Racializing Captain America: How Racial Attitudes Affect Perceptions of Affirmative Action and Diversity Initiatives in Media

Abstract

Is announcing a commitment to diversity enough to activate attitudes toward diversity initiatives? And what are the spillover effects of these programs? To address these questions, we conduct an experiment imbedded in a nationally representative survey of non-Hispanic White Americans (n = 1,519). We inform respondents that the White actor who plays Captain America will be replaced, while varying whether there is a reference to a diversity initiative and whether the replacement is White or Black. We find that reference to diversity initiatives on its own has no effect but the action of displaying diversity affects marketplace preferences and attitudes toward diversity initiatives.

Keywords

affirmative action diversity initiative experiment mass communication

Although affirmative action programs have long been studied in the social sciences, the related practice of diversity initiatives by private companies have received scant attention. We argue that this is notable gap for four reasons: first, most Americans do not have an extensive knowledge about politics, both because of various socioeconomic reasons (e.g., education) and because many Americans have neither the time nor energy to treat politics like a hobby (Carpini & Keeter, 1996; Stimson, 2015; Verba et al., 1995). Because of this, Americans without extensive political knowledge may form political opinions and attitudes in response to the apolitical content they regularly consume when such content begins to look political. Second, many White Americans have historically lived in segregated communities, and for these people, their only exposure to racial minorities is likely to occur in the form of media consumption (Entman & Rojecki, 2001). Third, these initiatives may be a yet-unstudied determinant of affirmative action attitudes; given the conceptual similarities of these practices, it is possible that some Americans may conflate the two. Finally, we do not yet know if merely stating a commitment to diversity and inclusion has the same effects as actually investing into such initiatives. And as these initiatives become increasingly common as the nation’s entertainment industry incorporates more racial minorities in their hiring practices, they will only become more important to understand.

There are a number of unresolved questions regarding how exposure to these diversity initiatives may affect political attitudes. We address three such questions. First, does exposure to such initiatives affect marketplace decision-making processes (e.g., what movie to see in theaters)? Second, do these responses spill over into attitudes toward diversity initiatives in general and affirmative action policies in other domains? Third, and most important, does communicating a commitment to diversity and inclusion garner support for that initiative? Or, conversely, is it possible that such “cheap talk” is insufficient? That is, must companies take action exhibiting a commitment to diversity to affect people’s attitudes toward these programs in general?

To assess these questions, we embedded an experiment in a nationally representative survey of White Americans. The experiment varied exposure to one of many descriptions of how the actor who plays Captain America in the Marvel Cinematic Universe (MCU) would be replaced following his retirement from the film franchise. We find strong evidence that simply mentioning diversity is not enough to activate people’s attitudes toward these programs; instead, there must also be evidence of diverse hiring practices at work (i.e., a person of color benefiting). Put simply, lip service is not enough; there must be action as well. Furthermore, we find that while attitudes on diversity initiatives are shaped by exposure to these practices, we do not find similar effects on attitudes toward affirmative action, suggesting that White Americans do see a distinction between the two practices. Finally, we show that marketplace attitudes toward these would-be films featuring either a White or Black Captain America are contingent on racial resentment while policy attitudes are not.

Diversity Initiatives and Affirmative Action

Affirmative action programs began in the 1960s to 1970s as a means of encouraging equal opportunity for racial minorities in college admission decisions and employment (Chen, 2009; Hirschman & Berrey, 2017). The term can refer to many policies and practices, ranging from “additional outreach and recruitment efforts, to preferential consideration of target-group members, to numerical targets or quotas” (Gilens et al., 1998, p. 167).

Affirmative action programs, generally mandated or incentivized by public policy, are primarily found in government agencies, companies that do significant work for the government, and higher education institutions. Affirmative action in education has begun to recede from public universities, instead becoming a mainstay of prestigious, private universities (Hirschman & Berrey, 2017). Furthermore, Americans have historically been rather unsupportive of the policy (Bobo, 1998; Kinder & Sears, 1981; Sidanius & Pratto, 2001; Sniderman & Carmines, 1997). However, recent data show that most Americans (71%), and most White Americans (66%), perceive affirmative action programs, at least on college campuses, as a “good thing” (Pew Research Center, 2017). Do note, however, that this change in attitudes is not likely due to generational replacement and young White Americans’ generally more-liberal attitudes, as many young Americans lack the contextual knowledge of why affirmative action policies are still necessary (DeSante & Smith, 2017, 2020).

We use the term “diversity initiative” to denote a certain voluntary policy that some companies in the entertainment industry adopt that encourage diversity in hiring and/or casting decisions. Although some may classify such initiatives as affirmative action, insofar that an organization is devoting resources to deal with possible discrimination, we draw a distinction. These diversity initiatives differ from what we might call formal affirmative action in three major ways. First, diversity initiatives can be announced and then never followed up on. Although this may be true for affirmative action as well, there is no public policy feedback mechanism for diversity initiatives that can potentially act as a means of accountability.

Second, these programs emerge not from public policy incentives, but rather from business decisions to respond to growing public and interest group pressure for greater representational diversity. The private politics literature sees such initiatives as preemptive efforts by a company to stave off potential boycotts from consumers (Baron, 2001; Baron & Diermeier, 2007; McDonnell et al., 2015). In recent years, examples abound of such negative consequences for the entertainment industry. For example, the “Oscars so White” controversy that surrounded the 2015 and 2016 Academy Awards led to the Academy nominating more diverse actors and directors in the years since (Schulman, 2018). In addition, outside of the threats presented by consumers, there can be significant economic incentives for entertainment companies to pursue these diversity initiatives. For example, the Marvel Studios film Black Panther, a superhero film that prominently features a majority Black cast, became the second-highest grossing superhero movie in North America soon after its release, grossing almost $700 million in domestic box offices alone (Mendelson, 2018). Of course, some business owners may pursue such initiatives in the interest of “doing the right thing,” but ultimately any private business must consider economic ramifications. We also are not interested here in the business decision per se but rather its effect on citizens’ preferences.

Finally, while attitudes toward affirmative action are well established, having been a mainstay of American political debate since their inception, attitudes toward diversity initiatives are nascent. We have decades of public opinion data on affirmative action attitudes, as discussed earlier, but very little on people’s attitudes toward diversity initiatives. It is entirely possible that people’s attitudes toward such initiatives are not established, allowing researchers the opportunity to glean from an ongoing learning process in the American public.

