Differentiating diversities : Moral diversity is not like other kinds

Diversity is widely celebrated and pursued in parts of American society, particularly within academe. Diversity is clearly associated with moral goods, such as justice, and with practical goods, such as the variety and quality of ideas. But from a social psychological point of view, diversity ought to cause a number of problems, such as divisiveness and conflict. A resolution of this paradox is proposed: there are several kinds of diversity, with different profiles of costs and benefits. In particular moral diversity is identified as being problematic and even selfcontradictory. Once moral diversity is distinguished from demographic diversity (e.g., race, gender, and national origin), it becomes possible to say that demographic diversity in an educational setting may be generally desirable, while moral diversity may have a more negative profile of effects. Three studies of attitudes and desires for interaction among college students confirm that moral diversity reduces desires for interaction more than does demographic diversity, and that both kinds of diversity are valued more in a classroom than in other social settings. These findings have important practical and ethical implications for discussions of diversity, multiculturalism, affirmative action, identity politics, and immigration policy. Moral Diversity -2 Differentiating diversities: Moral diversity is not like other kinds In recent years diversity has risen to the top of the list of American public goods. Bumper stickers exhort people to “celebrate diversity”. Universities hold diversity awareness days. Companies require employees to attend diversity workshops. Hiring and admission policies aim to increase the diversity of workplaces and schools. The use of quotas and affirmative action to achieve diversity is often controversial, but there is a general consensus that diversity, in and of itself, is good. Yet work in sociology and social psychology suggests that diversity should, in and of itself, have several potentially undesirable effects, and that diversity should be examined carefully before being promoted as a public good. The present study tests the possibility that there are different kinds of diversity, which have different profiles of good and bad effects. Discussions of whether diversity is good or bad therefore become similar to discussions of whether cholesterol, television, or shopping malls are good or bad: it depends on the kind, and on the context. The Benefits of Diversity Diversity in race and national origin has always been a fact of American life and a strength of American society. America’s economic and cultural success in the 20 century has been credited in part to the drive and creativity of its many waves of immigrants (Takaki, 1998), and there is historical evidence that foreign immigration generally increases the number of eminent individuals in a given society (Simonton, 1997). On a smaller scale, diversity is said to increase creativity in the workplace, and there is some evidence that diverse groups are better at creative problem solving because they have a broader base of experience (Cox & Blake, 1991; McLeod, Lobel, & Cox, 1996). In educational settings, a recent national survey of American adults commissioned by the Ford Foundation found that 75% of those surveyed believed that a diverse student body has more positive than negative educational effects (“Ford Foundation Survey”, 1999). Bowen and Bok (1998) found that majorities of both Black and White respondents favor even more diversity in educational settings than is now in place. A University of Michigan study found that students educated in diverse environments showed more complex thinking, more motivation to achieve, greater intellectual self-confidence and engagement, and the highest level of interest in graduate degrees (Gurin, 1999). There have even been claims that ethnic diversity is good for the health of the United States, since the nation has such a broad behavioral and gene pool to study when looking for the keys to health and longevity (Suinn, 1999). And finally, diversity is arguably a moral imperative when its promotion has the effect of rectifying past injustices and removing barriers to individual advancement. Even when no formal barriers exist for minority advancement, the lack of minority role models and mentors may be an obstacle that discourages talented individuals from entering a field. To the extent that the Moral Diversity -3 achievement of diversity throughout the work world “levels the playing field” for the next generation, diversity becomes closely linked to the basic virtues of liberty, justice, and equality (Ferdman & Brody, 1996; Fowers & Richardson, 1996). But despite the many claims made about the benefits of diversity, empirical evidence for these benefits is sparse. More troubling, there are empirical and theoretical reasons to worry about the potential costs of diversity. The Potential Costs of Diversity Multi-ethnic societies have always faced a high risk of dissension and civil war, and few such societies have been fully successful. Yet since the 18 century the United States has impressed foreign observers with its ability to unite and integrate people from diverse and even mutually hostile backgrounds (e.g., Crevecoeur, 1782/1997; de Tocqueville, 1835/1945). In the American motto “E Pluribus Unum,” (out of many, one) the “pluribus” has always pointed backwards, to the diverse backgrounds of Americans, while the “unum” has pointed forwards, referring to a common purpose, a common set of democratic values, and a shared future (Schlesinger, 1991). Beginning in the 1960s, however, American society has seen a major movement away from “unum” and toward “pluribus.” With the rise of identity politics, political correctness, and the multiculturalist movement in the 1980's, many historians and political scientists began to worry about new divisions and hostilities within American society. A variety of books appeared in the 1990's with titles such as The Twilight of Common Dreams (Gitlin, 1993), and The Culture of Complaint: The Fraying of America” (Hughes, 1993). In a widely cited book, The Disuniting of America, the liberal historian Arthur Schlesinger (1991, p.58) worried that "the cult of ethnicity exaggerates the differences, intensifies resentments and antagonisms, drives ever deeper the awful wedges between races and nationalities. The endgame is self-pity and self-ghettoization." Recent research and thinking in sociology supports these