Abstract
To date, research has primarily focused on the colorblind norms and behaviors of majority-White participants in majority-White contexts. Extending this work to more diverse samples and contexts, across four studies, we examine whether colorblind norms link to the colorblind behavior of racially diverse participants living in a racially diverse (i.e., heterogeneous) context. Findings suggest that participants living in a racially diverse context did not endorse colorblind beliefs (Study 1) and norms (Study 2), and instead behaved in race-conscious ways and overwhelmingly used race in a photo identification task. Furthermore, in Study 3, we find that colorblind norms are largely activated by the belief that talking about race is prejudiced. When participants were exposed to a social norm that linked talking about race to prejudice, colorblind behavior became more prevalent. Finally, in Study 4, we see that greater diversity of one’s context is correlated to less endorsement of colorblindness.
Within the US, the social norm of colorblindness prevails. The idea that discussing or mentioning race should be avoided is endorsed by many White individuals (Apfelbaum et al., 2012; Plaut, 2010). Despite automatically attending to and encoding race (Cosmides et al., 2003; Ito & Urland, 2003), White individuals within the US may feel social pressure to not “see” race. While colorblindness has been defined in a number of ways in the literature (see Rattan & Ambady, 2013; Rosenthal & Levy, 2010), we focus here on what Neville et al. (2013) call “color evasion”—avoiding race in an interaction context with a focus on similarities rather than differences. Engaging in colorblindness in interracial interactions may be a strategic attempt to, at least in part, appear unbiased (Apfelbaum et al., 2008). Indeed, individuals most concerned about appearing prejudiced are more likely to adopt colorblind norms and behaviors (Apfelbaum et al., 2008; Goff et al., 2013; Karmali et al., 2019; Pauker et al., 2015). Research evidence suggests that for White individuals in majority-White contexts, adopting colorblind norms appears to be driven by a concern that associates talking about race with being prejudiced (Apfelbaum et al., 2008; Pauker et al., 2015). However, colorblind norms may operate differently in racially diverse contexts because talking about race may be decoupled from prejudice, in other words, not talking about race may no longer be strategically motivated. The purpose of the current research is to examine whether norms surrounding talking about race are associated with colorblind behavior in racially diverse contexts.
Colorblind Norms in Racially Diverse Contexts
Research examining how individuals negotiate race-relevant situations has focused primarily on White participants (cf. Kohatsu et al., 2011; Pauker et al., 2015) in majority-White contexts, such as the continental US. For example, past research on colorblind behavior in interaction contexts (Apfelbaum et al., 2008; Norton et al., 2006) has been conducted in Massachusetts, where White people make up over 80% of the population (U.S. Census Bureau, 2019). The current research answers the call to include more racially diverse samples and contexts in research to expand our understanding of the factors that impact intergroup relations (Kitayama, 2017; Rattan & Ambady, 2013). Population projections for the US forecast that Whites will be the numerical minority by 2060, dropping to just 40% of the total population (Colby & Ortman, 2015). Given these shifting population demographics, it is imperative to understand the dynamics of intergroup relations in racially diverse contexts.
For the purposes of this paper, we define racially diverse contexts as having greater racial heterogeneity and specifically more equal proportions across many racial groups. Thus, to extend the study of intergroup relations beyond White participants in majority-White contexts, we examine colorblind norms within two racially diverse contexts where racial minority groups consist of at least 51% of the population and include more even distributions among groups. Specifically, we examine colorblind norms within two states known for their racial diversity: Hawai‘i (Studies 1–4) and California (Study 4). In both of these contexts, racial minorities account for over 60% of the population and have relatively equal proportions among various racial groups.
Although White people in majority-White contexts adopt colorblind norms and display colorblind behavior (Apfelbaum et al., 2008; Goff et al., 2013; Norton et al., 2006; Rattan & Ambady, 2013), research has demonstrated that racial minority individuals in majority-White contexts display more comfort talking about race compared to their White counterparts (Trawalter & Richeson, 2008). Racial minorities have more expertise with race-related discussions and thus may feel more comfortable talking about race in race-related situations. Indeed, developmental work indicates that White parents rarely talk to their children about race (Loyd & Gaither, 2018), whereas parents of racial minority children actively talk about race with their children from a young age (Lesane-Brown et al., 2010). Consequently, we expect that in a more racially diverse environment with larger numbers of racial minority individuals, race could be more readily used in conversations—making everyone more comfortable with discussing race. Thus, the diversity of the context may foster different social norms that support acknowledging and using race (Neville et al., 2014; Rattan & Ambady, 2013), and these social norms may extend to individuals in a context regardless of their own racial background. The goal of this paper is to empirically investigate whether colorblind norms and corresponding behavior in race-relevant situations persist in racially diverse contexts.
Colorblind Behavior in Racial Minorities
One reason for a reduction of colorblind behavior in racially diverse contexts may be that typical racial majority and minority members appear to diverge in how they use race. Many White individuals view colorblindness as an optimal strategy to deal with racial differences. For example, when compared to racial minorities, White individuals are more likely to endorse the power evasion dimension of colorblindness whereby they deny racism by focusing on equal opportunities (Awad et al., 2005; Neville et al., 2000; Offermann et al., 2014; Tawa et al., 2016; Worthington et al., 2008). Furthermore, color evasion can negatively impact minority group members who are on the receiving end of colorblind behavior (Apfelbaum et al., 2012; Neville et al., 2013; Plaut, 2010). For racial minority college students, the colorblind behavior of their White peers led to feelings of frustration, pain, and isolation (Lewis et al., 2000) as well as reduced cognitive functioning (Holoien & Shelton, 2012). Colorblindness can also impact racial minorities in the workplace—the more strongly their White coworkers endorsed colorblindness, the less psychological engagement racial minorities reported at work (Plaut et al., 2009).