Trusted Companies and Attitudes on Diversity

We argue that attitudes on diversity initiatives are not well established, relative to affirmative action attitudes. If this is the case, then these nascent attitudes are most likely affected by a process of Bayesian updating (Bartels, 1993; Bullock, 2009). Bayesian updating, which relies on the theory and application of Bayesian statistics, posits that people form attitudes on issues as more information comes to their attention (see Bullock, 2009 for a review). Bayesian updating, however, is often contingent on existing attitude strength and willingness to be open to new information (Barabas, 2004); therefore, it is more likely to operate for issues on which very little opinion exists.

Many White Americans have little knowledge about politics, and are rarely able to interact with peoples of color due to a history of segregation, both geographical and cultural (Entman & Rojecki, 2001). Indeed, “whiteness” is often a social construct maintained through repeated interactions with other White Americans (Frankenberg, 1993, 1997). Thus, it is likely that they rely on a combination of Bayesian updating and cue-taking from trusted sources to form attitudes on “new” issues like diversity initiatives (Zaller, 1992). Trusted sources are traditionally understood, in politics, to be elites of the same party or like-minded interest groups. However, when it comes to entertainment, such sources are of little use. Thus, in this situation, trusted companies are a more likely source of cues for those with low information to form attitudes on issues of diversity and inclusion.

We know a fair deal about how trusted various companies are in the United States. A study from the Baker Center (Georgetown University, 2018) finds that Amazon and Google are more trusted than any governmental institution aside from the military.¹ Given how well regarded many of these entertainment companies are, we posit that just one of these companies launching a diversity initiative may affect attitudes on such initiatives more generally, as exposure to an initiative from a trusted source is likely to serve as a novel cue to many White Americans. Thus, we offer our first hypothesis:

H1: Exposure to a single diversity initiative from a trusted company will result in greater support for such initiatives in general, relative to no exposure to such an initiative, all else constant.

As explained above, the process of Bayesian updating and cue-taking that we posit for H1 will likely not apply for an issue on which attitudes are already formed. Therefore, it is unlikely that exposure to a diversity initiative could result in spillover attitudes on affirmative action.

From Diversity Initiatives to Affirmative Action

Although we are able to offer a fairly straightforward prediction regarding exposure to a single diversity initiative, how such initiatives affect attitudes toward affirmative action is more unclear. The existing scholarship has not yet explored exposure to diversity initiatives as a potential source of affirmative action attitudes, so we must rely on the fact these two policies are conceptually similar to determine the possible spillover effects. Indeed, it may be possible that attitudes on affirmative action are so dominant in the American political consciousness that these attitudes “spillover” into anything conceptually associated, much like how distinct policy areas became racialized during President Barack Obama’s tenure (Tesler, 2012, 2015). To address this open question, we argue that an understanding of psychological schemata is especially useful.

Schemata, as often studied in psychology, are “knowledge structures that represent objects or events and provide default assumptions about their characteristics, relationships, and entailments under conditions of incomplete information” (DiMaggio, 1997, p. 269). Or, as one classic example goes, a small child thinks all four-legged animals are dogs or cats until they see a cow, simply by virtue that under their state of incomplete information, they rely on their prior experiences with four-legged creatures to form judgments. More complex concepts, like affirmative action policies, are similarly prone to schematic thinking (Nacoste, 1994).

Schemata largely operate psychologically, thanks to the availability heuristic, or one’s reliance on the ease at which relevant instances come to memory, usually to produce probabilities or frequencies of some event or phenomenon (e.g., four-legged animal = most likely a dog or cat; Tversky & Kahneman, 1973, 1974). Therefore, for novel experiences such as exposure to diversity initiatives, people likely rely on what most readily comes to mind, their prototypical conceptualization that comes closest to what they are being exposed to.

For diversity initiatives, the prototypical concept that most likely comes to mind is affirmative action, since the two are so similar, given that they both preference certain candidates for belonging to certain groups. However, as with the four-legged animals example above, this schemata for affirmative action policies does not quite apply to diversity initiatives. For one thing, they are run by private companies, which immediately conflicts with the definition of affirmative action as being “public” in nature. For those with economically conservative attitudes, for example, exposure to a diversity initiative challenges preconceived notions of what usually falls under the category of “affirmative action,” that is, publicly funded programs that encourage the hiring and acceptance of racial minorities in certain businesses and universities.

With this in mind, the case of how a diversity initiative in the film industry can affect policy attitudes becomes clearer, and we can offer a prediction: there should be no effect. This follows from the above detailing of how psychological schemata work. That is, schemata are formed based on available information formed from prior experience (e.g., affirmative action is good or bad because of X), but can be challenged when exposed to phenomenon that challenge the prototypical conceptualization (e.g., diversity initiatives lack feature X). When people are exposed to a diversity initiative, their schemata should be immediately challenged and result in those exposed to the information to understand that diversity initiatives are distinct from affirmative action. Therefore, we should see no effects on affirmative action attitudes as a result of being exposed to a diversity initiative:

H2: Exposure to a diversity initiative will not affect people’s attitudes toward affirmative action, all else constant.

Even so, some existing work on the effects of positive stereotypes in popular media find the opposite of what we predict in H2. Some of Fujioka’s (2000, 2005) work, for instance, finds in observational data that exposure to positive stereotypes about racial minorities can increase support for affirmative action among White (2000) and Black (2005) Americans. We posit differently because of our difference in substantive interest and subsequent choices in research methods. Fujioka’s work is largely concerned with attitudes toward affirmative action policies foremost, whereas we are more interested in studying nascent attitudes on private diversity initiatives. Furthermore, Fujioka’s research design relies on observational data as a result of relying on cultivation theory (Shrum, 1996), for which single experiments would be a poor test (i.e., there is no longitudinal element). In short, we predict a null here in contrast to Fujioka’s findings due to differences in research interests and necessary design choices, rather than because of some issue with Fujioka’s findings; indeed, we believe our hypotheses to be complementary to Fujioka’s.

Cheap Talk and Costly Action

Although we argue that exposure to a diversity initiative from a trusted company should affect attitudes toward such initiatives in general, and subsequently not spill over to attitudes on affirmative action, it is important to recognize that diversity initiatives can be “started” with little-to-no intention of following through. Diversity initiatives that are merely talked about can be seen as weak, or even disingenuous, commitments to the cause.