fears. Since the 1980's Americans have worried publicly and privately about the loss of a sense of community and of public life (Bellah, Madsen, Sullivan, Swidler, & Tipton, 1985). The 1990's saw the rise of several movements in response, such as communitarianism (Etzioni, 1994), and a new interest among social scientists in creating “healthy communities” (Putnam, 1995). It has recently been recognized that the “social capital” of a community can be as important for its health as its financial or intellectual capital. Social capital has many forms, but one important form is the dense network of obligations, expectations, and trustworthiness that grows up in stable communities in which people share a common past and an expectation of a common future (Coleman, 1988). Anything that divides people, encouraging them to separate into non-interacting or non-trusting subgroups, reduces the community’s social capital. Research in social psychology offers still more reasons for concern, showing just how easy it is to divide people against each other. Tajfel’s classic research on “minimal groups” demonstrated that people will divide themselves into groups at the drop of a coin (Tajfel, 1982; Moral Diversity -4 Tajfel & Billig, 1974; Turner, Brown, & Tajfel, 1979). When Tajfel and his collaborators created trivial similarities and differences among participants (based on having preferred the same abstract painter, or having chosen the same side of a coin in a coin flip), participants quickly formed affectively laden identifications with their fellow “group” members. Even though participants never directly interacted with each other, they liked their group members better, and they behaved spitefully towards “outgroup” members in a monetary distribution game. Other research has examined more meaningful conflicts, in which groups are pitted against each other in competing for resources. In their classic “summer camp” study, Sherif andpainter, or having chosen the same side of a coin in a coin flip), participants quickly formed affectively laden identifications with their fellow “group” members. Even though participants never directly interacted with each other, they liked their group members better, and they behaved spitefully towards “outgroup” members in a monetary distribution game. Other research has examined more meaningful conflicts, in which groups are pitted against each other in competing for resources. In their classic “summer camp” study, Sherif and collaborators randomly assigned boys at a summer camp to two groups. Conflict over resources (playing time on the baseball diamond, prizes won for athletic competition) eventually led to ingroup cohesiveness and outgroup hostility. The same affectively laden distinctions that were seen in the Tajfel studies were made far more strongly here between “us” and “them” (Sherif, Harvey, White, Hood, & Sherif, 1961). These findings demonstrate the affective consequences of arbitrary divisions made by an experimenter. When divisions are made on the basis of socially significant factors such as race, religion, sexual orientation, or country of origin, the resulting intergroup hostility can be far more serious. The most deadly riots in American history, from the draft riots of 1863 to the Los Angeles riots of 1992, have been race riots (Morris & Morris, 1976). Most American street gangs form along racial or ethnic lines (Shelden, Tracy & Brown, 1997). It seems that people, especially young men, will spontaneously form groups based on racial or ethnic similarity, and groups of young men will actively seek out other groups of young men for competition and conflict (Tiger, 1969). Different Kinds of Diversity We might at this point say that diversity, like most interesting things, has both costs and benefits. But there is another, more hopeful possibility: that diversity is like cholesterol, or like witches in the Wizard of Oz; that is, there is good diversity, and there is bad diversity. We might therefore be able to maximize the desirable form while minimizing the undesirable form. Several organizational behavior researchers have suggested that different kinds of diversity may have different kinds of effects. Jackson, Stone, and Alvarez (1992, p.56) distinguished demographic attributes from personal attributes. Demographic attributes are “those that are immutable, that can be readily detected during a brief interaction with a person, and for which social consensus can be assumed (e.g., sex, race, ethnicity, age).” Personal attributes, on the other hand, are “mutable and subjectively construed psychological and interpersonal characteristics (e.g. status, knowledge, behavioral style), which can change as a consequence of socialization processes." One personal attribute that they mentioned but did not discuss at length is values, including attitudes of all sorts. Moral Diversity -5 Williams and O’Reilly (1998), and Milliken and Martins (1996) specifically suggested that researchers should examine value diversity along with demographic diversity and other kinds of diversity. However their reviews of the literature found almost no empirical studies that have examined the effects of value diversity on group performance. The one study that addressed value diversity directly (Meglino, Ravlin, & Adkins, 1989) found that value congruence between workers and their supervisors was positively correlated with job satisfaction and commitment. A more recent study (Jehn, Northcraft, & Neale 1999) found that high value diversity predicted lower team effectiveness, efficiency, and morale, while high informational diversity (i.e., differences in knowledge bases and perspectives) was correlated with higher team effectiveness. However value diversity in this study referred not to moral values but to work-related values such as what the team’s real task, goal, target, or mission was thought to be. In the present article we suggest that the concept of value diversity is valuable, but that it should be reformulated as moral diversity, to focus on the kinds of important values upon which people’s world-views are based. We suggest that the key to resolving the paradox of diversity is to contrast moral diversity with demographic diversity. More precise definitions will allow an examination of this hypothesis. Demographic diversity. can be defined as the state of a group when a substantial percentage of its members (20%, perhaps?) fall into categories other than the modal category, on each of the principle demographic features. Demographic features are socially marked aspects of