Playing the Political Correctness Game

Two experiments explored the ramifications of endorsing color blindness as a strategy for appearing unprejudiced. In Study 1, Whites proved adept at categorizing faces on the basis of race, but understated their ability to do so. In Study 2, Whites playing the Political Correctness Game—a matching task that requires describing other individuals—were less likely to use race as a descriptor when paired with a Black partner than when paired with a White partner, a strategy that impaired communication and performance. In addition, avoidance of race was associated with Whites making less eye contact with and appearing less friendly toward Black partners. In many conversations, people are called upon to describe others solely on the basis of their appearance: Who was it that showed up late for the meeting? Which character in the movie said that? The new guy—what does he look like again? In such cases, the best information to use is that which is most diagnostic in distinguishing the target from other individuals—for example, hair color, height, gender, and race. We propose (and anecdotal experience suggests) that people vary in their willingness to use certain descriptors: Although having red hair and being Black might be equally diagnostic in a particular setting, we suggest that many individuals are more reluctant to use race than hair color in their descriptions. The present investigation explored how White individuals behave in situations in which the desire to appear unprejudiced leads to efforts to appear color-blind. Why would simply mentioning someone’s race serve as evidence of bias? There is, after all, nothing inherently racist about noticing race. But in a culture where motivations to avoid appearing prejudiced are increasingly pervasive (Gaertner & Dovidio, 1986; Plant & Devine, 1998) and few labels are as aversive as that of ‘‘racist’’ (Crandall, Eshleman, & O’Brien, 2002; Sommers & Norton, 2006), color blindness can serve as a useful stratagem: If I do not notice race, then I cannot be a racist. Certainly, noticing race is a necessary precursor to racism, but we propose that noticing race can be perceived as a sufficient indication of racism: People who do not notice race are not racist, whereas those who do notice race probably are. Indeed, noticing race does lead to the activation of stereotypic associations (Devine, 1989; Greenwald & Banaji, 1995), which can in turn lead to prejudicial behavior (Correll, Park, Judd, & Wittenbrink, 2002; Dovidio, Kawakami, & Gaertner, 2002). Thus, one mechanism for decreasing bias could be simply not to notice someone’s race, thereby forestalling the associations that result in prejudicial behavior. In most cases, however, perceivers do encode the race of other people relatively effortlessly and rapidly (Ito & Urland, 2003; Montepare & Opeyo, 2002; see Cosmides, Tooby, & Kurzban, 2003), and such information does affect judgments, though decision makers are reluctant to acknowledge this influence (Norton, Vandello, & Darley, 2004; Sommers & Norton, in press). We propose that the incongruity between trying to appear color-blind while automatically noticing color complicates strategic efforts to appear unbiased, creating an inevitable tension between efforts to achieve color blindness and actual success at doing so. The present research constitutes an empirical investigation of Whites’ efforts to fulfill their desire to appear unprejudiced by attempting to be color-blind. In Study 1, we examined Whites’ reluctance to admit to their facility at categorizing other people on the basis of race. In Study 2, we examined the consequences of this reluctance by introducing the Political Correctness Game, a two-player photograph-identification task that contrasts efficiency of communication with the desire to appear colorblind. Address correspondence to Michael I. Norton, Harvard Business School, Soldiers Field Rd., Boston, MA 02163, e-mail: mnorton@ hbs.edu, or to Samuel Sommers, Department of Psychology, Tufts University, 490 Boston Ave., Medford, MA 02155, e-mail: sam.sommers@ tufts.edu. PSYCHOLOGICAL SCIENCE Volume 17—Number 11 949 Copyright r 2006 Association for Psychological Science STUDY 1: CLAIMING COLOR BLINDNESS Though research suggests that individuals exhibit an own-race bias in memory for faces (e.g., Malpass & Kravitz, 1969), merely categorizing faces by race is a simpler task (see Stangor, Lynch, Duan, & Glass, 1992). In this study, we sought to demonstrate that Whites’ reports of difficulty in classifying other people on the basis of race are not borne out by their actual ability to do so. Method Fifty-seven White students (36 females) participated in exchange for $10. Photographs of volunteers were taken at a mall. We created a set of 24 photographs for each of seven categorical dimensions: race (Black/White), gender (male/female), age (over 30/under 25), color of the background in the photo (blue/ red), hair color (light/dark), facial