Shikha Dalmia just published a piece in The Week about Robert Putnam, professor of sociology at Harvard, which is a total smear job. It’s headlined “The alt-right’s favorite academic” and the opening lines set the tone:
“Diversity is overrated.”
That argument against immigration — once confined to the alt-right gutter— has climbed its way into respectable right-wing circles in the Trump era. The idea is apparently that people have a natural desire to be around their own, so there is nothing wrong with limiting “mass” immigration, especially from non-European countries that are too dissimilar from America.
So anyone who thinks “diversity” is overrated and opposes mass immigration belongs to the “alt-right”, but unfortunately this kind of view is gaining respectability, which is apparently making Dalmia very sad.
She goes on:
And who does the right invoke when making its case? Not the Nazi philosopher Carl Schmitt, who famously argued that maintaining healthy polities requires treating cultural strangers like enemies. No, they are increasingly dusting off the work of liberal Harvard political scientist Robert Putnam, whose research purports to show the pitfalls of diversity.
The trouble is that Putnam oversold his own research — and conservatives are overselling Putnam.
Now, Dalmia obviously hasn’t read Schmitt or didn’t understand him, but this is a topic for another post.
In this post, I want to focus on what she says about Putnam and how restrictionists have used his work, which is the topic of her article. Note that I am not going to discuss the merits of Putnam’s research per se, which is a complicated issue that would require a far more detailed post, but only what Dalmia says about it. (I also don’t think the case against multiculturalism crucially hinges on whether Putnam is right.) Thus, in this post, I will not attempt to show that Putnam didn’t oversell his research or that conservatives aren’t overselling Putnam, but only that Dalmia’s arguments to the contrary are bad. In fact, they are not just bad, they in effect amount to slander against both Putnam and conservatives who use his research to argue that diversity is overrated. I don’t know if Dalmia wrote that article out of dishonesty or stupidity, but since those aren’t mutually exclusive, I suspect it’s a bit of both. It’s still worth discussing because it’s a good illustration of how pro-immigration advocates use smear tactics to stifle the debate about immigration and multiculturalism.
First, I should briefly talk about Putnam’s research on diversity, which Dalmia summarizes as follows:
Putnam, whose 1990 landmark Bowling Alone bemoaning the growing atomization of Americans became a household name, published a paper 10 years ago showing that ethnically diverse communities in America suffered a loss of what he called “social capital” or solidarity. He studied 30,000 Americans across 41 communities — ranging from declining inner cities like Detroit, rural areas like South Dakota, and bustling metropolises like San Francisco. He found that regardless of income level or crime rate, the more diverse the community, the less it trusted not just other ethnic groups but also, remarkably, its own. People don’t riot in the streets, he found, they vacate them, retreating, turtle like, into their homes to watch TV rather than participate in community activities or neighborhood projects.
This is actually a pretty good summary and the only part of her piece that isn’t completely worthless. What is noteworthy about Putnam’s research isn’t that he found diversity reduced trust but that it does so within ethnic groups and not just between them.
As we have seen, Dalmia thinks Putnam oversold his research. She thinks so for two reasons, the first of which she describes in this passage:
It is not easy to reduce complex cultural phenomena to measurable metrics. And even though Putnam’s study is among the more thorough of its kind, his way of measuring trust — basically by asking people to rate on a three-point scale whether they would “say that most people can be trusted” — is arguably quite crude.
The article she links to in this passage is a discussion of this paper by Maria Abascal and Delia Baldassarri. I actually think it’s a very interesting paper, which identifies several problems with Putnam’s analysis, but they have nothing to do with the crudeness of the way in which he measured trust. It’s just that, in this paper, Putnam’s conclusion was only replicated when the dependent variable was trust between neighbors, but not with other measures of trust, which however were no less crude than the measure used for trust between neighbors.
In particular, Abascal and Baldassarri argue that the association between heterogeneity and trust found in Putnam’s analysis is not causal, but is a compositional effect resulting from the fact that non-whites report lower trust and are overrepresented in heterogeneous neighborhoods. In other words, ethnically diverse neighborhoods also tend to be less white and non-whites report lower trust, which according to Abascal and Baldassarri explains why diversity appears to lower trust in a community. I will not discuss the merits of this claim here, because even if Abascal and Baldassarri are correct, it will hardly be comforting for people who are trying to make the case for multiculturalism. Indeed, it would amount to saying that Putnam’s research gives us no reason to doubt the benefits of making society more diverse by importing more non-whites, since it’s not diversity per se that lowers trust but rather increasing the number of non-whites… So why didn’t Dalmia mention this small detail? It could be that she didn’t read Abascal and Baldassarri’s paper or that she did but failed to understand it. However, it could also be that she read it and understood this point, but didn’t mention it because it would have spoiled her pro-immigration party. Again, since she is clearly both stupid and dishonest, it’s hard to say.
