Ethnic Diversity and Social Cohesion

This summer one of my tasks as a research assistant has been to conduct a literature search on recent articles within political science and sociology that pertain to ethnic diversity and social cohesion or trust. While the literature still appears rather divided, the majority of recent publications appear to take a negative relationship between ethnic diversity and social cohesion as a given starting point, and then move to challenge that apparent consensus (e.g. Tolsma & Van de Meer, 2018; Douds & Wu, 2018) ¹. In this post, I briefly review six of the more pertinent recent articles that explore the relationship between ethnic diversity and social cohesion. While the results are somewhat mixed, there is a concern that without intelligent policymaking, increasing ethnic diversity can (through a variety of mechanisms) negatively impact social trust and cohesion. If that is true, it behooves social scientists and policymakers alike to identify strategies to mitigate any negative consequences. To preview two economists discussed later in the essay, ethnic diversity is here to stay–in and of itself, it is not the problem, it is how it is dealt with that can be problematic.

Of the six articles presented here, Dinesan, Schaeffer, and Sonderskov (2020), hereafter DSS, is the most useful for our purposes, as they give a wonderfully thorough summary of the state of the literature on ethnic diversity and trust, as well as a meta-analysis of existing articles. Their meta-analysis reveals that ethnic diversity is negatively correlated with social trust at a statistically significant level (see also Gundelach & Manatschal, 2017 for confirmatory evidence). DSS stress, however, that the relationship is of a “rather modest size” substantively speaking, and that the “apocalyptic claims regarding the severe threat of ethnic diversity…are exaggerated” (p.457).

Since “social trust” is such an amorphous concept, they separate it into four forms (p.442):

1. Generalized social trust (i.e., trust in strangers)

2. Outgroup trust

3. Ingroup trust

4. Trust in neighbours

Their meta-analysis finds that trust in neighbours is most negatively affected by ethnic diversity. The effect is roughly double that of generalized social trust (p.451). As an explanation for the finding that ethnic diversity erodes trust, DSS review common explanations from the literature. Firstly, that people partly infer the trustworthiness of others based on cues from their local environment (p.443). The ethnic background of other people can be one of these cues. When this ethnic cue is present, it triggers a phenomenon known as outgroup aversion (p.443). That is, people trust those who are different from themselves less because they are less similar. Similarity implies shared norms and other behavior-regulating features that encourage trust. While this may be more easily seen in local contexts, generalized trust may also be affected because people tend to evaluate the trustworthiness of generalized groups partly based on what they experience locally (p.443). So experiencing greater ethnic diversity locally can lead to less trust in a more general sense. They find some evidence for that, although it is not as severe an effect as it is at the local level.

Gundelach and Manatschal provide a slightly different explanation (2017, p.4). They theorize that within ethnically diverse contexts, social categorization based upon ethnicity becomes more salient. And since a well-known finding of social identity theory is that one trusts one’s own group more, this process of social categorization based upon ethnicity naturally leads to less trust in ethnically diverse areas.

Also different in terms of mechanisms, but the same in terms of implications, is Putnam’s constrict theory (as reviewed in Dinesan, Schaeffer, and Sonderskov, 2020, p.444). This suggests that ethnic diversity lowers trust because it leads to social isolation; people “hunker down” more in ethnically diverse areas. This isolation can take the form of ethnically diverse settings being less socially integrated, of there being more perceived helplessness to act collectively, or simply communication being less likely. DSS find some evidence for his theory, but also some evidence against as well. The implications for social trust are correct, but Putnam also stipulated that all levels of trust would be impacted. While that did happen, DSS state that Putnam’s theory does not account for why there is such variation in trust between the local and more general levels.

A different theory DSS touch on only briefly is ethnic diversity inducing group threat from intergroup competition (p.452). They find little evidence of this theoretical explanation in their meta-analysis.

Although some studies have found that positive contact dampens the negative effects of ethnic diversity on trust, the meta-analysis reveals that the results generally are inconsistent, and furthermore, that controlling for contact does reveal the same negative relationship between trust and diversity (p.454). They do recommend that further studies on contact are needed, however.

