Introduction
In the United States, diversity discourse is outwardly positive and celebratory. Most Americans praise the concept of diversity as an abstract ideal or a principle. However, research suggests that Americans may be less supportive about policies that promote racial diversity, and that they do not have a strong sense of personal closeness with others from diverse backgrounds. This study assesses and explores these contradictions and inconsistencies in Americans’ diversity attitudes via analysis of nationally representative survey data. First, we assess “principle-policy gaps” in Americans’ attitudes about diversity. This framework is commonly used in research about racial attitudes to illustrate how Americans’ support for racial equality in principle often outpaces their support for policies that redress racial inequality and promote racial diversity. Second, we assess “principle-personal gaps” in American’s diversity attitudes. This framework is less formally established in the scholarship, but several research literatures conjointly suggest that Americans’ support for diversity in principle outpaces their sense of personal closeness with others from diverse backgrounds.
We address core questions about these attitudinal gaps. Are they quantifiably tangible and prevalent in Americans’ diversity attitudes? How do they vary by individual-level demographics and other attitudinal factors? We operationalize the “principle-policy” gap by measuring differences between respondents’ abstract support for diversity in principle versus their support for policy associated with promoting diversity in institutional spaces. Similarly, we operationalize the “principle-personal” gap by measuring differences between respondents’ abstract support for diversity in principle versus their personal levels of private social distance towards people from different social out-groups. We analyze and combine several survey items that ask about these concepts with parallel Likert-type survey questions to generate additive scales with identical minimum and maximum, and we use these scales to analyze principle-policy gaps and principle-personal gaps in Americans’ diversity attitudes.
Findings show that these attitudinal gaps indeed exist and are substantial. Americans’ support for diversity in principle is measurably higher than their support for policies associated with promoting diversity and their sense of personal closeness with others from diverse backgrounds. Notably, both gaps have fairly normal distributions in the general population and are centered around a substantial level of attitudinal contradictions. We also find that the gaps have a substantial correlation; Americans with a greater principle-policy gap are likely to have a greater principle-personal gap.
Much of the research that informs this study has focused on White racial identity and political conservatism as core factors that relate to Americans’ attitudes and attitudinal gaps surrounding race, diversity, policy, and social difference in the United States. We find that White race and Republican partisanship have significant associations with these attitudinal gaps, albeit in contrasting ways. Racial identity and political partisanship are strong predictors of the principle-policy gap when modelled as unrelated items, but the interaction of the two is insignificant. The inverse is true, however; of the principle-personal gap. Racial identity and political partisanship prove weak predictors of this gap when modelled as unrelated items, but the interaction of the two is statistically significant. Essentially, race and partisanship seem to independently predict one’s principle-policy gap, yet the interaction of the two seems to predict one’s principle-personal gap.
That said, it would be reductive and inaccurate to say that only White Americans and political conservatives exhibit these attitudinal gaps. Regression models show that race, partisanship, and some other factors prove statistically significant predictors of the attitudinal gaps, but these attitudinal gaps are not necessarily overly-determined by any of the individual-level factors in our analysis. In fact, we find that Americans of various races and across the political spectrum exhibit diversity attitudes with substantial, tangible levels of principle-policy gap and principle-personal gap. Overall, these attitudinal gaps are fairly uniform and common the general population.
The study proceeds as follows. In our literature review, the first section describes research about diversity discourse and related attitudes in the United States, the second describes the framework of principle-policy gaps, and the third describes the framework of principle-personal gaps. The research design section details our research questions, hypotheses, and survey data we analyze. In the findings section, we demonstrate the existence of principle-policy gaps and principle-personal gaps in Americans’ diversity attitudes and analyze how these attitudinal gaps relate to individual-level factors, particularly racial identity and political partisanship. Our discussion section considers our core findings in conversation with previous research and suggests new directions for future study.
Literature Review
Diversity Discourse in the United States
The language of diversity and mainstream celebratory diversity discourse was institutionalized and popularized in the late twentieth century. In the latter years of the Civil Rights Movement, some universities and colleges voluntarily implemented policies to increase representation of underrepresented student groups, (e.g., people of color (POC), women, and the disabled). In the late 1970s and the early Reagan years, such policies became a politically polarized, hot-button issue (Kelly and Dobbin, Reference Kelly and Dobbin1998; Portocarrero and Carter, Reference Portocarrero and Carter2022). The 1978 Supreme Court case Bakke v. Regents of University of California banned racial quota systems in colleges, but a written opinion by Justice Lewis Powell legitimized a “subjective” assessment of applicant race in admissions procedures, stating that colleges’ pursuit of a diverse student body is constitutionally acceptable because of the benefits that diversity brings. This logic spurred the rise of diversity discourse and related policies among American colleges and corporations in the late twentieth century (Berrey Reference Berrey2015; Collins Reference Collins2011). This rationale was upheld by other Supreme Court cases Gratz v. Bollinger (2003), Grutter v. Bollinger (2003) and Fisher v. University of Texas (2016), cementing diversity as an institutional framework and common language during the twenty-first century.
Nationally representative survey data shows that most everyday Americans endorse the idea of diversity, at least when this concept is described in an abstract manner (Horowitz Reference Horowitz2019; Pew Research Center 2018; Rajasekar et al., Reference Rajasekar, Aguilar-Champeau and Hartmann2022). Diversity is lauded as a benefit for institutions and society at large, and outward support for diversity is entwined with social norms of denouncing classical racism and bigotry. Diversity discourse that celebrates racial difference by prioritizing the benefits of diversity was first popularized in colleges and corporations, but is now common in many settings, including churches (Barron Reference Barron2016), K-12 schools (Woody Reference Woody2020), and neighborhoods (Aptekar Reference Aptekar2015; Hoekstra and Gerteis, Reference Hoekstra and Gerteis2019; Mayorga-Gallo Reference Mayorga-Gallo2014). However, an overarching theme in the literature about diversity discourse is that such surface-level celebration of diversity can fall short of concretely pursuing racial equality, promoting inter-racial harmony, and challenging deep-seeded racial hierarchies in the United States.
Diversity discourse can act as “happy talk” (Bell and Hartmann, Reference Bell and Hartmann2007) that celebrates racial difference without acknowledging racial inequality, which can mask or ignore persistent racial inequalities in colleges and corporations (Berrey Reference Berrey2015; Collins Reference Collins2011; Luhr Reference Luhr2023; Thomas Reference Thomas2018). Diversity discourse and related attitudes are associated with colorblind racial beliefs and meritocratic narratives that deny the existence of racial inequality or blame such inequalities on a supposed lack of work ethic and proper values among people of color (Embrick Reference Embrick2011; Petts Reference Petts2020; Rajasekar et al., Reference Rajasekar, Aguilar-Champeau and Hartmann2022). Diversity discourse celebrates inter-racial harmony and cross-cultural exchange, but related policy and messaging can exoticize racial minorities as outsiders, reinforce Whiteness as the central American race, and fall short of fostering inter-racial social networks and harmony (Mayorga-Gallo Reference Mayorga-Gallo2019; Petts and Garza, Reference Petts and Garza2021; Woody Reference Woody2020).