Game theorists conceptualize issues like this as a distinction between cheap talk and costly action (Lupia & McCubbins, 1998) and argue that statements reflecting a commitment to action are seen as more credible than statements without such action. For example, Crawford and Sobel (1982) find that, under certain situations of information asymmetry, communicators who are trusted can affect an audiences’ attitudes. This is consistent with Hypotheses 1 and 2. However, without a strong perception of common interests—and it is not improbable that many people do not fully believe that they share the interests of economically fueled entertainment companies—any communication is likely ignored as cheap talk. What is needed in these cases, to have an effect, is a costly action, or what Spence (1974) refers to as a “market signal.” When the company takes an action, such as actually hiring in diverse ways, it signifies common interests with those who care about such things and would lead to more support.²

Given the cost calculus, companies may find it prudent to announce a commitment to diversity and inclusion, reap the benefits of such an announcement, and never follow through. The potential problem with this is that consumers may detect the difference between cheap talk and costly action and be unresponsive or punitive toward cheap talk. To test these possibilities, we offer the following hypothesis as an addendum to H1:

H3: Exposure to evidence of a diversity initiative (i.e., costly action) will result in more support of diversity initiatives in general, than exposure to a statement of commitment to diversity (i.e., cheap talk), relative to those unexposed, all else constant.

Racial Attitudes and Diversity Initiatives

We recognize that any attitudes formed in response to a diversity initiative are potentially moderated by racial attitudes. Specifically, the concept of symbolic racism (racial resentment hereafter, as this scale is the most commonly accepted operationalization of the theory) is useful for understanding White Americans’ attitudes on affirmative action policies. Racial resentment is argued to operate differently than outright racism, those attitudes that rely on ideas of biological inferiority. Rather, racial resentment attitudes rely more on people holding certain stereotypes of Blacks as lazy and not holding American values, as opposed to Blacks being marginalized by systemic discrimination (Kinder & Sears, 1981). The full battery can be found in Appendix E in the online supplement, but measures of racial resentment include items like, “it’s really a matter of some people just not trying hard enough; if blacks would only try harder, they could be just as well off as whites.”

However, there is a great deal of debate over the exact nature of opposition to affirmative action policies, such that scholars disagree on the potential mechanism: racial prejudice, group conflict, or political ideology. As explained above, Kinder and his various colleagues (Kinder & Mendelberg, 2000; Kinder & Sanders, 1990; Kinder & Sears, 1981) argue that racial resentment largely explains opposition to affirmative action policies among White Americans. Meanwhile, others argue that while racial hostility explains some of White Americans’ opposition to affirmative action, differences in political ideology better explain subsequent differences in support (Kuklinski et al., 1997; Sniderman & Carmines, 1997; Sniderman et al., 1991). And another group posits that racial resentment is one of several means by which White Americans justify their desire for social dominance and the perceived threat that minority groups pose to their group’s supremacy (Bobo, 1999; Sidanius et al., 1996; Sidanius & Pratto, 2001).

As stated from the outset, this study is concerned with how exposure to a diversity initiative can affect attitudes toward diversity initiatives and affirmative action. We are not, however, interested in looking for the psychological cause of opposition to diversity initiatives or affirmative action as the existing literatures are. Furthermore, all of the above literatures agree that racial resentment is a good predictor of affirmative action attitudes; they just disagree on the extent or the reasons why. That is fine for our purposes, and we offer the following:

H4: Racial resentment will moderate the effect of diversity statements and/or actions on diversity initiative attitudes, such that the positive effects will be higher among those with lower levels of racial resentment, all else constant.

In short, we should expect that exposure to an initiative that promotes minority inclusion will be assessed differently by those with high and low levels of racial resentment. For example, someone with high levels of resentment may be more hesitant to say that they would purchase a ticket for a film that was affected by a diversity initiative than someone with lower levels of resentment; indeed, we might expect the latter person to be more excited about the prospect of watching such a film.

Experimental Design

We leveraged a real-life trend of diversity initiatives in the film industry to craft realistic treatments in our experimental design. Specifically, we used Marvel’s wildly successful Cinematic Universe (MCU) as the subject matter, focusing on a “rumor” of potential changes to casting. This mimicked a real-life rumor around early 2018 that Chris Evans, who at the time played Captain America, would soon retire. Our experimental treatments use coverage of this rumor as a base and then apply one of four additions: (a) the idea that Sebastian Stan, a White man who played a different character in the MCU, would “replace” Evans in his role as Captain America (“White Man” condition, and our pseudo-control); (b) the idea that Anthony Mackie, a Black Man who played a different character in the MCU, would replace Evans (“Black Man” condition); (c) a nearly identical condition to the prior one, with an additional sentence about this change being done for the sake of diversity in the MCU’s cast (“Black Man Diversity” condition); and (d) a condition that did not name a specific replacement, but did mention the possible influence of a diversity initiative in the casting decision (“Unspecified Diversity” condition). All four conditions add legitimacy to the “rumor” by noting how a similar change occurred in the original comic books some years prior such that the change in the film universe would not be unprecedented. All treatment texts can be found in Appendix E in the online supplement.

The choice of this comic book film universe is apt for a number of reasons. Marvel, and its owner the Walt Disney Company, are trusted companies for many Americans (Piacenza, 2018; Valet, 2018). Choosing such a trusted and well-known entity will enable a robust test of H1 (effect on attitudes on diversity initiatives in general), especially given the earlier discussion on cues. Moreover, the varying use of diversity language across the conditions should allow us to test H3. By having a condition that merely talks about diversity, but names no specific actor, we can see if “cheap talk” is enough to affect people’s attitudes, or if the presence of a specific Black actor benefiting from such an initiative is necessary as well.

In addition, there are very few strong priors to any of the actors being replaced or doing the replacing. Although comic book heroes such as Superman, Batman, Spider-Man, and the like may be remembered fondly, either from one’s childhood or the recent string of superhero films, most people are unlikely to have particularly strong prior attitudes about these fictional characters (Martin, 2007). And the existing literature on psychological responses to superheroes is largely on the effect they have on one’s self-image, either as a figure of masculinity or weakness (Coyne et al., 2014; Pennell & Behm-Morawitz, 2015; Young et al., 2013). However, there is some literature that explains how people experience emotions related to fictional characters as if they were real (Hoorn & Konijn, 2003; Konijn & Hoorn, 2005; Oatley, 1995; E. S. Tan, 1996). This body of work would perhaps explain how someone could have a particularly strong reaction to the notion that Captain America would be replaced.

