Last November, William Saletan wrote a series for Slate on "Liberal Creationism," purporting to show that, according to the best evidence we have, there is good reason to think that observable racial disparities in scores on intelligence tests are irreducible to non-genetic explanations. Or, de-jargoned, that it's time to confront the uncomfortable truth that black people aren't as smart as white people (on average, of course).

Racial Harmony: The path to Saletanian paradise isn't paved with Saletanian contrarianism Saletan was widely panned, at Jewcy among many other venues, for drawing conclusions based on statistical innumeracy, flagrant misunderstanding of the literature on race, genes, and intelligence, and the "research" of J. Philippe Rushton, a discredited racist crank who years ago took to sending out unsolicited mass-mailings of his pamphlets to every member of the American Sociological Association.

Having brushed his shoulders off, Saletan is back on the race-and-intelligence beat today, and this time he's more circumspect. Much more circumspect. He's having an extended, self-serving dialogue with himself, in an effort to ensure his intended audience (William Saletan) that what motivated the original series wasn't racism, but the admirable conviction that the truth isn't any worse off for our discomfort with it; rather, we're worse off for not facing up to uncomfortable truths.

But of course, no persuasive critic of the "Liberal Creationism" pieces thought that Saletan was motivated by racism. The criticism was that knee-jerk contrarianism led him to present as "the truth" a case based on wafer-thin evidence and shoddy reasoning. Rather than confront the methodological lacunae that prevented him from giving a cogent public presentation of the state of the literature on race and intelligence --- some uncomfortable truths, you might say --- Saletan instead digs in, offering a general justification of his project of exploring "how to be an egalitarian in an age of genetic differences," by means of some unintentionally hilarious epistemological musings on truth and semantic musings on 'truth':

In retrospect, I was consumed by the wrong word. The flaw in my approach wasn't truth. It was the. Even if hereditary inequality among racial averages is a truth, it's less true, more unjust, and more pernicious than framing the same difference in nonracial terms. "The truth," as I accepted and framed it, was itself half-formed. It was, in that sense, a half-truth. And it flunked the practical test I had assigned it: To the extent that a social problem is genetic, you can't ultimately solve it by understanding it in racial terms.

Can you feel Saletan's pain yet? All he wanted to do was set right racial injustice, a noble goal if there ever was one, and might have succeeded if a pesky definite article hadn't tripped him up. But here's the thing. (And also, a thing.) As the Saletan of November might have put it, truths aren't any less true for being unjust or pernicious, nor does truth come in degrees. A proposition is true, in which case it's a truth, or it isn't, in which case it's a falsehood. Nor is it very difficult to distinguish the truth from a truth. The former is a complete, actual state of affairs, the latter is a proper part of an actual state of affairs. And the truth won't ever contradict a truth or vice versa, because any statement of the truth states all the truths. If you think you've found an intractable conflict between the two, check your work: something's gone wrong somewhere.

For example, the proposition that the best available evidence points strongly towards a genetic explanation of racial disparities in intelligence tests isn't true, hence is neither a truth nor a part of the truth. In other words, it's false. It's false despite being so counterintuitive in these politically correct days (which goes to show that counterintuition isn't foolproof --- aspiring contrarian journalists, take note!). Moreover, it's so clearly false that a brief conversation with credible expert in the field ought to suffice to convince you that it's false. And the cause of ameliorating racial inequalities, in which everyone should in principle be willing to join with Saletan, isn't served by promoting falsehoods, since a false theory of racial inequality is no more useful in reforming education and social policy, than a false physical theory is useful in building bridges and tunnels.

What's more, taking the time to do adequate background research in the first place relieves you of the effort involved in months of back-pedaling and TMI-laden internal dialogues about the nobility of your intentions --- effort that could be put to better use bringing whites and blacks together at the table of brotherhood.

I figured Julian Sanchez would have an interesting point to make about all this, and sure enough, he does:

A friend once told me about a study—perhaps a commenter can find me a reference?—in which two groups of students were casually informed in advance that previous studies had shown that women tend to perform less well on such tests. As I (fuzzily) recall, this happened to be true, and was true for the control group in the test. But the women in the group who'd been informed of this performed much worse, greatly magnifying the gender gap. Which is to say, if I'm getting the details right, that dwelling on certain kinds of differences—whatever their source—comes with its own costs.