Instead of adopting a colorblind approach to race, particularly given potential negative consequences for doing so, racial minority individuals might adopt other strategies for navigating race-relevant situations. For example, in stark contrast to color evasion, racial minorities may be more likely to endorse multicultural approaches to race, such that groups should not only be acknowledged but valued (Ryan et al., 2010; Ryan et al., 2007; Verkuyten, 2005). Extending these research findings to racially diverse contexts, where the proportions of racial groups are more evenly dispersed, we may find that the dominant norm may shift from colorblindness toward one that acknowledges race. Research has found that racial minorities appreciated a valuing diversity approach (e.g., race-conscious) as opposed to an equality approach (e.g., colorblind) when they perceived their context to have a moderate proportion of racial minorities (Apfelbaum et al., 2016). Thus, the goals of the current research are to examine whether social norms regarding talking about race impact colorblind behavior in race-relevant interactions within racially diverse contexts.
The Present Research
We examined the correspondence of colorblind norms and behavior expressed by Asian individuals in a racially diverse context. Specifically, in Study 1 we administered a photo identification task to measure Asian participants’ use of racial labeling and the extent to which their endorsement of colorblindness aligned with their behavior in the task. In Study 2 we aimed to replicate our findings from Study 1 with both White and Asian participants to test the alternative possibility that these behaviors stemmed from racial group membership alone, rather than social norms. Next, in Study 3 we demonstrated that colorblind behaviors are linked to the belief that talking about race is prejudiced, by experimentally manipulating norms that reinforce this belief. Lastly, in Study 4 we tested the possibility that the racial diversity of a context is associated with individuals’ endorsement of colorblindness and their perceptions that talking about race is prejudiced. To do this, we sampled from three locations across the US that differed in their racial diversity in order to investigate these relationships. Together, these studies help to illustrate the link between colorblind norms and subsequent behavior in race-related interactions when in racially diverse contexts.
Study 1
Despite well-meaning intentions to avoid prejudice, when individuals in majority-White contexts employ strategic colorblindness, this has negative psychological consequences for minority individuals (e.g., Plaut et al., 2009; Purdie-Vaughns et al., 2008). Given that colorblindness often does not benefit racial minorities, it is possible that in contexts where racial minorities are better represented, individuals use different strategies for negotiating race and diversity (i.e., multiculturalism or polyculturalism; Ryan et al., 2010; Ryan et al., 2007), resulting in colorblind norms and behaviors being less pervasive. The goal of Study 1 was to examine whether endorsement of colorblindness would extend to corresponding behavior (i.e., talking about race), using a photo identification task (e.g., Apfelbaum et al., 2008) with Asian individuals within the racially diverse context of Hawai‘i. We sampled individuals who identified as Asian (i.e., monoracial East or Southeast Asian) because they represent the largest racial group within this diverse context (the Asian population of Hawai‘i represents about 37%; U.S. Census Bureau, 2019). We anticipate that Asian participants in this diverse context would be more likely to acknowledge race, and their rationale for doing so will emphasize that race is a functional category that helps them complete the task’s objective. We also expect that participants’ endorsement of colorblindness would correspond to their behavior demonstrated in the task.
Method
Participants
We recruited 118 participants from the University of Hawai‘i Psychology Department’s participant pool. The responses of 26 individuals were removed from the study due to not meeting our race preselection criteria (monoracial East or Southeast Asian), and one because their testing session was interrupted by a fire alarm. The final sample included 91 undergraduates who participated in exchange for extra course credit or a $5.00 Starbucks gift card. The sample consisted of East or Southeast Asian undergraduates (53 females) who were 18–48 years old (Mage = 20.75 years, SD = 5.18). The rationale for why one participant chose (not) to use race was not provided due to experimenter error. A sensitivity power analysis using G*Power (Faul et al., 2009) was conducted; with our sample size, we found that with 80% power and α = .05, we would be able to detect an effect of W = 0.35 for a chi-square with df = 3.
Materials and procedure
Participants completed a photo identification task to measure their acknowledgement of race (Apfelbaum et al., 2008). An Asian experimenter welcomed the participant into a quiet room located in the lab and asked them to sit in front of 30 4 × 6 in. photographs of faces arranged in three rows of 10. Participants were told that the goal of the task was to identify a target photo randomly selected by the experimenter by asking as few yes/no questions as possible, that the trial would end once they had correctly identified the target photo, and that they would be asked to complete four trials in total. Photos differed along a range of perceptual cues but varied systematically by race (Black vs. White), gender (female vs. male), and background color (blue vs. red). Thus, asking questions about race, gender, or background color would facilitate task performance by eliminating roughly half of the photos in the array. While the participants familiarized themselves with the array, the experimenter turned on the video camera. After completing all trials, participants were asked to explain why they did or did not use race during the task (i.e., “Why did you choose [not] to use racial labels?”). Participants then moved to a computer cubicle and completed items that assessed whether they personally endorsed colorblindness (adapted from Pauker et al., 2015), followed by a demographic questionnaire.
