On the Varieties of Diversity: Ideological Variations in Attitudes Toward,and Understandings of Diversity

Abstract

Three studies explore the possibility that attitudes toward “diversity” are multidimensional rather than unidimensional and that ideological differences in diversity attitudes vary as a function of diversity subtype. Study 1 (n = 1,001) revealed that the factor structure of attitudes toward 23 diverse community features was bidimensional. Factors involving demographic and viewpoint diversity emerged. Conservatives reported more positive attitudes toward viewpoint diversity, and liberals more positive attitudes toward demographic diversity. Study 2 (n = 1,012) replicated Study 1 findings, and extended Study 1 results by showing attitudes toward the general concept of diversity predicted attitudes toward demographic diversity but not viewpoint diversity. In Study 3, 386 participants rated how relevant a set of features was to their prototypical understanding of diversity. A confirmatory factor analysis (CFA) revealed people discriminate between viewpoint, demographic, and consumer diversity. Conservatives perceived viewpoint features as more relevant to “diversity,” whereas liberals perceived demographic features as more relevant.

Keywords

diversity ideology attitudes prototypes politics

Contemporary American life is marked by two major trends, one sociopolitical and one sociodemographic. In the sociopolitical realm, recent decades have seen increased political and affective polarization (Bail et al., 2018; Iyengar et al., 2019; Mason, 2015; Tucker et al., 2018). Republicans and Democrats are increasingly divided on ideological issues and report greater animosity toward opposing party members (Iyengar et al., 2019). The sociodemographic trend is increased diversity (Frey, 2020; Lichter, 2013). Neighborhoods, schools, and businesses are increasingly populated by individuals with diverse ancestries (Craig et al., 2018), gender identities, sexual-orientations (Schmidt, 2021), religious beliefs (Mohamed, 2020), and ideological identities (Serhan, 2020).

Although both conservatives and liberals likely agree that diversity affects all levels of society, the two parties often disagree on how diversity-related policy should be crafted, obstructing significant societal change. Thus, the social significance of these contemporary trends highlights the need for research at their intersection. We report three studies that, in complementary ways, explore the intersection between political ideology, attitudes, and beliefs toward diversity.

Ideology and the Meanings of “Diversity”

Investigators working at this intersection of political ideology and diversity attitudes face a challenge. Although “few words in the current American lexicon are as ubiquitous” as diversity, “the actual meanings and function of the term . . . are difficult to pinpoint” (Bell & Hartmann, 2007, p. 895). Note that Bell and Hartmann (2007) wisely say “meanings,” plural. Like many words, “diversity” may mean different things to different people, and these differences in meaning may be systematic.

Different conceptions of “diversity” may derive from experiences with media. People selectively expose themselves to online and offline resources that are attitude confirming (Motyl et al., 2014). Liberals and conservatives thus experience distinctly different social media worlds and news networks (Barberá et al., 2015; Tucker et al., 2018), with many citizens obtaining information solely from ingroup media (Conover et al., 2011). As polarization increases, the news sources people encounter are increasingly partisan (Bail et al., 2018). In particular, the rhetoric surrounding diversity significantly differs between liberal (e.g., Hansler et al., 2021) and conservative (e.g., Carlson, 2020) new sources (Coe et al., 2008). Exposure to qualitatively different media sources could contribute not only to different attitudes toward diversity but also to different understandings of precisely what “diversity” means.

Any such variability in meaning would have implications for research on diversity attitudes. Extant findings suggest that, compared with conservatives, liberals have more positive attitudes toward diversity and policies that increase diverse representation (Federico & Sidanius, 2002; Sidanius et al., 1996). One interpretation of this finding is that liberals and conservatives hold different attitudes toward the same, fixed thing: “diversity.” An alternative interpretation is that a survey question about “diversity” (e.g., Danbold & Unzueta, 2020) activates different mental representations in the minds of liberals and conservatives. In linguistic terms, the word “diversity” may have different referents for the different ideological groups. In the language of categorization research, liberals and conservatives may differ in the category exemplars that come to mind when the category label “diversity” is presented (Rosch, 1999). Differences in expressed attitudes thus may, in part, reflect differences not only in feelings about diversity but also in what the term “diversity” means.

The Dimensionality of Diversity

If people assign varying meaning to the term diversity, then it follows that diversity is a multidimensional rather than unidimensional concept. Multidimensionality implies that people possess different understandings and attitudes toward different subtypes of diversity; attitudes toward one type of diversity would not necessarily predict attitudes toward other types. For example, in an interview study, when participants who initially expressed positive views toward the general concept of diversity were questioned about the diversity of specific social groups, they reported conflicting attitudes: positive attitudes toward some groups, and more negative appraisals of others (Bell & Hartmann, 2007).

Methods employed in extant studies often are not sensitive to this possible multidimensionality. Investigators have asked participants their opinion on diversity without specifying a type (Hentschel et al., 2013; Hofhuis et al., 2015; Hostager & De Meuse, 2008), or have used single attitude ratings to represent a person’s attitudes toward diversity across all groups and social contexts (Danbold & Unzueta, 2020; Guillaume et al., 2013; Roberson et al., 2017). Such methods cannot detect varying understandings of the term or possible ideological variation in feelings toward different diversity subtypes.

The potential multidimensionality of diversity may contribute to inconsistencies in the empirical literature (Guillaume et al., 2013). For example, in research relating diversity to satisfaction and productivity in work and academics, some studies find diversity to be helpful (Gates & Mark, 2012; Herring, 2009; E. T. Parker et al., 2016; Sommers, 2006), whereas others suggest it can be harmful (Gates & Mark, 2012; Hentschel et al., 2013; Pearsall et al., 2008; Thatcher & Patel, 2011). In research evaluating the impact of diversity training, findings vary depending on the culture where the training occurs (Guimond et al., 2013) and whether the company establishes an organizational responsibility to promote diversity (Kalev et al., 2006). Findings regarding the types of people who are most open to diversity also are mixed. Some studies report that liberals are more open to interacting with people from different backgrounds (Chirumbolo & Leone, 2010; Verkuyten & Yogeeswaran, 2020), whereas others find liberals and conservatives to be equally closed off to outgroup members (Chambers et al., 2013; Conway et al., 2018; Crawford & Pilanski, 2014).

We therefore conducted three studies at the intersection of political ideology, diversity attitudes, and beliefs. Studies 1 and 2 explored the multidimensionality of diversity and ideological differences in attitudes toward diversity features. Study 3 investigated possible variations in the perceived meaning of “diversity” by asking participants to judge the relevance of a set of features to diversity.

Study 1

Study 1 investigated ideological differences in attitudes toward a wide variety of diversity features. Participants rated how much diversity or homogeneity they would desire in 23 different community features that could be considered relevant to diversity. Because liberal constituents and politicians tend to endorse more prodiversity statements and policies in politics and social media compared with conservatives (Federico & Sidanius, 2002; Iyengar et al., 2019), their stored representations and attitudes toward diversity should be more positive than conservatives. Thus, we predicted that liberals would report more favorable attitudes toward diversity than conservatives. Furthermore, liberals tend to be more open to experience and uncertainty, and therefore may also be more comfortable interacting with people who do not share their identity (Chirumbolo & Leone, 2010). Indeed, political conservativism positively correlates with some constructs opposed to diversity, such as right-wing authoritarianism, ethnocentrism (Pratto et al., 1994), and implicit and explicit racism (Hodson et al., 2009).

Second, we predicted that diversity preferences would be multidimensional, such that a high preference for diversity in one community feature (e.g., ethnicity) would not necessarily be indicative of a high preference for another community feature (e.g., viewpoints about abortion). In addition, we predicted that individuals would rate similar features equivalently, meaning people who reported preferences toward diversity in attitudes about the rich would report preferences toward diversity in attitudes about the poor.

To test these hypotheses, we ran an exploratory factor analysis (EFA). Because diversity preferences and attitudes may be unidimensional (i.e., if you like one type of diversity, you like all types), or multidimensional (i.e., liking one type of diversity does not mean you like another type), an EFA is the ideal analysis to deduce how many factors, or dimensions, exist within our set of diversity feature items (see Kim et al., 1978).