Furthermore, by not mentioning affirmative action policies in the treatments, any effects observed on attitudes on affirmative action will likely be due to exposure to the diversity initiative, although recall that we predict no such effects (H2); what remains to be seen is if White Americans really do differentiate the two policies as we have argued. And in line with the H4, the “Black Man” and “Black Man Diversity” treatments ought to encourage heterogenous effects on market preferences and attitudes on diversity initiatives as they feature a Black man prominently in the text.

Survey and Measures

The survey was conducted by Bovitz, Inc. from August 21 to 28, 2018.³ Bovitz collected the data from a nonprobability-based, but representative (on all key census demographics), sample of the United States, using the American Community Survey as its benchmarks to construct the sample. The survey was administered via the Internet. The sample (n = 1,519) is a nationally representative sample of non-Hispanic, White adults in the United States.⁴ Overall, the sample demographics compare favorably with the 2016 American Community Survey, the most recently available data collected by the U.S. Census Bureau (see Appendix A in the online supplement for more details). The sample skews slightly more male and more educated and is slightly younger and less wealthy than the population of White Americans.

Prior to treatment, the survey asked basic demographic questions and the standard racial resentment battery (DeSante, 2013; Valentino et al., 2002, 2017). Almost every question involving a scale was on a scale from 1 to 7, including the posttreatment variables. Responses to the four racial resentment questions were additively scaled into a single variable (α = 0.82). In addition, respondents were asked if they ever read/watched any comics-related media during their childhood or currently, and if they watched any of the films in the MCU. If they said yes to any of these three questions, they were then asked how important these media were to them. We use these prior exposure and importance questions as controls in some of the supplemental models (Appendix D in the online supplement) and find that they do not significantly alter the effects. We drop them from the analyses presented in this article because they significantly decrease the statistical power we can leverage in our interactive models. Respondents were then assigned to treatment, with randomization resulting in: White Man (n = 394), Black Man (n = 364), Black Man Diversity (n = 412), and Unspecified Diversity (n = 346).

After treatment, respondents were then asked a series of questions related to their attitudes that fell into two distinct areas: marketing questions and policy support. The marketing questions assess attitudes toward this specific diversity initiative (i.e., Captain America being replaced), ask if the change discussed in the article is a good one, and whether the respondent would purchase a ticket to a hypothetical Captain America film after the described change. We expect these attitudes to be moderated by racial resentment, along with the diversity initiative attitudes (H4). The policy support questions are our main dependent variables of interest and assess attitudes toward diversity initiatives in general and affirmative action. Of primary concern is the question, “To what extent do you oppose or support diversity initiatives in which private businesses prioritize hiring minorities?” The affirmative action version of this question is are worded almost identically. It and all other questions asked in the survey can be found in Appendix E in the online supplement.

Results

All analyses were conducted in R, as recently as version 3.6.2. We first conduct difference-in-means tests before using ordinary least squares (OLS) for all interaction models; all analyses use two-tailed tests for calculating statistical significance. Variables in the difference in means are on a 1 to 7 scale while OLS values are recoded 0–1 for ease of interpretation. All analyses use the White Man condition as the reference category. We also conduct bivariate and multivariate models for all analyses as robustness checks, which are included in Appendix B.6 and B.7 in the online supplement; they do not differ significantly from the models presented here.

Overall, we find support for Hypotheses 1, 2, and 3. Recall that H1 predicted exposure to one diversity initiative would result in greater support for initiatives in general. Table 1 summarizes a series of difference-in-means tests comparing values of the two outcome variables across experimental conditions; note that all comparisons are made to the White Man conditions and that positive coefficients indicate higher support. In this table, we can see a positive and statistically significant (p = .059) effect of being exposed to the Black Man Diversity treatment text. The effect sizes are fairly small, but this is a one-off exposure to a weak treatment, so small effects should be expected. This indicates that exposure to this condition results in more positive assessments of diversity initiatives more generally. Although the finding reported from the difference-in-means test is marginally statistically significant, we do find that this is a consistent finding across numerous robustness checks, many of which can be found in the online supplemental appendix, indicating it is not by chance.

Table 1.

Difference in Means Across Conditions.

Experimental condition	Affirmative action support	Diversity initiative support
White Man (control condition) (n = 394)	4.137 (SD = 1.626)	4.195 (SD = 1.528)
Black Man (n = 364)	4.262 (SD = 1.637) (df = 749; t = 1.05; p = .294)	4.36 (SD = 1.583) (df = 746; t = 1.453; p = .147)
Black Man Diversity (n = 412)	4.104 (SD = 1.744) (df = 803; t = −0.275; p = .783)	4.407 (SD = 1.658) (df = 803; t = 1.89; p = .059)*
Unspecified Diversity (n = 346)	4.058 (SD = 1.695) (df = 717; t = −0.647; p = .518)	4.202 (SD = 1.619) (df = 713; t = 0.059; p = .953)

Note. All comparisons made to White Man conditions.

p < .10.

Our second hypothesis predicted null effects on affirmative action attitudes, which we also find support for, as shown in Table 1; that is, we do not observe spillover effects of the treatments on attitudes toward affirmative action. These null findings for affirmative action support are consistent in our analyses even when using proxies for support such as favoring increased or decreased spending for the policy (Appendix B.5 in the online supplement). The lack of a spillover effect is especially interesting when one considers that the correlation between support for diversity initiatives and affirmative action is quite high (0.75) and that the difference in means and in standard deviations are rather small (0.15 and 0.078, respectively).

This potentially indicates two things: (a) people do see a difference between diversity initiatives and affirmative action and thus do not conflate the two, as we argue, and (b) attitudes on diversity initiatives do not necessarily predict affirmative action attitudes, and vice versa. Some may argue that it is a stretch to argue that White Americans see a total conceptual distinction between government programs and initiatives undertaken by private companies. Even so, our survey evidence does suggest that White Americans make a conceptual distinction between this particular diversity initiative in the domain of popular media and affirmative action in employment or education. This is further evidence, we argue, in favor of a Bayesian updating process at work in the case of diversity initiatives, a process that cannot operate in the case of affirmative action.