As I said above, there is another reason why Dalmia thinks Putnam oversold his research:
Furthermore, George Mason University’s Bryan Caplan notes, Putnam conveniently forgot to highlight that part of his research that showed that many other factors, particularly homeownership, correlate far more strongly with social trust than homogeneity. So why did Putnam bury them and highlight a less important factor instead? Essentially because it’s more in line with his thesis in Bowling Alone. It’s a classic case of “confirmation bias.”
Caplan is very intelligent, but when it comes to immigration, he regularly says not so intelligent things and the post Dalmia cites in this passage is a case in point.
In this post, Caplan points out that, in Putnam’s study, other variables beside homogeneity, such as home-ownership, have a stronger association with trust. This is true, but it doesn’t justify Dalmia’s ridiculous claim that Putnam “buried” this fact by highlighting the relationship between homogeneity and trust instead. Since the relationship between trust and most other variables in Putnam’s model was already well-known or at least not particularly surprising, whereas the association between trust and homogeneity wasn’t, it made perfect sense for him to highlight the latter. Perhaps more importantly, Caplan says that since home-ownership is more strongly associated with trust than homogeneity, you would increase trust far more by encouraging home-ownership than by reducing diversity. But this absolutely doesn’t follow and there is no way Caplan doesn’t know that, so it’s hard not to conclude he is counting on the fact that most of his readers, such as Dalmia, don’t know enough about statistics and causal inference to realize it.
There are several reasons why this inference is fallacious. First, just because people who currently own their home report more trust than people who don’t even when you control for a variety of socio-economic variables, it doesn’t follow that increasing the rate of home-ownership will also increase self-reported trust. For instance, it could be that having more trust in your neighbors makes it more likely you will buy a home, in which case the causal relation would actually go the other way. Moreover, from the fact that, when you control for a variety of socio-economic variables at both the individual and neighborhood level, you find a relatively weak association between diversity and trust, it doesn’t follow that increasing homogeneity would have a small effect on trust. Indeed, if diversity has a negative causal effect on trust, the most obvious causal proxies are things like poverty, crime, etc. But these are precisely the variables that Putnam included as covariates, so if diversity reduce trust by increasing poverty, crime, etc., the coefficient of the homogeneity variable in the regression could be small even though homogeneity has a large causal effect.
Now, I don’t know if diversity per se increases poverty, crime, etc., but increasing the number of poor, uneducated immigrants from certain regions definitely does. Hence, even if Putnam is wrong and diversity per se does not lower trust, this kind of immigration does and therefore it will again bring little comfort to people like Dalmia, who claim it does not. (I know people often claim studies have shown that immigration doesn’t increase crime, but they don’t actually show that and the irony is that, if people believe they do, it’s in large part because they commit exactly the fallacy I have just criticized. I plan to discuss the effect of immigration on crime in detail when I have more time, so I won’t say more on that issue for the moment, but if you don’t understand what I’m saying now you will then.) In any case, even if we put aside this point, it should be clear that Caplan’s post rests on extremely bad reasoning. Of course, you wouldn’t know that by reading Dalmia’s piece, probably because she has no idea herself.
As we have seen, she also suggests that Putnam highlighted the effect of diversity on trust because she claims it’s more in line with the view he defends in Bowling Alone, a book in which he argues that Americans are increasingly disconnected from one another. As I have noted above, there is a much better reason why it made sense for Putnam to highlight the relationship between diversity and trust, namely that unlike the relationship between e. g. crime and trust it was unexpected. But Dalmia’s explanation also suggests that Putnam was enthusiastic about this finding, which couldn’t be further from the truth. Indeed, while Putnam collected the data and noticed the relationship between diversity and trust in 2001, he did not publish the paper until 2007, precisely because the results were not what he wanted. Here is what the Financial Times said about this at the time:
[Putnam’s finding that diversity lowers trust] is a contentious finding in the current climate of concern about the benefits of immigration. Professor Putnam told the Financial Times he had delayed publishing his research until he could develop proposals to compensate for the negative effects of diversity, saying it “would have been irresponsible to publish without that”.
In fact, if Putnam can be criticized for something, it’s for letting politics interfere with science in such a way. To be fair, he published his data and noted the relationship on a website in 2001, but as we shall see shortly the delay in publication wasn’t the only way in which he let politics interfere with scholarship.