Method

Participants

Participants were 1,001 U.S. residents (41.7% female, 58.3% male, age: 18–85 years; M_age = 40.26, SD_age = 16.67), who registered with the website YourMorals.org to complete the study. Eighty-two percent of participants were White, 2.7% African American, 2.3% East Asian, 3.1% Latino, 0.9% Middle Eastern, 2.4% Native American, 1.2% South Asian, and 1.8% identified as “Other.” Participants ranged in political orientation from very liberal to very conservative (194 very liberal, 333 liberal, 143 slightly liberal, 131 moderate, 54 slightly conservative, 98 conservative, 48 very conservative). One standard for estimating appropriate sample size for EFA states that there should be five to 10 participants for every variable measured when N = 300 (Tinsley & Tinsley, 1987). This proportion becomes smaller as the sample size increases (Kyriazos, 2018). We measured 23 variables, or items, leaving us with 41.71 participants for each variable. Because we have greater than 10 participants for each item and greatly exceed N = 300, we achieved a sufficient sample size to obtain at least 80% power to detect distinct factors in the data if they exist.

Materials and Procedures

All data and materials were downloaded from Yourmorals.org. Participants completed a brief demographic survey, in which they reported their age, gender, race, and overall political orientation, on a scale from 1 (very liberal) to 7 (very conservative). Measures were selected that most closely assessed preferences for different forms of diversity. Participants completed the 23-item “Preference for Similarity vs. Difference” questionnaire which measured how much similarity or homogeneity participants wanted on a variety of characteristics in their community on a 1 (everyone is the same as me) to 7 (everyone is different from me) scale. Example characteristics included education, musical preferences, religiosity, age, ethnicity, beliefs about global warming, and beliefs about the role of government. No studies in the current work are preregistered, but all study materials, data, codebook, and data scripts can be found at our Open Science Framework (OSF) page (https://osf.io/4h35g/).

Results

First, a series of EFAs were conducted to determine the dimensionality of diversity attitudes. A Kaiser-Meyer-Olkin (KMO) test on all 23 diversity preference items yielded a value of .88, ensuring that the data were suitable for EFA. An examination of the scree plot, eigenvalues, and parallel analysis suggested that two or five factors would best explain the data. Therefore, two factors were extracted first, followed by five factors.

Two-Factor Model

An EFA with a promax rotation was implemented using a maximum likelihood extraction, extracting two factors. Overall, the two-factor model obtained a simple structure, explaining 32.5% of variance (see Table 1 for factor loadings). All but three items loaded primarily on one factor, with loadings ranging from .422 to .748. Both religion items did not load clearly or strongly on a single factor, and therefore were not included in the final two-factor model. In addition, the item “education” loaded only on Factor 2 at .253, and therefore was not included in the final two-factor model.

Table 1.
EFA Factor Loadings for 23 Community Feature Preference Items.

Factor loadings two-factor structure

Community feature items Viewpoint Demographic

Political orientation .50 .25

Attitudes toward:

Abortion .70 .01

Same-sex marriage .65 −.09

Global warming .75 −.16

Evolution .60 −.03

The rich .49 .15

The poor .68 −.15

Role of government .59 .09

Social safety net .67 −.06

Religious identification .25 .40

Ethnicity −.20 .53

Spoken language −.12 .47

Age −.18 .52

Marital status −.10 .54

Number of children .01 .42

Musical preferences −.01 .64

Architecture preferences −.004 .49

Restaurants −.03 .52

Shops −.03 .53

Movie preferences −.04 .61

Reading preferences .09 .55

Religiosity* .34 .34

Education* .16 .25

Note. Bold numbers indicate the item was included in the model as loading on the corresponding factor. EFA = exploratory factor analysis.

*
The item was not included in the final model.

When examining the items that loaded on each factor, the viewpoint-based and ideological items loaded on Factor 1, and therefore, Factor 1 was labeled viewpoint diversity (see Table 1). Items that loaded on Factor 2 were demographic and cultural in nature (i.e., race, age, musical preferences), and therefore, this factor was labeled demographic diversity. The two-factor structure was simple, explained a moderate amount of variance, and supported previous work that emphasized the importance of analyzing demographic and viewpoint diversity separately (Haidt et al., 2003).

Five-Factor Model

An EFA with a promax rotation found a moderately simple structure, with the model explaining 41.7% of the variance in diversity preference (see Online Supplemental Table 6 for loadings). When examining item loadings, Factor 1 seemed to represent consumer-type preferences, and included diversity in restaurants, stores, and preferences for music, movies, and reading (factor loadings: .35–.80). Factor 2 represented economic attitudes, and included political orientation, and attitudes about the rich, the poor, social safety nets, and the role of government in everyday life (loadings: .42–.82). Factor 3 represented moral attitudes, and included attitudes about abortion, same-sex marriage, global warming, and evolution (loadings: .50–.75). Factor 4 contained demographic items, and included ethnicity, spoken language, age, education, number of children, and marital status (factor loadings: .37–.75). Finally, Factor 5 included religiosity and religious identification (factor loadings: .72–.84). The items education and architectural preferences did not load above .30 and therefore were not included in the model.

Overall, the two- and five-factor models explained the data well, and possessed theoretically relevant loading patterns. Although the five-factor model explained more variance than the two-factor model, the two-factor model provided a simpler structure, and had an equal number of items in each factor. Because composite scores of the factors would be created to assess group differences, the two-factor model was selected as it had a greater number of items in each factor.

To assess ideological differences regarding diversity preferences, demographic diversity composite scores (M = 3.20, SD = 0.65), a = .80, and viewpoint diversity composite scores (M = 4.56, SD = 0.75), a = .80, were created. There was a small positive correlation between demographic and viewpoint diversity, r(999) = .18, p < .001, 95% confidence interval (CI) = [.12, .24], and ideological differences emerged: Greater liberalism was associated with more positive attitudes toward demographic diversity in communities, r(999) = .19, p < .001, 95% CI = [.13, .25], and greater conservativism was associated with more positive attitudes toward viewpoint diversity in communities, r(999) = −.22, p < .001, 95% CI = [−.27, .16]. Ideology was a continuous variable, with higher scores indicating greater conservativism and lower scores indicating greater liberalism.

In summary, although “diversity” is a single word in the dictionary, it was not a unitary construct in the minds of our participants. An EFA revealed two distinct factors in participants’ ratings of community features: (a) demographic diversity (i.e., diversity in features such as age, religion, and ethnicity) and (b) viewpoint diversity (i.e., diversity in political orientation and attitudes toward sociopolitical topics). Results indicated that two to five factors best explained the structure of individual difference in diversity preferences. These findings strongly suggest that diversity preference is not a unidimensional construct but, instead, has psychometrically distinct facets. Preliminary support was also found for ideological differences in viewpoints toward diversity—more conservative participants preferred viewpoint diversity and more liberal participants preferred demographic diversity.

Study 2

As with many terms in the natural language, “diversity” has multiple referents, and the Study 1 results documented the need for a multifactor approach to the concept. Study 2 addressed a related factor involving semantics—in this case, one that concerned the wording of the scale anchor employed in the “Preference for Similarity vs. Difference” measure. Specifically, the anchor “no one is the same as me” could be interpreted as indicating a preference for diversity through one’s community or, alternatively, as a preference of personal uniqueness, with everyone except the individual doing the rating being the same. It is possible that some participants construed the scale this way, producing additional error around community characteristic ratings. Therefore, an additional version of the Preferences for Similarity vs. Differences scale was added in Study 2 that more closely captured the construct of diversity.

As previously discussed, studies often investigate attitudes toward diversity without providing a definition of the term. It is possible that people possess unique understandings of diversity, and use only a fragment of the construct to make diversity-related judgments. For example, when asked their opinion about diversity in an interview, people tended to report generally positive attitudes. However, when the same people were questioned about diversity attitudes regarding specific cultural or ethnic groups, viewpoints were inconsistent, sometimes directly conflicting with previous positive evaluations of diversity (Bell & Hartmann, 2007). Adapting this paradigm, Study 2 assessed participants’ attitudes toward the general concept of diversity without providing a definition of the term. By doing so, we were able to investigate whether general attitudes toward diversity actually predict how people feel about specific types of diversity.