The diversity discourse literature has often focused how White Americans think about diversity. In college settings, some White college students speak very positively about the importance of diversity on campus in the abstract, but they also express mixed feelings about diversity policy and diversity programming pursuing that goal (Hikido and Murray, Reference Hikido and Murray2016; Thomas Reference Thomas2018; Warikoo Reference Warikoo2016). Research about White managers’ and supervisors’ attitudes towards diversity policy finds a similar theme of theme of praising workplace diversity in the abstract while also expressing misgivings about policy meant to promote diversity and increase representation (Berrey Reference Berrey2015; Embrick Reference Embrick2011; Smith and Hunt, Reference Smith and Hunt2021). In diversifying communities, White residents can also exhibit contradictory attitudes, valorizing diversity in principle while also expressing reservations about the rising number of racial and ethnic minorities in the community (Aptekar Reference Aptekar2015; Hoekstra and Gerteis, Reference Hoekstra and Gerteis2019; Mayorga-Gallo Reference Mayorga-Gallo2014).
Additionally, political conservatism has been a focus in this literature. Diversity policy has historically been opposed by conservative political mobilization and Republican-led efforts since the Reagan era and continues today (Berrey Reference Berrey2015; Kelly and Dobbin, Reference Kelly and Dobbin1998; Portocarrero and Carter, Reference Portocarrero and Carter2022). In June 2023, the Supreme Court ruled on the case Students for Fair Admission v. Harvard. The Republican-majority court broke precedent with the other Supreme Court cases described above and ruled that colleges’ subjective consideration of applicant race in pursuit of promoting campus diversity is an unconstitutional practice. Nationally representative surveys about Americans’ attitudes about diversity find that White Americans and political conservatives hold less favorable views overall than non-Whites and liberals (Horowitz Reference Horowitz2019; Pew Research Center 2018; Rajasekar et al., Reference Rajasekar, Aguilar-Champeau and Hartmann2022). Following these themes, our analysis considers how racial identity and political partisanship relate to principle-policy gaps and principle-personal gaps in Americans’ diversity attitudes.
The Principle-Policy Gap
The framework of the “principle-policy gap” is well-known in literature about racial attitudes in the United States. After the Civil Rights Era, classical bigotry and explicit racism greatly declined. But newer prejudicial discourse emerged, using coded language about culture, morals, and meritocratic values to denigrate racial minorities; such discourse also can trivialize, deny, or justify ongoing racial inequalities in the post-Civil-Rights Era (Bobo and Charles, Reference Bobo and Charles2009; Bobo et al., Reference Bobo, Charles, Krysan, Simmons and Marsden2012; Kinder and Sanders, Reference Kinder and Sanders1996; Kuklinski et al., Reference Kuklinski, Sniderman, Knight, Piazza, Tetlock, Lawrence and Mellers1997; McClosky and Zaller, Reference McClosky and Zaller1984; Sidanius et al., Reference Sidanius, Pratto and Bobo1996). Several scholarly theories have conceptualized such racial attitudes and related discourse, (e.g., “modern racism” (McConahy 1986), “racial resentment” (Kinder and Sanders, Reference Kinder and Sanders1996), “symbolic racism,” (Brandt and Reyna, Reference Brandt and Reyna2012; Sears and Henry, Reference Sears and Henry2003), “laissez-faire racism” (Bobo et al., Reference Bobo, Kluegel, Smith, Tuch and Martin1997), and “colorblind racism,” (Bonilla-Silva Reference Bonilla-Silva2003; Bonilla-Silva and Dietrich, Reference Bonilla-Silva and Dietrich2011; Burke Reference Burke2017)). Such theories and others have their own unique complexities and details. But, as Steven A.Tuch and Michael Hughes (Reference Tuch and Hughes2011) write, “a topic of long-standing interest in racial attitudes scholarship is the discrepancy between whites’ strong support for principles of racial equality on one hand and their intransigence on policies designed to redress that inequality on the other” (p. 135).
The principle-policy gap framework is well-established in the racial attitudes literature, and research consistently finds that supportive attitudes about racial equality in principle can outpace support for policy meant to promote racial diversity and equality, such as affirmative action or school busing programs (Federico and Sidanius, Reference Federico and Sidanius2002; Kinder and Sanders, Reference Kinder and Sanders1996; Kuklinski et al., Reference Kuklinski, Sniderman, Knight, Piazza, Tetlock, Lawrence and Mellers1997; Taylor and Parcel, Reference Taylor and Parcel2019; Tuch and Hughes, Reference Tuch and Hughes1996, Reference Tuch and Hughes2011), sometimes referred to as a “principle-implementation” gap (e.g., Dixon et al., Reference Dixon, Durrheim and Thomae2017), This framework is also evident in research about diversity discourse. Corporate employees and college students endorse diversity in principle, but they can also express ambivalence about diversity policies, efforts, or programs in their institutions (Berrey Reference Berrey2015; Embrick Reference Embrick2011; Hikido and Murray, Reference Hikido and Murray2016; Warikoo Reference Warikoo2016). In research regarding how White Millennials think about diversity, Candis Watts Smith and Sarah Mayorga-Gallo (Reference Smith and Mayorga-Gallo2017) uncover “a new iteration of the principle-policy gap. Young people support diversity and the presence of people of color in predominantly white spaces—in principle—but they do not necessarily support the policies that are aimed to increase diversity” (p. 904). Similarly, studying beliefs about diversity in the workplace, Jamillah Bowman Williams and Jonathan M. Cox (Reference Bowman Williams and Cox2022) find evidence of a principle-practice gap, (i.e., “inconsistency between general support for the principle of diversity… and the actions one is willing to take to actually promote diversity” (p. 7)).
Research about the principle-policy gap has often focused on how White racial identity and/or conservative political belief relate to these attitudinal inconsistencies (Bobo and Charles, Reference Bobo and Charles2009; Bobo et al., Reference Bobo, Charles, Krysan, Simmons and Marsden2012; Hunt and Smith, Reference Hunt and Smith2021; Sidanius et al., Reference Sidanius, Pratto and Bobo1996; Tuch and Hughes, Reference Tuch and Hughes2011). But racially progressive and politically liberal Whites can also express contradictions in their attitudes about race and race-related issues. In Jennifer Chudy’s study (2023) of White racial justice activists, participants who are genuinely and deeply invested in addressing inequalities faced by Black Americans still suggest individual-level solutions rather than structural-level changes for addressing racial inequality and discrimination. Geneva Cole (Reference Cole2020) studies how patterns in Whites’ self-identification with their racial identity relates to their thoughts about the Black Lives Matter movement. Whites who recognize and acknowledge their White privilege express support for the movement’s goals, but some are less supportive of how the movement pursues those goals. Experimental-survey research by Efrén O. Pérez and colleagues (Reference Pérez, Kuo, Russel, Scott‐Curtis, Muñoz and Tobias2022) finds that when White Democrats are primed with a sense of racial threat, they express more-conservative attitudes that protect Whiteness and Whites when it comes to racially-coded topics such as immigration and legacy admissions in college. Notably, being primed with a sense of racial threat does not lead White Democrats to adopt more-conservative attitudes to non-racialized topics such as federal spending on scientific education and public infrastructure.
Relatedly, while much of the research has focused on Whites’ attitudes in racial attitudes literature, non-White Americans can also subscribe to negative racial stereotypes (Bobo et al., Reference Bobo, Charles, Krysan, Simmons and Marsden2012; Gay Reference Gay2006; Hunt Reference Hunt2007; King et al., Reference King, Phillips, Kivisalu and O’Toole2015) and colorblind racial attitudes (Chan Reference Chan2020; Gonlin and Cambell, Reference Gonlin and Campbell2017; Hartmann et al., Reference Hartmann, Croll, Larson, Gerteis and Manning2017). Additionally, non-White Americans also can subscribe to explanations for racial inequality that focus on individualistic rather than structural frames, as well as express misgivings about policy associated with promoting racial equality and diversity (Croll Reference Croll2013; Hunt Reference Hunt2016). Following such themes in previous research, our analysis is attentive to how racial identity and political partisanship relate to principle-policy gaps in Americans’ diversity attitudes.