As we have seen, Dalmia doesn’t just claim that Putnam oversold his research, she also says that critics of multiculturalism do. Here is the passage where she explains why:
Furthermore, as Putnam forthrightly acknowledges — but his right-wing appropriators ignore — the loss of trust due to increasing diversity is a short-term phenomenon. Over the long run, people reconstitute new identities and bonds based on other shared characteristics. Yesterday’s “them” become tomorrow’s “us.” For example, Putnam notes, in the 1920s, Americans were acutely conscious of divisions among European sub-groups — the Irish, Italians, Germans, Eastern Europeans — and in the 1950s of various Protestant denominations — Methodists, Lutherans, Baptists. None of these distinctions matter anymore.
If you haven’t read Putnam’s study, but only this, you would probably conclude that Putnam also found evidence that, in the long-run, the effect of diversity on trust disappears. Except he did not.
Indeed, in this passage, Dalmia is referring to the last section of Putnam’s paper. But this section is not based on any serious data analysis and was aptly described by Ilana Mercer as a “stern pep talk”. (It even closes with an inspirational quote by then-Senator Obama.) Putnam just makes the point that, as a result of immigration during the 19th and early 20th century, there was a lot of diversity in the US that also generated friction, which didn’t matter in the long-run because the immigrants and their descendants blended in the American “melting-pot”. This is a common argument, but it’s also totally underwhelming. There are many reasons why, but I will only mention two. First, we don’t actually know that, to the extent that immigrants reduced trust back then, it has recovered since then. For all we know, trust in the US may have been permanently reduced by 19th and early 20th century immigration. At least, Putnam’s paper doesn’t present any evidence that it was not, despite what someone who has only read Dalmia’s piece may think.
Moreover, even if Putnam were right that trust in the US recovered as immigrants and their descendants integrated, this would not say much about the long-term effect of immigration today. Indeed, both the immigrants and the way in which receiving countries are dealing with them have changed a lot, and many other things have also changed, so we can’t really infer much from the past about what’s going to happen with immigration today. In particular, immigrants to the US in the 19th century and the early 20th century were mostly European, which is no longer the case. It’s pretty obvious that it could affect how easily people who immigrate to the US today integrate. It’s also the case that people in the US no longer deal with immigrants in the same way. For instance, xenophobia is probably far less intense, which reduces the incentives to assimilate. If you regularly get beat up and called a “wop” because you speak Italian, that’s obviously a motive to assimilate.
Moreover, the economic, technical, etc. circumstances are completely different. For instance, thanks to the progress of communications, it’s much easier for immigrants to stay in touch with their country of origin today than it was a century ago, which could also get in the way of integration. Indeed, there is evidence that, whatever the reason, immigrants are not integrating as fast as they used to. Even if Putnam is right that trust recovered after 19th and early 20th century immigration increased diversity in the US, this could mean that it won’t be the case with the increase of diversity that results from immigration today. It’s also worth noting that, after 1924, immigration essentially came to a halt in the US, which may have something to do with the fact that integration was successful, to the extent that it was successful and didn’t affect trust in the long-run. Of course, I’m not claiming that any of this proves that, to the extent that diversity lowers trust, it does so permanently and not just temporarily. But it shows how silly the argument from the past is. If you want to argue that, in view of what happened in the past, we can be optimistic about the long-term effects of immigration today on trust, you have a lot more work to do.
At this point in the article, having distorted his work and motivations, Dalmia stops talking about Putnam and explains why immigration is indispensable:
Right-wing diversity critics argue that race is different. Unlike religion and nationality, it is an immutable fact of life and given the inherently tribal nature of humans, ignoring it to build a racially eclectic society means inviting conflict (which Putnam’s study did not find, incidentally). In other words, the defining project of American liberalism to transcend the tribal ties of “blood and soil” through a commitment to a universalistic creed of liberty and equality is a farce in the eyes of these self-styled American patriots.
But a society dedicated to slicing and dicing people to ensure demographic homogeneity will wind up far more fractious. It is actually quite hilarious to witness the bitter disputes among alt-righters about who exactly is genetically pure enough to qualify as a member of their tribe. While in liberal societies big differences become irrelevant over time, in an ethnically homogeneous society small differences inevitably grow to oversized importance. Indeed, hierarchies over minor impurities or cultural differences will become magnified, breeding not harmony but strife. In 17th century Europe, people were almost entirely white. But it wasn’t exactly a picture of comity. They simply replaced the internecine warfare among various European tribes in previous centuries with sectarian warfare among Protestants and Catholics. It was liberalism, with its commitment to protecting the freedom of all without regard to caste, class, or ethnic and religious differences, that finally brought a halt to Europe’s centuries-long state of constant war.