First, we hypothesized that at least two factors would emerge, with one being a demographic factor and another a viewpoint diversity factor. Second, we hypothesized that greater liberalism would be associated with more positive attitudes toward demographic diversity and greater conservativism would be associated with more positive attitudes toward viewpoint diversity. Third, because diversity preferences are multidimensional, we hypothesized that general diversity preferences would not predict viewpoint or demographic diversity preferences. Fourth, because liberals tend to support prodiversity policy (e.g., Federico & Sidanius, 2002), we predicted liberals would be more likely to report positive attitudes toward the general concept of diversity than conservatives.

Method

Participants

One thousand twelve MTurk workers in the United States (55.1% female, 44.6% male; M_age = 36.82, SD_age = 11.63) took part in the study. Participants were 74.0% White, and ranged in political orientation from very liberal to very conservative (147 very liberal, 225 liberal, 155 slightly liberal, 213 moderate, 122 slightly conservative, 104 conservative, 42 very conservative). The EFA for Study 2 was conducted on 23 items (i.e., the minority-majority scale and then homogeneity-diversity scale), leaving 44 participants for each item. Because we have greater than 10 participants per each item and the sample size greatly exceeds 300, we achieved a sufficient sample size to obtain at least 80% power to detect distinct factors in the data if they exist (Kyriazos, 2018; see methodology files).

Materials and Procedures

Participants completed the same demographic survey that was completed in Study 1. Political orientation was measured on an 11-point scale, ranging from 1 (very conservative), 6 (middle of the road), to 11 (extremely liberal). As in Study 1, participants completed the same Preferences for Similarity and Differences questionnaire; however, the scale was renamed “Majority-Minority preferences” to more accurately reflect the construct assessed in the scale.

Participants also completed a preference for homogeneity and diversity questionnaire (homogeneity-diversity questionnaire). The homogeneity-diversity questionnaire was identical to the majority-minority scale, with the exception of the scales’ anchors. Participants reported the degree to which they wanted all people to be different from each other on each of the features (in contrast to different “from them”).

Next, participants completed a five-item measure of the general concept of diversity (called general diversity) that assessed preferences for diversity when it was framed as a general concept without a specified context. Participants indicated how much they agreed with statements such as “diversity is a good thing” and “my ideal community would be very diverse” on a scale from 1 (strongly disagree) to 7 (strongly agree). The five items were combined and averaged into a composite general diversity preference score for each participant.

Study 2 Results

Before completing the EFA and testing our predictions, we first had to assess whether the two versions of the diversity scale captured different constructs. Although correlations between equivalent items were only moderately correlated, the same factor structure emerged for each scale with near equivalent loadings, and therefore, factor information will only be reported for the homogeneity-diversity scale (see Online Supplemental Material Table 7 for the minority-majority factor loadings).

EFA for Homogeneity and Diversity Scale

As in Study 1, an EFA was used to determine the factor structure of diversity preferences. A parallel analysis suggested a two-factor structure would best fit the data. An EFA with a promax rotation found that the two-factor structure very closely resembled the factor structure and loadings from Study 1 and explained 53.9% of variance. Half of the items loaded on Factor 1, and half on Factor 2, creating the demographic and viewpoint diversity factors (see Table 2).

Table 2.
Exploratory Factor Analysis Factor Loadings for 23 Community Feature Preference Items for Homogeneity-Diversity Scale.

Factor loadings 2-factor structure

Community feature preference items Viewpoint Demographic

Political orientation .58 .21

Religious identification* .38 .44

Attitudes toward:

Abortion .86 −.18

Same-sex marriage .87 −.20

Global warming .81 −.18

Evolution .74 −.05

The rich .60 .13

The poor .82 −.16

Role of government .70 .03

Social safety net .73 .003

Morality .67 .01

Ethnicity .10 .55

Spoken language .33 .31

Age −.01 .69

Marital status −.05 .68

Number of children −.03 .73

Architecture preferences −.09 .76

Restaurants −.12 .81

Shops and stores −.11 .83

Movie preferences −.17 .89

Reading preferences −.12 .83

Note. Bold numbers indicate the item was included in the model as loading on the corresponding factor.

*
The item was not included in the final model.

Ideological Differences in Diversity Preference

Composite scores were created for general, viewpoint, and demographic diversity (see Table 3 for descriptives). As predicted, correlational analyses revealed greater liberalism was associated with more positive attitudes toward demographic diversity, r(492) = −.20, p < .001, 95% CI = [−.28, −.11], and greater conservativism was associated with more positive attitudes toward viewpoint diversity, r(494) = .11, p = .003, 95% CI = [.04, .22].

Table 3.
Descriptive Statistics for Each Diversity Subscale.

Diversity type Alpha Overall mean Conservative mean Democratic mean

General .95 [.95, .96] 4.64, 1.21 4.04, 1.33 5.08, 0.89

Demographic .90 [.89, .91] 4.29, 0.95 3.17, 1.12 4.43, 0.94

Viewpoint .92 [.91, .93] 3.17, 1.12 3.35, 1.13 3.03, 1.08

Note. Standard deviations are reported after means.

General Diversity Attitudes

We hypothesized that liberals would be more likely than conservatives to endorse diversity in general terms, but that general diversity preferences would not predict preferences for viewpoint diversity or demographic diversity. General diversity was regressed on political orientation, and greater liberalism was associated with more positive attitudes toward general diversity (b = −.18, p < .001, R² = .26), supporting predictions. Furthermore, general diversity did not significantly predict viewpoint diversity preferences, b = .08, p = .065, R² = .005, but contrary to predictions, did predict greater support for demographic diversity, b = .34, p < .001, R² = .19.

If liberals and conservatives possess different understandings of the general concept of diversity, then general diversity attitudes may predict demographic and viewpoint diversity preferences for each ideological group differently. When examining only liberal participants, general diversity significantly predicted demographic diversity, b = .41, p < .001, R² = .14, but did not predict viewpoint diversity preferences, b = −.05, p = .513, R² = −.002. For conservatives, general diversity significantly predicted both demographic diversity, b = .32, p < .001, R² = .21, and viewpoint diversity preferences, b = .29, p < .001, R² = .11.

Study 2 Discussion

As hypothesized, liberals were more likely than conservatives to endorse the general concept of diversity. Furthermore, general diversity did not predict viewpoint diversity, but did significantly predict demographic diversity preferences. Thus, it may be that when people think of diversity in the abstract, people primarily imagine differences in ethnic and cultural groups, and do not necessarily consider diversity in attitudes. These findings suggest that demographic features may be central to peoples’ prototypes of diversity.

When investigating ideological differences in the relationship between general and specific diversity, general diversity predicted demographic diversity preferences for both liberals and conservatives. However, general diversity only predicted viewpoint diversity preferences for conservatives, but not for liberals. Thus, if liberals like general diversity, it is probable they like demographic diversity, and less probable they like viewpoint diversity. On the contrary, if conservatives like general diversity, it is probable they like both demographic and viewpoint diversity.

Study 3

Study 2 found that positive attitudes toward the general concept of diversity predicted demographic diversity preferences for both conservatives and liberals. Thus, without a provided context, people may use their attitudes toward demographic diversity to make diversity-relevant judgments. However, general diversity predicted viewpoint diversity for conservatives, but not for liberals. Therefore, it may be that liberals and conservatives possess different understandings of the general concept of diversity. Study three took up the challenge of investigating these differences in understanding.

One possibility is that liberals and conservatives have different categorization rules, indicating they differ in what objects they believe belong to the “diversity” category. For some categories, membership is clear-cut because there are explicit rules dictating membership (i.e., whether someone is or is not a licensed MD). Other categories such as “good citizen” or “diversity” do not have clear, objective membership rules, but rather, are “fuzzy sets” (McCloskey & Glucksberg, 1978). Once a category is fuzzy, people likely differ in what they see as central members, or prototypes, of the category, and thus categorize objects differently. In general, people categorize objects based on their knowledge and beliefs about the world (Murphy & Medin, 1985). Thus, people with different beliefs, experiences, and exposure to the concept likely have different categorization rules.