The Principle-Personal Gap
The terminology of the “principle-personal gap” is less formally established in social science scholarship, but we feel such a framework is highly relevant to the contemporary context. America is experiencing increasing ethnoracial and religious diversity (Frey Reference Frey2018; Hout and Fischer, Reference Hout and Fischer2014), growing political and affective polarization (Baldarassi and Park, Reference Baldassarri and Park2020; Iyengar et al., Reference Iyengar, Lelkes, Levendusky, Malhotra and Westwood2019; Perry 2020), ongoing social distance among Americans from different backgrounds (Bishop Reference Bishop2009; Smith et al., Reference Smith, McPherson and Smith-Lovin2014; Stewart et al., Reference Stewart, Edgell and Delehanty2018), and reinvigorated White nationalism in the political arena (Craig et al., Reference Craig, Rucker and Richeson2018; McVeigh and Estep, Reference McVeigh and Estep2019; Mutz Reference Mutz2018). Based on this social and political context, we describe how several research literatures suggest that Americans’ support for diversity in principle likely outpaces their sense of personal closeness with people from diverse backgrounds, which has important implications in a diversifying United States.
Our framework is informed by literature about social distance and homophily. Research has found that Americans are generally socially distant from others of different backgrounds, and that they prefer associating with those of similar race, religion, and political belief (Bishop Reference Bishop2009; Gimpel and Hui, Reference Gimpel and Hui2015; Karakayali Reference Karakayali2009; Smith et al., Reference Smith, McPherson and Smith-Lovin2014). Relatedly, research about neighborhood preferences finds that despite widespread support for racial integration, Americans from a variety of backgrounds express mixed feelings about actually living in racially diverse neighborhoods themselves (Hall and Krysan, Reference Hall and Krysan2017; Havekes et al., Reference Havekes, Bader and Krysan2016; Krysan et al., Reference Krysan, Carter and van Londen2017). A similar theme appears in research regarding how a diverse college environment and college diversity programs impact diversity in students’ friendships, relationships, and overall social network. Overall, depending on student and/or institutional factors, the abstract celebration of diversity is no ironclad guarantee that a campus becomes a utopia of inter-racial friendships and harmony, nor that college graduates will create diverse social networks and maintain them after graduation (Bowman and Park, Reference Bowman and Park2014; Fischer Reference Fischer2008, Reference Fischer2011; Stearns et al., Reference Stearns, Buchmann and Bonneau2009).
The principle-policy gap framework is also informed by research about Americans’ reactions to changing demographics. Even though everyday narratives of an impending “majority-minority” nation can be overblown (Alba Reference Alba2018), American politics, media, and culture have popularized this talking point. Research has studied how Americans react to the idea of such demographic change in the nation and the community (Craig et al., Reference Craig, Rucker and Richeson2018; Craig and Richeson, Reference Craig and Richeson2014; Craig and Richeson, Reference Craig and Richeson2018; Danbold and Huo, Reference Danbold and Huo2015; McConnell and Rodríguez-Muñiz, Reference McConnell and Rodríguez-Muñiz2023), especially as related to the 2016 election and the Trump campaign (Knowles and Tropp, Reference Knowles and Tropp2018; Major et al., Reference Major, Blodorn and Blascovich2018; McVeigh and Estep, Reference McVeigh and Estep2019; Mutz Reference Mutz2018; Myers and Levy, Reference Myers and Levy2018). Overall, despite everyday Americans’ outward celebration of diversity, their attitudes towards the browning of America appear mixed at best.
Relatedly, an international literature has investigated Robert D. Putnam’s “constrict thesis” (Reference Putnam2007), which suggests that increasing community diversity in Western societies creates a “hunkering-down effect,” negatively impacting local social cohesion and community social capital. Subsequent research has found mixed evidence, revising and nuancing this initial thesis by pointing to how factors such as local economic conditions and racial attitudes shape this picture (for reviews, see Abascal and Baldarassi, Reference Abascal and Baldassarri2015; Portes and Vickstromm, Reference Portes and Vickstrom2011; van der Meer and Tolsma, Reference van der Meer and Tolsma2014). This ongoing inquiry also gives us reason to suspect that Americans’ celebration of diversity in principle is higher than their sense of personal closeness with people from diverse backgrounds.
Finally, the framework of the principle-personal gap is evident in studies about diversity discourse in diversifying American communities. Many residents are supportive of community diversity in principle, but some also express misgivings and coded prejudices about newcomers in their locale. Additionally, local government, civic organizations, and everyday interactions can create and reinforce symbolic boundaries that mark White residents as belonging to the community and non-White residents as interloping, unwelcome outsiders (Aptekar Reference Aptekar2015; Berrey Reference Berrey2015; Mayorga-Gallo Reference Mayorga-Gallo2014; Hoekstra and Gerteis, Reference Hoekstra and Gerteis2019).
The research cited in this section has also focused on White race and political conservatism as core factors related to diversity attitudes and attitudinal inconsistences. For example, in research about how college diversity facilitates inter-racial friendships, White students are sometimes less likely to create social connections than non-White students, particularly depending on institutional context (Bowman and Park, Reference Bowman and Park2014; Fischer Reference Fischer2008). Additionally, White Americans are more likely to express negativity or mixed feelings about changing demographics in the nation and the community (Craig and Richeson, Reference Craig and Richeson2014; Danbold and Huo, Reference Danbold and Huo2015; McConnell and Rodríguez-Muñiz, Reference McConnell and Rodríguez-Muñiz2023). In their review article of such research, Maureen A. Craig and colleagues (Reference Craig, Rucker and Richeson2018) write, “collectively, this work suggests that whites are threatened by these changes, which is likely to reduce support for racial and ethnic integration and race-conscious efforts to redress racial inequality” (p. 191). The idea of increasing ethnoracial diversity in the nation and/or community can activate status threat among White Americans, which can be linked to conservative political belief and Republican partisanship, especially in the context of the 2016 election and the Trump campaign (Craig and Richeson, Reference Craig and Richeson2018; Major et al., Reference Major, Blodorn and Blascovich2018; McVeigh and Estep, Reference McVeigh and Estep2019; Mutz Reference Mutz2018; Knowles and Tropp, Reference Knowles and Tropp2018).
To summarize, our literature review has described how Americans’ outward celebration diversity in principle belies underlying complexities and attitudinal contradictions. Americans generally seem less enthusiastic about policies meant to facilitate racial equality and promote diversity, as described by the established framework of the principle-policy gap. Additionally, Americans might not feel a sense of personal closeness with people from diverse backgrounds, which we conceptualize as a principle-personal gap. The current study analyzes nationally representative survey data to explore and assess these attitudinal gaps in Americans’ diversity attitudes. Following the literature, we analyze how individual-level factors, particularly racial identity and political partisanship, relate to these gaps.
Research Design
Research Questions and Hypotheses
Our literature review and theoretical discussion yields a core set of research questions and corresponding hypotheses, with specific attention to how racial identity and political partisanship relate to these gaps in diversity attitudes.
