's behavior and how it is disturbing to oneself rather than making inferences about the offender's intentions or impugning the person's character will be relatively well received. These considerations may be especially important when confronters wish to preserve the quality of their relationships with those who are confronted. For instance, Czopp et al. (2006) found that confronters were evaluated more favorably when their confrontation emphasized equal treatment and fairness compared to when they were threatening and focused on the offender.

Conclusions

As research that we reviewed earlier in this article has revealed, people are often silent in the face of prejudice rather than confrontational, enduring the costs of remaining silent rather than suffering possible costs of speaking up (Kaiser & Miller, 2004; Shelton & Stewart, 2004). Our hope is that research such as the present work that identifies factors that influence confrontation effectiveness will provide a theoretically grounded understanding of how to maximize confrontation effectiveness. Our findings suggest that Whites may be perceived as legitimately confronting injustices more than Blacks, whereas Blacks are more likely to be perceived as just complaining, and these perceptions undermine confrontation effectiveness. Thus, Whites who challenge others’ biases through confrontation may be relatively powerful agents of change in the case of racism.

Additional information

Notes on contributors

Margo J. Monteith

Authors contributed equally. Author order is alphabetical.

Notes

Authors contributed equally. Author order is alphabetical.

¹ There were not enough non-Black participants to consider ethnicity as a factor in analyses. Results held the same pattern when analyzing only White participants as when non-Blacks were also included.

² One outlier more than 3 SD from the mean of the acceptance index was removed.