The meta-analysis also finds that possible confounders such as social disadvantage and crime are not plausible alternative explanations. The effect of ethnic diversity remains even when controlling for such factors.

While DSS’s prognosis is grim, several authors suggest there are intervening variables that might lessen (and sometimes inadvertently strengthen) the effects of diversity on social cohesion. Watters, Ward, and Stuart (2020) offer a proposed solution at dampening the loss to social cohesion that ethnic diversity stimulates. Specifically, they suggest that multiculturalism² may be a viable strategy to encourage social cohesion in the face of ethnic diversity. In a survey that included measures related to multiculturalism and social cohesion, the authors found that perceptions of an environment of multiculturalism did increase trust and national attachment (which they view as two dimensions of social cohesion) for Latinos , but not for Whites. The reason for this discrepancy, the authors speculate, is because the presence of a widespread valuing of cultural diversity can make Latinos feel that their social identity is safe, whereas for Whites, there is a perception that multiculturalism is only for ethnic minorities. To remedy this, the authors think there needs to be a focus on an “all-inclusive” approach to encourage Americans to believe that multiculturalism belongs to everyone (p.90). So, in essence, multiculturalism may represent a good coping strategy for Western societies moving forward in order to cope with the increasing ethnic diversity from immigration.

Similar to the multiculturalism proposed by Watters, Ward, and Stuart (2020), Mintchev and Moore (2018) speculate that super-diverse neighbourhoods might not be penalized in terms of social cohesion like more traditionally diverse neighbourhoods. I use the term ‘speculate’ because the authors do not include any method of testing besides a very half-hearted inclusion of “ethnographic material from East London” as described by other authors. I think characterizing their work as “ethnographic” does a real disservice to actual ethnographies. Furthermore, they don’t speak to social cohesion per se, but to a more amorphous “prosperous society.” Nevertheless, the logic they lay out for why super-diversity might work differently from regular diversity is compelling.

Super-diversity involves migrant communities from multiple countries of origin who have some degree of internal fragmentation of groups, as in new versus old immigrants etc. (Mintchev & Moore, 2018, p.6). Mintchev and Moore think super-diversity may not erode trust for several reasons (see p.10). In traditional diverse environments, two or three well-consolidated ethnic groups may be a source of anxiety linked to specific fear of prejudice or persecution, but this anxiety is diffused with super-diversity. When surrounded by super-diversity one cannot afford to be uncivil towards others who are different because everyone is different. Furthermore, the knowledge that everyone has to be accommodating can make the neighbourhood feel more safe for everyone, knowing that each person is free to express their own cultural identity without seeming threatening. Also, since extreme diversity is typically seen as a normative good for liberals and young people, being part of such a neighbourhood can be a source of pride for those groups.

Gundelach and Manentschal offer an additional policy-based solution (2017). As noted earlier, they rely on a social identity explanation for ethnic diversity negatively impacting general social trust. They suggest that liberal integration policies have the potential to redefine who belongs to the ingroup by granting immigrants greater access to the political community, which may help increase trust (p.6). Or, it can also (according to Realistic Group Conflict Theory) lead to natives seeing immigrants gain access to resources and a place in the community that they would prefer for natives, thus increasing hostility and reducing trust (p.6).They test these competing hypotheses in Switzerland using surveys that ask about generalized trust, and matching that with ethnicity and policy data. They find a moderating effect for integration policy and evidence for both competing hypotheses. Some liberal integration policies (such as increasing immigrant access to employment) seem to increase generalized trust (p.14). Others, such as family reunification rights and naturalization rights, actually decrease trust even more. The authors speculate that these policies (which increase the number of ethnically-diverse immigrants) evoke feelings of threat to native-born Swiss citizens (p.14). So, not a silver bullet, but certain government policies can somewhat mitigate the negative effects of ethnic diversity on trust.

Fitting into this category of methods to reduce the negative impact of ethnic diversity on social cohesion is Meir and Fletcher’s paper (2019). This paper is about the potential of sports to promote social cohesion in ethnically diverse communities. The theory, although light, makes sense; unfortunately, they do not actually test their theory, as too many participants dropped out, rendering their study unfeasible.