Measures
Colorblind behavior
Trained research assistants blind to the purpose of the study coded the video recordings for whether participants used race-related terminology (e.g., “African American,” “dark skin,” “White,” “light complexion,” etc.) to identify the target photo in each trial (coded as 0 = no, 1 = yes). Two raters independently coded each video, and a third independent rater resolved discrepancies (Cohen’s κ = 0.88). Responses were summed across the trials (ranging from 0 = did not mention race in any trial, to 4 = mentioned race in every trial) and divided by the total number of trials, resulting in an index of the frequency with which race was mentioned. We analyzed (a) whether a participant used race one or more times (not colorblind behavior) compared to whether they never used race (colorblind behavior), and (b) the frequency of acknowledging race across all trials (where higher scores represent less colorblind behavior).
Rationale for behavior
Two research assistants independently coded participants’ rationale for why they did or did not use race during the task. Coding discrepancies were resolved independently by a third rater (Cohen’s κ = 0.86). Previous research found that individuals provide either task- or social-focused reasons for (not) using race during the task (see Pauker et al., 2015). Building on this scheme, responses were coded as aligning with one of four strategies. For two of the strategies, participants provided task-focused reasoning, which indicated that race was acknowledged because this dimension was (1) functional and a good strategy to use (e.g., “it was a faster way to identify different pictures,” “it helped narrow it down”), or (2) perceptually salient and apparent (e.g., “visually easy to identify,” “it’s the most obvious labels to see”). For the third strategy, participants provided social-focused reasoning, which indicated that race was avoided because of (3) social concerns (e.g., “it didn’t seem appropriate to use racial words,” “because I thought it was racist”). The final strategy consisted of (4) idiosyncratic responses (e.g., “I don’t know,” “I’m not good at differentiating them”).
Personal colorblindness
Four items modified from Pauker et al. (2015) were used to assess whether participants personally endorsed a colorblind approach to race (e.g., “I am uncomfortable talking about race,” “I bring up race in [my] everyday conversations” [reverse-scored]). Agreement with the statements was rated on a 6-point scale (1 = very strongly disagree, 6 = very strongly agree). Responses were averaged such that higher scores indicate greater personal colorblindness (α = .69).
Results
Colorblind behavior
The majority of Asian participants did not exhibit colorblind behavior: only 14 (15.4%) participants failed to acknowledge race in the task. Conversely, 77 participants (84.6%) mentioned race in at least one trial. Across all four trials, on average, participants acknowledged race 67.6% (SD = 37.91) of the time.
Rationale for behavior
As expected, participants who chose to mention race versus not, provided different rationales for their behavior, χ2(3) = 48.50, p < .001, V = .73. Those that mentioned race largely used task-focused rationales, whereas the few that did not mention race used idiosyncratic rationales followed by social concerns. Overall, functional (45.5%) and perceptual (37.8%) reasons were mentioned more by participants than social concerns (7.7%) and idiosyncratic (8.9%) reasons (see Figure 1).
Use of rationale for mentioning race (or not) during the photo identification task: Study 1.
Personal colorblindness
The frequency with which race was acknowledged was negatively associated with personal colorblindness (M = 3.25, SD = 0.65), r = −.26, p = .03; those who were less likely to endorse colorblindness mentioned race more frequently in the photo identification task.
Discussion
As one of the first studies to examine colorblind behavior in racial minorities living in a racially diverse context, we found that Asian participants overwhelmingly acknowledged race; over 80% mentioned race at least once during the photo identification task. In addition, task-focused rationales (functional or perceptual) were most frequently provided to justify acknowledging race, whereas social concerns (e.g., concerns about appearing prejudiced) were rarely mentioned. Finally, lower endorsement of colorblindness was related to less colorblind behavior. This pattern of results is consistent with the possibility that in this racially diverse setting, people may not adopt a colorblind norm and instead adopt alternative social norms that encourage acknowledging race.
Study 2
One alternative explanation for the results of Study 1 is that Asian participants are simply more comfortable talking about race. Although people of color and White individuals may both feel social pressure to adopt colorblind norms (Neville et al., 2013), Asian people, due to their racial minority status in the broader context of the US may be more comfortable talking about race (Sue, 2013) and therefore less likely to exhibit colorblind behavior. Past research has found that racial minority children adhere to social norms (e.g., colorblindness in a majority-White context), despite their racial identity (Pauker et al., 2015). However, it remains unclear whether talking about race more openly is driven by the social norms in a context, regardless of a person’s racial identity.
To examine this possibility, in Study 2 we included White participants for comparison, as the majority of the research conducted to date has focused on this demographic. If the lack of colorblind behavior in Study 1 was based solely on racial minority group membership, then we would expect Asian participants to acknowledge race and provide task-focused reasons for doing so, but we would expect White participants to display colorblind behavior (avoid mentioning race) and provide social-focused reasons for doing so, replicating research conducted in less racially diverse contexts. However, if, as hypothesized, this racially diverse context features social norms that encourage acknowledging race, then both White and Asian participants in Hawai‘i should mention race with the same frequency during the photo identification task and provide similar rationales for doing so.
Since we anticipate that perceptions of larger social norms may differ in this racially diverse context, in Study 2 we measured participants’ perceptions of colorblind norms in Hawai‘i. We expect that participants will behave in line with perceived social norms, and both White and Asian participants will endorse similar social norms.
