A few weeks ago I got a phone call from a fellow writing an article for the Boston Globe on Harvard political scientist Robert D. Putnam's research on ethnic diversity's impact on social capital, which I've been writing about, every now and then, since 2001 (here and here). But the journalist seemed at least as interested in asking about David Duke, of all people, as about Putnam's data, so it didn't seem like a very productive conversation. Anyway, here's his article:
The downside of diversity
A Harvard political scientist finds that diversity hurts civic life. What happens when a liberal scholar unearths an inconvenient truth?
By Michael Jonas | August 5, 2007
IT HAS BECOME increasingly popular to speak of racial and ethnic diversity as a civic strength. From multicultural festivals to pronouncements from political leaders, the message is the same: our differences make us stronger.
But a massive new study, based on detailed interviews of nearly 30,000 people across
"The extent of the effect is shocking," says Scott Page, a
Meanwhile, by drawing a portrait of civic engagement in which more homogeneous communities seem much healthier, some of Putnam's worst fears about how his results could be used have been realized. A stream of conservative commentary has begun -- from places like the Manhattan Institute and "The American Conservative" -- highlighting the harm the study suggests will come from large-scale immigration. But Putnam says he's also received hundreds of complimentary emails laced with bigoted language. "It certainly is not pleasant when David Duke's website hails me as the guy who found out racism is good," he says.
This reminds me of when Mearsheimer and Walt's essay on "The Israel Lobby" came out and the neocon NY Sun immediately solicited an endorsement of it from David Duke, which then showed up in 56,000 of the first 177,000 references to it on Google.
So how to explain
Diversity, it shows, makes us uncomfortable -- but discomfort, it turns out, isn't always a bad thing. Unease with differences helps explain why teams of engineers from different cultures may be ideally suited to solve a vexing problem. Culture clashes can produce a dynamic give-and-take, generating a solution that may have eluded a group of people with more similar backgrounds and approaches. ...
So that's why
The image of civic lassitude dragging down more diverse communities is at odds with the vigor often associated with urban centers, where ethnic diversity is greatest.
This is another version of the theory that Richard Florida gets $35,000 a lecture for propounding, that the reason high tech centers like Silicon Valley are rich is because they attract a lot of gays, bohemians, artistes, and immigrants. That appears to get the causality backwards -- Dr.
It turns out there is a flip side to the discomfort diversity can cause. If ethnic diversity, at least in the short run, is a liability for social connectedness, a parallel line of emerging research suggests it can be a big asset when it comes to driving productivity and innovation. In high-skill workplace settings, says Scott Page, the
"Because they see the world and think about the world differently than you, that's challenging," says Page, author of "The Difference: How the Power of Diversity Creates Better Groups, Firms, Schools, and Societies." "But by hanging out with people different than you, you're likely to get more insights. Diverse teams tend to be more productive."
In other words, those in more diverse communities may do more bowling alone, but the creative tensions unleashed by those differences in the workplace may vault those same places to the cutting edge of the economy and of creative culture.
Look, there is a theoretical upside to having both object-oriented Western engineers and context-oriented East Asian engineers, but the friction costs imposed by diversity (e.g., language difficulties and culture differences) make that hard to achieve profitably in the real world. Anyway, that's not what Americans mean by "diversity." Here, the word means hiring more underperforming minorities (e.g., blacks and Latinos), which, not surprisingly, doesn't improve your organization's performance.
What people often get mislead by when they claim that diversity improves performance is a simple selection effect: if you have a highly selective, big money organization, you are often going to end up with people from exotic places, but that doesn't mean that -- all else being equal -- diversity makes your organization work better. It just means that not all else is equal: world-class talent is found in several parts of the world (although not necessarily all parts of the world).
For example, consider the Top 10 Golfers in the World. For most of this decade, one member of that group has been a black-skinned, white-featured Indian from the
So, what happens is that people notice that a glamorous world-class organization like, say, the New York Yankees has a diverse set of outstanding employees from all over the world, so, therefore, the way to make, say, your Dunder-Mifflin regional paper products office in Scranton more successful is to increase its diversity.
Well, no, that doesn't follow, for reasons that are obvious if you are allowed to articulate them without calling the wrath of the Equal Employment Opportunity Commission down on your head: your organization probably isn't world-class so it can't select world-class talent. If you are currently irrationally discriminating in hiring from the local market, then stop it: you can attract better talent by hiring meritocratically. But if you are already hiring meritocratically, then "seeking diversity" (i.e., hiring more blacks and Hispanics by lowering qualifications required for them) will only make your talent level worse.