identity that one did not choose, and that cannot be easily changed. Race, gender, and ethnic or national origin are the three prototypical demographic features that are at stake in modern discussion of diversity, but the above definition admits less prototypical features such as social class, religion, sexual orientation, handicapped status, and age. Moral diversity can be similarly defined as the state of a group when a substantial percentage of its members (20%?) do not value the most valued moral goods of a community. Moral goods are social, personal, or spiritual obligations (e.g., justice, social harmony, self-actualization, piety, chastity) to which one appeals to justify or criticize the practices and behaviors of others, and which are felt to be binding on all people (or at least on all people in a particular role or position; see Shweder & Haidt, 1993; Shweder, Much, Mahapatra, & Park, 1997). Moral goods are experienced as affectively laden self-evident truths, or intuitions; people care strongly about them, and find it difficult to explain their goodness to someone who does not share their intuition (Haidt, in press). A simpler but equivalent way of describing moral diversity is as the state of a group when many different ideas of right and wrong are represented, and there is no widespread consensus about which moral goods should be pursued. Why Moral Diversity May Be Less Desirable than Demographic Diversity The above definition shows moral diversity to be a fraternal twin of anomie. Durkheim (1897/1951) argued that people need society to provide a normative order, with regulatory Moral Diversity -6 constraints, within which people can set and pursue goals. When the normative order breaks down and there is no moral consensus, the resulting state is known as anomie, or normlessness. It is a state in which suicide rates rise and the sense of meaning or purpose in life becomes harder to find. From a Durkheimian perspective, moral consensus is essential for a healthy community, while moral diversity is a threat. Social psychologists have come to similar conclusions. An enormous body of research demonstrates the importance of similarity, particularly shared attitudes, for interpersonal attraction and cooperation (Byrne & Clore, 1970; Byrne & Nelson, 1965; Heider, 1958; Newcomb, 1961, 1978). Interacting with people who hold dissimilar attitudes raises skin conductance levels (Clore & Gormly, 1974), providing a visceral cue that may damage further interactions. Disagreements that challenge one’s cultural and moral worldview lead to desires for ostracism and punishment (Solomon, Greenberg, & Pyszczynski, 1991). Byrne et al. (1975, p.206) noted that “the response to the threat raised by disagreement is to denigrate those who disagree; not only are they rejected, but they are also seen as lacking in intelligence, knowledge, morality, and psychological adjustment.” Rokeach specifically contrasted race and belief as determinants of liking, and found that shared belief always trumped shared race. White participants, even in the segregated South, said they would prefer to be friends with a Black person who shared their beliefs on important issues (e.g., God, communism, and desegregation) than with a White person who held opposing beliefs (Rokeach, Smith, & Evans, 1960; see similar findings in Anderson & Cote, 1966). Rokeach and Mezei (1966) replicated these results in a behavioral study. White and Black participants chose to go on a coffee break with discussion group members (confederates) who shared their beliefs but not their race more often than with discussion group members who shared their race but not their beliefs. Subsequent research has demonstrated, however, that the relative magnitude of racial similarity and belief similarity effects depends on other factors, such as the presence or absence of norms about cross-race association, and the degree of intimacy of the relationship, i.e., shared race matters more for dating, while shared belief matters more for general liking (Hyland, 1974; Insko, Nacoste, & Moe, 1983; Stein, Hardyck, & Smith, 1965; Triandis & Davis, 1965). Why Wanting Moral Diversity may be Incoherent The above findings suggest practical reasons for avoiding moral diversity: if one wants to build cohesive, non-anomic communities, moral diversity is problematic. But there are philosophical arguments that go further, implying that one cannot even coherently want moral diversity. Most philosophical attempts to define morality include as a necessary feature that moral rules apply universally (Hare, 1981; Kant, 1959/1785). If one says “I value gender equality, but others need not value gender equality,” then gender equality is a matter of personal taste. If one says “We in our culture value gender equality, but people in other cultures need not value gender equality,” then one is treating gender equality as a social convention (Turiel, 1983). Moral Diversity -7 But if one sees gender equality as a moral good or a moral truth, then one is committed to saying “I value gender equality, and everyone else should too, even in other cultures.” The mere act of saying “I value X, but I would prefer to live in a world where there is diversity with respect to X” is to deny that X is a moral good. Indeed, what would you think of a person who declared “I value tolerance (or abortion rights, or the Ten Commandments), but I would prefer to live in a world where many people do not.” Such a statement is either incoherent, or else it trivializes tolerance (or abortion rights, or the Ten Commandments) as a personal taste. The Present Research The present studies were designed to determine whether moral and demographic diversity have different profiles of desirability. In the first two studies we examined college student preferences for diversity on 17 issues in three social domains (at the university as a whole, in a small seminar class, and in a roommate) using a within-participants design. In Study 1 we surveyed almost all members of a male fraternity, and in Study 2 we surveyed a broader and more diverse cross-section of non-fraternity college students. We began with the following hypotheses: 1) Diversity preferences will vary by diversity-type such that demographic diversity will be more valued than moral diversity. 