Recall that our third hypothesis stated that we should expect stronger effects from treatments that emphasized costly action, rather than mere lip service or cheap talk, which we also find support for. Table 1 shows that simply mentioning diversity (i.e., cheap talk), as is the case in the Unspecified Diversity condition, results in no effects. Those exposed to the Black Man Diversity condition were more likely to support diversity initiatives (p = .059). Again, the effect size is fairly small, and this is a conservative test of the hypotheses. Even so, there are effects that indicate the potential for ongoing learning through additional exposures, and the real-world effects would be on a magnitude of potentially millions of people.⁵ These findings demonstrate that it is not simply enough to talk about diversity to affect people’s attitudes about a diversity initiative. Rather, there must be some sort of concrete action associated with the initiative.

Next, we assess H4, which posits that attitudes on diversity initiatives will be moderated by racial resentment. We first present some results that help validate that the treatments worked as designed and produced results in line with our general expectations, especially as it relates to racial resentment moderation. This, of course, implies an interactive model. We conducted OLS regression with the marketing questions as our main outcome variables, and with the treatments interacted with racial resentment. This OLS model uses the White Man condition as the control to be compared with. Also, we ran these analyses without the interactions, and with party ID (PID), and still found treatment effects on whether it was a good idea or not, similar to what is found here (Appendix B.1 and B.2 in the online supplement). The results of these models can be found in Table 2; note that positive values indicate greater support. As one can see, being exposed to either condition featuring a Black man results in thinking that would-be change is a good idea and that one would be likely to purchase a ticket for said hypothetical film (p < .001). This makes some sense given the tremendous success of the majority-Black-cast Black Panther film in early 2018; White Americans seemingly want to support these types of films when Blacks are cast in nonstereotypical roles. However, what is more revealing are the interactions between the treatment and racial resentment.

Table 2.

Interactive Treatment Effects on Marketplace Preferences.

Variable	Good or bad idea?	Buy a ticket?
Black Man	0.154*** (0.030)	0.189*** (0.048)
Black Man Diversity	0.139*** (0.030)	0.161*** (0.045)
Unspecified Diversity	−0.0005 (0.027)	0.026 (0.047)
Racial resentment	0.036 (0.040)	0.044 (0.067)
Party ID	−0.013 (0.018)	−0.011 (0.028)
Age	0.036*** (0.014)	−0.144*** (0.022)
Income	0.009 (0.020)	0.063** (0.029)
Black Man × racial resentment	−0.298*** (0.059)	−0.324*** (0.092)
Black Man Diversity × racial resentment	−0.221*** (0.059)	−0.333*** (0.086)
Unspecified Diversity × racial resentment	−0.059 (0.054)	−0.122 (0.091)
Observations	1,514	1,514
Adjusted R²	.052	.073
F statistic (df = 10; 1,503)	9.247***	12.988***

Note. Reference category = White Man condition.

p < .1. **p < .05. ***p < .01.

We present these interactions graphically in Figures 1 and 2, allowing for a clearer interpretation of the interaction between the treatments and racial resentment. The figures, from left to right, represent going from the weakest levels of racial resentment to the strongest levels. The different lines represent the different experimental conditions and those assigned to them. The coefficient value, thus, is the effect on the outcome of interest for each treatment group, at each level of racial resentment. The far left of Figure 1, for the Black Man and Black Man Diversity conditions, shows that those with the weakest racial resentment attitudes express the strongest support for the proposed change in actors. However, the far right of Figure 1, in those same conditions, shows that those with the strongest racial resentment attitudes express the strongest opposition to the proposed change in actors. And looking at the Unspecified Diversity treatment group, one can see that the line and its 90% confidence intervals overlap with the intercept, indicating null effects. When taking the trend lines together in Figure 1, we see that as one’s racial resentment grows stronger, so too does one’s opposition to the proposed change in actors, but only when exposed to the two conditions wherein a Black man is the replacement (p < .001). Furthermore, we see a similar trend in Figure 2, such that as one’s racial resentment grows stronger, we see reduced intention to buy a ticket to any Captain America film starring the proposed Black actor (p < .001).

Figure 1.

Interactive relationship between racial resentment and whether the change is a good idea.

Figure 2.

Interactive relationship between racial resentment and intent to buy a movie ticket.

In addition, we ran these models with PID as the interaction and find similar patterns of effects, with stronger Republicans expressing more opposition and Democrats expressing greater support. It is important to note, however, that lower levels of racial resentment do not necessarily indicate greater levels of racial acceptance or sympathy (see Chudy, 2017). These findings are strong evidence in favor of H4: When exposed to a diversity initiative, marketplace preferences are contingent on levels of racial resentment.

Finally, we assess the moderation effects of racial resentment on policy support attitudes. Although Table 2 and Figures 1 and 2 show evidence that racial resentment strongly moderated the treatment effect on marketplace decision-making, we do not find similar effects for our policy support outcomes, as shown in the models in Table 3. The effects of the treatments on support for diversity initiatives discussed above are not moderated by racial resentment (or even PID). The lack of a treatment effect on support for affirmative action persists even after accounting for potential moderation by racial resentment, further supporting H2. For the sake of brevity, the interaction plots for Table 3’s interactions are excluded here and can be found in Appendix B.3 in the online supplement; analyses for a PID interaction were also conducted and are also null (Appendix B.4 in the online supplement). This is very interesting, given that the attitudes toward this particular diversity initiative were shown to be contingent on racial attitudes. To see neither of these findings in the interactive model for supporting either affirmative action or diversity initiatives is surprising. Thus, we can only report mixed support for H4. Although racial resentment helps explain divergent marketplace decision-making among White Americans, it does not explain the effect of the treatments on support for diversity initiatives.

Table 3.

Interactive Treatment Effects on Program Support.