One such study (pilfered from Julian's comments thread), is here. It's a phenomenon very similar to the bandwagon effect in polling, and underscores just how difficult it is to prise apart the various factors determining IQ scores according to their aetiologies. So then, why not, as a number of folks have suggested, including commenters here at Jewcy, just say, the IQ gap could be environmental or could be a combination of genetic and environmental factors, we just don't know, and leave it at that?

First, let's consider the epistemic situation we're in. We already know that there are environmental factors that have a role in determining how individuals do on IQ tests --- no one disputes that. What we have, as far as evidence of race-linked genetic differences goes, is the simple prima facie evidence that there is an IQ gap, and despite a lot of furious (and fraudulent) obfuscation, not a jot of, shall we say, secunda facie evidence. Taking the epistemic position that it could be environmental, it could be environmental and genetic, we just don't know, is not really much more justified than the epistemic position that it could be environmental, it could be (earthly) environmental and also caused by perturbations in tbe orbit of Saturn, we just don't know. In fact, it wouldn't be terribly difficult to come up with prima facie evidence that some utterly contrived, gruesome, and gerrymandered phenomenon is tied to differential performance on IQ tests --- just specify some disjunction of events that occurs only when and where black people take IQ tests, e.g. the Brownian motion of particles not contiguous with but in the immediate spatiotemporal vicinity of black test-takers, observe the correlation between the data sets, and posit a causal relationship. But of course, my contrived example doesn't survive a moment's scrutiny, and likewise, the hereditarians' evidence doesn't survive more than a few moments' scrutiny.

Second, we should be careful to distinguish two claims: (1) Genes have something to do with determining mature cognitive abilities. (2) Race is a reliable predictor of average intelligence. (1) is undeniable (and I've been a little dismayed to see some people assuming that I deny it). No two people are equal in their cognitive abilities, and of course genes are part of the causal history of those inequalities. But what the evidence resoundingly fails to warrant is a jump from (1) to (2) --- there is no secunda facie or deeper evidence that inequalities in cognitive abilities have anything to do with race. When spelled out in explicit terms, the empirical weakness of the hereditarian position can't be concealed, but because (1) and (2) are superficially similar --- because, in short, our commonplace notions of race cross-cut popular understandings of genetics --- a sufficiently clever rhetorician can move from (1) to (2) and make the move appear innocuous. John Holbo has a name and analysis of the general rhetorical strategy:

I call it ‘the two-step of terrific triviality’. Say something that is ambiguous between something so strong it is absurd and so weak that it would be absurd even to mention it. When attacked, hop from foot to foot as necessary, keeping a serious expression on your face. With luck, you will be able to generate the mistaken impression that you haven’t been knocked flat, by rights. As a result, the thing that you said which was absurdly strong will appear to have some obscure grain of truth in it. Even though you have provided no reason to think so.

And that, folks, is basically what we're dealing with. Because we can't, and shouldn't jettison our belief that people in general possess unequal cognitive abilities, and that genomic differences are part of the story, some people are unwilling to let go completely of the idea that the differentials in average IQ scores exist in part in virtue of the ethnic identities of the test-takers, an entirely different claim. No matter how many times one points out that the hereditarians' evidence doesn't support their claims, the argument just will not give up the ghost.

There's one final point that I ought to have made sooner and really ought to make now. Namely: Race is an extraordinarily confused, ill-defined, and scientifically useless concept. For the purposes of my articles I abided by the naming conventions Saletan was employing, but it's really time to set the record straight. The average genomic difference between two individuals attributable to what we would commonly think of as a racial difference is vanishingly small, far, far less, for exmple, than the genomic difference between men and women (who nevertheless have the same average IQ). Moreover, since, as Robert Farley puts it, "up until about 200 years ago, the vast vast vast vast vast vast vast vast majority of people in Africa, Asia, South America, North America, and Europe lived in essentially similar economic conditions," meaning that it's preposterous even in prima facie terms to suggest that homo sapiens broke apart into distinct subspecies due to divergent evolutionary pressures that have existed for just 200 years.

And last of all, Cosma Shalizi makes a point about peer review and its limitations that reinforces my own, and is particularly pertinent to the commenter who populated a now several-pages-long comment thread with links to the collected works of J. Phillipe Rushton:

A journal's peer review is only as good as the peers it uses as reviewers. If everyone, or almost everyone, who referees for some journal is in the grip of the same mistake, then they will not catch it in papers they review, and the journal will propagate it. In fact, since journals usually recruit new referees from their published authors or people recommended by old referees, mistakes and delusions can become endemic and self-confirming in epistemic communities associated with particular journals. To give a concrete example, the community using Physica A is pretty uniformly (and demonstrably) mistaken about how to tell when something is a power-law distribution, so what that journal publishes about power laws is unreliable, and those who derive their training and information from that journal go on to propagate the errors. It would be easy to find even more extreme examples from the physical and mathematical sciences (especially, I must say, among journals published by Elsevier), but it would take too long to explain why they are wrong.  