Method
Participants
We aimed to recruit 30 participants of each racial background based on past research that examined colorblind behavior (Apfelbaum et al., 2008; Norton et al., 2006). Sixty-six undergraduates from the University of Hawai‘i Psychology Department’s participant pool and members of the community participated in exchange for extra credit in psychology courses or a $5.00 Starbucks gift card. The sample included 34 East or Southeast Asian (24 females; Mage = 20.35 years, SD = 4.66, age range: 17–44 years old) and 32 White (20 females; Mage = 30.80 years, SD = 15.40, age range: 18–71 years old) participants. 1 Because of experimenter error, six White participants did not complete the colorblind norms questionnaire and are not included in the regression analyses. A sensitivity power analyses using G*Power (Faul et al., 2009) was conducted, and determined we would be able to detect an effect size of d = 0.70 with 80% power and α = .05 when conducting comparisons across our two racial groups.
Materials and procedure
Participants completed the measures as outlined in Study 1, with the following exception: instead of measuring personal colorblindness, four items were used to assess perceptions of contextual colorblind norms (e.g., “In Hawai‘i, people bring up race in their everyday conversations” [reverse-scored]; adapted from Pauker et al., 2015). Agreement with the statements was rated on a 6-point scale (1 = very strongly disagree, 6 = very strongly agree), and responses were averaged together such that higher scores indicate greater perceived colorblind norms (α = .60).
As in Study 1, the photo identification task was completed with an experimenter who belonged to the participant’s racial ingroup (i.e., Asian participants interacted with an Asian experimenter, and White participants with a White experimenter). Video recordings of the photo identification task were coded for colorblind behavior (Cohen’s κ = 1.00), and for the rationale for using race (Cohen’s κ = 0.78) using the same procedures as Study 1.
Results
Colorblind behavior
Replicating the results of Study 1, the majority of participants did not exhibit colorblind behavior: only two (5.9%) Asian and six (18.8%) White participants failed to acknowledge race in the photo identification task. In contrast, 32 (94.1%) Asian participants and 26 (81.3%) White participants asked about race at least once during the photo identification task. Averaged across all four trials, Asian participants acknowledged race 77.2% (SD = 31.60) of the time, and White participants acknowledged race 64.1% (SD = 39.10) of the time. As anticipated, Asian and White participants did not reliably differ in their tendency to mention race, t(64) = 1.51, p = .14, d = 0.37.
Rationale for behavior
Replicating Study 1, participants’ rationales differed by whether they mentioned race or not, χ2(3) = 10.80, p = .01, V = .41. Rationales did not differ by participant race, χ2(3) = 3.57, p = .31. Those that mentioned race largely used task-focused rationales, whereas the few that did not mention race used perceptual rationales followed by social concerns and idiosyncratic rationales. Overall, participants used functional (25.8%) and perceptual (60.6%) rationales, and less frequently mentioned social concerns (9.0%) or idiosyncratic rationales (4.6%; see Figure 2).
Use of rationale for mentioning race (or not) during the photo identification task collapsed across participant race: Study 2.
Colorblind norms
Somewhat unexpectedly, White participants (M = 3.27, SD = 0.50) were more likely to perceive contextual colorblind norms as compared to Asian participants (M = 2.66, SD = 0.75), t(57) = −3.52, p < .001, d = −0.93. To examine if colorblind norms or participant race predicted frequency of acknowledging race, we regressed frequency of acknowledging race onto participant race, colorblind norms, and their interaction. Participant race was effect-coded as Asian (1) versus White (−1), and perceived colorblind norms was mean-centered. We found an effect for perceived colorblind norms such that lower perceived colorblind norms were related to greater frequency of use of race, b = −0.27, SE = 0.08, t = −3.59, p < .001. We found no effect for participant race, b = 0.002, SE = 0.05, t = 0.04, p = .96. However, these effects were qualified by a significant interaction, b = 0.21, SE = 0.08, t = 2.80, p = .007. A simple slopes analysis revealed that for Asian participants, perceived colorblind norms did not influence their use of race, b = −0.06, SE = 0.08, p = .43. However, for White participants, perceiving lower colorblind norms related to increased frequency of use of race, b = −0.49, SE = 0.13, p < .001.
Discussion
Consistent with the results of Study 1, we find that in this racially diverse context, participants overwhelmingly made use of race; over 80% of our participants mentioned race at least once during the photo identification task. Further, the tendency to acknowledge race did not differ by race of our participants. Consistent with our prediction that social norms in a racially diverse context (rather than racial identity) would sway participants’ behavior, both White and Asian individuals mentioned race often and to a similar extent. Replicating Study 1, participants were more likely to use functional or perceptual rationales, as compared to social concerns or idiosyncratic rationales, during the photo identification task, and this did not differ by participant race.
In support of a social norms explanation, participants who perceived less of a colorblind norm in Hawai‘i were more likely to acknowledge race. However, unexpectedly, the extent to which perceived norms were related to participants’ behavior differed based on participant race.
For Asian participants, perceived colorblind norms were not related to whether they acknowledged race or not. Yet, for White participants, those who perceived lower colorblind norms in Hawai‘i, more frequently acknowledged race. As compared with Asian participants, it may be that White participants are more sensitive to perceived colorblind norms due to relatively greater concerns about appearing prejudiced. Given these findings, in Study 3 we aimed to tease apart colorblind norms and how they relate to strategies to appear nonprejudiced.
Study 3
We provide evidence from diverse samples of participants that lower endorsement of colorblind norms (Study 1) and lower perception of colorblind norms (Study 2) were associated with lack of colorblind behavior. In order to directly test whether colorblind norms impact colorblind behavior, we next manipulated social norms that highlight talking about race as prejudiced, to see whether these beliefs also mapped onto people’s behavior. Specifically, we exposed participants to one of three conditions: (a) colorblind norms that explicitly link talking about race to prejudice, (b) colorblind norms with no additional information, or (c) a control condition. We expect that, similar to research conducted in the continental US (e.g., Apfelbaum et al., 2008; Pauker et al., 2015), even in a racially diverse context, social norms that link talking about race to prejudice would trigger colorblind behavior.