Variable	Affirmative action support	Diversity initiative support
Black Man	0.059* (0.035)	0.053 (0.034)
Black Man Diversity	0.037 (0.034)	0.048 (0.034)
Unspecified Diversity	−0.005 (0.035)	0.030 (0.033)
Racial resentment	−0.453*** (0.051)	−0.406*** (0.048)
Party ID	−0.077*** (0.024)	−0.102*** (0.023)
Age	−0.068*** (0.018)	−0.049*** (0.018)
Income	−0.047* (0.026)	0.004 (0.024)
BlackMan × racialresentment	−0.085 (0.070)	−0.055 (0.068)
BlackManDiversity × racialresentment	−0.070 (0.069)	−0.007 (0.067)
UnspecifiedDiversity × racialresentment	−0.009 (0.071)	−0.047 (0.069)
Observations	1,513	1,514
Adjusted R²	0.264	0.243
F statistic	55.114*** (df = 10; 1,502)	49.653*** (df = 10; 1,503)

Note. Reference category = White Man condition.

p < .1. **p < .05. ***p < .01.

Discussion and Conclusion

This study is perhaps one of the first to document the effects of exposure to diversity initiatives in the entertainment industry, as opposed to exposure to information about affirmative action that the literature has studied to date (Kinder & Sanders, 1990; Kuklinski et al., 1997; Sidanius et al., 1996). We argue that studying these initiatives and their effects is important because the modal White American is far more likely to encounter them than some formal affirmative action policy in their daily life. And if affirmative action policies continue to diminish in number as previously documented (Hirschman & Berrey, 2017), then the need to understand the effects of these diversity initiatives will increase in tandem.

Our most important finding is that exposure to a single diversity initiative leads to changes in attitudes regarding diversity initiatives, writ large. In particular, our findings stress the importance of not only talking about diversity, but that such language must be acted upon to garner the support we observe in our experiment. Actions speak louder than words in this case, as evidenced by the fact that our condition that merely mentions diversity (Unspecified Diversity) does not generate results, both in terms of policy support and marketplace attitudes. This key finding is something of a warning for companies trying to communicate their diversity initiatives and capitalize on the good will we find in our study: paying lip service to diversity in hiring practices is not enough to actually change attitudes and there must be action that follows the verbal commitment. In short, if companies want the profits that come from capitalizing on these sentiments (e.g., Black Panther and its enormous success), cheap talk is not enough; they have to invest in hiring practices that feature people of color in prominent roles.

Interestingly, we also find that these attitudes on diversity initiatives do not translate to similar attitudes on affirmative action, perhaps indicating that White Americans understand the two programs to be distinct, as we argue. This is interesting given how similar the programs might appear. We argue that this happens because of two distinct processes at work. Affirmative action attitudes are largely set, having been the subject of political debate for decades, meaning that it is exceedingly difficult to “move” these attitudes in either direction. Conversely, attitudes on diversity initiatives are relatively nascent, and thus malleable to a certain extent. Attitudes that are not yet set can be influenced by trusted sources of information, or in our case, trusted entertainment companies. When these companies, like the Walt Disney Company, send a cue to consumers that diversity and inclusion efforts are worth pursuing, people are likely to follow that cue, barring any strong predisposition. This is a process of Bayesian updating wherein new information is incorporated to update one’s attitudes, and where companies can provide the new information (Bartels, 1993). That said, further research is needed to analyze how ordinary White Americans conceptualize diversity initiatives, affirmative action programs, and the relationship between them. For example, scholars working in this area could explore which aspects of affirmative action schemata matter most to people: the preferentially treatment of those belonging to a certain group, or the emphasis on public spending.

Furthermore, we find that exposure to information about a diversity initiative results in marketplace assessments that are moderated by racial attitudes, a finding in line with much of the literature on affirmative action. In real terms, those with high racial resentment do not want a Black Captain America and those with low racial resentment are at least more supportive of the idea. However, we do not find, as a result of treatment exposure, that racial resentment moderates support for diversity initiatives (or affirmative action), writ large, contrary to those same expectations. One possible explanation for this surprising finding (or lack thereof) is that racial resentment attitudes, in respect to affirmative action attitudes, reflect a reaction to a perceived zero-sum game in which Blacks are perceived as “stealing” jobs away (Bobo, 1998; Sidanius et al., 1996). For diversity initiatives in the entertainment industry, it can hardly be said that most people have material interests in the hiring practices of major film studios. This lack of material interest may diminish the relevance of certain racial attitudes, at least in this domain. Ultimately, however, this possible explanation cannot be tested with the data we have and future work should pursue this line of inquiry further, especially since the expectation set by the literature did not pan out.

What we present here is the first attempt at understanding attitudes stemming from exposure to diversity initiatives in popular media and elsewhere, and how those attitudes differ from what we traditionally understand about affirmative action attitudes. There is still much work to be done, and we suggest some of those future avenues of research above. We have shown that diversity initiatives have real effects on White Americans’ attitudes and intended behaviors in the marketplace, and these attitudes differ from those on affirmative action. Furthermore, we find that cheap talk about diversity initiatives is insufficient; if companies wish to leverage support for diversity among White Americans with low racial resentment (let alone people of color), their commitment to diversity must include meaningful action.

It seems that researchers can no longer afford to assume that White Americans conflate diversity initiatives and affirmative action; this assumption must be tested, as we have done here. But perhaps what is most striking about this work is that companies can no longer afford to merely announce diversity initiatives and then never follow through; they must actually start hiring people of color to reap the sort of support we find in this study. In real terms, actions speak louder than words and people can tell the difference between the two.

Supplemental Material

DS_10.1177_1077699020909308 – Supplemental material for Racializing Captain America: How Racial Attitudes Affect Perceptions of Affirmative Action and Diversity Initiatives in Media

Supplemental material, DS_10.1177_1077699020909308 for Racializing Captain America: How Racial Attitudes Affect Perceptions of Affirmative Action and Diversity Initiatives in Media by S. R. Gubitz and Denzel Avant in Journalism & Mass Communication Quarterly

Footnotes

Acknowledgements

We are grateful for suggestions and feedback from: Jamie Druckman, Mary McGrath, Tabitha Bonilla, Josh Robison, Don Green, Don Kinder, Lukas Otto, Amanda d’Urso, Matt Nelsen, Kumar Ramanathan, and Jessica Roden; participants of the 2019 MPSA, ICA Political Communication PhD student preconference, and Chicago Area Behavioral Workshop annual meetings; the Northwestern American Politics Student Workshop; the Northwestern Political Behavior Workshop; and two anonymous reviewers.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Support for this research was provided by Northwestern University.

ORCID iD

S. R. Gubitz

Supplemental Material

Supplemental material for this article is available online.