Now then, unless some glossy mag wants to pay for a feature piece on this stuff, this is the last you'll hear of it from me.

I'm surprised and glad to see that William Saletan has (semi-)recanted his series on race, genes, and IQ. Those who have read my response in these pages, and especially the supplement on J. Philippe Rushton, will understand why it was incumbent upon Saletan to make this apology:

But the thing that has upset me most concerns a co-author of one of the articles I cited...

For the past five years, J. Philippe Rushton has been president of the Pioneer Fund, an organization dedicated to "the scientific study of heredity and human differences." During this time, the fund has awarded at least $70,000 to the New Century Foundation. To get a flavor of what New Century stands for, check out its publications on crime ("Everyone knows that blacks are dangerous") and heresy ("Unless whites shake off the teachings of racial orthodoxy they will cease to be a distinct people"). New Century publishes a magazine called American Renaissance, which preaches segregation. Rushton routinely speaks at its conferences.

This is gracious of Saletan, and clear enough evidence that his motive was a contrarianism that got the better of his sense of moderation, and not racism. Indeed, scanning Saletan's whole series, from even the third entry he appears to have noticed that he bit off more than he could chew, and began carefully disavowing his prior claims.

That said, I think it's a bit rich for Saletan to pretend that his missives were simply descriptive pieces in which he eschewed the chance to take a position, writing "I outlined the evidence primarily to illustrate the limits of the genetic hypothesis." Really? I was under the impression that those who deny hereditarianism are "liberal creationists" whose belief that the Flynn effect and the narrowing of the IQ gap demonstrate the fundamental plasticity of achievement on standardized tests is as groundless and faith-based as the belief of Iraq hawks in the success of the surge.

In other words, nobody forced Saletan to take a position, or to escalate his rhetoric in doing so. The caveats and to-be-sures and on-the-other-hands bracketing his pieces don't make his basic claims mean anything other than what they mean. Still, it's good to see Saletan own up to his mistake, even if belatedly and in a qualified way. (He could have mentioned my role in bringing attention to Rushton, but no hard feelings.)

The same, unfortunately, cannot be said for Andrew Sullivan, who still seems to regard his flacking for The Bell Curve as an exercise in heroic truth-telling. The bottom line is that it can in fact happen that a politically correct belief is also a factually correct belief, and when it does, advancing the opposite argument makes you gullible, not courageous.

What's more, there is a long political tradition to racial ev-psych, and it's one I doubt Sullivan would want to identify with if he really understood it. (Its history helpfully explodes the idea that the hereditarians are just a band of dispassionate scientists, who alone of all scientists in their field, have no axe to grind.) As Jonathan Marks put it in an e-mail to me shortly after my piece originally went up:

[O]ne part of the scientific racist’s arsenal, which came up during the segregation era (notably in Carlteon Putnam’s Race and Reason) is that there is a conspiracy of silence about racial inequalities being led by communists, anthropologists, and Jews. It’s not too far from the surface in a lot of contemporary discourse, either. Except now, the creationists are part of the conspiracy as well.

Whistle-blowers or cracked conspiracy theorists? I report, you decide.

Are differences in performance on standardized tests indicative of differences in intelligence? Are differences in intelligence, in turn, rooted in genetic difference? Are the genetic determinants of intelligence distributed in predictable patterns along racial lines? Finally, are race-based discrepancies in mean intelligence brute biological facts, impervious to social or technological efforts to alleviate them?

These are hardly new questions. It wasn't very long ago when Richard J. Herrnstein and Charles Murray answered all the foregoing questions affirmatively in The Bell Curve, thereby inciting fresh brawls in the public policy world and in academia over political correctness, institutional and societal racism, academic freedom, the limits of social policy, meritocracy, the ethics of biological and anthropological research, the validity of intelligence testing—and on and on.