Method
Participants
An a priori power analysis to detect effects for a one-way ANOVA with three levels (social norm condition: talking about race is prejudiced, colorblind, and control) with achieved power of 0.80 and effect size of f = 0.30 (which is similar to past effect sizes found in Norton et al. [2006], and the smallest effect we could detect given our ability to recruit participants at the time of data collection) indicated we required a sample size of 111 (G*Power; Faul et al., 2009). Therefore, we recruited 112 participants (66 females; Mage = 21.10, SD = 6.02) from University of Hawai‘i at Manoa’s undergraduate student population to participate in exchange for extra course credit or a $5.00 gift card. Since we found that colorblindness endorsement and the tendency to acknowledge race did not differ by participant race (Study 2), we recruited only Asian participants for this study.
Materials and procedure
Participants were randomly assigned to one of three conditions: talking about race is prejudiced norm, colorblind norm, or the control. First, participants were led to a computer where they reviewed the instructions alone in a cubicle while the experimenter set up the game in the experiment room. In all conditions, participants were given identical instructions on how to play the photo identification game; however, in our two experimental conditions, participants were given an example video of a past participant (a confederate) playing the game to facilitate their understanding of how the game worked. In the talking-about-race-is-prejudiced condition, the participant did not use a race-related question to identify the target photo. When the experimenter asked “Why did you choose not to use racial labels in this task?” the ostensible participant responded that they did not use race because “here in Hawai‘i, we don’t use race because it’s racist.” The video was identical in the colorblind condition, with the exception that the ostensible participant responded that they were not sure why they did not use race. The important distinction between these two conditions is that while both are modeling colorblind behavior, in the talking-about-race-is-prejudiced condition, participants hear an explicit rationale for another person’s colorblind behavior that invokes a broader social norm in Hawai‘i and links talking about race to prejudice. We included a colorblind condition to examine whether merely modeling of colorblind behavior was enough to shift norms and behavior. In the control condition, participants were not shown an example video, but were given the same instructions as in all other conditions on how to play the game (e.g., “You will be asked to guess what photo your partner has with as few yes/no questions as possible”).
Participants were then led to the experiment room and played four rounds of the photo identification task with the experimenter. Video recordings of the photo identification task were coded for the frequency with which race was acknowledged (Cohen’s κ = 1.00) and the rationale for using race (Cohen’s κ = 0.84) using the same procedures as in Study 1. After participants completed the task, they were moved to a computer to complete questionnaires. Afterwards, they were debriefed about the purpose of the experiment and given information on how colorblind strategies may be ineffective at improving race relations.
The questionnaires consisted of endorsement of colorblindness (Study 1; α = .67), perceptions of colorblind norms in Hawai‘i (Study 2; α = .70), and demographic questions.
Results
Colorblind behavior
Participants demonstrated more colorblind behavior (i.e., they did not mention race in any of the trials) in the talking-about-race-is-prejudiced condition (25 participants were colorblind; 67.6%), as compared to the colorblind condition (18; 48.7%) and the control condition (1; 2.63%), χ2(2) = 35.17, p < .001, V = .56 (see Figure 3). The frequency with which participants acknowledged race averaged across all four trials also differed by condition, F(2, 109) = 35.98, p < .001, ηp2 = .40. Post hoc comparison using Tukey’s correction found that participants in the talking-about-race-is-prejudiced condition acknowledged race less often (14.2% of trials, SD = 26.10) than those in the colorblind condition (35.8%, SD = 40.20) and the control condition (75.7%, SD = 27.60), ts(109) > 5.41, ps < .001. There was also a significant difference in frequency of acknowledging race for participants in the colorblind compared to the control condition, t(109) = 2.92, p = .004.
Percentage of participants who exhibited colorblind behavior across conditions: Study 3.
Rationale for acknowledgment of race
As expected, participants’ rationale for acknowledging race differed across conditions, χ2(6) = 19.11, p = .004, V = .29. Participants in the talking-about-race-is-prejudiced condition most frequently mentioned a perceptual (48.7%) rationale for (not) using race, followed by social concerns (27.0%), and a functional (24.3%) rationale. For those in the colorblind-norm condition, the most reported rationale was functional (46.0%) followed by perceptual (35.1%), social concerns (10.8%), and idiosyncratic (8.1%). Lastly, replicating our findings from Studies 1 and 2, those in the control condition most often reported a functional (50.0%) and perceptual (47.4%) rationale for acknowledging race, followed by one (2.6%) idiosyncratic response.
Endorsement of colorblindness
We conducted a one-way ANOVA comparing endorsement of colorblindness across conditions, F(2, 109) = 3.80, p = .03, ηp² = .07. Those in the talking-about-race-is-prejudiced condition endorsed colorblindness to a greater extent than those in the control condition, t(73) = 2.50, p = .01. The colorblind condition did not significantly differ from either the talking-about-race-is-prejudiced or the control condition, ps > .19 (see Table 1 for means and standard deviations). Furthermore, endorsement of colorblindness was significantly related to colorblind behavior (r = .31, p < .001), such that those who endorsed colorblindness to a greater extent, exhibited more colorblind behavior.
Means and standard deviations of colorblind measures across conditions: Study 3.
Notes. Standard deviations are shown in parentheses. Significant comparisons are indicated by shared superscripts. CB = colorblind; HI = Hawai’i.