2) Diversity preferences will vary by domain such that diversity of all types will be most desired in the least intimate domain (the university), and least desired in the most intimate domain (roommate). That is, we expected diversity to show the profile of a NIMBY (Not In My Back Yard) issue: a public good that people may support, as long as they do not have to have frequent and close contact with it. 3) In the intimate setting of a fraternity, admission decisions will select against moral diversity, but not demographic diversity. That is, because attitude dissimilarity leads to repulsion (Rosenbaum, 1986), we expected that demographic diversity stripped of moral diversity would be fully acceptable. In Study 3 we used a between-participants manipulation to look more closely at the features (moral and demographic) of a target person that affect people’s desires for interaction, in a low intimacy and a high intimacy context. We predicted that Hypotheses 1 and 2 would again be supported. The general approach used in these studies was similar to that used by Rokeach, Smith, and Evans (1960) in asking participants about potential interaction partners who varied on demographic factors as well as on morally laden beliefs. However that work was mostly performed in the 1960's and 1970's, before diversity became a public good. The present research incorporates several new features that make it more relevant for current discussions of diversity and multiculturalism. Most importantly, it moves beyond preferences for dyadic interactions to directly ask students about their ideals for the diversity of the groups that they live and study in. Moral Diversity -8 Since college admissions is the most important battleground for debates over diversity and affirmative action, it is important to know what kinds of diversity college students want, and where they want it. Secondly, the present study examines desires for diversity within a college fraternity, an institution often accused of creating pockets of low diversity on college campuses. By surveying almost all members of a real residential group we were able to find out how desires for different kinds of diversity might shape the creation of such a group. Third, the present study examined a wide variety of attitudes and values, including issues of lifestyle and personal taste; it did not limit itself to the heavily moralized values studied by Rokeach et al. Study 1 Study 1 examined male fraternity members’ preferences for diversity by having them complete three questionnaires, assessing: 1) their own attitudes and demographic features; 2) how variation on those attitudes and demographic features would affect the admission of potential candidates to their fraternity, and 3) how much variation they prefer on those attitudes and demographic features at the university as a whole, in a seminar class, and in a roommate. Method Participants. Thirty-two members of an all-male social fraternity at the University of Virginia participated. The second author was a member of this fraternity, and he solicited their voluntary cooperation. All but three members of the fraternity participated. All participants were undergraduates at the University of Virginia, including 13 sophomores, 11 juniors, and 8 seniors. The mean age was approximately 20 years. Three participants identified themselves as African American, 28 as Caucasian, one as Hispanic, and none as Asian-American. The religions they were raised with included 14 reports of Protestant denominations, 9 of Catholicism, 5 of Judaism, and 1 of Greek Orthodox. Materials. Three questionnaires were stapled together into a packet. The Self questionnaire asked the participant about his own attitudes and activities on 16 items (e.g., “how much do you enjoy participating in athletics?”, “how do you feel about abortion?”) using a 9-point Likert scale. The Self questionnaire then asked about a variety of demographic factors, including the participant’s social class, ethnicity, religion, political views, urban/ruralness of hometown, and favorite sports teams. The Self questionnaire was followed by the Admissions questionnaire, which began: “Each of the following people represents a potential member of your fraternity. For each person, please check whether the characteristics mentioned would make you more likely or less likely to admit the person into your fraternity. If the characteristics would have absolutely no effect on you, then check ‘indifferent’.” Forty-three potential candidates were then described who varied on one of the attitudes, traits, or demographic factors that were assessed in the Self questionnaire (e.g., “An individual who is extremely pro-life and thinks that all abortion doctors are murderers”). Moral Diversity -9 The third questionnaire was the Diversity questionnaire, which directly assessed preferences for 17 types of diversity in three social domains, beginning with the least intimate University domain. Participants read the following instructions: If you were considering potential students to come to UVA, how diverse would you want the students to be? For each issue, on a scale of 0 to 9, circle the amount of diversity you would want for each question. A"0" means you want no diversity (everyone is the same as you), while a "9" means you want maximum diversity (the majority of the people are not like you and are spread out among all possible categories). Participants were then asked about 17 specific issues, each of which had been assessed on both the Self questionnaire and the Admissions questionnaire (e.g., “the ethnicity of the students”, “the views on abortion of the students”). The instructions were then repeated but with the first sentence changed to read “If you were considering what type of students you would prefer to have in a small seminar class (10 people), how diverse would you want the students to be?” The 17 issues were then repeated, with the same 10 point Likert scale. The instructions were then repeated a third time, with the first sentence changed to read “If you were considering a potential roommate, how similar/different from you would you want him to be?” The scale was redefined to be used for a single person by saying “A ‘0' means you want that person exactly the same as you, while a ‘9' means you want your roommate being very different from you.” The three versions were always given in this order because it was thought that participants would be most familiar with thinking about diversity at the University as a whole, and would use those ratings as a baseline when thinking about the seminar and roommate versions. The items on all three questionnaires were chosen to represent three different kinds of issues. The two kinds we were most interested in were demographics (ethnicity, SES, and religion), and attitudes about politico-moral issues (abortion, gun control, environmental protection, and affirmative action). In addition, to compare demographic and moral diversity to other factors that might be relevant to fraternity admissions, we asked about a variety of personal tastes (in music and recreation), social assets (social skill level, physical attractiveness), and attitudes about activities that might affect the social interactions of fraternity brothers (views on drinking, hazing, marijuana use, sexual promiscuity, and sexual orientation). We can group these factors together temporarily as “social-interactional” issues. Procedure. Each participant was handed the three questionnaires and asked to fill them out in a room without anyone else present. Upon returning their completed questionnaires participants were debriefed and asked to avoid discussing the study with other fraternity members. Results Results are organized around three questions: 1)What kinds of diversity do participants want? 2)Where do they want it? 3)What factors affect fraternity admissions? Moral Diversity -10 What kinds of diversity do participants want? Table 1 shows the mean preferences for diversity, on a 0-9 scale, for the 17 kinds of diversity. The table is sorted by declining desire for diversity, averaged across the three domains. This ordering shows that the most valued kinds of diversity are the three demographic factors (SES, ethnicity, and religion), followed by a few social-interactional issues, followed by the politico-morality issues. The least valued kinds of diversity are the social-interactional issues involving sexuality and drug use. The fraternity members generally want the people around them, particularly their roommates, to share their attitudes about sex and drugs, but they do not want everyone around them to be like themselves in SES, ethnicity, or religion. Where do they want it? Table 1 shows that preferences for diversity varied considerably by domain. A one-way repeated-measures MANOVA was performed on all 17 diversity items, using domain as a within-subjects measure, and a significant overall effect of domain was found, F(34, 90) = 3.65, p < .001. Univariate tests found a significant effect of domain for all 17 items (see Table 1 for F and p values). However the ordering of the domains was unexpected. We had predicted that preferences for diversity would be inversely proportional to the intimacy of the domain (Hypothesis 2), such that diversity would be most valued at the university and least valued in a roommate, with the seminar classroom falling in between. Yet Table 1 shows that diversity was most valued in the seminar classroom, followed closely by the less intimate university domain, and then preferences for diversity drop off sharply in the roommate domain. This pattern holds for 16 of the 17 items (all but Social Skill Level), and paired-samples t tests show that the difference between the seminar and roommate domains was significant in every case (at p < .001, except for attractiveness, which was significant at p < .05). Differences between the classroom and university domains were smaller and less frequent; they were significant at p < .01 for gun control, affirmative action, marijuana use, alcohol use, and sexual orientation, and they were significant at p < .05 for ethnicity, religion, athleticism, environment, and sexual promiscuity. The interaction of diversity-type and domain. To facilitate the analysis of the interaction of diversity-type and domain we reduced the number of variables by creating summary variables for different types of diversity. A principal components factor analysis of the average diversity preferences (averaged across three domains) for each of the 17 items yielded four factors with eigenvalues greater than 1, and a Scree test confirmed that a four factor solution was reasonable. These four factors accounted for 70% of the variance, and when rotated they gave an interpretable solution. We took the three highest loading items from each factor to create the following four subscales, which we labeled as 1) Demographics: Ethnicity, Religion, and SES. 2) Politico-moral: Environment, Affirmative Action, and Gun Control. 3) Socio-Sexuality: Promiscuity, Sexual Orientation, and Attractiveness. 4) Activities: Recreation, Athleticism, and Social Skill Level. We note that the first 2 subscales matched our a priori categories, although it Moral Diversity -11 is interesting to note that Abortion (which we had thought of as a politico-moral issue) ended up loading most heavily on the demographics factor (perhaps because the abortion debate is so closely linked to religious ideologies), while attitudes about marijuana and alcohol loaded highest on the politico-moral factor. Figure 1 shows how desires for each of the four diversity-types varied by domain. Demographic diversity was most highly valued in all three domains, while diversity in SocioSexuality was least valued. Diversity in Activities showed the greatest sensitivity to domain, plunging from a relatively high 5.8 in the University domain to 2.9 in the roommate domain. Apparently participants are happy to be at a University where people pursue diverse activities, but they would prefer to share activities with their roommates. A 3 (domain) x 4 (diversity-type) repeated measures MANOVA, where both factors were within-subjects, confirmed the features that are visible by eye in Figure 1: there is a significant effect of domain, F(2, 68) = 129.89, p < .001; there is a significant effect of diversity type, F(3, 67) = 39.19, p < .001, and there is a significant interaction between domain and diversity type, F(6, 65) = 7.16, p < .001. To test Hypothesis 1 directly, the repeated measures MANOVA was performed again as a 3x2 MANOVA, using only the two focal diversity types: politico-moral, and demographic. The results confirm that demographic diversity was more desired than moral diversity, F(1, 31) = 15.70, p < .001, supporting Hypothesis 1. The effect of domain was again significant, F(2, 30) = 28.76, p < .001, although this time the interaction between domain and diversity type was only marginally significant, F(2, 30) = 2.69, p = .08. Fraternity admissions. Given that fraternity members value different kinds of diversity differently, what kinds of people do they actually want to admit to their fraternity? The Admissions questionnaire presented 43 potential applicants to the fraternity for whom a salient trait was described. Participants were asked to say if the trait would make them more likely to admit the applicant (scored as +1), less likely to