This principle is exemplified in research in the personality domain of psychology. For example, Cervone (1997) developed a research paradigm in which individuals “map” personality qualities to everyday circumstances. The mappings reveal participants’ subjective beliefs about the meaning of the given personality term. Results show that different people rating the same term construe the meaning of the term differently. One person may view “independence” as an ability to withstand social pressure, whereas another may view it as an ability to carry out professional and social responsibilities (Cervone, 2004; Figure 5). Thus, people differ in what they consider to be the “prototype” of independence (Rosch, 1999).

Likewise, liberals and conservatives may learn about and be exposed to diversity in different contexts, and therefore may possess different prototypes of diversity. The documented rise in political and affective polarization makes it increasingly likely that people are solely exposed to diversity-related issues from their respective news channels, peer groups, and social media sources (Iyengar et al., 2019), effectively creating different environments to learn these prototypes.

For example, when discussing diversity-related topics, such as multiculturalism, affirmative action, or creation of “safe spaces” for minority groups, conservative news networks are more likely to speak critically about these topics than liberals (e.g., Fox News: Carlson, 2020; Gonzalez, 2018). Efforts to increase diversity or provide extra opportunities to minority groups are often perceived as violations of equity, as they seemingly prioritize diverse people over the “most qualified” or those who “worked the hardest” (Sidanius et al., 1996). Furthermore, conservatives are more likely to promote color-blind ideology (Brown et al., 2003), and support notions of “All Lives Matter” or “Blue Lives Matter” (Hansler et al., 2021). Therefore, they may perceive majority groups, such as White, conservative, protestants, as equally belonging to the “diversity” category (Danbold & Unzueta, 2020).

In addition, conservatives preferred viewpoint diversity more than liberals in Studies 1 and 2. Preferences are tied closely to the self-concept, and because the self-concept is the most rehearsed, efficient schema, preferred features may be rated as more relevant to one’s diversity schema than nonpreferred features (Markus et al., 1982). Thus, we predicted conservatives would consider viewpoint differences as more relevant to diversity than liberals.

In contrast, liberal news media outlets often openly support efforts at increasing multiculturalism and diversity, and tend to speak about diversity in terms of advancing the representation and power of minority racial groups (Bauder, 2020). Thus, liberals may understand diversity as being primarily composed of minority racial and ethnic group identities. Furthermore, liberals tended to report more positive attitudes toward demographic diversity in Studies 1 and 2 than conservatives. Therefore, we predicted liberals would consider demographic differences as more prototypical, to diversity than conservatives.

Because Studies 1 and 2 found that diversity is multidimensional and contains at least two distinct factors, viewpoint and demographic diversity, we ran a confirmatory factor analysis (CFA) to test whether diversity is truly multidimensional. We hypothesized that perceptions of diversity contain at least a viewpoint and demographic factor, and a two-factor structure would fit the data significantly better than a unidimensional structure.

Method

Participants

The sample included 386 MTurk workers in the United States (65% female, 35% male; M_age = 36.01, SD_age = 10.75). Participants were 13.99% African American, 5.70% Asian American, 70.47% White, 5.96% Latino, 2.33% Native American, and 1.55% identified as “Other,” and ranged in political orientation from very liberal to very conservative (205 liberals, 110 conservatives, 71 moderates). Monte Carlo simulations find that models with three factors, five continuous items for each factor, and sample sizes of 150 are able to achieve power of .81 (Kyriazos, 2018; Muthén & Muthén, 2002). The hypothesized model in Study 3 contains two factors, with at least nine items per factor. Thus, a sample size of 386 should be more than sufficient in achieving power of at least .81.

Materials

We devised a novel measure, the Prototypical Diversity Questionnaire, to assess participants’ implicit perceptions and understandings of diversity. Following methodology from Kosmitzki and John (1993), participants rated the relevance of a series of community features to their prototypical understanding of diversity.

Specifically, participants were asked to imagine “a very diverse community” and to think about “the types of people and places that exist in a very diverse community.” Participants then determined how relevant 29 different community features were to their image of a diverse community on a 1 (completely irrelevant to my thinking) to 8 (very relevant to my thinking) scale. The 23 community features used in Studies 1 and 2 were used, along with six additional neutral community features (i.e., attitudes about favorite colors, insects, month, etc.). Participants were encouraged to answer the questions as quickly as possible. The neutral features were written to be irrelevant to diversity and were included to act both as a neutral comparison for demographic and viewpoint diversity and as distractor items to prevent demand characteristics.

Procedure

After providing informed consent, participants completed a practice session for the Prototypical Diversity Questionnaire (see Online Supplemental Material). Next, participants completed the full Prototypical Diversity Questionnaire, which contained 29 different features to rate. Finally, participants completed the same demographics section used in Study 1.

Results

We hypothesized that diversity is a multidimensional construct, with at least one factor representing demographic diversity and another factor representing viewpoint diversity. To test our hypothesized dimensionality, we used a CFA to compare the fit of a one-factor model with the fit of the hypothesized two-factor model. If diversity is perceived as multidimensional, the two-factor model should fit the observed data better than the one-factor model.

CFA: Comparing One-Factor With Two-Factor Model

The Lavaan package in R was used to complete the following analyses (Rosseel, 2012). First, all items were entered as loading on a single factor to obtain fit indices. The neutral items were not included as they were intended to be irrelevant to diversity. We used a full information maximum likelihood model, with likelihood estimations to account for missing data. Fit indices indicated that the single-factor model demonstrated poor fit, χ²(170) = 1,360.29, p < .001, with Tucker–Lewis index (TLI) of .60, comparative fit index (CFI) of .64, and root mean square error of approximation (RMSEA) of .14, 90% CI = [.13, .14]. Conventions from Browne and Cudeck (1992) were used to interpret the adequacy of fit indices.

To create the two-factor model, the 11 demographic items were entered as loading on Factor 1 and the nine viewpoint items were entered as loading on Factor 2. Fit indices revealed that the two-factor model fit was better than the one-factor model fit, but was still only fair, χ²(169) = 1,047.44, p < .001, TLI = .70, CFI = .74, and RMSEA = .12, 90% CI = [.10, .12]. Supporting predictions, a chi-square difference test between the two models found that the two-factor model fit the data significantly better than the one-factor model, supporting the argument that diversity is not unidimensional, χ²(1) = 312.85, p < .001. However, the relatively poor fit of both models suggests that neither a one-factor nor the hypothesized two-factor model best describes understandings of diversity.

Because participants made qualitatively different evaluations between the first two studies and this study, it is not surprising that a different factor structure emerged in this study. For example, a person may have a preference for architectural diversity, but may not rate architecture as relevant to diversity. In addition, the EFA from Study 1 found that a five-factor model may best fit the data, strongly suggesting that diversity perceptions and preferences are likely influenced by more than two latent constructs.

When examining item loadings on each factor, the demographic factor possessed substantial variability in values. Specifically, the items diversity in “restaurants,” “shopping,” “architecture,” “movies,” and “reading” all loaded between 1.34 and 1.58 on the demographic factor, with the rest of the items loading below 0.58, with the exception of the number of children and marital status loading at 1.02. This clustering of high and low loadings suggests that the demographic factor may actually be composed of at least two factors, with one factor consisting of diversity in consumer-type features and the second factor consisting of diversity in more typically demographic features. On the viewpoint factor, all items loaded above 1.37 except for political identity, which loaded at 0.83. It may be that people perceive political orientation as more of an identity or demographic feature than a viewpoint-based form of diversity.

Because relevance ratings appeared to follow a different loading pattern than diversity feature preferences (from Studies 1 and 2), an EFA was run (not including neutral items). A visual examination of the scree plot and a parallel analysis led us to extract three factors using a promax rotation.

The three-factor model possessed a moderately simply structure (see Table 4). The first factor contained mostly viewpoint-relevant features, and therefore was labeled viewpoint diversity (item loadings: .65–.87). The only item that did not load clearly on the viewpoint factor was “morality” (.36), and therefore, it was not included in the final model for subsequent analysis.

Table 4.
EFA Factor Loadings for 23 Community Feature Relevance Ratings.