Perception of colorblind norms in Hawai‘i
We found a significant difference in perceptions of colorblind norms in Hawai‘i by condition, F(2, 109) = 4.21, p = .02, ηp² = .07. Those in the talking-about-race-is-prejudiced condition perceived others in Hawai‘i to endorse colorblindness to a greater extent as compared to those in the control condition, t(109) = 2.76, p = .007 (see Table 1). None of the other comparisons were significant (ps > .09). As expected, perceptions of colorblind norms were also related to colorblind behavior (r = .25, p = .008), such that those who perceived colorblind norms in Hawai‘i were more likely to exhibit colorblind behavior.
Discussion
The results of Study 3 support our hypothesis that race-related social norms influence whether participants acknowledge race. As predicted, when participants were exposed to a talking-about-race-is-prejudiced norm, they exhibited colorblind behavior and tended to avoid acknowledging race. This effect was stronger than in the colorblind-norm-only condition, where the talking-about-race-is-prejudiced norm was not made salient. Lastly, our findings in the control condition replicated the results from Studies 1 and 2 where participants in this diverse context overwhelmingly used race and gave functional and perceptual rationales for doing so.
It is interesting to note that participants overwhelmingly reported functional and perceptual rationales across all conditions (not only the control condition), which means that many people used these types of rationales to also support why they would not talk about race. Although the talking-about-race-is-prejudiced and colorblind conditions impacted behavior, the prevailing norm in Hawai‘i may be to notice and utilize salient characteristics, such as race, in social interactions (as supported by the findings of Studies 1 and 2 and the control condition in this Study 3). Therefore, when normative social influence pressured participants to avoid using race (in the noncontrol conditions), their rationales reported that race was no longer functional or perceptually useful, in order to justify their colorblind behavior (e.g., “I feel like I can’t really tell what someone is just by looking at them”). These rationales may have helped to solve any cognitive dissonance experienced due to a difference between the highlighted social norms that impacted their behavior and their actual beliefs. Further research is needed to uncover the process by which colorblind norms are internalized.
Finally, we found that those exposed to a talking-about-race-is-prejudiced norm perceived others in Hawai‘i to also endorse colorblind strategies, and they themselves endorsed colorblindness to a greater extent. Together, the findings of Study 3 suggest that introducing an explicit colorblind norm that invoked the strategic nature of colorblindness (i.e., highlighting the belief that talking about race is prejudiced) shifted participants’ use of colorblind behavior.
Study 4
Our findings from Studies 1–3 demonstrate a strong link between social norms and colorblind behavior; the lack of colorblind endorsement in the diverse context of Hawai‘i was associated with less colorblind behavior. One possibility is that racial diversity in the population guides how people use race, which in turn shifts social norms away from colorblindness. In Study 4 we examine this possibility by measuring personal endorsement of colorblindness across a few contexts that differ in the population diversity of racial majority (White) and minority (Asian) participants. Building on our previous findings, we anticipate that, regardless of their racial identity, participants living in more racially diverse (i.e., heterogeneous) contexts would be less likely to endorse colorblindness as compared to participants in more racially homogenous contexts. Additionally, if colorblind endorsement is motivated by the belief that talking about race is prejudiced, we expect to see a relationship between endorsement of colorblindness and the belief that talking about race is prejudiced. Such findings would support the potential relationship between population diversity and social norms moving away from colorblindness extending beyond the context of Hawai‘i.
We recruited White and Asian participants from contexts that varied in their racial diversity: Hawai‘i, California, and Massachusetts. Hawai‘i is the most racially diverse state in the US as calculated by the diversity index (higher scores indicate greater racial heterogeneity and equal proportions across groups; Logan, 2014); thus, Hawai‘i leads in racial diversity on a state level. Asians (37%), Whites (25%), and multiracials (24%) each make up almost one third of the population in Hawai‘i (U.S. Census Bureau, 2019). To extend our research beyond a single context, in Study 4 we included participants living in California. This state ranks second in the US for racial diversity and has racial demographics similar to Hawai‘i (e.g., large Asian population in certain counties, a majority-minority state; Logan, 2014). Lastly, we recruited participants from Massachusetts because this state is low in racial diversity (Logan, 2014) and is where the seminal research demonstrating the link between colorblind-norm endorsement and behavior in predominantly White samples was conducted (i.e., Apfelbaum et al., 2008; Norton et al., 2006). By sampling participants from these locations, we were able to examine how the racial diversity of people’s context corresponds to their endorsement of colorblindness.
Method
Participants
We recruited Asian and White participants from Hawai‘i, California, and Massachusetts. We preselected participants from the following California counties to most closely mimic the population of Hawai‘i: Orange County, Santa Clara County, Alameda County, San Francisco County, and San Mateo County. According to the U.S. Census Bureau (2019), these counties were majority-minority, with the largest minority group being Asian. We aimed to collect a sample of 50 participants per location and race. We collected data from 100 Hawai‘i (50 White, 50 Asian; 60 females, 40 males; Mage = 45.80, SD = 16.80), 102 California (52 White, 50 Asian; 64 females, 38 males; Mage = 42.50, SD = 15.90), and 104 Massachusetts (52 White, 52 Asian; 67 females, 37 males; Mage = 40.20, SD = 15.80) participants. A sensitivity power analysis using G*Power (Faul et al., 2009) conducted with our sample size found that, with 80% power and α = .05, we would be able to detect an effect size of f2 = .05 in multiple regression with seven predictors. A survey containing the measures was distributed to participants via a Qualtrics panel.