admit (scored as -1), or if the trait would make no difference (scored as 0). Table 2 shows how the 43 potential applicants would fare. Onesample t tests compared each mean to 0, to evaluate the null hypothesis that the trait in question did not affect the likelihood of admission. With a more stringent alpha level of .01 (because of the use of multiple t-tests) 22 of the 43 traits exerted a significant effect on hypothetical admissions decisions. Consistent with widely held stereotypes about fraternities, Table 2 shows that this particular fraternity is a men’s social club, not a political club. Members are looking to admit athletic, socially successful men whose lifestyle will fit in with the current members. Members do not want to admit people who are gay, Afrocentric, or who hold strong religious views. It should further be noted that the fraternity as a whole does not seem to care about potential members’ politico-moral beliefs, including whether the applicant is a liberal or conservative. But are traits valued or feared because they exemplify or threaten a fraternity ideal, or are fraternity members Moral Diversity -12 each trying to pick people identical to themselves? If the latter, then the fraternity admissions process would appear to be a diversity-reducing mechanism. To determine whether fraternity members were trying to admit people who resembled themselves individually we calculated the correlations between the traits given in Table 2 and the self-ratings on the same traits that participants made in the first questionnaire, in which they were asked for their own views on gun control, abortion, homosexuality, etc., as well as for demographic information about themselves. For eight of the traits no such match was possible (e.g., we did not know how “successful” our participants were with women, and there were no Asian or Buddhist participants). The remaining 35 correlations are given in the right-most column of Table 2. All but three of the correlations are positive, indicating that individual fraternity members are more likely to value potential applicants that resemble themselves, both on attitudes and on demographic background. However a closer inspection reveals that the largest correlations occurred on the politico-moral items. Of the nine correlations that were significant at p < .01, six of them were being for or against gun control, being for or against affirmative action, and being politically liberal or conservative. Two of the remaining three were issues with moral overtones for many Americans (being gay, or being a regular marijuana user). In other words, if a fraternity member was pro-gun control he wanted a potential fraternity member to be pro-gun control. However, the fraternity member’s social class, religion, and liking for books and sports did not make him more likely to favor someone of his class, religion, or recreational preference. It appears, then, that fraternity members as individuals show an anti-diversity preference primarily on moral issues, but these preferences are overridden by the more important shared preference for athletic, socially successful men who like to drink. Discussion The results confirm hypotheses 1 and 3, and suggest a modification of Hypothesis 2. Diversity preferences did indeed vary by type (Hypothesis 1). Demographic diversity was the most valued form of diversity, in all three domains, and it did not adversely affect fraternity admissions. Politico-moral diversity was less highly desired in all three domains, and it was disliked by individual members in the fraternity admissions process. Diversity preferences also varied by domain (Hypothesis 2), but not in the way that we had expected. All kinds of diversity were more highly valued in a seminar class than in the less intimate university setting. This finding supports advocates of diversity who claim that diversity enriches education for everyone. Members of a mostly White and not very liberal fraternity said they want to be exposed to people who are different from themselves in a seminar class. They did not see diversity as a burden or obligation to be spread around the university, but best kept far from themselves. The only item in Table 1 that was not ranked above the midpoint of the scale in the seminar domain was Sexual Orientation. But given the nature of the scale, giving a high rating on this item would have meant wanting to be in a class that was mostly comprised of gay Moral Diversity -13 students. The fact that the fraternity brothers gave a mean rating of 3.97 when the midpoint of the scale was 4.5 indicates that they were not generally averse to having a few gay classmates, although they were extremely averse to having a gay roommate (M = 0.84, the lowest rating in all of Table 1). The findings on the Admissions questionnaire confirm, at least at this one fraternity, many of the widely held stereotypes about fraternities: members value a certain lifestyle, involving drinking, sports, loud music, and the pursuit of women. These values appear to guide fraternity admissions, and members do not want to admit anyone who will interfere with that lifestyle, e.g., born-again Christians, homosexuals, afrocentric African-Americans, or teetotalers. In the intimate and highly interactive environment of the fraternity, lifestyle diversity is strongly disliked, while demographic diversity (particularly in matters of race and SES) is not problematic, as long as it does not bring with it cultural diversity. Black and Asian candidates faced no discrimination when they were presented as members of mainstream American culture. The Black upper-middle class candidate was in fact one of only 5 candidates whose description made him significantly more likely to be admitted. Moral issues showed an unusual pattern of effects. On average politico-moral beliefs had little impact on admissions decisions (most means were near 0 in Table 2, except for the strongly prolife candidate). Again, the fraternity is a social club, not a political club, and shared activities are more important than shared political views. However it was primarily on the politico-moral issues that individual members’ self-ratings correlated with their admission decisions. Liberal members preferred liberal candidates, while conservative members preferred conservative candidates. It appears to be valuable, then, to distinguish among types of diversity, and among domains of interaction. But the group examined in Study 1 is an unusual group -all-male, and self-selected for compatibility. It may not be representative of the