Loadings 3-factor structure

Community feature items Viewpoint Consumer Demographic

Attitudes toward:

Abortion .87 −.13 −.01

Same-sex marriage .76 −.12 .08

Global warming .86 .03 −.13

Evolution .71 .16 −.14

The rich .67 .11 −.06

The poor .85 −.07 −.15

Role of government .71 −.03 .11

Social safety net .65 .04 .05

Morality* .36 −.09 .35

Political orientation .34 −.14 .48

Religious identification .06 .05 .63

Ethnicity −.30 −.06 .78

SES .34 −.08 .05

Age −.17 .12 .64

Marital status .14 .25 .38

Number of children* .19 .32 .20

Architecture preferences −.03 .76 −.09

Restaurants −.18 .80 .10

Shops and stores −.17 .73 .10

Movie preferences .11 .67 −.16

Reading preferences .23 .55 −.07

Note. Bold numbers indicate the item was included in the model as loading on the corresponding factor. EFA = exploratory factor analysis; SES = socioeconomic status.

*
The item was not included in the final model.

Item loadings on the second factor ranged from .55 to .80 and captured activities and goods that often require people to spend money. Thus, this factor was labeled “Consumer preferences.” The item “number of children” loaded on the consumer factor at .320, and because this item is unlike the other five items, it was not included in the final model for subsequent analyses.

Loadings on the third factor ranged from .38 to .78 and included demographic items. Therefore, this factor was labeled demographic preferences. Differing from Studies 1 and 2, political orientation actually loaded on the demographic factor rather than the viewpoint factor. Thus, people may perceive political orientation as more of an identity-relevant feature than a viewpoint feature.

To test the three-factor model against the one- and two-factor models, the three-factor model was entered into a CFA (see Figure 1). Although fit indices improved compared with the one- and two-factor models, overall fit was still relatively poor, χ²(149) = 589.58, p < .001, with TLI = .84, CFI = .86, and RMSEA = .09, 90% CI = [.08, .10]. Modification indices suggested that diversity in shopping and eating should be correlated, and therefore, this modification was made, significantly improving fit, χ²(1) = 43.56, p < .001, TLI = .86, CFI = .88, and RMSEA = .08, 90% CI = [.08, .10]. The modified three-factor structure fit the data significantly better than the two-factor, χ²(1) = 501.42, p < .001, and the one-factor model, χ²(22) = 814.27, p < .001.

Figure 1.
Three-factor model.

Ideological Differences

The three-factor structure was used for all following analyses. To create composite scores for demographic, consumer, viewpoint, and neutral features, the items that loaded on each factor were averaged (see Table 5). Although predictions were not originally made for consumer preferences, this category was included in the analysis of variance (ANOVA) for exploratory purposes.

Table 5.
Descriptive Statistics for Each Diversity Relevance Rating Subscale.

Diversity type Alpha Overall mean Conservative mean Democratic mean

Demographic .76 6.46, 1.13^a 6.29, 1.16 6.60, 1.06

Viewpoint .90 4.71, 1.43^b 4.90, 1.37 4.63, 1.47

Consumer .82 4.67, 1.73^b 5.01, 1.65 4.58, 1.69

Neutral .92 2.90, 1.91^c 3.48, 2.08 2.62, 1.79

Note. Shared superscripts indicate means are not significantly different. Standard deviations are reported after means.

A one-way repeated-measures ANOVA comparing demographic, ideological, consumer, and neutral features indicated significant variation among feature type, F(2.21, 799.46) = 462.74, p < .001, ges = .40. Supporting the hypothesis, Tukey honestly significant difference (HSD) comparisons found that demographic features were rated as significantly more relevant to diversity than viewpoint, t(1,083) = 19.23, p < .001, and neutral t(1,083) = −37.25, p < .001. Although not hypothesized, demographic features were significantly more relevant to diversity than consumer features, t(1,083) = 18.82, p < .001. Both viewpoint, t(1,083) = 19.23, p < .001, and consumer features, t(1,083) = −28.43, p < .001, were significantly more relevant to diversity than neutral features. However, viewpoint features were not rated differently than consumer features, t(1,083) = .80, p < .001. Overall, people rated demographic features as more relevant to diversity than viewpoint, consumer, and neutral features.

We hypothesized that there would be ideological differences in the relevancy ratings of viewpoint and demographic features. As predicted, more conservative participants rated viewpoint features as relevant to diversity, r(374) = .12, p = .018, and more liberal participants rated demographic features as relevant to diversity, r(379) = −.11, p < .031. Although not hypothesized, conservative participants rated consumer features, r(377) = .14, p = .005, and neutral features, r(378) = .26, p < .001, as more relevant to diversity than liberals.

Study 3 Discussion

A two-factor model of diversity features fit the data significantly better than a one-factor model, supporting the first hypothesis. However, a CFA found that a three-factor model described the data better than the two-factor model. Thus, people do not perceive diversity as a unidimensional, or even bidimensional construct, but rather likely perceive at least three categories of diversity. Furthermore, people rated demographic features as most relevant to diversity, followed by viewpoint and consumer features. Finally, conservatives rated viewpoint features as more relevant to diversity than liberals, and liberals rated demographic features as more relevant to diversity than conservatives.

General Discussion

The United states is both increasingly diverse (Frey, 2020; Lichter, 2013) and increasingly politically and affectively polarized (Bail et al., 2018; Iyengar et al., 2019; Mason, 2015). The current work explored the intersection of these two trends and investigated ideological differences in both attitudes and understandings of diversity to arrive at a more complete definition of the term. We found that diversity is perceived as a multidimensional concept, and peoples’ preferences and understandings toward these different dimensions varied as a function of context and political ideology.

In Studies 1 and 2, participants rated how much diversity they would desire on 23 different community features. Two factors emerged, demographic diversity (e.g., race, ethnicity, marital status) and viewpoint diversity (e.g., politics and beliefs), with conservatives reporting more positive attitudes toward viewpoint diversity in their communities, and liberals reporting more positive attitudes toward demographic diversity. Furthermore, positive attitudes toward general diversity significantly predicted attitudes toward demographic diversity for both ideological groups. Thus, when asked their general opinion on diversity, people likely first reference diverse demographic features. In contrast, general diversity predicted positive attitudes toward viewpoint diversity for conservatives, but not for liberals. Therefore, conservatives who agreed with the statement “diversity is a good thing” were also more likely to prefer living in communities where people possess different attitudes toward topics such as global warming or abortion. This finding implies that conservatives may include viewpoint features into their general understanding, or prototype, of diversity, whereas liberals may not.

Following findings from Study 2, Study 3 explored ideological differences in the average prototype of diversity and whether people perceive multiple dimensions of diversity. A CFA found that diversity is indeed multidimensional, and people think about diversity in at least three categories: demographic, viewpoint, and consumer based. The original demographic factor broke down into two subtypes, demographic and consumer-based diversity.

The emergence of three factors in Study 3 is a notable divergence from the two factor solutions found in Studies 1 and 2. This difference likely emerged because attitudes toward objects (measured in Studies 1 and 2) and perceptions of objects (measured in Study 3) are not equivalent constructs. For example, a person may be equally fond of a high degree of diversity in restaurants and racial identities in their community, but they would likely distinguish race as belonging to the “demographic” category, and restaurants as belonging to the “consumer” category.

Similar attitudes may be reported for these two diversity types because consumer diversity is essentially a by-product of demographic diversity. The food, clothing, music, art, and architectural styles in a neighborhood are largely influenced by the types of cultural and ethnic groups that live in that community. If a neighborhood possesses many diverse ethnic restaurants, this likely indicates the population in that neighborhood is also ethnically diverse. Indeed, research finds that the types of stores and restaurants in a community can signal the identities of its residents (Motyl et al., 2020). People were able to correctly assume that neighborhoods with a high concentration of hybrid cars and organic food stores had a majority liberal population, whereas neighborhoods with a high concentration of protestant churches and gun stores had a majority conservative population of conservatives.

