It’s now been more than a decade since Robert Putnam, the Harvard political scientist behind the famous book Bowling Alone, reported a troubling finding that he had tried really, really hard to make go away: In census tracts with higher levels of ethnic diversity, people tended to report trusting each other less. Putnam didn’t use this as an argument for reducing immigration — to the contrary, he bent over backward to avoid that conclusion — but it was nonetheless clear that eroding social trust was a downside to immigration that we should pay attention to.
Putnam’s finding has continued to pop up in the immigration debate, even sparking a column last year from Noah Smith at Bloomberg View pointing out that, while some other research does back Putnam, “there are plenty of studies that don’t support Putnam’s conclusion — or even find the opposite.”
When there are lots of studies of the same topic, the time is ripe for a “meta-analysis” — a study of all the studies, synthesizing their findings. And that’s what three Danish researchers do in a new paper, available online now and set to be published next year in the Annual Review of Political Science. The paper combines 1,001 results published in 87 different studies in numerous countries.
On the one hand, the review finds that people who live in diverse areas are indeed less trusting. Putnam’s results were no outlier, and in fact they were slightly on the weak side.
On the other hand, any impact of diversity on trust is tiny. The economist Bryan Caplan once did some back-of-the-envelope math based on Putnam’s results and found that even sizable increases in diversity would yield only small losses in trust. And the new meta-analysis leads to the same conclusion.
To prepare readers to be underwhelmed, the authors point out that only 5 to 10 percent of the variation in people’s trust levels stems from differences across contexts like neighborhoods, as opposed to differences among individuals within those contexts. In other words, it’s common for someone to be much less trusting than his neighbor, but comparatively rare for a whole neighborhood to be much less trusting than another one. Further, diversity levels are just one feature that differentiates one context from another, alongside more directly pertinent ones such as crime rates.
On average across all the studies, diversity accounts for . . . something like 0.7 percent of the variation, above and beyond what we can explain with other variables. I don’t think a finding that small should play much of a role in the immigration debate. But the paper also notes some interesting patterns within these results and suggests avenues for further research.
For one thing, the effect of diversity on trust is most apparent at lower levels of geography. Diverse neighborhoods see lower levels of trust much more reliably than diverse countries do. Presumably this is because the demographics of someone’s neighborhood affect his day-to-day life a lot more than the overall demographics of his country do.
For another, different types of trust have different relationships with diversity. Trust in neighbors is most clearly lower when diversity is high. Interestingly, trust in members of out-groups is least clearly affected, suggesting that the phenomenon here is not just about racist backlash to immigration, though some people may hesitate to admit to survey-takers that they distrust out-groups specifically.
There are other interesting complications to this research as well. Different ethnic groups often have different levels of trust, so if you control for the ethnicities of the people who answer a trust survey, you separate out two different effects. You might find, as Caplan summarized Putnam’s results, that “‘diverse’ communities have low trust, but the reason isn’t that diversity hurts trust; it’s that non-whites — especially blacks and Hispanics — have low trust.”
Relatedly, one might separate the effects of diversity per se from the ways that some specific groups react to the presence of other specific groups — something the study doesn’t dwell on as much as I would have liked. (It mainly points out that, for technical reasons, this is hard to do.) For example, there’s some evidence that majority groups respond better to immigrant groups with higher skill levels and lower crime rates. If there’s similar variation in terms of how diversity affects trust, the ill effects of, say, low-skilled immigration could be much more severe than the overall average indicates.
It’s also worth pointing out that these studies do not use particularly rigorous methods. These studies mainly just check to see if diverse areas have lower levels of trust after statistically “controlling” for other variables, such as the income and education of the people responding to the survey being analyzed, as well as income levels and whatnot in the areas studied. Since we can never be sure we’ve measured and accounted for all the important variables, this approach is doomed to be less than convincing, even when a thousand different findings are all analyzed simultaneously. (After all, they might all be biased by the same mistakes and omissions.)
In addition, many question the value of the survey questions typically used to measure trust. On that point, a worthy companion to the new lit review is the just-released working paper from the Cato Institute’s Alex Nowrasteh and Andrew C. Forrester, which aims to blast holes in the way social scientists analyze the concept of trust, especially in the context of its importance to economic growth.
One study noted in the Danish lit review did confirm the basic “more diversity, less trust” finding by comparing changes in diversity with changes in trust within areas, but we could use a lot more work like that. “Natural experiments,” where areas are suddenly exposed to a great increase in diversity in a short time, would be especially valuable.
Immigration advocates hoping to “debunk” Putnam’s findings will be disappointed in this new review, because it finds that his results are entirely mainstream. Those hoping to use his work as evidence that diversity viciously tears us apart, though, should pay close attention to exactly how small these effects are and how much we don’t know yet.
Something to Consider
If you enjoyed this article, we have a proposition for you: Join NRPLUS. Members get all of our content (including the magazine), no paywalls or content meters, an advertising-minimal experience, and unique access to our writers and editors (conference calls, social-media groups, etc.). And importantly, NRPLUS members help keep NR going. Consider it?
If you enjoyed this article, and were stimulated by its contents, we have a proposition for you: Join NRPLUS.