broader college population. Before claims about the valuation of different kinds of diversity can be made, a replication study is needed. Study 2 Study 2 was designed to replicate Study 1 using a non-fraternity sample that would be more representative of students at the University of Virginia. Method Participants. Thirty nine students in an introductory psychology class at the University of Virginia completed the study in exchange for a half hour of experimental credit. Members of fraternities and sororities were excluded from participation. The 23 women and 16 men included 26 freshmen, 10 sophomores, 1 junior and 2 seniors, so the mean age was approximately 19 years. Twenty-nine participants identified themselves as Caucasian, 4 as African-American, and 6 as Asian or Asian-American. The religions they were raised with included 14 reports of Moral Diversity -14 Protestant denominations, 6 of Catholicism, 4 of Judaism, 2 of Islam, 1 each of Hinduism and Paganism, and 9 of “none”. Materials. The Self questionnaire from Study 1 was used again, with several minor changes for a non-fraternity population (e.g., questions about hazing, year of pledging, and whom you “hang out with” were dropped). The Self questionnaire was followed by the Diversity questionnaire from Study 1, with the omission of questions about “views on hazing”. Design and Procedure. Participants were given the two questionnaires and an informed consent form at the end of one class, and were asked to return the completed forms at the beginning of the next class. Results Results are organized around three questions: 1)What kinds of diversity do people want? 2) Where do they want it? And 3) How does a non-fraternity population compare to a fraternity population in its preferences for diversity? What kinds of diversity do people want? Table 3 shows the mean preferences for diversity on 16 issues across three social domains. The table is sorted by declining average desire for diversity, showing a similar ordering to that shown by the Fraternity sample in Table 1. This time, however, the ordering by diversity-type is conceptually neater, corresponding almost exactly to the four factors derived from the factor analysis of Study 1: Participants want the most diversity on demographic variables, followed by activities variables. They want less diversity on the morally loaded variables: politico-moral attitudes are all grouped together at the midpoint of the 0-9 scale, while items about sexuality and drug use elicit the lowest preferences for diversity. This pattern supports Hypothesis 1 (preference for demographic diversity over moral diversity). Where do they want it? Table 3 shows the same pattern of domain effects that was found in study 1: participants want the most diversity in a seminar setting and the least in a roommate. The ordering of Seminar > University > Roommate held for 13 of the 16 kinds of diversity (all but physical attractiveness, social skill level, and sexual orientation). A one-way repeatedmeasures MANOVA using domain as a within-subjects factor found an overall effect of domain, F(32, 4) = 5.13, p < .001. Univariate tests found a significant effect of domain for all 16 diversity types (see Table 3). Paired-samples t tests confirmed that the difference between the Seminar and Roommate domains was significant for all 16 diversity types (all at p <.01), although the differences between the University and Seminar domains were only significant for Athletics, Gun Control, Environment, and Affirmative Action (all at p < .05). The interaction of diversity-type and domain. As in Study 1 we sought to reduce the number of diversity-types by creating summary variables for different types of diversity. A principal components factor analysis of diversity preferences (averaged across three domains) for the 16 items yielded three factors with eigenvalues greater than 1, however a Scree test showed a sharp elbow after 2 factors, which accounted for 59% of the variance. When a varimax rotation was Moral Diversity -15 applied to a two factor solution two easily interpretable factors were found. The first factor merged activities and demographic variables (loadings in declining order: Athleticism, Recreational Interests, Sexual Orientation, Music, SES, Attractiveness, Religion, Ethnicity, and Social Skill). The second factor appeared to be a morality factor, mixing political morality with attitudes about drugs and sex (loadings in declining order: Abortion, Alcohol, Gun Control, Marijuana, Environment, Sexual Promiscuity, and Affirmative Action). We computed subscale scores for each participant using the seven highest loading items on each factor (i.e., all items with loadings above .60). Figure 2 shows how preferences for diversity varied across the three domains for each of the two subscale scores. Figure 2 shows the same general picture as did Figure 1: Demographic and activities diversity is most valued, especially in the public setting of the university and the educational setting of a seminar class. Moral diversity is less valued in all domains, although the difference between diversity categories shrinks among roommates. A 2 (diversity-type) x 3(domain) within-subjects MANOVA confirmed the effect of diversity type, F(1, 38) = 24.99, p < .001, the effect of domain, F(2, 37) = 80.69, p < .001, and the interaction of diversity-type and domain, F(2, 37) = 9.52, p < .001. These findings support Hypothesis 1 (demographic diversity more valued than moral diversity), and they support the same modification of Hypothesis 2 that was found in study 1: desires for diversity are highest in a seminar class, and lowest in a roommate. Fraternity vs. non-fraternity responses. Differences between samples on individual items can be directly calculated by subtracting cells in Table 3 from the corresponding cells in Table 1. However to facilitate an overall comparison we computed subscale scores for the non-fraternity sample using the same four subscales used in Study 1, across each of the 3 domains. We then subtracted the 12 values obtained for the fraternity sample from the 12 values obtained for the non-fraternity sample. Table 4 shows the 3x4 table of differences between the samples, where positive numbers indicate that the non-fraternity sample wanted more diversity. Differences between the two samples were small. A 2x3x4 (Sample x Domain x Diversity-type) repeated measures MANOVA found no main effect of sample (fraternity vs. non-fraternity, F(1, 68) = 2.24, n.s.), and only one interaction involving sample, with diversity-type, F(3, 