Notes

Author Biographies

S. R. Gubitz is a PhD candidate at the Department of Political Science, Northwestern University. He is the Mellon Cluster Fellow in Rhetoric and Public Culture. His research centers on political incivility, racial attitudes, and issues of race and media.

Denzel Avant is an independent researcher interested in racial attitudes and political incorporation of marginalized peoples. He received his Master of Arts in Public Policy and Master of Arts in Political Science from Northwestern University.

References

Barabas

(2004). How deliberation affects policy opinions. American Political Science Review, 98(4), 687–701. https://doi.org/10.1017/S0003055404041425

Baron

D. P.

(2001). Private politics, corporate social responsibility, and integrated strategy. Journal of Economics & Management Strategy, 10(1), 7–45. https://doi.org/10.1111/j.1430-9134.2001.00007.x

Baron

D. P.

Diermeier

(2007). Strategic activism and nonmarket strategy. Journal of Economics & Management Strategy, 16(3), 599–634. https://doi.org/10.1111/j.1530-9134.2007.00152.x

Bartels

L. M.

(1993). Messages received: The political impact of media exposure. American Political Science Review, 87(2), 267–285. https://doi.org/10.2307/2939040

Bobo

(1998). Race, interests, and beliefs about affirmative action: Unanswered questions and new directions. American Behavioral Scientist, 41(7), 985–1003. https://doi.org/10.1177/0002764298041007009

Bobo

(1999). Prejudice as group position: Microfoundations of a sociological approach to racism and race relations. Journal of Social Issues, 55(3), 445–472. https://doi.org/10.1111/0022-4537.00127

Bullock

J. G.

(2009). Partisan bias and the Bayesian ideal in the study of public opinion. The Journal of Politics, 71(3), 1109–1124. https://doi.org/10.1017/S0022381609090914

Carpini

M. X. D.

Keeter

(1996). What Americans know about politics and why it matters. Yale University Press.

Chen

A. S.

(2009). The fifth freedom: Jobs, politics, and civil rights in the United States, 1941-1972. Princeton University Press.

10.

Chudy

(2017). Racial sympathy in American politics (Doctoral dissertation, University of Michigan). https://deepblue.lib.umich.edu/bitstream/handle/2027.42/137000/jchudy_1.pdf?sequence=1&isAllowed=y

11.

Coyne

S. M.

Linder

J. R.

Rasmussen

E. E.

Nelson

D. A.

Collier

K. M.

(2014). It’s a bird! It’s a plane! It’s a gender stereotype! Longitudinal associations between superhero viewing and gender stereotyped play. Sex Roles, 70(9–10), 416–430. https://doi.org/10.1007/s11199-014-0374-8

12.

Crawford

V. P.

Sobel

(1982). Strategic information transmission. Econometrica, 50(6), 1431–1451. https://doi.org/10.2307/1913390

13.

DeSante

C. D.

(2013). Working twice as hard to get half as far: Race, work ethic, and America’s deserving poor. American Journal of Political Science, 57(2), 342–356. https://doi.org/10.1111/ajps.12006

14.

DeSante

C. D.

Smith

C. W.

(2017, April 6–9). Fear, institutionalized racism, and empathy: The underlying dimensions of whites’ racial attitudes [Paper presentation]. Presented at the Annual Meeting of the Midwest Political Science Association, Chicago, IL, United States.

15.

DeSante

C. D.

Smith

C. W.

(2020). Racial stasis: The millennial generation and the stagnation of racial attitudes in American politics (1st ed.). University of Chicago Press.

16.

DiMaggio

(1997). Culture and cognition. Annual Review of Sociology, 23(1), 263–287. https://doi.org/10.1146/annurev.soc.23.1.263

17.

Entman

R. M.

Rojecki

(2001). The black image in the white mind: Media and Race in America. University of Chicago Press.

18.

Frankenberg

(1993). Growing up white: Feminism, racism and the social geography of childhood. Feminist Review, 45(1), 51–84. https://doi.org/10.1057/fr.1993.38

19.

Frankenberg

(Ed.). (1997). Displacing whiteness: Essays in social and cultural criticism. Duke University Press.

20.

Fujioka

(2005). Black media images as a perceived threat to African American ethnic identity: Coping responses, perceived public perception, and attitudes towards affirmative action. Journal of Broadcasting & Electronic Media, 49(4), 450–467. https://doi.org/10.1207/s15506878jobem4904_6

21.

Georgetown University. (2018). American institutional confidence poll. https://bakercenter.georgetown.edu/aicpoll/

22.

Gilens

Sniderman

P. M.

Kuklinski

J. H.

(1998). Affirmative action and the politics of realignment. British Journal of Political Science, 28(1), 159–183. https://doi.org/10.1017/S0007123498000143

23.

Hirschman

Berrey

(2017). The partial deinstitutionalization of affirmative action in U.S. higher education, 1988 to 2014. Sociological Science, 4, 449–468. https://doi.org/10.15195/v4.a18

24.

Hoorn

J. F.

Konijn

E. A.

(2003). Perceiving and experiencing fictional characters: An integrative account. Japanese Psychological Research, 45(4), 250–268. https://doi.org/10.1111/1468-5884.00225

25.

Kartik

Ottaviani

Squintani

(2007). Credulity, lies, and costly talk. Journal of Economic Theory, 134(1), 93–116. https://doi.org/10.1016/j.jet.2006.04.003

26.

Kinder

D. R.

Mendelberg

(2000). Individualism reconsidered. In Sears

D. O.

Sidanius

Bobo

(Eds.), Racialized politics: The debate about racism in America (pp. 44–74). University of Chicago Press.

27.

Kinder

D. R.

Sanders

L. M.

(1990). Mimicking political debate with survey questions: The case of white opinion on affirmative action for blacks. Social Cognition, 8(1), 73–103. https://doi.org/10.1521/soco.1990.8.1.73

28.

Kinder

D. R.

Sears

D. O.

(1981). Prejudice and politics: Symbolic racism versus racial threats to the good life. Journal of Personality and Social Psychology, 40(3), 414–431. https://doi.org/10.1037/0022-3514.40.3.414

29.

Konijn

E. A.

Hoorn

J. F.

(2005). Some like it bad: Testing a model for perceiving and experiencing fictional characters. Media Psychology, 7(2), 107–144. https://doi.org/10.1207/S1532785XMEP0702_1

30.