They equate social cohesion with a term called “community cohesion.” A cohesive community is one in which there is a “common vision, sense of belonging, [and a] positive appreciation of diversity.” (p.1). That definition, while normatively pleasing, leaves much to be desired analytically. It implies that only diverse communities (or communities that value diversity) can be cohesive. So presumably Meir and Fletcher would consider any community outside of the modern era to lack cohesion by definition? I’m sure there were some cohesive communities even in reprehensible societies like Nazi Germany. Or for that matter, any community before 1900, as diversity had not yet gained any positive cachet as a term or concept. In any case, the authors go on to identify 8 different ‘domains’ of community cohesion:

1. Common values

2. A civic culture

3. Social order

4. Solidarity

5. Lack of income inequality

6. Social capital

7. Place attachment

8. Identity

Although they don’t delve deeply into this list, some of these domains are worth thinking about for future research, as they have been ignored by other researchers. For example, place attachment is not discussed enough in terms of being a marker of social cohesion. Meir and Fletcher take less time to describe how sports could help achieve that cohesion, simply saying that allegiance to a sport’s team could create a larger, more inclusive social identity that can eclipse other identities (p.5). Furthermore, sports provides opportunities for social contact in a meaningful and cooperative way, which can increase trust (p.5). In this, they fit into the strain of research (reviewed by DSS) that posits that meaningful social contact could help dampen the negative effects of ethnic diversity on social cohesion.

For what it’s worth, I think that as disappointing a paper as Meir and Fletcher present here, I love their tack. I have long wanted to carry out a study looking at the intersection of gender, sports, and what I would term “team unity”, but could easily be considered as social cohesion. Having been part of many sports teams, I know first-hand their ability to create a powerful sense of solidarity and unity, and I suspect that may be even more pronounced in girls’ and women’s teams. I think the Volden et al. (2013) piece about women legislators being more effective in many situations than men because of a greater commitment to “consensus-building activities” points the way theoretically. I could see extensions of this study including whether communities who have women in leadership positions being more cohesive.

For a completely different conclusion, the economists Van Staveren and Pervaiz suggest that social scientists have completely misidentified the relationship between ethnic diversity and social cohesion (2017). In their view, it is not ethnic diversity that is the problem, it is how diversity is dealt with. More specifically, they identify social exclusion as the mechanism that explains why increased ethnic diversity leads to lower trust. It is not the diversity in and of itself, it is the greater likelihood of social exclusion between groups that arises from diversity. When social exclusion is controlled for in regression analyses, the negative impact of diversity on social cohesion disappears (or at least becomes statistically insignificant). It is worth noting, however, that these authors measure both social exclusion and cohesion quite differently, or at least quite differently from political scientists. Social exclusion is measured from the proportion of survey respondents indicating that they have experienced discrimination (an index of various measures of discrimination); and social cohesion is measured by the number of riots, assassinations, terrorist acts, violent demonstrations, and other indicators of civil disorder (p.719). So it is not the more typical measures of generalized trust that most samples use.

Despite their rhetoric, it is also up for debate whether their conclusion is actually markedly different. They acknowledge that social exclusion and ethnic diversity often go hand in hand, since an increase in ethnic diversity leads to more chance of social exclusion between different ethnic groups. So greater ethnic diversity does lead to lower social cohesion in most instances, not because of diversity necessarily, but because of the social exclusion that accompanies diversity. The mechanism is indeed different, but the sentiment is actually somewhat similar to the other authors reviewed here. Both DSS and Gundelach and Manentschal would agree that it is not diversity by itself, but rather that diversity triggers something else that reduces social trust. In DSS’s estimation it is outgroup aversion, whereas in Gundeach and Manentschal’s it is a greater tendency to socially categorize based on ethnicity. Whatever the mechanism, greater ethnic diversity is more likely to reduce social trust and cohesion (at least in the short term). The more useful part of Van Staveren and Pervaiz’s argument (as well as Watters, Ward, and Stuart (2020) and Gundelach and Manentschal (2017)), is that it rejects the essentialism that can creep into the literature. Diversity does typically lead to reduced social trust through one of various mechanisms, but need not always if conditions can be created that do not include social exclusion, but rather promote multiculturalism or other inclusive policies.