66) = 3.39, p < .05. Independent samples t-tests were conducted on each of the 12 pairs of values that went into Table 4, as well as on each of the row and column means. There were no significant differences between the samples on either of the two focal kinds of diversity -demographic and politicomoral. The few differences found were concentrated in the areas of activities and socio-sexuality. In all cases the fraternity sample wanted less diversity than did the non-fraternity sample, and this difference may in part explain why some people seek to live in fraternities in the first place, Moral Diversity -16 Discussion of Study 2 Moral Diversity -17 Study 2 confirmed the findings of Study 1. Once again, demographic diversity was more valued than moral diversity (Hypothesis 1). And once again all kinds of diversity were most valued in the seminar class and least valued in a roommate, supporting a modified version of Hypothesis 2: Diversity preferences vary by domain of interaction, being highest in educational contexts and lower in personal contexts. College students see an advantage to being exposed to people different from themselves in their classes. Study 3 Studies 1 and 2 established that when asked to think directly about how much diversity they want in different contexts, college students make distinctions among diversity types and among contexts. However by highlighting the issue of diversity these studies may have activated participants’ explicit pro-diversity attitudes. Yet one of the paradoxes of American college life is that despite high levels of support for diversity in theory, social groupings often remain racially balkanized and segregated in fact (Willis, Reeves, & Buchanan, 1977). Students may endorse diversity in general while choosing to live, eat, date, and spend time primarily with those like themselves. To examine more directly how different kinds of diversity might affect desires for social interaction, study 3 avoided the use of the word “diversity” in the main instrument, and instead asked students to say how much they would want to interact with a variety of potential partners, in either an academic or dating context. Once again we predicted that demographic diversity would be more valued than moral diversity (Hypothesis 1), and that diversity in an educational context would be more valued than in a more intimate dating context (Hypothesis 2, as modified by studies 1 and 2). The basic finding that race differences become more important than belief differences as the intimacy of contact increases was well established in the 1960's (Triandis & Davis, 1965), but we thought it worthwhile to revisit this issue to see if times had changed. Method Participants. Participants were 248 students (66% female) enrolled in an introductory psychology class at the University of Virginia. All were undergraduates ranging in age from 17 to 21 years. Eighty one percent of the students identified themselves as Caucasian, 8% as AfricanAmerican, 6% as Asian-American, and 5% as “Other.” Three students reported having one Caucasian parent and another of a different race. By religious background, 55% reported having been raised in a branch of Protestant Christianity, 25% reported Roman Catholicism, 6% Judaism, 4% other religions, and 10% reported “none” or “atheist.” Materials. Participants were randomly assigned to receive one of two versions (Lecture or Dating) of a questionnaire that assessed their willingness to interact with different others. The two versions were identical except for their opening vignettes. The Lecture version began with the following vignette: Moral Diversity -18 Suppose you are just starting a new lecture class. The workload for the class is very heavy, and the Professor suggests that people team up in pairs for the semester so they can share the reading load and the research for the final project. You don’t know anybody in the class, and you’re talking about this with your roommate when your roommate says, “Hey, I know someone in the class who’s really smart, and who might be a good partner for you.” You are interested, so you ask your roommate for more details about the person. How much would each of the following things affect your desire to work with that person? The Dating version provided a similar vignette: Suppose you are single at the start of a new semester, and you are interested in meeting someone. Your roommate’s boyfriend or girlfriend is coming to UVA for the weekend from William and Mary, and is bringing along another friend (of the sex to which you are attracted). Your roommate has seen this friend, and tells you that the friend is single and very attractive. You are interested, and so you ask your roommate for more details about the person. How much would each of the following things affect your desire to go out with that person? Following the vignette, participants were presented with 15 pieces of information (prompts) about the stranger. Each of the 15 prompts addressed one of six social issues, three of which were demographic issues (race, religion, and SES), and three of which were moral issues (attitudes about abortion, attitudes about gay rights, and political orientation). Specifically, the prompts presented the stranger in turn as being Pro-Choice; Pro-Life; from a wealthy, upper-class family; from a working-class, blue-collar family; a strong critic of gay rights; a strong supporter of gay rights; African-American; White; Chinese-American; Baptist; Jewish; Muslim; atheist; a liberal Democrat; and a conservative Republican. After each prompt the participant was asked to rate, on a seven point Likert scale, how each piece of information, taken separately, would affect the participant’s desire to interact with the stranger. The scale endpoints were defined as: 1 = “would make me very reluctant to work with [date] this person”, 7 = “would make me very enthusiastic about working with [dating] this person”, and the midpoint (4) was defined as “would make no difference.” Finally, participants were asked to identify their own race, ethnicity, religion, political orientation, and stance on abortion and gay rights. Procedure. The questionnaire was distributed at the beginning of a lecture, and was completed and collected within 8 minutes. At the top of the questionnaire participants wrote their private, in-class identification numbers, a regular feature of the course, which allowed responses from the present questionnaire to be related to responses from other questionnaires done in class, including the Marlowe-Crowne Social Desirability Scale.