These attitudinal gaps simply may not exist in the general population, or they might be minimal at best. To assess this, we construct three scale items representing (1) support for diversity in principle, (2) attitudes about policy associated with racial diversity, and (3) feelings of personal closeness with people from diverse backgrounds. We then construct two scales that measure the principle-policy and principle-personal gaps, the subject of Research Question 1 and Research Question 2.
RQ 1: Is there a principle-policy gap in Americans’ diversity attitudes?
RQ 2: Is there a principle-personal gap in Americans’ diversity attitudes?
H1.0: There is no substantial difference between Americans’ attitudes about diversity in principle and their attitudes about policy associated with racial diversity.
H1.A: Americans’ attitudes about diversity in principle are substantially higher than their attitudes about policy associated with racial diversity.
H2.0: There is no substantial difference between Americans’ attitudes about diversity in principle and their feelings of closeness with people from diverse backgrounds.
H2.A: Americans’ attitudes about diversity in principle are substantially higher than their sense of personal closeness with others from diverse backgrounds.
We also consider how these attitudinal gaps relate to participant individual-level factors. Much of the literature has analyzed how White race and political conservatism relate to the gaps, contradictions, and underlying dimensions of Americans’ diversity and racial attitudes, informing our focus on racial identity and political partisanship in Research Question 3.
RQ 3: How do personal factors relate to the scale items and the gap items?
H3.0: Individual-level factors prove insignificant predictors of the scale and/or gap items.
H3.A1: Individual-level factors such as race and partisanship prove significant predictors of the scale and/or gap items, and levels of attitudinal gap items will prove very different across Americans of varying racial identity and political partisanship.
H3.A2: Individual-level factors such as race and partisanship prove significant predictors of the scale and/or gap items, but levels of attitudinal gap will not prove very different across Americans of varying racial identity and political partisanship.
The gaps may be very similarly distributed among Americans from different backgrounds and with different beliefs, and regressions may not uncover any significant associations between individual-level factors and the attitudinal gaps, as considered by H3.0. By contrast, the regressions may reveal that individual-level factors, particularly racial identity and political partisanship, have statistically significant associations with the attitudinal gaps and prove highly relevant to predicting who exhibits more attitudinal gap, as considered by H3.A1. On the other hand, even if regressions uncover statistically significant associations, the attitudinal gaps may not actually exhibit large, substantial differences across participants with different racial identities and political partisanship, as considered by H3.A2. This third hypothesis considers the above-described recent research about how white liberals can express contradictory and conservative-leaning attitudes about race-related issues (Chudy Reference Chudy2023; Cole Reference Cole2020; Pérez et al., Reference Pérez, Kuo, Russel, Scott‐Curtis, Muñoz and Tobias2022). This hypothesis also considers how people of color can sometimes subscribe to racial prejudices or negative stereotypes (Bobo et al., Reference Bobo, Charles, Krysan, Simmons and Marsden2012; Hunt Reference Hunt2007; King et al., Reference King, Phillips, Kivisalu and O’Toole2015), as well as how people of color can exhibit dimensions of colorblind racism and identify as colorblind themselves (Chan Reference Chan2020; Gonlin and Campbell, Reference Gonlin and Campbell2017; Hartmann et al., Reference Hartmann, Croll, Larson, Gerteis and Manning2017)
Data and Measures
We analyze the Boundaries in the American Mosaic (BAM) survey. This nationally representative survey data is well-suited for the current study because a large number of survey items are relevant to studying principle-policy and principle-personal gaps in Americans’ diversity attitudes. Contracted through GfK with funding from the National Science Foundation, the survey was fielded during a two-week period in early 2014. Participants were selected from GfK’s nationally representative panel sampling frame, which is based on recruiting respondents in English- and Spanish-speaking households via probability-based random address sampling from U.S. postal service records. Participant recruitment was completed via direct mail, telephone follow-up, and online registration (Couper Reference Couper2017). GfK supplies households with laptop and internet access as needed for survey completion. Participants received a cash incentive or credit for computer and Internet access as compensation for survey participation.
The recruitment rate for the primary sample was 13.9%, and the profile rate was 64.1%. The BAM sample was drawn from panel members via a probability proportional to size (PPS) weighted sampling approach. Of the 4353 people contacted, 2521 responded and took the survey, leading to a completion rate of 57.9%. Based on the GfK’s recruitment and profile rates for the panel sampling frame, the cumulative response rate was 5.2% (Callegaro and DiSogra, Reference Callegaro and DiSogra2008; DiSogra et al., Reference DiSogra, Callegaro and Hendarwan2009). The BAM sampling strategy was designed to slightly oversample POC participants to ensure diverse racial and ethnic representation within the survey sample. Our regressions are weighted to reflect the actual population of the United States.Footnote 1
Constructing The Key Measures
Our analysis is based on creating three additive scales and using them to create measures of the attitudinal gaps. These three scales are referred to as “principle-”, “policy-,” and “personal-scale.” Our key consideration was to select and recode measures so that the additive scales would have an identical minimum (0) and maximum (15) score, allowing us to meaningfully compare differences across each of the scales and calculate the gap measures.
Principle Scale
Based on five survey questions, this scale measures Americans’ support for diversity in principle. The first three items ask if it is “very important,” “somewhat important,” “not very important,” and “not important at all” that We value racial diversity, People can practice whatever religion they choose, and Everyone is treated equally. We recoded each of these measures on a 0–3 scale for importance. Another item asks participants if they “Strongly Agree,” “Somewhat Agree,” “Somewhat Disagree,” or “Strongly Disagree” with I value having people who are different from me in my community, which we also recoded on a 0–3 scale for agreement. The fifth item asks The United States is one of the most socially and culturally diverse nations in the world. Do you see this as mostly a strength, mostly a weakness, or equally a strength and a weakness? with five responses: “mostly a strength,” “somewhat a strength,” “equally a strength and a weakness,” “somewhat a weakness,” and “mostly a weakness.” We recoded this measure into a comparable 0–3 scale, assigning a score of 1.5 to the central, neutral category. The resulting additive scale has a minimum of 0 points and a maximum of 15, with good internal consistency (ɑ = 0.72).
Policy Scale
Based on five survey questions, this scale captures attitudes about policies associated with racial diversity. Three survey items ask participants if they “Strongly Agree,” “Somewhat Agree,” “Somewhat Disagree,” or “Strongly Disagree” with the following statements: African Americans should receive special consideration in job hiring and school admissions, African Americans should get economic assistance from the government, and Preferential treatment for racial minorities violates the principle of equal opportunity. Items were coded so that higher values represented support for race-conscious policy on our common 0–3 scale. Two additional items ask participants if they support affirmative action admission programs at colleges and universities and if they support the enforcement of anti-discrimination law. Item responses included reasons for support or opposition, which we recoded on our 0–3 scale. Respondents selected one of the four following responses to the affirmative action item, recoded with the corresponding scores in parentheses: “Maximizing diversity,” (3) “Ensuring equal access to education,” (2) “I do not support affirmative action programs,” (0) and “I have no opinion about affirmative action” (0). Then, respondents selected one of the four following responses to the anti-discrimination law item, recoded with the corresponding scores in parentheses: “Purposes of diversity and inclusion,” (3) “Legal compliance,” (2) “I do not support the enforcement of anti-discrimination law,” (0) and “I have no opinion about anti-discrimination law” (0) received the corresponding scores in parentheses. The resulting additive scale has a minimum of 0 points and a maximum of 15, with good internal consistency (ɑ = 0.67).