Perhaps the most remarkable fact about The Bell Curve, in light of the controversy surrounding it, is that, as Stephen Jay Gould put it in his New Yorker review, the book "contains no new arguments and presents no compelling data." Or to be somewhat more charitable, Herrnstein and Murray's work is a straightforward retread of studies by psychometricians, psychologists, and anthropologists stretching back at least to the late nineteenth century, from Charles Spearman's purported discovery of the famous "general factor of intelligence" (on which more in a little while), to Lewis Terman's invention of the Stanford-Binet IQ test and hereditary interpretation of its results, to Carleton Coon's lifelong pursuit of a scientifically valid hierarchical taxonomy of races. Neither Herrnstein and Murray's arguments nor, for that matter, the arguments of The Bell Curve's critics, are anything new under the sun.

Which brings us to William Saletan's "Created Equal" series, which has been running in Slate since Sunday. Taking James Watson's ignominious career-suicide as his point of departure, Saletan argues that the longitudinal data concerning racial disparities in IQ are recalcitrant in the face of well-meaning beliefs in fundamental genetic equality —a view Saletan caricatures as "liberal creationism"—and that if we are unwilling to acknowledge this stark reality, we can't begin to address racial gaps in academic or professional success. In other words, liberal fantasies of a purely environmental explanation of differential achievement stand squarely in the way of the actual amelioration of such differences.

There are several tangled logical threads upon which Saletan's argument precariously hangs, including a barely questioned assumption that IQ is a reasonable approximation of intelligence, a related and similarly barely questioned assumption that there is a general intelligence factor (g) that standardized tests can measure, several rather hopelessly contorted interpretations of contradictory studies, and a seeming utter ignorance of the significance of intra-group changes in IQ scores over the last hundred years. I'll come back to all these points and more momentarily; but I feel it's my journalistic duty not to bury the lede, which so far none of Saletan's blogospheric critics seem to have noticed. Namely, the principal study on which Saletan rests his case is a two-year old paper by J. Phillippe Rushton and Arthur Jensen.

To put this as fairly as it can be put: Rushton and Jensen are anything but a new wave of scholars come to shed light on a heretofore intractable problem, as Saletan presents them. On the contrary, they have spent nearly a century combined harping on the same theme again and again, in paper after paper, and that theme is black racial inferiority. (Care for a taste of just how old-fashioned they are? They group human beings into a tripartite classificatory scheme of "Caucasoids," "Mongoloids," and "Negroids." It's in the 2005 paper, and it's roughly as credible as the Shem/Ham/Japheth theory of race.)

Jensen, as Melvin Konner noted in The Tangled Wing, has been tossing up one-sided hypotheses about the relationship between race and IQ since the 1960s that have consistently been swatted away by Gould, Howard Gardner, and others, but the fact that Jensen's findings have since been debunked did not prevent them from seeping into The Bell Curve. (Thus Saletan's articles come full circle, to say nothing of the geometric configuration of the chain of research supporting the hereditarian position.)

As for Jensen's co-author, in the excellent book Human Biodiversity: Genes, Race, and History, Jonathan Marks described the nature and integrity of Rushton's scholarly puruits as follows:

J. Phillipe Rushton calculated, on the basis of crude skull measurements of army inductees, that the average brain size of Asian males was 1403, of whites 1361, and of blacks 1346 cubic centimeters...

Have we thus discovered the biological basis for the differences in intelligence that previous generations have always assumed were there?...[T]he scientific issues and assumptions are as false as they have always been. First, we must admire the apparent cranial expansion of Asians over the last half-century, when researchers consistently reported their having smaller brains than whites. Obviously this implies the possibility of a comparable expansion in blacks. More likely, it implies the possibility of scientists finding just what they expect when the social and political stakes are high.

Meanwhile, in his review of Rushton's Race, Evolution, and Behavior in the Canadian Journal of Sociology Online, neuroscientist Douglas Wahlstein maintains an air of scholarly understatement, writing "I believe that great harm could be done to both the social and natural sciences if the standards for evidence and proof advocated in this book were to gain wider acceptance." Wahlstein further quotes Rushton replying to his critics to the effect that they "have failed to show an opposite predicted ordering in brain size, intelligence, sexual restraint.'' Apropos of that last point, Marks is delicate enough not to mention Rushton's companion studies of average penis size by race—and there, as the saying goes, you have it.

The foundation of Saletan's hereditarian argument is the pseudo-science of a man obsessed not only with his preconceptions of the inferior intelligence of blacks, but their lack of "sexual restraint" as well. Pity Saletan could not contact D.W. Griffith for comment.