Consumer-based attitudes may influence how people seek out, select, or avoid certain communities. For example, if a highly conservative person was interested in buying a home, they may avoid neighborhoods with many diverse restaurants, markets, and bookstores, as these features likely indicate a diverse, liberal-minded population. This selective migration process poses a potential dilemma, as enhanced residential mobility (Oishi, 2010) enables people to move more easily to communities where primarily like-minded and demographically similar others live. This process may form ideological enclaves (Motyl et al., 2014), which in turn may exacerbate the already rising political and affective polarization dominating the United States.

When investigating relevancy ratings of the three factors from Study 3, all participants rated demographic features as most relevant to diversity, followed by viewpoint features, and then consumer features. Therefore, when people think about diversity in the abstract, they most likely initially think about people of different ethnicities, ages, and SES backgrounds. If future studies ask about diversity attitudes in the abstract, researchers must be aware that responses are likely based on only a sliver of the diversity spectrum. These responses should not be used to represent a person’s attitude toward all forms of diversity.

Study 3 also found that liberal participants were more likely to rate demographic features as relevant to diversity, and conservatives were more likely to rate ideological features as relevant to diversity. Not only do liberals and conservatives differ in preferences but they also perceive the meaning of diversity differently. This difference in understanding may be attributable to the increasingly divergent political and social media worlds to which liberals and conservatives are exposed (Barberá et al., 2015; Tucker et al., 2018). Within their respective ideological worlds, party leaders and pundits may “teach” party-specific definitions of diversity through rhetoric and framing. Increased political and affective polarization likely increases the probability that people seek out attitude-confirming news networks and discredit information that challenges attitudes and beliefs.

It is likely that these differences in attitudes and understandings uniquely predict numerous important outcomes. First, being open to different ideological viewpoints may have the potential to ameliorate political and affective polarization. Simply hearing an opposing viewpoint without immediately delegitimizing the information may be an important first step, as cooperation is required to pass legislation and create positive societal change for either ideological group. Furthermore, negative attitudes toward viewpoint diversity can create hostile and nonproductive work environments, potentially causing those with minority viewpoints to feel threatened and isolated (Duarte et al., 2015; Peters et al., 2020). Indeed, some work finds that a portion of ethnic or racial prejudice may be explained by the fact that people assume outgroup members possess different ideological viewpoints (Chambers et al., 2013). Thus, it is important to investigate how attitudes toward viewpoint diversity affect intergroup relations, as well as interventions to ameliorate conflict created by diverging ideological attitudes.

Regarding demographic diversity, people who support racial diversity are more likely to support policies aimed at increasing representation and opportunities for minority groups, such as affirmative action (Federico & Sidanius, 2002; Reyna et al., 2006) and multiculturalism (Wilkinson, 2017). In addition, countries that adopt procultural diversity policies tend to be more successful at improving intergroup relations (Guimond et al., 2013), and companies with more pro-gender and racial diversity policies are more likely to have greater innovative efficiency (Mayer et al., 2018), increased managerial diversity, and feelings of inclusion for all (Kalev et al., 2006).

Limitations and Future Directions

Because Study 1 used precollected data, we were not able to tailor questions to meet the specific goals of the study. Rather than asking participants directly about diversity, participants rated how much they wanted to live in a community where everyone was the same or different from them on the given characteristic. Participants may have perceived this measure as asking about preferences for minority versus majority status, rather than a desire to live in a diverse or homogeneous community. However, Study 2 attenuated this concern, as an addition of a more deftly written diversity scale achieved a near equivalent pattern of results as the original Study 1 measure.

The slight difference in outcomes between the two types of scales may be due to the majority-minority scale being framed in reference to the self. Specifically, participants reported how much they wanted to live in a community where people were “the same as me” or all “different from me,” whereas the diversity scale asked whether they desired communities where people were “all the same” or “all different from each other.” The self-concept is likely the best rehearsed schema (Markus et al., 1982), and therefore, activation of this schema may have led people to utilize more personal knowledge and experiences when answering the majority-minority scale compared with the homogeneity-diversity scale. Thus, the activation of the self-schema may have explained a portion of divergence between the outcomes of the two measures.

In addition, the current work only investigated peoples’ explicit attitudes toward diversity and relied on self-report measures. As previous work has documented, self-reported attitudes do not consistently predict behavior, and people can explicitly report positive attitudes toward a given group, while simultaneously possessing negative implicit attitudes toward the group they rated positively (Dovidio et al., 2004). For example, liberals who initially rated themselves as favorable to diversity actually reported more negative attitudes toward ethnic outgroups after experiencing an increase in diversity in their own neighborhoods (Enos, 2014). Therefore, it is possible that liberal participants in our study who responded favorably toward demographic diversity in their communities have never lived in diverse neighborhoods themselves, and thus cannot accurately predict their preferences. Future research should investigate whether those who report positive attitudes toward each type of diversity actually engage in positive interactions with diverse groups.

Conclusion

Diversity is not a single psychological construct, it is multidimensional. At the least, people perceive demographic, viewpoint, and consumer diversity, and peoples’ attitudes toward each diversity type differ as a function of ideology. Conservatives appear to be more tolerant of viewpoint diversity in their communities, and liberals appear to be more tolerant of demographic diversity. Although peoples’ prototypes of diversity were primarily composed of diverse demographic groups, conservatives rated viewpoint features as more relevant to diversity than liberals, and liberals rated demographic features as more relevant to diversity than conservatives. This difference in understanding is likely due to increased political and affective polarization, and the ensuing different political and social media worlds in which liberals and conservatives learn their respective definitions of diversity. Therefore, studies and diversity training must account for the multiple dimensions and different understandings of diversity in their work. Doing so will likely lead to an enhanced understanding of the impact of diversity and more effectively teach people to cooperate with dissimilar others.

Supplemental Material

sj-docx-1-psp-10.1177_01461672211028141 – Supplemental material for On the Varieties of Diversity: Ideological Variations in Attitudes Toward, and Understandings of Diversity

Supplemental material, sj-docx-1-psp-10.1177_01461672211028141 for On the Varieties of Diversity: Ideological Variations in Attitudes Toward, and Understandings of Diversity by Kathryn A. Howard, Daniel Cervone and Matthew Motyl in Personality and Social Psychology Bulletin

Supplemental Material

sj-docx-2-psp-10.1177_01461672211028141 – Supplemental material for On the Varieties of Diversity: Ideological Variations in Attitudes Toward, and Understandings of Diversity

Supplemental material, sj-docx-2-psp-10.1177_01461672211028141 for On the Varieties of Diversity: Ideological Variations in Attitudes Toward, and Understandings of Diversity by Kathryn A. Howard, Daniel Cervone and Matthew Motyl in Personality and Social Psychology Bulletin

	Factor loadings two-factor structure
Political orientation	.50	.25
Attitudes toward:
Abortion	.70	.01
Same-sex marriage	.65	−.09
Global warming	.75	−.16
Evolution	.60	−.03
The rich	.49	.15
The poor	.68	−.15
Role of government	.59	.09
Social safety net	.67	−.06
Religious identification	.25	.40
Ethnicity	−.20	.53
Spoken language	−.12	.47
Age	−.18	.52
Marital status	−.10	.54
Number of children	.01	.42
Musical preferences	−.01	.64
Architecture preferences	−.004	.49
Restaurants	−.03	.52
Shops	−.03	.53
Movie preferences	−.04	.61
Reading preferences	.09	.55
Religiosity*	.34	.34
Education*	.16	.25

	Factor loadings 2-factor structure
Political orientation	.58	.21
Religious identification*	.38	.44
Attitudes toward:
Abortion	.86	−.18
Same-sex marriage	.87	−.20
Global warming	.81	−.18
Evolution	.74	−.05
The rich	.60	.13
The poor	.82	−.16
Role of government	.70	.03
Social safety net	.73	.003
Morality	.67	.01
Ethnicity	.10	.55
Spoken language	.33	.31
Age	−.01	.69
Marital status	−.05	.68
Number of children	−.03	.73
Architecture preferences	−.09	.76
Restaurants	−.12	.81
Shops and stores	−.11	.83
Movie preferences	−.17	.89
Reading preferences	−.12	.83