Personal Scale
This measure is based on Americans’ sense of personal closeness with others from diverse social backgrounds. Several survey questions ask participants how they would feel if their child married someone from certain groups, a measure of private social distance (Stewart et al., Reference Stewart, Edgell and Delehanty2018) that captures a more personal dimension than measures of public and civic expressions of social distance. The item reads, People can feel differently about their children marrying people from various backgrounds. Suppose your son or daughter wanted to marry someone from the different backgrounds listed here. Would you approve of this choice, disapprove of it, or wouldn’t it make any difference at all one way or the other? There are several groups listed following this introduction, and participants responded “Approve,” (3) “No Difference,” (1) or “Disapprove” (0) per group.Footnote 2 Our personal scale is based on participants’ responses regarding their child marrying a Muslim, an atheist, an African American person, a Hispanic person, a White person, and a recent immigrant, but each participant’s score is determined by their responses to five of these six groups.
For all respondents, their score on the personal diversity scale includes their attitudes about marrying (1) atheists, (2) Muslims, and (3) recent immigrants. The remaining two groups in the scale change based on the respondent’s self-reported race and ethnicity. White respondents have their attitudes about marrying (4) African Americans and (5) Hispanics included in the scale, but not for Whites. Similarly, Black respondents have their attitudes about marrying (4) Whites and (5) Hispanics included, but not for Blacks. Finally, Hispanic and respondents with other racial identities and/or mixed racial identities have their attitudes about (4) White Americans and (5) African Americans included, but not for Hispanics. This last group is aggregated for sample size considerations, given the limited number of those who identified racially as other or mixed-race among survey respondents. Thusly, participants’ scores on the personal scale were based on their responses to five of the six groups listed above, and the resulting additive scales each have a minimum of 0 points and a maximum of 15. All showed a high degree of internal consistency (White respondents’ alpha = 0.81, Black respondents’ alpha = 0.79, Hispanic and other respondents’ alpha = 0.80). Higher scores indicate greater favorability towards one’s child marrying outside the in-group.
Participant Individual-Level Factors
We use regressions to consider how individual-level factors predict principle-policy and principle-personal gaps. We describe the independent variables in our regressions below.
Demographic Measures
We consider demographic items, as is standard in research about racial attitudes (Bobo et al., Reference Bobo, Charles, Krysan, Simmons and Marsden2012; Hunt Reference Hunt2007; Quillian Reference Quillian2006) and principle-policy gaps (Taylor and Parcel, Reference Taylor and Parcel2019; Tuch and Hughes, Reference Tuch and Hughes2011). Drawing on self-reported information from BAM participants, we begin with Age in years, coded so that higher values reflect older age. We consider participant Race as a factor item via dummy codes for White non-Hispanic (the referent group), Black non-Hispanic, Hispanic, Other, or 2+ races. We use a binary measure of Gender (woman = 1) and a binary indicator of Sexuality (not-heterosexual =1)Footnote 3. We use a factor measure of Education with four categories: “Less than high school” (the referent group), “High School,” “Some college,” and “Bachelor’s degree or higher.” We measure Income based on nineteen possible income-brackets; higher scores represent more income. We draw on an item asking participants to describe their political identification as “Strong Democrat,” “Not Strong Democrat,” “Leans Democrat,” “Undecided/Independent/Other,” “Leans Republican,” “Not Strong Republican,” or “Strong Republican.” Ranging 1–7, the categorical measure Political Partisanship is coded so higher scores represent greater support for the Republican party and lower scores reflect greater support for Democrats. In regressions, Age, Income, and Political Partisanship are mean-standardized.
Attitudinal and Behavioral Measures
Following research about racial resentment, symbolic prejudices, and modern racism (Brandt and Reyna, Reference Brandt and Reyna2012; Kinder and Sanders, Reference Kinder and Sanders1996; McConahay Reference McConahay, Dovidio and Gaertner1986; Sears and Henry, Reference Sears and Henry2003), as well as research about anti-Muslim attitudes and anti-Semitism in the United States (Gerteis et al., Reference Gerteis, Hartmann and Edgell2020; Hobbs et al., Reference Hobbs, Lajevardi, Li and Lucas2023), we account for participants’ prejudices towards racial and religious minority groups. A BAM item reads, “Here is a list of potential problems in American society. For each problem, please mark all of the groups that contribute to the problem.” The list of problems reads, “They are a threat to order and public safety,” “They don’t share my morals or values,” “They take jobs and resources that should go to others, “They are dependent on welfare and government assistance,” “They are intolerant of others,” “They want to take over our political institutions,” and “They don’t contribute to my community.” We create an additive index of Prejudicial Attitudes towards racial and religious minority groups based on combining participants’ responses about Blacks, Hispanics, Asians, Muslims, and Jewish people (ɑ = 0.83).Footnote 4
Informed by scholarship about colorblind racism (Bonilla-Silva Reference Bonilla-Silva2003; Bonilla-Silva and Dietrich, Reference Bonilla-Silva and Dietrich2011; Burke Reference Burke2017), we measure participants’ colorblind racial attitudes based on three BAM items. The first asks if “Race no longer matters” in the United States. The second asks if “Racism is or will soon be a thing of the past.” The third reads, “For the most part, I’m color-blind, that is, I don’t see race.” Each item has four possible responses: strongly agree, somewhat agree, somewhat disagree, and strongly disagree. Not Colorblind is an index combining these items, coded so higher scores reflect lower colorblind racial attitudes (ɑ = 0.67).
We consider long-standing theories of racial threat (Blalock Reference Blalock1967) and group contact (Allport Reference Allport1954) about racial attitudes. Racial threat theory suggests that increasing minority group size and inter-racial interaction in the community can lead to racial animosity, especially if groups see themselves in competition for cultural power and material resources; conversely, contact theory suggests that increased inter-group contact and interaction can sometimes facilitate harmony and friendship, thereby decreasing racial animosity (Aberson Reference Aberson2015; Craig et al., Reference Craig, Rucker and Richeson2018; Quillian Reference Quillian2006; Richeson and Sommers, Reference Richeson and Sommers2016). The item Lower Racial Threat reads, “I often feel threatened by other racial groups,” coded so that higher values reflect lower racial threat. The item Lower Diverse Friends reads, “there IS a lot of social and cultural diversity among my friends,” coded so that higher values reflect lower inter-group contact. Both items have four Likert-type responses ranging from strongly agree through strongly disagree.
One’s economic self-interest and financial insecurity may drive negativity towards affirmative action and diversity policy (Crosby et al., Reference Crosby, Iyer and Sincharoen2006; Oh et al., Reference Oh, Choi, Neville, Anderson and Landrum-Brown2010). Therefore, we include an item that asks participants how their finances today compare to five years ago. Financially Worse Off marks those who chose “Worse” as 1 and those who chose “Better” or “About the Same” as 0. Additionally, research has considered how “social dominance orientation” (SDO; i.e., the acceptance of social hierarchy, stratification, and inequalities as natural) can predict opposition to affirmative action (Crosby et al., Reference Crosby, Iyer and Sincharoen2006; Ho and Unzueta, Reference Ho and Unzueta2015). A BAM item asks participants if the “Increasing gap between the rich and poor” represents a problem in America; participants chose from “Very Serious Problem,” “Somewhat serious problem,” “Not a very serious problem,” and “Not a very serious problem at all.” This item, Stratification Endorsement, is coded so that higher values indicate greater SDO and acceptance of inequality.