The leading critic of Putnam's research from the left is
I happened to come across your blog comments on the Erica Goode article about Putnam's work. I realize this may be thin skinned on my part, but am writing to explain my views. The NYT article did not make any effort to communicate why I felt Putnam's claim that integrated neighborhoods can produce lowered trust/social capital (as expressed in the article they sent me) was questionable. And this is despite numerous complaints when I found out what was going to appear in the article-I did not see why anyone would care that I disagreed without knowing why. In any event, here is my argument:
The difficulty in moving from a correlation between individual attitudes and neighborhood ethnic diversity to a causal statement that neighborhood diversity affects individual attitudes is that one needs to appropriately control for the possibility that individuals located in different neighborhoods may systematically differ with respect to various characteristics that affect attitudes. Putnam fails to control for these potential differences in anything approaching a statistically adequate way. (His empirical analysis uses some control variables in running "trust" regressions, but these are not adequate for addressing this problem. In economics, the work of James Heckman pioneered the recognition that unobservable sources of self-selection are a first order issue in empirical work of this type. A classic example is the analysis of the effects of job training programs.)
Further, there is a logical basis for expecting this problem to arise. Neighborhood membership is the outcome of an individual's preferences, beliefs (about his future prospects, the neighborhood's development, etc.) and the constraints he faces (prices, borrowing capacity, etc.) Suppose that I observe two caucasian heads of household, with identical incomes such that one lives in an ethnically diverse neighborhood whereas the other lives in an ethnically homogeneous neighborhood. The former gives survey answers indicating less trust of others than the latter. Should I conclude that diversity has caused the lower trust that has been measured for the former, or is it reasonable to interpret the difference as reflecting that the person living in the diverse neighborhood is simply not comparable to the other person because they differ with respect to some other factors than those I have listed (i.e. are observable to the data analyst)?
In my opinion, it is easy to think of reasons why the latter is the case. For example, trust may be associated with contentment. If both individuals prefer to live in the homogeneous neighborhood, but the former has not been able to move to it for some reason (inability to borrow, competing claims on income, simply has yet to move!), it would not be a surprise that he is less trusting. I could imagine a similar scenario if one considers the role of beliefs about the future in affecting choices. Presumably, for example, I will spend more on a house if I am optimistic about my economic prospects than otherwise. The point is that if two individuals are comparable with respect to observable characteristics yet make different choices, one has to take a stand on why the choices differ. Putnam's analysis assumes the sources of the differences are unrelated to attitudes on trust and I do not find this assumption plausible.
Is this an unfair standard of evidence for Putnam's claim? I think it is a reasonable standard because (forgive my repeating myself) an individual's membership in a neighborbood is a choice and hence a function of the the collection of preferences, constraints, and beliefs that characterize him. In his article, Putnam dismisses this type of concern on the basis that one would have to believe that curmudgeons deliberately choose to live in more diverse communities. This dismissal does not address the constraint and belief aspects of neighborhood choice. His answer also confuses causality and correlation, the correlation of curmudgeonly attitudes with a taste for integration can occur for many reasons. The issue is not whether curmudgeons choose diverse neighborhoods because they are curmudgeons, but whether the set of preferences that produce low trust are correlated with diverse neighborhood choices. Putnam's argument presupposes some neat division in preferences that I think is psychologically implausible. As it happens, for me personally, I put high value on a diverse community but have little interest in personal relationships with my neighbors.
Let me jump in here with an example that I think illustrates Durlauf's point. My late mother moved from St. Paul, Minnesota to Los Angeles (which in 2000 came out as very distrustful in Putnam's study) during WWII, while her sister still lives in St. Paul. (Midwestern communities tend to rank higher in trust).
Perhaps there was a selection effect going on: maybe my aunt was happier with their neighbors in St. Paul than my mother was, or my aunt felt more connected to their neighbors than my my mother did, so that contributed to why one sister stayed "back East" and one left for LA. Aggregate that over millions of examples, and you might come up with a reason why white Angelenos are less neighborly than their relatives in the Midwest.
I also get the impression that LA was physically designed to be less neighborly than many older cities, even before it became all that ethnically diverse. For instance, most homes in LA have tall, solid fences around their backyards, while in some other built-up areas of the country, it's common to have no more than a low chain-link fence separating your backyard from your neighbor's. As a native Angeleno, this kind of lifestyle I saw in Chicago struck me as lacking in privacy, but there is a tradeoff between privacy and trust.