Measures
Diversity index
We collected participants’ zip codes for where they currently resided. These were later matched to the participant’s respective city or county. Using data from the U.S. Census, a diversity index was calculated such that representation of many and more equal-sized racial/ethnic groups would result in a higher score. A score of 0 indicates the lowest diversity, with complete homogeneity (i.e., all White or all Asian), and 100 indicates the highest diversity, with equal distribution amongst many groups (i.e., 25% White, 25% Asian, 25% Black, 25% Pacific Islander; Lee et al., 2012).
Endorsement of colorblindness
Two items were used to assess colorblindness (Norton et al., 2006): “When I interact with other people, I try not to notice the color of their skin” and “If everyone paid less attention to race and skin color, we all would get along much better” (α = .84). We also measured endorsement of a colorblind approach by presenting participants with a passage about colorblindness and asking how much they agreed with a colorblind approach as an effective strategy for improving equality (α = .86; Richeson & Nussbaum, 2004; Wolsko et al., 2000). Both measures were anchored from 1 (strongly disagree) to 6 (strongly agree), and highly correlated (r = .63, p < .001), therefore we combined these two measures to form an index of endorsement of colorblindness (α = .87).
Talking about race is prejudiced
We constructed a measure to capture whether participants believed talking about race is prejudiced. Responses were made on a 6-point scale (1 = strongly disagree, 6 = strongly agree). To capture location-specific norms, participants in each location received instructions to think about how people in either Hawai‘i, in (California county), or in Massachusetts would answer these questions. Five items assessed perceptions of whether talking about race was perceived as prejudiced: “Someone who mentions someone’s race/ethnicity is racist,” “To be culturally sensitive, it is best not to mention someone’s race or ethnicity,” “Talking about race/ethnicity is not offensive” (reverse-coded), “People can talk about race/ethnicity without being concerned about appearing prejudiced” (reverse-coded), and “Talking about someone’s race/ethnicity is not prejudiced” (reverse-coded; α = .76).
Results
Endorsement of colorblindness
We conducted a general linear model with participant race (Asian = 1 vs. White = −1), racial diversity (continuous; mean-centered), belief that talking about race is prejudiced (continuous), and their interaction terms as factors on our outcome measure of endorsement of colorblindness. We found no significant effect for race (p = .41). There was a main effect for diversity, b = −0.02, SE = 0.004, 95% CI [−0.03, −0.01], t(291) = −4.40, p < .001, such that those who lived in more racially diverse places endorsed colorblindness to a lesser extent. Similarly, the belief that talking about race is prejudiced was significantly related to colorblind endorsement, b = 0.21, SE = 0.09, 95% CI [0.04, 0.38], t(291) = 2.40, p = .02, such that those who believed that talking about race is prejudiced endorsed colorblindness to a greater extent. None of the interaction terms were significant, ps > .05 (see Table 2 for parameter estimates and Table 3 for correlations among variables).
Parameter estimates: Study 4.
Note. TARP = Talking About Race is Prejudiced Scale.
Correlations across variables: Study 4.
Note. TARP = Talking About Race is Prejudiced Scale.
p < .01. **p < .001.
Discussion
The results from Study 4 support the link between colorblind-norm endorsement and population diversity. Across two contexts known for being racially diverse (Hawai’i, California) and one known for being largely homogeneously White (Massachusetts), increasing context diversity corresponded to decreased colorblind-norm endorsement. Importantly, this pattern emerged regardless of participant race, highlighting the role of social norms in participants’ context, rather than their racial identity. Lastly, we found that the belief that talking about race is prejudiced was significantly related to people’s endorsement of colorblind beliefs, which further supports the notion that colorblind norms are linked to lay beliefs that talking about race is prejudiced behavior.
General Discussion
In four studies we examined the relation between race-related social norms and the tendency to exhibit colorblind behavior in racially diverse contexts. When examining people’s actual behaviors, we found that both Asian (Studies 1 and 3) and White (Study 2) individuals living in Hawai‘i tended to use race in a photo identification task, and endorsed functional and perceptual rationales for doing so. In Study 3, we found that when exposed to a colorblind norm that explicitly tied talking about race to prejudice, participants were less likely to acknowledge race and more likely to mention social concerns related to talking about race (i.e., that it is not socially appropriate), as compared to participants exposed to colorblind-norm and control conditions. Lastly, in Study 4 we provide evidence for the possibility that the endorsement of colorblind beliefs is related to the racial diversity in people’s environment, and that endorsement of colorblindness is linked to their beliefs that talking about race is prejudiced. In Study 4 we found a relationship between colorblind endorsement and perceptions that talking about race is prejudiced. It may be that colorblind norms are less prevalent in diverse contexts and, consequently (due to racial minorities being more comfortable with this topic), race is not considered to be a taboo topic of conversation. Together, these studies provide insight into how race-related social norms may operate in diverse settings, with diverse participants. Moving away from colorblindness, it is possible that as societies grow more racially diverse, social norms surrounding race relations will foster race-conscious norms when it comes to intergroup relations.
