“Diversity” became the announced goal of schools and employers and liberal activists once American voters and courts turned against “affirmative action” for black Americans (never mind the idea of reparations). Earlier, in the 1960s and ‘70s, the Johnson and Nixon administrations had pushed racial “goals” (not quotas, they stressed) in a not-so-transparent effort to redress some of the economic disadvantages accrued from centuries of slavery and Jim Crow. However, with votes such as the 1996 passage in California of a constitutional amendment outlawing state institutions from considering race in employment, contracts, or education, and with Supreme Court cases reaching almost as far, liberals retreated from affirmative action to promoting “diversity.” Ethnic diversity, they argue, is good for everyone, not just minorities; it makes learning, working, neighboring, and deliberating better (e.g., here). Thus was born a defense for legally considering race just a bit, as well as a set of careers in diversity promotion, management, training, and law.
Opinion leaders from school teachers to corporate CEOs now promote, with some support from research, the virtues of diversity. Yet, out of view from most public discussion of the topic, a line of scholarly research emerged that implies the opposite. It suggests that the more diverse neighborhoods, cities, or countries are, the less people cooperate to common ends and the more they socially disengage; they “hunker down,” in one colorful rendition. A new paper by Maria Abascal and Delia Baldassarri in the latest issue of the American Journal of Sociology revisits this academic line of research and forces us to think back to why diversity was important in the first place.
The research I refer to involves calculating for some area–a census tract, county, state, or country–a measure of ethnic-racial heterogeneity. If everyone is of the same ethnic group, then ethnic heterogeneity is zero. If each person is of his or her unique group, heterogeneity is 1.00. In reality and depending on the specific kind of statistic, places vary in a narrower range from low to high heterogeneity. Importantly, as Abascal and Baldassarri remind us, whether residents of a community are 70 percent black, 20 percent white, and 10 percent Latino or are instead 10 percent black, 70 percent white, and 20 percent Latino, the community will score the same on “heterogeneity,” but surely would be different otherwise, a point to which I will return.
Studies have found that the more heterogeneous an area, the lower the level of common public goods, such as expenditures on social needs. And there is some evidence that the more heterogeneous the place, the greater the distrust and alienation among the residents. The most prominent statement along these lines is a highly cited 2007 paper by political scientist and author of Bowling Alone, Robert Putnam.
Using large national surveys conducted in 2000 to study “social capital,” Putnam reported that, with just about everything else held equal, the higher the heterogeneity index in someone’s neighborhood (census tract), the lower the average level of “trust” the respondents expressed toward neighbors, members of other ethnic groups, and even members of their own group. This central finding, like those from prior studies, helped create an undercurrent of pessimism about what diversity produces even as diversity was gaining such approval in the public discussion.
(A side note on “trust”: Researchers in the last few decades have generated a huge literature on what correlates with survey respondents saying they do or do not “trust” people–usually people in general or sometimes specific kinds of people. It is an odd literature built around an odd–and sort of accidental–question whose meaning is not clear and probably shifting. [See the discussion on pp. 192-94 and notes in Made in America.] Nonetheless, this literature has influence.)
Abascal and Baldassarri reanalyze the Putnam surveys. They initially find the same sorts of results: the more ethnically heterogenous a census tract–based on the percentages of white, black, Hispanic, and Asian residents–the less likely residents were to say that they trusted people in general, their neighbors, members of their own racial group, and of other racial groups, and the less likely they were to report that people in their communities would cooperate in an emergency.
Then, Abascal and Baldassarri make two important moves. One, anticipated by Putnam, is to point out that different kinds of people live in different kinds of neighborhoods. In a majority-white society, you cannot have diverse neighborhoods without adding in racial minorities. And it turns out that blacks and Hispanics, on average, report less trusting attitudes than do whites. (Why that might be so is another subject.) However, once the authors statistically adjust for this pattern, for ethnically-influenced “self-selection,” the negative effect of heterogeneity on all these measures evaporates, except marginally perhaps for distrusting neighbors. (In additional tables that Maria Abascal sent to me, neighborhood heterogeneity was not systematically and independently associated with respondents’ reports of giving to charity, the number of friends they claimed, or how much they said they trusted “the local government.”)
The other move Abascal and Baldassarri make, in some ways more meaningful, is to point out that neighborhoods that are heterogeneous are systematically different from those that are not and differently so for different groups. For most whites, a smattering of nonwhites is enough to make the neighborhood relatively diverse; for blacks, who commonly live in highly segregated and often quite poor neighborhoods, it takes an unusual number of other types of people to diversify the neighborhood. Once this distinction between kinds of homogenous or heterogeneous neighborhoods is taken into account as well as residents’ self-selection, heterogeneity is not connected to distrust, except ….
Except that when the authors break down the analysis by race–that is, by who is asked about trusting–they find that “[O]nly for whites, living among out-group members [by itself] predicts lower levels of trust. However, it is not ethnoracial diversity per se that makes whites apparently ‘hunker down’ but rather the presence of nonwhites, particularly blacks and Hispanics.” That is, the authors discover that a distrust response to a specific kind of diversity seems specific to whites. This finding is consistent with much other research showing that whites try to avoid and react negatively to non-white neighbors, especially blacks. Indeed, the general findings in other studies that greater heterogeneity seems to lead to fewer public goods can usually be better understood as findings that larger percentages of blacks leads whites to resist the provision of public goods more.
So, perhaps we are back to where we started circa 1965. Diversity–that is, racial integration–matters not necessarily because it might create wonderful “social capital,” but because it might counteract an historical racial caste system.