Following Putnam’s “constrict thesis” (Reference Putnam2007), researchers have investigated whether the ethnoracial composition of neighborhoods impacts community engagement, social capital, and neighborhood attachment. Therefore, we created the additive index Neighborhood Disconnection from four BAM, each with four Likert responses ranging from strongly agree to strongly disagree: “I feel safe walking alone at night in my neighborhood,” “The people in my community really care about their neighbors,” “Most people in my community share the same basic values that I do,” and “It would not bother me too much to move from here into some other community” (alpha = 0.76). Higher scores reflect greater disconnect from one’s neighborhood.
Finally, research finds that anxieties and resentments about immigration are linked to racial backlash, status threat, and political mobilization, especially in relation to how Americans react to changing demographics and the idea of a majority-minority nation (Craig and Richeson, Reference Craig and Richeson2014; Mutz Reference Mutz2018; Myers and Levy, Reference Myers and Levy2018). We therefore include the item Permit Immigration, which asks if “The U.S. should do more to limit immigration” with four Likert-style responses from strongly agree through strongly disagree, coded so that higher values indicate more favorability towards immigration.
See Table 1 for summary statistics for our core measures and Table 2 for the correlations between the three diversity scales. See our supplementary materials for descriptive statistics of the predictor variablesFootnote 5.
Descriptive Statistics of Core Measures
Correlations Between Diversity Scale Measures
The Finding section begins with visualizations of the three scale items and the two gaps. We then present OLS regressions that model the gaps as predicted by the individual-level factors described above, and we analyze how adding interaction items for racial identity and political partisanship affects regression results. We also present visualizations that further explore how racial identity, political partisanship, and their interaction items relate to the attitudinal gaps.Footnote 6
Findings
Figure 1 presents kernel density estimates of Americans’ attitudes towards the three scale items. We see that Americans’ scores on the principle-scale substantially differ from scores on the policy-scale and the personal-scale. The principle-scale has a mean of 12.06 and a median of 12.5, substantially larger than the policy-scale (mean of 5.25 and median of 5) and the personal-scale (mean of 5.44 and median of 5). Then, distribution of the principle-scale has the least normal shape in Figure 1 with a skew of -1.24, more pronounced and negative than the policy-scale (skew of +0.36) and the personal-scale (skew of +1.14). Overall, these initial results show that the American general population is more favorable towards the principle-scale than towards the policy-scale and the personal-scale.
Distributions of Scale Measures.
Figure 2 presents kernel density estimates of the principle-policy and principle-personal gaps, derived by subtracting respondents’ scores on the policy-scale and the personal-scale from their scores on the principle-scale. Respondents with positive scores exhibit more favorability towards diversity in principle than their support for policy that promotes diversity and their sense of personal closeness with others from diverse backgrounds. A large majority of respondents have positive scores on both curves in Figure 2, and the distributions center around a tangible, substantial level of attitudinal gap corresponding to over a standard deviation of change on each of the individual scales. Notably, both attitudinal gap distributions are fairly symmetrical. While neither are a perfect normal curve, they are quite close. This suggests that most Americans hold some level of the principle-policy gap (mean of 6.80) and the principle-personal gap (mean of 6.49). Overall, Figure 2 shows that principle-policy gaps and principle-personal gaps clearly exist in Americans’ attitudes about diversity, which rejects hypotheses H1.0 and H2.0.
Distributions of Gap Measures.
Notes: Diversity scale gaps are calculated by taking the difference of the “Diversity in Principle” scale measure and the “Diversity in Policy” and “Diversity in Personal Life” scale measures. Higher positive values indicate a respondent agreed more strongly with diversity in principle than they agreed with either policy or intermarriage survey items.
Table 3 compares model fit of OLS regression models predicting the scales and gaps based on the independent variables described earlier. We explore the OLS results for the gaps more deeply below, but at the moment we highlight the differences in model fit (see our supplementary materials for full results of OLS models predicting the scale items). Table 3 shows an R2 of 0.31 for the principle-scale, 0.32 for the policy-scale, and 0.17 for the personal-scale. But, we see an R2 of 0.166 for the principle-policy gap and 0.093 for the principle-personal gap. Across the board in Table 3, regressions predicting the scales have superior model fit than regressions predicting the gaps derived from those scales. This is notable because all five models use the same list of independent variables and because they all include over 91% of respondents in the BAM survey, meaning their samples are mostly the same respondents. Overall, these differences in explained variance suggest that the attitudinal gaps are more difficult to explain based on the predictors derived from the literature than the three scale items.
Comparing Model Fit of OLS Regressions of Scales and Gaps
Figure 3 assesses how the gaps relate to one another. The scatterplot and the corresponding fit-line demonstrate a clear association between the two gaps, with a substantial correlation (r=0.30). Essentially, people who exhibit one of these attitudinal gaps in their diversity attitudes are likely to also exhibit the other attitudinal gap.
The Association Between The Attitudinal Gaps.
Notes: Jittered scatterplot suggests a reasonably strong association (r=0.30) between Principle-Policy and Principle-Personal gaps in diversity attitudes. Respondents with higher scores on each measure tend to express weaker support for tangible diversity outcomes, relative to their support for diversity in principle.
We now explore how individual-level factors relate to the attitudinal gaps. Table 4 presents results from two OLS models that predict the principle-policy gap based on the independent variables described earlier, as depicted in Model 1. Model 2 uses the same predictor variables but also adds interaction terms for racial identity and political partisanship.
OLS Regression Results Predicting Principle-Policy Gaps
Notes: Two-tailed tests of significance, with *p<.05, **p<.01, ***p<.001. Standard errors in parentheses. Models incorporate BAM 2014 post-stratification and sampling weights. Model Ns vary because scale items were constructed using different survey outcome measures, with varying amounts of nonresponse.
In Model 1 of Table 4, Age has a strong, positive association with the principle-policy gap at the p<.001 level. Then, Black and Hispanic Americans are less likely than White Americans to exhibit a principle-policy gap (both at p<.001). Non-heterosexual people are less likely to exhibit this gap than heterosexual people (p<.05). Those with a high school degree (p<.01) and some college (p<.001) are more likely to exhibit this gap than Americans without a high school degree. Income has a positive association with the principle-policy gap (p<.05). Political partisanship has a positive association with this gap (p<.001), with Republicans more likely to exhibit this gap. Those with lower colorblind racial attitudes are less likely to exhibit this gap than those with higher colorblind racial attitudes (p <.01). Then, lower level of racial threat has a positive association with this gap, while a lower level of diversity in one’s friend group has a negative association with this gap (both at p<.001). Finally, those who are more disconnected from their neighborhood and those who hold favorable immigration attitudes are less likely to exhibit this gap (both at p<.001).
Notably, the results of Model 1 and Model 2 are virtually identical in Table 4. Adding the interaction terms for race and politics does very little to the model and none of these interaction terms are statistically significant. These results suggest that when it comes to predicting the principle-policy gap, racial identity and political partisanship are significant predictors, but the interaction terms for these items do not significantly relate to the principle-policy gap. To explore this further, Figure 4 provides box-and-whisker charts of mean principle-policy gap scores across Americans of different racial identity (left panel) and political partisanship (right panel). Then, based on Model 2 of Table 4, Figure 5 provides predicted principle-policy gaps based on the interaction of racial identity and political partisanship for Whites, Blacks, and Hispanics, net of the other predictors.
Mean Principle-Policy Gaps across Race and Political Partisanship.
Notes: Boxplots indicate differences in the distribution of attitudinal gap sizes by race and political party ID. Differences in principle-policy gaps are significant in both cases (Race F=48.12, p<.001; Party F = 25.04, p<.001).