In any case, Rushton's enquiries into differential skull and brain-sizes among the races, and the problems attendant upon them, are particularly relevant for present purposes, because Saletan himself places some significant weight on the implications of brain size, writing that magnetic imaging studies show " at least a 40 percent correlation of brain size with IQ. One analysis calculates that brain size could easily account for five points of the black-white IQ gap." To outline just one of Saletan's numerous misunderstandings of what correlations do, and do not, imply, let's turn back to Marks:

[B]asic scientific protocol requires that all relevant variables be controlled before drawing conclusions about the cause of an observed difference between samples. But in this case we do not even know what those variables are, or what the appropriate statistical corrections (for example, for body size) may be. Brain size correlates, for example, with age and with nutritional state in early life...

[T[hough there was agreement that women have smaller average brains than men (assuming their brains don't grow in subsequent studies!), they apparently do not have lower average IQs. This obviously would undermine the strict determination of intelligence by brain size, which should already be common sense...

By now, this approach to the determination of the average intellectual abilities of group members has degenerated into sophistry. The populations within each "race" vary widely in measured cranial capacity, with the four largest sets of skulls deriving from the aboriginal males of Hawaii, Tierra del Fuego, France, and South Africa, respectively.

The upshot is twofold: First, there is no good reason to suppose that differentials in brain size are not attributable to dietary and other non-genetic variables, and second, whatever correlation between brain size and IQ exists must be weak enough to allow for women to have the same average IQ as men despite having smaller brains. Furthermore, Saletan's reliance on differential brain size stands in direct contradiction to his take on another of his data points, a study of children born in post-war Germany showing that children of white GIs had the same average IQ as children of black GIs.

Saletan explains this finding away by noting that military recruits are already pre-screened for intelligence. The trouble is that the very same brain-size data Saletan uses to bolster his hereditary interpretation of the IQ gap is originally based on measurements of soldiers, hence the fact that the offspring of black and white soldiers have the same average IQ implies that brain size differentials are either not genetically-determined, or else can be positively ruled out as explaining the black-white IQ gap.

Similarly, Saletan concedes that the black-white IQ gap is narrower than it used to be, points out the hereditarian explanation that "the gap closed fractionally in the middle decades of the 20^th century, but...scores in the last two to three decades show no improvement," and dismisses the import of this narrowing of the gap in extraordinarily blithe terms:

On the one hand, the IQ surge is hugely exciting. If it closes the gap to zero, it moots all the putative evidence of genetic barriers to equality. On the other hand, the case for it is as fragile as the case for the Iraq surge. You hope it pans out, but you can't see why it would, given that none of the complicating factors implied by previous data has been adequately explained or taken into account.

Here is where Saletan's argument really comes off the rails, even leaving aside the hopelessness of his analogy. The question before us is whether IQ scores are (strongly or weakly) genetically determined in patterns that roughly correspond to racial designations. The evidence for the affirmative is that there has always been and remains a gap between blacks and whites. (Recall that Asians currently score better than whites, but that, as Marks notes, early intelligence tests found higher scores among whites than Asians. One might conclude, as a hereditarian would no doubt be tempted to, that those early generations of intelligence tests were tests of social fit rather than intelligence. On what grounds, in that case, would one base one's confidence that current supposed intelligence tests do not in fact test various other qualities?) Never mind the fact that the black-white gap used to be greater than it is now; suppose instead, for the sake of argument, that the difference in average IQ scores of blacks and whites had remained the same throughout the entire history of IQ tests. Would that finding, at least, lend credence to the hereditarian argument?

It would not, absent relevant information about absolute changes in the average scores of blacks and whites in the same period. Only if blacks and whites not only maintained a nearly constant difference in scores relative to each other, but a nearly constant average score as well, would data about black and white IQ scores begin to support the case that IQ is in part racially determined. Alternatively, if, for example, the average IQ of whites improved from 50 to 100, while the IQ of blacks improved from 36 to 86, observers would note precisely the same absolute difference in average scores between the two groups at each end point, but a substantially different percent difference (half, in fact), not to mention intragroup changes in IQ that are completely inconsistent with the suggestion that racial genetics determine IQ.

And on this score, the data are unequivocal. According to Brad Delong's summary of recent studies:

The average IQ score of America's "white" population today is 100. According to Ulric Neisser, America's "white" children in 1932 had an average today's-test IQ score of 80. Dutch army conscripts in 1952 scored 30 IQ points lower than conscripts in 1982... [T]he African-American IQ test average rose by 6 points relative to the "white" average between 1972 and 2002. According to Brierley (1970), in the 1960s African-Americans from Ohio had an average IQ score greater than that of whites from Arkansas by 10 points.

Continue reading...