Different Contexts, Different Race-Related Norms
We provide the first evidence that the strategies used to negotiate race-relevant situations in racially diverse contexts may diverge from the strategic colorblindness largely adopted in racially homogenous contexts across the US (e.g., Apfelbaum et al., 2008; Norton et al., 2006; Pauker et al., 2015). Our results from Studies 1 and 2 and from the control condition in Study 3 indicate that, instead of colorblind behavior, in the racially diverse context of Hawai‘i, White and Asian individuals overwhelmingly acknowledge race. The tendency to use strategies which acknowledge race, regardless of participant race, suggests that the normative precedent in the racially diverse context of Hawai’i may encourage individuals to talk about race. Further supporting this possibility, in the current research, participants who were more likely to acknowledge race were less likely to personally endorse colorblindness (Study 1), perceive a colorblind social norm in Hawai‘i (Study 2), and consider talking about race to be prejudiced behavior (Study 3).
An interesting point to note is that perception of colorblind norms was not consistently related to acknowledgment of race. For Asian participants in Study 1, decreased personal endorsement of colorblindness was related to decreased colorblind behavior. However, perceptions of colorblind norms were only meaningfully related to White participants’ (but not Asian participants’) behavior in Study 2 (i.e., if they perceived others followed colorblind norms, they also adopted the norm). Yet, Asian individuals are susceptible to normative social influence, as demonstrated in Study 3. One possibility is that White individuals’ minority status in Hawai‘i may increase pressure to follow social norms and behave accordingly. As a “minority” in this context, White individuals may more strongly oppose colorblind ideology, as a function of assimilating to dominant group norms (Plaut et al., 2009). While we believe the social context is important for shaping individuals’ race-related beliefs irrespective of their racial group membership (see Study 4), it is important to acknowledge that Study 2 may have been underpowered, and thus results should be interpreted with caution. It is clear that further research is needed to more fully examine how racial identity interacts with social norm endorsement in diverse contexts.
Regardless, our findings highlight the important role that perceptions that talking about race is prejudiced have for colorblindness. In Studies 3 and 4, we begin to examine potential explanations for why we might see a general lack of colorblind endorsement in this racially diverse context. Due to a larger racial minority population and comfort with talking about race, this behavior may not be considered taboo in the same way it may be considered in more homogenously White contexts. We provide support for this possibility in Study 3 by demonstrating that when contextual norms set talking about race to be prejudiced behavior, participants in Hawai‘i were less likely to acknowledge race and more likely to endorse colorblind norms as compared to a control condition. In Study 4 we extend this beyond one racially diverse setting and provide evidence that an increase in racial diversity in one’s context is related to a decrease in colorblind endorsement. Furthermore, those who perceived talking about race as prejudiced were more likely to endorse colorblind beliefs. Together, the results of these studies provide initial evidence that social norms concerning race are susceptible to manipulation, even with racial minority participants in diverse contexts. Extending these findings, it may be possible to shift people’s behavior to begin acknowledging race in more positive ways.
Limitations and Future Research
Following past research, we used the photo identification task from Apfelbaum et al. (2008) and Norton et al. (2006), which depicted Black and White individuals. The original studies focused on White participants whose racial ingroup was included in the photo identification task. In our research, we focused on the perspective of non-White (Asian) participants whose racial ingroup was not included in the task. That Asian participants only viewed outgroup members might provide a more stringent test of our hypotheses. Given that Asian participants in Studies 1–3 did not hesitate to acknowledge race when examining two outgroup targets, we anticipate our results would be maintained with the inclusion of ingroup targets.
Building on this limitation, another potential issue is that none of our behavioral experiments included an outgroup experimenter. It is plausible that the presence of an outgroup experimenter may exacerbate anxiety during the task and promote more colorblind behavior (Apfelbaum et al., 2008). Future research should address this gap to gain a better understanding of how social norms influence face-to-face interracial interactions in diverse settings.
Lastly, it is important to note that while we found a significant relationship between endorsement of colorblind beliefs and the racial diversity of one’s context in Study 4, we primarily tested these hypotheses about colorblind behavior in the context of Hawai‘i. It is possible that other cultural factors explain people’s behaviors surrounding race in Hawai‘i. For example, Hawai‘i differs from other states in many ways, including having a recent history of colonization, being geographically isolated, having unique demographics, and having only gained statehood within the past 60 years. Despite this concern, we did find a relationship between endorsement of colorblindness and beliefs that talking about race is prejudiced within our California sample, as well as evidence that increased contextual diversity was related to less endorsement of colorblindness. However, future research should directly test whether colorblind behaviors differ across diverse samples of participants from a variety of diverse contexts, to verify the generalizability of these results.
Conclusion
Despite the projected growth in racial diversity within the US (Colby & Ortman, 2015), little research has investigated the dynamics of intergroup relations amongst racial minorities in racially diverse contexts. As suggested by our findings, it is possible that different social norms operate (e.g., talking about race is not prejudiced) in racially diverse contexts, thus allowing people to feel comfortable talking about race in more functional ways. If people in racially diverse contexts feel no hesitancy to mention race, it may be that their concerns about appearing prejudiced are mitigated in some other way. Furthermore, by reinforcing norms that encourage the use of race, racially diverse contexts may support conversations that are necessary to address inequities. The failure to acknowledge race only reinforces racial hierarchies that contribute to the continued unfair treatment of historically disadvantaged groups and the perpetuation of racial bias in society (Dovidio et al., 2015). Explicit mention and labeling of race may be necessary to achieve equity (Plaut et al., 2018); therefore, understanding contexts in which acknowledging race is not linked to negative outcomes such as appearing prejudiced is vital for progress in race relations. Armed with insights into what “works” to promote social harmony in diverse contexts, we may be able to develop interventions for use in other contexts that ease the tensions typical of interracial interactions (e.g., Richeson & Shelton, 2007), and eventually foster more positive intergroup relations for our increasingly diverse society.
