Another arrow problem...  

Just as the R & R uproar in the economics community has reminded us correlation is not cause and sometimes it's really hard to see which way the causal arrow points, another landmark study receives a startling new interpretation.

From the introduction to a refreshingly contrarian article by Stephanie Stern, a law professor at Chicago-Kent College of Law:

In 2007, accompanied by a firestorm of publicity, Robert Putnam announced that residential racial diversity causes declines in social capital. Social capital is a prominent theory, popularized by Putnam, of the aggregate value of citizen participation in associations and organizations, social ties and networks, civic engagement, trust, and norms of reciprocity. In a study of forty-one U.S. communities, Putnam found that people living in racially diverse communities were less likely to work on a community project or volunteer, less likely to expect others to cooperate to solve collective problems, reported lower trust in others, had fewer close friendship ties, expressed less confidence in local government, and registered to vote lessfrequently. Most provocatively, Putnam found a strong “hunker[ing] down” effect, contrary to both the constrict and contact hypotheses of integration, where racial diversity caused residents of diverse communities to withdraw from social and civic life and report lower trust in members of other races and their own race. Unsurprisingly (to all but Robert Putnam it seems), his research provoked a torrent of political commentary and academic response. Conservative commentators argued that the findings called into question the value of racial mixing, headlines trumpeted the conclusion that “greater diversity equals more misery,” and Putnam’s research featured in a recent amicus brief as evidence against the value of affirmative action in college admissions. Sociologists and economists reanalyzed Putnam’s data and conducted their own empirical studies to assess his findings (these studies indicate that the diversity decrement is statistically significant, but small). Legal scholars accepted, albeit unhappily, the conclusion that racial diversity diminishes local social capital.

Curiously, in the handwringing about the harms to social capital and the ensuing debate, no one questioned whether the problem was social capital. From a property scholar’s perspective, one plausible interpretation of the correlation between high social capital and low diversity is that high social capital reduces the costs of excluding minorities (i.e., the non-dominant race in a community) and maintaining racial homogeneity. Holding preferences for racial homogeneity constant and positive, there may be reverse causation: high social capital, in the form of close social networks and strong tastes for organizational participation and voluntary action, may facilitate community organizing to exclude by race or class through both informal and legal mechanisms. The motivation for exclusion may be preferences for homogeneity, increased property values from exclusionary land use policies, or in predominantly minority, lower-income areas, concerns that white gentrification will make housing unaffordable. Conversely, low social capital may make it difficult for residents to organize to exclude and may result in greater racial fractionalization.

More succinctly: social capital facilitates NIMBYism.
The whole point of social capital, after all, is to facilitate collective action. It's usually bandied as a cure for commons-type problems like littered parks or crime-infested streets, but there's no reason to assume a neighborhood will deploy  its social capital only for wise and benevolent ends.  A neighborhood that is adept at organizing litter patrols and crime watches will also likely be adept at organizing opposition to real estate developments that threaten to add economic or racial diversity. If this is the case, then we should expect, rather than be surprised, to see high social capital negatively correlated with diversity. [A little more after my generous excerpt - sorry]

I find this idea plausible at least. Our automatic reaction to ascribe the logical flow of cause and effect makes our initial reactions "sticky", I guess. Being able to devise tests that can detect the chicken from the egg is crucial to preventing the establishment of false world-views and even faultier solutions for problems.