Kuklinski

J. H.

Sniderman

P. M.

Knight

Piazza

Tetlock

P. E.

Lawrence

G. R.

Mellers

(1997). Racial prejudice and attitudes toward affirmative action. American Journal of Political Science, 41(2), 402–419.

31.

Lupia

McCubbins

M. D.

(1998). The democratic dilemma: Can citizens learn what they need to know? Cambridge University Press.

32.

Martin

J. F.

(2007). Children’s attitudes toward superheroes as a potential indicator of their moral understanding. Journal of Moral Education, 36(2), 239–250. https://doi.org/10.1080/03057240701325381

33.

McDonnell

M.-H.

King

B. G.

Soule

S. A.

(2015). A dynamic process model of private politics: Activist targeting and corporate receptivity to social challenges. American Sociological Review, 80(3), 654–678. https://doi.org/10.1177/0003122415581335

34.

Mendelson

(2018, August 4). Box office: “Black Panther” to top $700 million today. Forbes. https://www.forbes.com/sites/scottmendelson/2018/08/04/box-office-black-panther-to-top-700-million-today/

35.

Nacoste

R. W.

(1994). Policy schemas for affirmative action. In Heath

Tindale

R. S.

Edwards

Posavac

E. J.

Bryant

F. B.

Henderson-King

. . . Myers

(Eds.), Applications of heuristics and biases to social issues (Vol. 3, pp. 205–221). https://doi.org/10.1007/978-1-4757-9238-6

36.

Oatley

(1995). A taxonomy of the emotions of literary response and a theory of identification in fictional narrative. Poetics, 23(1), 53–74. https://doi.org/10.1016/0304-422X(94)P4296-S

37.

Pennell

Behm-Morawitz

(2015). The empowering (super) heroine? The effects of sexualized female characters in superhero films on women. Sex Roles, 72(5–6), 211–220. https://doi.org/10.1007/s11199-015-0455-3

38.

Pew Research Center. (2017). Race, immigration and discrimination. http://www.people-press.org/2017/10/05/4-race-immigration-and-discrimination/

39.

Piacenza

(2018, April 26). Superhero movies possess staying power with viewers. Morning Consult. https://morningconsult.com/2018/04/26/superhero-movies-possess-staying-power-with-viewers/

40.

Schons

Steinmeier

(2016). Walk the talk? How symbolic and substantive CSR actions affect firm performance depending on stakeholder proximity. Corporate Social Responsibility and Environmental Management, 23(6), 358–372. https://doi.org/10.1002/csr.1381

41.

Schulman

(2018, January 23). Is the era of #OscarsSoWhite over? The New Yorker. https://www.newyorker.com/culture/cultural-comment/is-the-era-of-oscarssowhite-over

42.

Shrum

L. J.

(1996). Psychological processes underlying cultivation effects further tests of construct accessibility. Human Communication Research, 22(4), 482–509. https://doi.org/10.1111/j.1468-2958.1996.tb00376.x

43.

Sidanius

Pratto

(2001). Social dominance: An intergroup theory of social hierarchy and oppression. Cambridge University Press.

44.

Sidanius

Pratto

Bobo

(1996). Racism, conservatism, affirmative action, and intellectual sophistication: A matter of principled conservatism or group dominance? Journal of Personality and Social Psychology, 70(3), 476–490.

45.

Sniderman

P. M.

Carmines

E. G.

(1997). Reaching beyond race. Harvard University Press.

46.

Sniderman

P. M.

Piazza

Tetlock

P. E.

Kendrick

(1991). The new racism. American Journal of Political Science, 35(2), 423–447. https://doi.org/10.2307/2111369

47.

Spence

A. M.

(1974). Market signaling: Informational transfer in hiring and related screening processes (1st ed.). Harvard University Press.

48.

Stimson

J. A.

(2015). Tides of consent: How public opinion shapes American politics. Cambridge University Press.

49.

Tan

Fujioka

Tan

(2000). Television use, stereotypes of African Americans and opinions on affirmative action: An affective model of policy reasoning. Communication Monographs, 67(4), 362–371. https://doi.org/10.1080/03637750009376517

50.

Tan

E. S.

(1996). Emotion and the structure of narrative film: Film as an emotion machine. Routledge.

51.

Tesler

(2012). The spillover of racialization into health care: How President Obama polarized public opinion by racial attitudes and race. American Journal of Political Science, 56(3), 690–704. https://doi.org/10.1111/j.1540-5907.2011.00577.x

52.

Tesler

(2015). Priming predispositions and changing policy positions: An account of when mass opinion is primed or changed. American Journal of Political Science, 59(4), 806–824. https://doi.org/10.1111/ajps.12157

53.

Tversky

Kahneman

(1973). Availability: A heuristic for judging frequency and probability. Cognitive Psychology, 5(2), 207–232. https://doi.org/10.1016/0010-0285(73)90033-9

54.

Tversky

Kahneman

(1974). Judgment under uncertainty: Heuristics and biases. Science, 185(4157), 1124–1131. https://doi.org/10.1126/science.185.4157.1124

55.

Valentino

N. A.

Hutchings

V. L.

White

I. K.

(2002). Cues that matter: How political ads prime racial attitudes during campaigns. American Political Science Review, 96(1), 75–90. https://doi.org/10.1017/S0003055402004240

56.

Valentino

N. A.

Neuner

F. G.

Vandenbroek

L. M.

(2017). The changing norms of racial political rhetoric and the end of racial priming. The Journal of Politics, 80(3), 757–771. https://doi.org/10.1086/694845

57.

Valet

(2018, April 17). America’s most reputable companies 2018. Forbes. https://www.forbes.com/sites/vickyvalet/2018/04/17/americas-most-reputable-companies-2018/

58.

Verba

Schlozman

K. L.

Brady

H. E.

(1995). Voice and equality: Civic voluntarism in American politics. Harvard University Press.

59.

Young

A. F.

Gabriel

Hollar

J. L.

(2013). Batman to the rescue! The protective effects of parasocial relationships with muscular superheroes on men’s body image. Journal of Experimental Social Psychology, 49(1), 173–177. https://doi.org/10.1016/j.jesp.2012.08.003

60.

Zaller

(1992). The nature and origins of mass opinion. Cambridge University Press.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

1.23 MB