Not only is she attacking a straw man, but if the example she uses to make her point says anything about the merits of multiculturalism today, it’s hard to see how it could be in favor of it.
First, as she has done throughout the article, she is suggesting that anyone who is skeptical about the benefits of diversity is really a white supremacist obsessed with racial purity. But most critics of multiculturalism don’t really care about race, except insofar as it’s correlated with culture. So it’s patently false that right-wing critics of diversity focus on racial homogeneity and think racial diversity, as opposed to cultural/religious diversity, is uniquely bad. It’s just that, unlike libertarians of the Reason/Cato variety, such as Dalmia, most conservatives aren’t stupid enough to think you can import millions of people from countries that have little to no liberal tradition and expect they will magically assimilate with no adverse effects on Western societies in the long-run. Perhaps it’s because, unlike that kind of libertarian, they are not entirely devoid of common sense. Some of them also know about the evidence that immigrants affect the economic/political institutions of a country even in the long-run. Libertarians tend to disregard that evidence because they have a totally unreasonable belief in the magical powers of ideas to transform people. Dalmia’s attempt to reduce any critics of immigration to white supremacists is also typical of the way in which pro-immigration advocates conduct the debate about immigration. They have to smear because they know that, if there were a honest debate about it, they would get crushed.
Dalmia’s sophomoric understanding of 17th century European history certainly doesn’t bolster her case. Sure, liberalism arguably contributed to the end of sectarian warfare in Europe, but it mostly stopped because Europeans were tired of killing each other. Furthermore, many other factors, such as the advent of the Westphalian order (which enshrined the principle of the sovereignty of nation-states in international law), also played a role. In any case, 17th century European history doesn’t show that, as long as we remain committed to liberal principles, multiculturalism won’t create serious problems in the future. In fact, if you’re trying to make the case for multiculturalism, you really have to be a special kind of stupid to bring up the history of religious warfare in Europe. I understand that Dalmia only brought that up in response to the view that racial diversity, as opposed to cultural/religious diversity, is uniquely bad. But since almost no critic of multiculturalism ever held that view, it’s not particularly interesting. On the other hand, anyone who has not totally lost touch with reality understands that, insofar as 17th century European history says anything about the merits of multiculturalism today (which may not be very far given how completely different the context is), it doesn’t exactly increase the probability that it will turn out great…
Dalmia goes on by arguing that diversity is essential to innovation:
The pursuit of ethnic and cultural homogeneity won’t make America less conflict prone, but it will make it collectively dumber. Putnam’s Harvard colleague, anthropologist Joe Henrich, notes that cultural evolution works much like biological evolution. It needs an environment rich in variation and complexity where different cultures can challenge, compete, and combine with each other to generate new ideas while jettisoning old habits and counterproductive traditions. Indeed, small homogeneous societies inhabited by ethnic clones become less dynamic and even regress. America led the world in cutting-edge innovations in the 20th century, especially the IT revolution, precisely because its relatively free immigration and trade policies allowed for a much more rapid and free flow of information among diverse groups.
This is another talking point which not only doesn’t prove anything and flies in the face of the facts.
Between 1924 and 1965, there was virtually no immigration in the US, yet it didn’t prevent it from putting a man on the moon in 1969. Perhaps it would have done so even sooner if immigration had continued after 1924, but this is hardly obvious and, in any case, this fact should make Dalmia’s assertion that immigration is responsible for America’s leading role in the IT revolution somewhat dubious. After all, the transistor was invented in AT&T’s Bell Labs in 1947, when there was no immigration, by John Bardeen, Walter Brattain and William Shockley, all of whom were natural-born US citizens. Of course, this doesn’t show anything, but neither does making some vague gestures at the role of diversity in innovation. Japan is probably the most ethnically/culturally homogenous nation in the world, yet it still leads the world in terms of innovation. Now, few people would deny that, for example, attracting people from India who did their PhD in computer science at Harvard helps the American IT industry. But that’s not to say that importing millions of uneducated Mexicans has the same effect, although it may allow Dalmia to pay less for strawberries than she otherwise would. If she wants to make that claim, she can’t just assert it, she has to present evidence in favor of it. But I wouldn’t hold my breath if I were you.