Predicted Principle-Policy Gaps by Race & Politics Partisanship Interaction.
Notes: Figure illustrates differences in predicted principle-policy gap by partisanship and race. White, Black, and Hispanic Republicans do not exhibit significantly different associations between partisanship and gaps. Predictions are generated from Model 2 in Table 4 and include all controls, and slope differences are not statistically significant.
In Figure 4, we see statistically significant differences in the mean level of principle-policy gap across racial identity and political partisanship at the F<.001 level. White Americans exhibit higher levels of principle-policy gaps (mean = 7.45) than Black Americans (mean = 4.66) and Hispanic Americans (mean = 6.40). But, even though statistically significant differences exist between racial groups’ mean principle-policy gap, most Americans in all racial categories exhibit tangible, substantial levels of principle-policy gap. Similarly, there are statistically significant differences in mean principle-policy gap across Americans of different political partisanship; principle-policy gap rises with stronger support for the Republican party, with the strongest Republicans exhibiting the highest mean score of 8.02. That said, those who lean to the Democratic party also exhibit substantial levels of the principle-policy gap. The strongest Democrats exhibit a mean score of 5.77, which is far from minimal.
In Figure 5, results illustrate that as support for the Republican party rises, the predicted principle-policy gap rises uniformly for White, Black, and Hispanic respondents. There are no statistically significant differences between White, Black, and Hispanic respondents across political partisanship. Together, results from Table 4, Figure 4 and Figure 5 demonstrate that White racial identity and Republican partisanship predict a higher principle-policy gap in Americans’ diversity attitudes, but the effect of partisanship does not vary by race.
Table 5 presents results from two OLS regression models that predict the principle-personal gap based on the independent variables described earlier. There are two columns, corresponding to Model 1 and Model 2; the former includes the baseline list of predictor variables, and the latter includes interaction terms for race and political partisanship.
OLS Regression Results Predicting Principle-Personal Gaps
Notes: Two-tailed tests of significance, with *p<.05, **p<.01, ***p<.001. Standard errors in parentheses. Models incorporate BAM 2014 post-stratification and sampling weights. Model Ns vary because scale items were constructed using different survey outcome measures, with varying amounts of nonresponse.
In Model 1 of Table 5, Age has a strong positive association with the principle-personal gap (p<.001). Americans in the Other racial category are more likely than White Americans to express this gap (p<.05). Women are more likely to express this gap than men and non-heterosexual people are less likely to express this gap than heterosexual people (both at p<.001). Those with a high-school degree are more likely to express this gap than those without a high school degree (p<.05). Higher levels of prejudice are strongly, positively associated with expressing the principle-personal gap (p<.001). Those with lower diversity in their social network are less likely to express this gap (p<.01). Those who do not see stratification as a problem are less likely to express this gap, and those who support immigration are less likely to express this gap (both at p<.001).
Regression results predicting the principle-personal gap Model 1 and Model 2 in Table 5 are not exactly alike. Notably, the coefficients for Black race and Other race (when compared to White race) are more statistically significant in Model 2 than in Model 1. In Model 2, Black Americans are less likely to express this gap than White Americans (p<.05), and those in the Other racial category are more likely to express this gap than White Americans (p<.01 level). Political partisanship was an insignificant predictor in Model 1, but in Model 2, increased support for the Republican party is a positively associated with the principle-personal gap (p<.01). Then, the interaction terms for racial identity and political partisanship show that Black Republicans are less likely to express a principle-personal gap than White Republicans (p <.05) and that Hispanic Republicans are less likely to express this gap than White Republicans (p <.01). These results suggest that when it comes to predicting Americans’ principle-personal gap, the interaction of racial identity and political partisanship are at least somewhat a part of this picture.
To explore this further, Figure 6 provides box-and-whisker plots of the mean scores for principle-personal gap across Americans of different racial identity and political partisanship. Then, based on Model 2 of Table 5, Figure 7 provides predicted principle-policy gap scores for the interactions of racial identity and political participation for Whites, Blacks, and Hispanics, net of other predictor variables.
Principle-Policy Gaps across Race and Political Partisanship.
Notes: Boxplots indicate differences in the distribution of attitudinal gap sizes by race and political party ID. Differences in principle-personal gaps are marginally significant in both cases (Race F=2.83, p<.05; Party F = 2.65, p<.05), but of much smaller magnitude than personal-policy gaps.
Predicted Principle-Personal Gaps by Race & Political Partisanship Interaction.
Notes: Figure illustrates differences in predicted principle-personal gap by partisanship and race. White Republicans exhibit higher gaps (More support for diversity in principle than personal intermarriage) than Black or Hispanic Republicans. Predictions are generated from Model 2 in Table 5 and include all controls, and slope differences are statistically significant (Black vs White p<.05; Hispanic vs White p<.01).
Figure 6 shows that the magnitude of difference between Americans of differing race and partisanship in mean principle-personal gap is much smaller than the earlier findings regarding mean principle-policy gap in Figure 4. Significance tests show that while the mean level of gap differs by racial identity and political partisanship, those differences are only significant at the p<.05 level. Overall, Figure 6 demonstrates that mean levels of principle-personal gap are fairly substantial and uniform across Americans of differing racial identity and political partisanship, and there are minimal differences between non-Whites versus Whites and strong Democrats versus strong Republicans (see our supplementary materials for exact details about these means).
However, Figure 7 shows clear differences in the association between conservative political partisanship and principle-policy gaps for White, Black, and Hispanic respondents, and the interaction effect estimating these slope differences is statistically significant. For Whites, the predicted level of principle-personal gap increases as support for the Republican party increases. But, for Blacks and Hispanics, the predicted level of principle-personal gap decreases as support for Republicans increases. That is, non-White conservatives express more consistent rates of support for both diversity in principle and diversity in their personal lives than non-White liberals. Taken together, Table 5, Figure 6, and Figure 7 demonstrate that racial identity and political partisanship do not greatly predict differences in one’s principle-personal gap, but effect of partisanship on principle-personal gaps does vary by race.
That said, model results and their previously described lower explained variance show that principle-policy gaps and principle-personal gaps in Americans’ diversity attitudes are not over-determined by racial identity, political partisanship, or any of the other items in our analysis. Despite some statistically significant differences that the models uncover, model coefficients have fairly small magnitude of predicted difference in gap based on our predictor items. Overall, our findings demonstrate that these attitudinal gaps are fairly prevalent among Americans who vary by race, politics, and several other individual-level factors. Whites and non-Whites, as well as conservatives and liberals, exhibit substantial levels of both attitudinal gaps. This rejects hypotheses H3.0 and H3.A1.
In our discussion section, we summarize our main findings, and we situate them in in conversation with previous literature. We also consider directions for future research.
Discussion
Americans generally celebrate and endorse the concept of diversity as an abstract principle, but this outward positivity can belie a lack of support for policy meant to promote racial diversity and little sense of personal closeness towards others from diverse backgrounds. The current study has used nationally representative data to examine such principle-policy gaps and principle-personal gaps in Americans’ diversity attitudes. These attitudinal gaps indeed exist and are substantial among the general population. Notably, both have fairly normal distributions, even though the scales from which they were made had fairly pronounced skews. We also analyzed how a variety of potentially relevant individual-level factors variables predict these gaps. The explained variance of the regressions modeling the scales was substantially higher than that of the regressions modeling the gaps, suggesting that the gaps are fairly consistent across Americans from all walks of life. While racial identity, political partisanship, and some other items are associated with the attitudinal gaps, they are not overly deterministic of one’s predicted attitudinal gaps. Overall, a core takeaway of our analysis is that most Americans from varying backgrounds seem to hold these attitudinal gaps. Our study points to several considerations for future research, described below.