expression (smiling/not smiling), and facial hair (present/absent). In the sorting task, participants completed seven blocks of a computer task; in each block, they categorized 24 photos on one of the seven dimensions. When a photo appeared, participants used the mouse to click on the correct classification (e.g., ‘‘White’’ or ‘‘Black’’ when the photos were to be categorized by race). In the hypothetical task, participants read a questionnaire that included sample photos. The questionnaire asked them to imagine themselves performing the computer-based sorting task. Participants ranked the seven dimensions in terms of how quickly they believed they would be able to categorize the photos on the basis of that dimension. Results and Discussion Participants in the sorting task were at least 95% accurate in their categorizations across all seven dimensions, and there was no difference in accuracy across dimensions, F(6, 114) 5 1.65, n.s.; accuracy for the race categorization was 99.1%. In order to compare results across the two tasks, we converted response times in the sorting task to rank scores from 1 through 7, similar to the self-reported ranks from the hypothetical task. Participants in the sorting task were quickest to categorize faces by background color; categorizing by gender came next, followed by categorizing by race (M 5 3.60). In the hypothetical task, in contrast, participants estimated that race would be their second slowest dimension (M 5 4.43), ranking only age lower. Whites thus underestimated the speed with which they would be able to categorize by race, t(47.3) 5 2.07, prep 5 .93, d 5 0.55, via Welch’s separate-variance t test. The only other dimension to vary by task was gender: Participants overestimated their speed in categorizing by gender (M 5 2.19 in the hypothetical task) compared with their actual performance (M 5 3.05), t(31.4) 5 2.09, prep 5 .92, d5 0.60. In essence, participants substituted a less controversial dimension—gender—for a more controversial one, race (Rodin, Price, Bryson, & Sanchez, 1990). Of course, it is possible that individuals are simply generally unaware of their facility at racial categorization. We addressed this alternative explanation by conducting a follow-up study with Black individuals (N 5 22). Although Black participants’ ability to categorize by race (M 5 3.67) was comparable to that of Whites in Study 1, t < 1, their estimates of this ability (M 5 2.78) were significantly higher than Whites’, t(36.7) 5 3.68, prep 5 .99, d 5 1.04. Rather than supporting a general tendency for individuals to lack awareness of their facility with racial categorization, these data support the conjecture that Whites’ underestimation of this ability results from a specific desire— one not shared by our Black participants—to appear unprejudiced. Finally, the fact that no underestimation emerged in Whites’ reports of ability to categorize by background color (red/ blue) suggests that it is not color itself but rather the meaning of color when associated with ethnicity that led to the observed discrepancies among Whites. STUDY 2: THE POLITICAL CORRECTNESS GAME Having demonstrated Whites’ reluctance to admit the extent to which they use race to differentiate other people, we examined some of the possible consequences of this reluctance. In Study 2, White participants engaged in a two-player game (see Weber & Camerer, 2003) in which they asked a partner questions in order to identify a target face in a set of photos that varied on the same dimensions used in Study 1. We manipulated whether participants worked with a White or Black confederate, expecting participants to be particularly likely to underutilize race as a descriptor when interacting with a Black partner because of increased pressure to appear unprejudiced. We further expected this avoidance of race to impede communication, resulting in less efficient performance on the task. We videotaped the sessions and coded participants’ behavior to capture the interpersonal costs of efforts to appear color-blind. Method Thirty White students (20 females) participated in exchange for $10. Participants were randomly assigned to work with either a White or a Black confederate. Individuals arrived at the lab expecting to participate in a study with a partner, who was in fact a confederate. The confederate arrived 2 min after the participant, at which point a White male experimenter explained the two roles for the study: questioner and answerer. An ostensibly random drawing was rigged so that the participant became the questioner and the confederate the answerer. The questioner received an array of 32 photographs from Study 1 that varied orthogonally on three Confederates were female undergraduates instructed to answer only ‘‘yes’’ or ‘‘no.’’ In a pretest (N 5 16), participants’ ratings of their partner’s performance were similar for Black and White confederates (Ms 5 8.25 and 8.63), t 5 1.21, n.s. 950 Volume 17—Number 11 The Political Correctness Game