Moreover, while attracting highly skilled individual from all over the world is arguably good for the US economy, one can doubt that it has anything to do with cultural diversity. Dalmia suggests that ethnic/cultural diversity translates into diversity of ideas. Pro-immigration advocates make that claim all the time, but for some reason, they never bother presenting any evidence for it. Perhaps it’s because there isn’t any. Indeed, I don’t know of any evidence that it’s true, but I know of plenty that makes it doubtful. For instance, the Cato Institute, which is not exactly known for its support of restrictionism, recently published a report about the state of free speech in America that suggests non-whites are significantly less liberal than whites. In particular, according to the survey on which this report is based, they are far more likely to support hate speech laws.Between-race variation seems comparable in magnitude to between-party variation.
This result was also clear when the survey asked questions about how free speech operates.The differences in the way people of different races think about free speech are pretty striking, but not really surprising given what the previous graph showed about support for hate speech laws.
Moreover, even when you control for the race targeted by hate speech, the pattern doesn’t change.Note that, for all the talk about the evil of whiteness, whites are the only group that seem more willing to ban hate speech against other groups than against their own. Only in the case of flag burning were blacks more liberal than whites, but even in that case hispanics were more strongly in favor of banning it and, in any case, the differences were much smaller. (63% of hispanics were in favor of a ban, whereas 58% of whites and 50% of blacks were.) Of course, race could be confounded by other variables such as ideology and socio-economic status, but I doubt it completely explains the difference.
The report does not discuss the issue of confounding and the survey probably doesn’t have a large enough sample to settle that question anyway, but it still found that large between-group differences existed even among Democrats.Moreover, even if the difference between whites and other groups were entirely the result of confounding by socio-economic variables, it’s not clear what difference it would make. Unless you can also show that immigrants and their descendants are going to converge in terms of socio-economic status, it’s still the case that immigration will undermine the support for liberal norms such as free speech. Needless to say, the results of this poll did nothing to undermine the confidence of people at Cato that multiculturalism was a great idea.
There is also evidence from California that Asians do not support free speech as much as whites.In the case of Asians, the gap presumably can’t be explained by socio-economic differences, since Asians in California have a much higher median income than whites and, in the US as a whole, they are far more educated. If pro-immigration advocates want to claim that cultural diversity will translate into diversity of ideas, they clearly have some work to do, because prima facie the evidence certainly doesn’t support that claim and even seems to support the opposite claim.
Dalmia concludes her piece by using the example of Japan to warn of the dangers of homogeneity:
Ethno-state purists like to point to Japan as an example of a society that has successfully combined a radical program of ethnic homogeneity with modern standards of living. But Japan is paying a heavy price for its cultural purity and restrictive immigration policies. It’s in the midst of a demographic collapse as its population rate plummets far below replacement levels. And if social trust and isolation is a problem in large diverse societies, it can be an even bigger problem in small homogeneous ones. Because the Japanese don’t have a critical mass of people to find the right mate or friend, they are hunkering down like Putnam’s turtles, unable to form strong bonds. Indeed, Japanese people are slightly less trusting than Americans — and a whole lot less so than “high trust” Australia, a poster child for diversity.
Upon reading that, one can only hope that she will eventually open a statistics textbook, so she can learn that correlation isn’t causation and avoid making a fool of herself by writing that kind of nonsense.
Even if immigration were the kind of magical solution to the problems caused by population ageing most pro-immigration advocates imagine, which it isn’t because, among other things, immigrants also grow old, it wouldn’t mean that we should have more of it. Indeed, even if immigration could totally offset the effects of lower fertility and higher longevity, it might also have adverse effects on countless other things, such as crime, trust, culture, etc. If you want to argue that immigration will make the West better off, as Dalmia claims, you have to address this possibility. She tried to do that with trust, but as we have seen, the result was, shall we say, less than convincing… She made up for it by smearing Putnam and critics of immigration, but this will only fool people who are already convinced. Everyone else understands that the debate about multiculturalism can’t be settled by the kind of lazy arguments Dalmia made in that piece. Right-wing critics of multiculturalism understand that, but so do left-wing critics of multiculturalism (there are plenty of them), and even many people who actually support multiculturalism. Is this because they are white supremacists? No, it’s just because they have a brain. Of course, for people like Dalmia, it can be hard to see the difference, but that’s only because they don’t have one.
EDIT: I have added a graph that shows non-whites remain more supportive of hate speech laws even when you control for the race targeted by hate speech. I also mentioned the results of the survey about flag burning for the sake of completeness.
ANOTHER EDIT: On Twitter, Garett Jones points out that Putnam’s model also includes another proxy for diversity (% US citizens), which presumably biases downward the coefficient of the variable he used for homogeneity.