First, our study further highlights how the established scholarly framework of principle-policy gaps, commonly used in research about racial attitudes (Bobo et al., Reference Bobo, Charles, Krysan, Simmons and Marsden2012; Hunt and Smith, Reference Hunt and Smith2021; Tuch and Hughes, Reference Tuch and Hughes1996, Reference Tuch and Hughes2011), also applies to research about diversity attitudes and diversity discourse in the United States. Following other authors who have made this connection (e.g., Bowman Williams and Cox, Reference Bowman Williams and Cox2022; Smith and Mayorga-Gallo, Reference Smith and Mayorga-Gallo2017), our nationally representative analysis further highlights the gap between Americans’ celebration of diversity in principle versus their support for policy associated with addressing racial inequality and promoting racial diversity. As related to research that considers the nature and impacts of differing diversity policy, programming, and approaches in institutions (e.g., Dobbin et al., Reference Dobbin, Schrage and Kalev2015; Kalev et al., Reference Kalev, Dobbin and Kelly2006; Warikoo and Deckman, Reference Warikoo and Deckman2014), future authors will have to consider how to transform college administrators’ and corporate executives’ outward support for diversity in principle into concrete, tangible support for specific diversity initiatives in their organizations.
Second, this study has helped formalize the concept of a “principle-personal gap” as related to diversity attitudes. Our theorization is informed by research about social distance and homophily, Americans’ reactions to demographic change, inquiry following Putnam’s (Reference Putnam2007) constrict thesis, and how residents talk about diversity in changing neighborhoods. We hope future research helps further formalize and explore the framework of a principle-personal gap, as American society is experiencing growing racial and religious diversity (Frey Reference Frey2018; Hout and Fischer, Reference Hout and Fischer2014), ongoing social distance (Smith et al., Reference Smith, McPherson and Smith-Lovin2014; Stewart et al., Reference Stewart, Edgell and Delehanty2018) and rising polarization (Iyengar et al., Reference Iyengar, Lelkes, Levendusky, Malhotra and Westwood2019; Perry 2020). In this context, understanding and addressing the principle-personal gap will be necessary to foster diversity in everyday Americans’ social networks and build positive relations between different social groups.
Third, our study has considered how several individual-level factors relate to gaps in Americans’ diversity attitudes, and our findings could inform future inquiry. Participants’ age, diversity in their social networks, and immigration attitudes have fairly strong associations with both the principle-personal and principle-personal gaps. Following findings in previous literature, we have also found that White race and Republican partisanship relate to these attitudinal gaps, albeit in differing ways. White Americans and strong Republicans have a higher mean principle-policy gap when these items are treated as unrelated, and the predicted effect of partisanship does not vary by race. The inverse is true, however, for the principle-personal gap. Americans of differing racial identities and political partisanship exhibit similar mean levels of this gap, but White Republicans have a higher predicted gap than Black and Hispanic Republicans. Future research will be necessary in order to understand the mechanisms and consequences of how such individual factors relate to these attitudinal gaps, particularly the contrasting behavior of racial identity and political partisanship.
That said, and fourth, our analysis illustrates that the attitudinal gaps are not over-determined by racial identity, partisanship, or any of other individual-level factors in our models. Much of the research cited in this study has considered White race and political conservatism as core drivers of principle-policy gaps and principle personal gaps. But, it would be empirically inaccurate to say that only Whites and/or conservatives hold these gaps in Americans’ diversity attitudes; non-Whites and liberals also exhibit substantial levels of attitudinal gap. Future work will be needed to understand the causal mechanisms and potential impacts of these dynamics. Authors who tackle these questions should consider recent research about how White liberals can express mixed attitudes about social movements and structural-level policies that redress racial inequality (Chudy Reference Chudy2023; Cole Reference Cole2020; Pérez et al., Reference Pérez, Kuo, Russel, Scott‐Curtis, Muñoz and Tobias2022). Similarly, future research that builds on our work should consider how people of color can sometimes hold negative attitudes about other racial groups and even their own group (Bobo et al., Reference Bobo, Charles, Krysan, Simmons and Marsden2012; Cummings and Lambert, Reference Cummings and Lambert1997; Gay Reference Gay2006; Hunt Reference Hunt2007; King et al., Reference King, Phillips, Kivisalu and O’Toole2015), as well as how some people of color also express support for the underlying dimensions of colorblind racism and post-racial ideology (Chan Reference Chan2020; Gonlin and Campbell, Reference Gonlin and Campbell2017; Hartmann et al., Reference Hartmann, Croll, Larson, Gerteis and Manning2017)
This study does have some limitations. A survey analysis such as ours can only study attitudes and beliefs. Authors such as Bowman Williams and Cox (Reference Bowman Williams and Cox2022) have emphasized the importance of studying how inconsistencies in Americans’ attitudes about diversity predict the actual actions and practices they undertake for promoting organizational diversity. Then, our measure of the principle-personal gap is based on assessing Americans’ attitudes towards a variety of different out-groups, but our list of social groups and diverse backgrounds list is not necessarily exhaustive. Additionally, we lacked the measures to study how sexism, homophobia, ablism, and related prejudices relate to these attitudinal gaps, but this important to consider, especially as diversity discourse is the main framework for institutional policies that address the underrepresentation of women, sexual minorities, and the disabled (Dobbin et al., Reference Dobbin, Schrage and Kalev2015; Herring Reference Herring2009; Portocarrero and Carter, Reference Portocarrero and Carter2022). We hope future research using different research designs can address these unanswered questions.
Despite limitations, our core findings contribute to existing inquiry about American diversity discourse. Diversity is widely celebrated as an abstract principle, but Americans from all walks of life can exhibit (1) lower support for polies that redress racial inequality and promote diversity and (2) lower sense of personal closeness with others from diverse backgrounds. As the two gaps are substantial and fairly correlated, future research about diversity discourse, attitudes, and policy would benefit by being attentive to them both. Authors have highlighted the need to transform common ideas, everyday messages, and policy approaches surrounding diversity so as to promote a focus on concretely redressing ongoing racial inequality and underrepresentation in contemporary society (Berrey Reference Berrey2015; Embrick Reference Embrick2011; Mayorga-Gallo Reference Mayorga-Gallo2019; Portocarrero and Carter, Reference Portocarrero and Carter2022). Our study also highlights that diversity discourse must be transformed in ways that overcome social distance and foster social connections among people from diverse backgrounds. Otherwise, these gaps in everyday Americans’ diversity attitudes will likely persist, which has serious implications for a nation that is still rife with persistent racial inequalities and hierarchies yet is also becoming more diverse and polarized along several social dimensions. At the time of this writing, recent policy and legislation suggests that diversity discourse may be facing a new line of backlash; examples include the infamous “Stop Woke” bill in Florida and the 2023 Supreme Court case Students for Fair Admission vs. Harvard in which the Court ruled against race-conscious admissions policy. In this backdrop, we hope future research continues to explore the contradictions and complexities of Americans’ attitudes about diversity.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/S1742058X24000079.
Acknowledgements
The authors acknowledge and appreciate support for data collection and research assistance from the National Science Foundation (Grant Numbers 1258926 and 1258933) and the Edelstein Family Foundation.