Diversity type	Alpha	Overall mean	Conservative mean	Democratic mean
General	.95 [.95, .96]	4.64, 1.21	4.04, 1.33	5.08, 0.89
Demographic	.90 [.89, .91]	4.29, 0.95	3.17, 1.12	4.43, 0.94
Viewpoint	.92 [.91, .93]	3.17, 1.12	3.35, 1.13	3.03, 1.08

	Loadings 3-factor structure
Attitudes toward:
Abortion	.87	−.13	−.01
Same-sex marriage	.76	−.12	.08
Global warming	.86	.03	−.13
Evolution	.71	.16	−.14
The rich	.67	.11	−.06
The poor	.85	−.07	−.15
Role of government	.71	−.03	.11
Social safety net	.65	.04	.05
Morality*	.36	−.09	.35
Political orientation	.34	−.14	.48
Religious identification	.06	.05	.63
Ethnicity	−.30	−.06	.78
SES	.34	−.08	.05
Age	−.17	.12	.64
Marital status	.14	.25	.38
Number of children*	.19	.32	.20
Architecture preferences	−.03	.76	−.09
Restaurants	−.18	.80	.10
Shops and stores	−.17	.73	.10
Movie preferences	.11	.67	−.16
Reading preferences	.23	.55	−.07

Diversity type	Alpha	Overall mean	Conservative mean	Democratic mean
Demographic	.76	6.46, 1.13^a	6.29, 1.16	6.60, 1.06
Viewpoint	.90	4.71, 1.43^b	4.90, 1.37	4.63, 1.47
Consumer	.82	4.67, 1.73^b	5.01, 1.65	4.58, 1.69
Neutral	.92	2.90, 1.91^c	3.48, 2.08	2.62, 1.79

Footnotes

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) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Kathryn A. Howard

Matthew Motyl

Supplemental Material

Supplemental material is available online with this article.

References

Bail

C. A.

Argyle

L. P.

Brown

T. W.

Bumpus

J. P.

Chen

Hunzaker

M. F.

. . . Volfovsky

(2018). Exposure to opposing views on social media can increase political polarization. Proceedings of the National Academy of Sciences, 115(37), 9216–9221.

Barberá

Jost

J. T.

Nagler

Tucker

J. A.

Bonneau

(2015). Tweeting from left to right: Is online political communication more than an echo chamber? Psychological Science, 26(10), 1531–1542.

Bauder

(2020, July 09). NBC news Sets goal of a workforce that is HALF MINORITY. Retrieved March 10, 2021, from https://www.washingtonpost.com/business/nbc-news-sets-goal-of-a-workforce-that-is-half-mnority/2020/07/09/aed55b48-c224-11ea-8908-68a2b9eae9e0_story.html

Bell

J. M.

Hartmann

(2007). Diversity in everyday discourse: The cultural ambiguities and consequences of “happy talk.” American Sociological Review, 72(6), 895–914.

Brown

M. K.

Carnoy

Currie

Oppenheimer

D. B.

Wellman

Shultz

M. M.

(2003). Whitewashing race: The myth of a color-blind society. University of California Press.

Browne

M. W.

Cudeck

(1992). Alternative ways of assessing model fit. Sociological methods & research, 21(2), 230–258.

Carlson

(2020, December 02). Tucker Carlson: The “DIVERSITY” delusion and the destruction of the AMERICAN MERITOCRACY. https://www.foxnews.com/opinion/tucker-carlson-end-of-meritocracy-biden-diversity-delusion

Cervone

(1997). Social-cognitive mechanisms and personality coherence: Self-knowledge, situational beliefs, and cross-situational coherence in perceived self-efficacy. Psychological Science, 8(1), 43–50.

Cervone

(2004). The architecture of personality. Studia Universitatis Babes-Bolyai-Psychologia-Paedagogia, 49(1), 3–44.

10.

Chambers

J. R.

Schlenker

B. R.

Collisson

(2013). Ideology and prejudice: The role of value conflicts. Psychological Science, 24(2), 140–149.

11.

Chirumbolo

Leone

(2010). Personality and politics: The role of the HEXACO model of personality in predicting ideology and voting. Personality and Individual Differences, 49(1), 43–48.

12.

Coe

Tewksbury

Bond

B. J.

Drogos

K. L.

Porter

R. W.

Yahn

Zhang

(2008). Hostile news: Partisan use and perceptions of cable news programming. Journal of Communication, 58(2), 201–219.

13.

Conover

Ratkiewicz

Francisco

Gonçalves

Menczer

Flammini

(2011, July). Political polarization on twitter [Conference session]. International AAAI Conference on Web and Social Media (Vol. 5, No. 1), Palo Alto, CA, United States.

14.

Conway

III.

L. G. Houck

S. C.

Gornick

L. J.

Repke

M. A.

(2018). Finding the Loch Ness monster: Left-wing authoritarianism in the United States. Political Psychology, 39(5), 1049–1067.

15.

Craig

M. A.

Rucker

J. M.

Richeson

J. A.

(2018). The pitfalls and promise of increasing racial diversity: Threat, contact, and race relations in the 21st century. Current Directions in Psychological Science, 27(3), 188–193.

16.

Crawford

J. T.

Pilanski

J. M.

(2014). Political intolerance, right and left. Political Psychology, 35(6), 841–851.

17.

Danbold

Unzueta

M. M.

(2020). Drawing the diversity line: Numerical thresholds of diversity vary by group status. Journal of Personality and Social Psychology, 118(2), 283–306.

18.

Dovidio

J. F.

Gaertner

S. L.

Smith

E. R.

Semin

G. R.

Leung

Bond

M. H.

. . . Kirk

E. R.

(2004). Aversive racism. Advances in Experimental Social Psychology, 36, 1–438.

19.

Duarte

J. L.

Crawford

J. T.

Stern

Haidt

Jussim

Tetlock

P. E.

(2015). Political diversity will improve social psychological science. Behavioral and Brain Sciences, 38, 1–58.

20.

Enos

R. D.

(2014). Causal effect of intergroup contact on exclusionary attitudes. Proceedings of the National Academy of Sciences, 111(10), 3699–3704.

21.

Federico

C. M.

Sidanius

(2002). Sophistication and the antecedents of Whites’ racial policy attitudes: Racism, ideology, and affirmative action in America. Public Opinion Quarterly, 66(2), 145–176.

22.

Frey

(2020, July 16). The nation is diversifying even faster than predicted, according to new census data. https://www.brookings.edu/research/new-census-data-shows-the-nation-is-diversifying-even-faster-than-predicted/

23.

Gates

M. G.

Mark

B. A.

(2012). Demographic diversity, value congruence, and workplace outcomes in acute care. Research in Nursing & Health, 35(3), 265–276.

24.

Gonzalez

(2018, August 20). Yes, it’s time for affirmative action to end—Finally. https://www.foxnews.com/opinion/yes-its-time-for-affirmative-action-to-end-finally

25.

Guillaume

Y. R.

Dawson

J. F.

Woods

S. A.

Sacramento

C. A.

West

M. A.

(2013). Getting diversity at work to work: What we know and what we still don’t know. Journal of Occupational and Organizational Psychology, 86(2), 123–141.

26.

Guimond

Crisp

R. J.

De Oliveira

Kamiejski

Kteily

Kuepper

. . . Sidanius

(2013). Diversity policy, social dominance, and intergroup relations: Predicting prejudice in changing social and political contexts. Journal of Personality and Social Psychology, 104(6), 941–958.

27.

Haidt

Rosenberg

Hom

(2003). Differentiating diversities: Moral diversity is not like other kinds 1. Journal of Applied Social Psychology, 33(1), 1–36.

28.

Hansler

Atwood

Gaouette

(2021, January 19). Pompeo attacks MULTICULTURALISM, saying it is “not who America is.” https://www.cnn.com/2021/01/19/politics/pompeo-multiculturalism-tweet/index.html

29.

Hentschel

Shemla

Wegge

Kearney

(2013). Perceived diversity and team functioning: The role of diversity beliefs and affect. Small Group Research, 44(1), 33–61.

30.

Herring

(2009). Does diversity pay?: Race, gender, and the business case for diversity. American Sociological Review, 74(2), 208–224.