One reason for the big backyard fences in LA is to keep neighborhood kids from sneaking into your swimming pool and drowning. Most LA homes don't actually have pools, but the ones that do need to put up fences, and everybody else kind of wants to look like they might have a pool behind their fence.
Moreover, celebrities set the styles in LA, and big movie stars are, by necessity, very unneighborly. Streets in the Hollywood Hills typically are too narrow and winding to have sidewalks, so it's practically impossible to stroll down the street to visit your neighbors without risking being run over. You have to get in your car and drive, so why bother hanging out with your physical neighbors when you can just drive a little farther and visit somebody you already know? Stars like this kind of lifestyle because it keeps doofus fans from camping out their (nonexistent) sidewalks.
A lot of these celebrity styles infected the non-celebrity neighborhoods down in the flat lands, such as the annoying fashion in the plebeian San Fernando Valley for not having sidewalks.
Also, the Spanish-style of home architecture that has periodically been in fashion here in Southern California is unneighborly -- there's a rather blank facade on the street with an internal courtyard or some other private feature. Mexico is of course a classic a low-trust society.
All this raises chicken or egg questions about whether the physical layout of LA contributes to the unneighborliness of Angelenos or whether Angelenos chose the physical layout because of their orneriness, but in either case, it can raise the correlation between distrust and diversity when comparing LA to Minnesota.
Does this explain much about Putnam's results -- I don't know. It might for LA, but LA is far from the whole study of 40 communities.
Of course, this is all an argument as to why the evidence does not support the claim, not that the claim is wrong. I have only read one article on Putnam's new work (sent to me by the NYT as background for their interview) providing an overview of what he argues in the book, not the book manuscript itself. And to be clear, the specific claim that I understand to be original to Putnam in the new new work is that the presence of different ethnic groups makes a person less trusting of his own group as well as others. This is what I find hard to believe. The claim does not correspond to any social psychology studies of which I am aware (not that I am well read here), nor does it correspond to any social science theory which I find plausible, nor is the empirical evidence in Putnam's paper persuasive, as I have tried to argue. In contrast, there is no end of evidence on the ubiquitous potential for intergroup conflicts e.g. the Robbers' Cave experiment. Hence my presumption that Putnam's claim is incorrect and that the burden of proof is on the person making the claim.
Also, I don't think that that there is any reason why folks on the left should be put off if Putnam's claim about the effect of integration on attitudes is true. One reason is that the defense for many race-related policies (I am thinking of policies whose advocates regard them as promoting integration, and I would include affirmative action in this category) is based on ethical considerations for which the validity of Putnam's claim has little import as these considerations (if valid) trump side effects of the type Putnam suggests can occur.
In fact, one could read Putnam's findings as justifying government interventions of various types. If integrated communities induce a loss of social capital, this is an example of an externality. To push this further and in an empirical direction, ethnic conflicts in the
One interesting thing I learned from Moskos and Butler's book is how the Army uses IQ testing to nearly equalize the average IQs of white and black recruits -- which, of course, makes for social more equality among soldiers than is possible among citizens in general, where the ethnic IQ gaps constantly get in the way. The most obvious implication for civilian policy is to not let in so many low IQ illegal immigrants.
A related point: Putnam's writings tend to be utopian when it comes to public policy. And as it happens, I argued to the NYT writer that the suggestion that ethnic conflict can be resolved by benign government policies is dubious without careful attention to context and recognition of how little we really know about the determinants of race relations, trust attitudes, and the like. (Much of my own current research is on the question of how to choose policies when the policymaker knows relatively little about the determinants of the phenomenon that one wishes to affect.)
Anyways, enough venting. I hope this note finds you and your family well.
The most obvious response to uncertainty in policymaking would be ... prudence. For example, we don't know what the results of the current massive unskilled immigration will be, so it would seem reasonable to cut back on it: the potential upside is limited and potential downside is much larger, so why do it?
Instead, we constantly hear things like, "Well, yes, I suppose now that you mention it, that admitting millions of illegal aliens puts big stresses on education, but ... all we have to do is fix the public schools!"
Well, swell, except that we don't know how to fix the public schools, and even if we did, we aren't likely to do it.
Or fix our crumbling infrastructure, or fix our health insurance crisis, or fix our excessive carbon emissions, or fix a whole bunch of things that we aren't likely to fix. Cutting back on immigration won't be all that easy, but it's a lot more manageable many other problems that immigration puts additional strains upon.