31.

Hodson

Hogg

S. M.

MacInnis

C. C.

(2009). The role of “dark personalities”(narcissism, Machiavellianism, psychopathy), Big Five personality factors, and ideology in explaining prejudice. Journal of Research in Personality, 43(4), 686–690.

32.

Hofhuis

van der Zee

K. I.

Otten

(2015). Measuring employee perception on the effects of cultural diversity at work: Development of the benefits and threats of diversity scale. Quality & Quantity, 49(1), 177–201.

33.

Hostager

T. J.

De Meuse

K. P.

(2008). The effects of a diversity learning experience on positive and negative diversity perceptions. Journal of Business and Psychology, 23(3–4), 127–139.

34.

Iyengar

Lelkes

Levendusky

Malhotra

Westwood

S. J.

(2019). The origins and consequences of affective polarization in the United States. Annual Review of Political Science, 22, 129–146.

35.

Kalev

Dobbin

Kelly

(2006). Best practices or best guesses? Assessing the efficacy of corporate affirmative action and diversity policies. American Sociological Review, 71(4), 589–617.

36.

Kim

J. O.

Ahtola

Spector

P. E.

Mueller

C. W.

(1978). Introduction to factor analysis: What it is and how to do it (No. 13). SAGE.

37.

Kosmitzki

John

O. P.

(1993). The implicit use of explicit conceptions of social intelligence. Personality and Individual Differences, 15(1), 11–23.

38.

Kyriazos

T. A.

(2018). Applied psychometrics: Sample size and sample power considerations in factor analysis (EFA, CFA) and SEM in general. Psychology, 9(08), 2207–2230.

39.

Lichter

D. T.

(2013). Integration or fragmentation? Racial diversity and the American future. Demography, 50(2), 359–391.

40.

Markus

Crane

Bernstein

Siladi

(1982). Self-schemas and gender. Journal of Personality and Social Psychology, 42(1), 38–50.

41.

Mason

(2015). “I disrespectfully agree”: The differential effects of partisan sorting on social and issue polarization. American Journal of Political Science, 59(1), 128–145.

42.

Mayer

R. C.

Warr

R. S.

Zhao

(2018). Do pro-diversity policies improve corporate innovation? Financial Management, 47(3), 617–650.

43.

McCloskey

M. E.

Glucksberg

(1978). Natural categories: Well defined or fuzzy sets? Memory & Cognition, 6(4), 462–472.

44.

Mohamed

(2020, May 30). A new estimate of U.S. Muslim population. https://www.pewresearch.org/fact-tank/2018/01/03/new-estimates-show-u-s-muslim-population-continues-to-grow/

45.

Motyl

Iyer

Oishi

Trawalter

Nosek

B. A.

(2014). How ideological migration geographically segregates groups. Journal of Experimental Social Psychology, 51, 1–14.

46.

Motyl

Prims

J. P.

Iyer

(2020). How ambient cues facilitate political segregation. Personality and Social Psychology Bulletin, 46(5), 723–737.

47.

Murphy

G. L.

Medin

D. L.

(1985). The role of theories in conceptual coherence. Psychological Review, 92(3), 289–316.

48.

Muthén

L. K.

Muthén

B. O.

(2002). How to use a Monte Carlo study to decide on sample size and determine power. Structural equation modeling, 9(4), 599–620.

49.

Oishi

(2010). The psychology of residential mobility: Implications for the self, social relationships, and well-being. Perspectives on Psychological Science, 5(1), 5–21.

50.

Parker

III.

E. T. Barnhardt

C. L.

Pascarella

E. T.

McCowin

J. A.

(2016). The impact of diversity courses on college students’ moral development. Journal of College Student Development, 57(4), 395–410.

51.

Pearsall

M. J.

Ellis

A. P.

Evans

J. M.

(2008). Unlocking the effects of gender faultlines on team creativity: Is activation the key? Journal of Applied Psychology, 93(1), 225–234.

52.

Peters

Honeycutt

De Block

Jussim

(2020). Ideological diversity, hostility, and discrimination in philosophy. Philosophical Psychology, 33(4), 511–548.

53.

Pratto

Sidanius

Stallworth

L. M.

Malle

B. F.

(1994). Social dominance orientation: A personality variable predicting social and political attitudes. Journal of Personality and Social Psychology, 67(4), 741–763.

54.

Reyna

Henry

P. J.

Korfmacher

Tucker

(2006). Examining the principles in principled conservatism: The role of responsibility stereotypes as cues for deservingness in racial policy decisions. Journal of Personality and Social Psychology, 90(1), 109.

55.

Roberson

Ryan

A. M.

Ragins

B. R.

(2017). The evolution and future of diversity at work. Journal of Applied Psychology, 102(3), 483–499.

56.

Rosch

(1999). Principles of categorization. In Margolis

Laurence

(Eds.), Concepts: Core readings (pp. 189–206). MIT Press.

57.

Rosseel

(2012). Lavaan: An R package for structural equation modeling and more. Version 0.5–12 (BETA). Journal of Statistical Software, 48(2), 1–36.

58.

Schmidt

(2021, February 26). 1 in 6 Gen z adults are LGBT. and this number could continue to grow. https://www.washingtonpost.com/dc-md-va/2021/02/24/gen-z-lgbt/

59.

Serhan

(2020, January 06). Populism is morphing in insidious ways. https://www.theatlantic.com/international/archive/2020/01/future-populism-2020s/604393/

60.

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.

61.

Sommers

S. R.

(2006). On racial diversity and group decision making: Identifying multiple effects of racial composition on jury deliberations. Journal of Personality and Social Psychology, 90(4), 597–612.

62.

Thatcher

Patel

P. C.

(2011). Demographic faultlines: A meta-analysis of the literature. Journal of Applied Psychology, 96(6), 1119–1139.

63.

Tinsley

H. E.

Tinsley

D. J.

(1987). Uses of factor analysis in counseling psychology research. Journal of Counseling Psychology, 34(4), 414–424.

64.

Tucker

J. A.

Guess

Barberá

Vaccari

Siegel

Sanovich

Nyhan

(2018). Social media, political polarization, and political disinformation: A review of the scientific literature. Political Polarization, and Political Disinformation. Advance online publication.

65.

Verkuyten

Yogeeswaran

(2020). Interculturalism and the acceptance of minority groups: The role of political orientation. Social Psychology, 5(12), 135–140.

66.

Wilkinson

(2017, December 07). U.S. multiculturalism or cultural assimilation? https://www.huffpost.com/entry/us-multiculturalism-or-cultural-assimilation_b_8218490

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

0.03 MB

0.09 MB

	Factor loadings two-factor structure
Community feature items	Viewpoint	Demographic
Political orientation	.50	.25
Attitudes toward:
Abortion	.70	.01
Same-sex marriage	.65	−.09
Global warming	.75	−.16
Evolution	.60	−.03
The rich	.49	.15
The poor	.68	−.15
Role of government	.59	.09
Social safety net	.67	−.06
Religious identification	.25	.40
Ethnicity	−.20	.53
Spoken language	−.12	.47
Age	−.18	.52
Marital status	−.10	.54
Number of children	.01	.42
Musical preferences	−.01	.64
Architecture preferences	−.004	.49
Restaurants	−.03	.52
Shops	−.03	.53
Movie preferences	−.04	.61
Reading preferences	.09	.55
Religiosity*	.34	.34
Education*	.16	.25

	Factor loadings 2-factor structure
Community feature preference items	Viewpoint	Demographic
Political orientation	.58	.21
Religious identification*	.38	.44
Attitudes toward:
Abortion	.86	−.18
Same-sex marriage	.87	−.20
Global warming	.81	−.18
Evolution	.74	−.05
The rich	.60	.13
The poor	.82	−.16
Role of government	.70	.03
Social safety net	.73	.003
Morality	.67	.01
Ethnicity	.10	.55
Spoken language	.33	.31
Age	−.01	.69
Marital status	−.05	.68
Number of children	−.03	.73
Architecture preferences	−.09	.76
Restaurants	−.12	.81
Shops and stores	−.11	.83
Movie preferences	−.17	.89
Reading preferences	−.12	.83