Tuesday, May 19, 2009

Leaky Jonathan has discovered Nisbett

Jonathan Leake is the science writer for the London "Times" who can always be relied upon to draw the most politically correct conclusions -- no matter how much he has to distort the facts. Some of his global warming reports are classics of deliberate distortion. He has now discovered Nisbett's work on the flexibility of IQ and has taken Nisbett's determined optimism fully on board. Since I have already fisked a lot of Nisbett's assertions (See here and here), I won't say a lot here but a few comments are in order.

The more you know about a subject, the more laughable some of the assertions are. One of the recommendations for making your kid smarter, for instance, is to teach them delay of gratification. Yet my careful psychometric research on that subject (psychometrics is notoriously absent from delay of gratification research) showed that such a thing hardly exists. The popular tasks for assessing a kid's tendency to delay gratification correlate with one-another hardly at all. Delay of gratification is highly situational. It is not a stable trait like IQ is. Yet it is precisely the hope of those who recommend delay of gratification training that delay of gratification IS a stable and generalizable disposition.

I think you should already be getting the impression that the conclusions Leftists such as Leake and Nisbett come to are more driven by hope than facts. Much of the argument put by readers of Nisbett is a straw man argument. They seem blissfully unaware that nobody has ever claimed that IQ is solely determined by genetics. Everybody has always conceded that a good or bad environment can also make a difference. A good environment can help you make the most of what genetics has given you. The usual estimate is about two thirds of the determination of IQ is genetic while about one third is environmental. Nisbett himself concedes that. His only innovation is to make the split more equally. On rather specious grounds he reduces the causation to about 50/50 genetic/environment. So the upshot is that the small improvements in IQ that have been demonstrated by various environmental interventions are no surprise to anybody and upset no assumptions or generalizations.

Leake overstates the significance of the Flynn effect. Flynn found that average IQ rose through the 20th century. Yet from the beginning it was always held that an environment which is both visually and verbally stimulating was ideal for getting the best out of a child's genetic potential for intelligence. So what happened beginning in the second half of the 20th century? TV! And huge exposure of kids to TV -- which is a VERY rich source of verbal and visual stimulation. Much as the Leftist elite hate the thought, TV alone could potentially explain the rise in overall IQ. There is more to it than that but I think one can see from that alone that the Flynn effect disturbs no prior theory or generalization.

And we also see in the Leake article a rather unfortunate mention of the fact that the black/white IQ gap is small during childhood but is large during adulthood. That finding is held to support the view that blacks are made dumber because of their poorer educational experiences. There is however a simple biological explanation that requires no reference to education. It is HIGHLY politically incorrect, however, so I will just give the link here. It starts from the fact that baby chimps are just as smart as human infants.

Perhaps under the influence of Flynn, who seems quite obsessed with it, Leake also brings up the absurd Eyferth study, as proof of equal black/white genetic potential for intelligence. The study looked at mixed race children fathered by black U.S. servicemen in Germany after WWII. As the obvious implications of the study are a bit brutal, I will again simply refer readers to some prior comments of mine here and here

Perhaps I should take this opportunity for a comment on another publication by Nisbett -- one that seems at first crushing but which in fact suggests that Nisbett has not the faintest idea of what a correlation coefficient means. Rushton & Jensen commented on Nisbett's work by drawing attention to the overall correlation between IQ and head size -- which strongly suggests a significant genetic contribution to IQ, as nobody argues that head size can be influenced by better education etc.

Head size is however only one of the factors affecting IQ. There is absolutely no doubt that the causation of IQ is polygenetic and what some of the genes are is now emerging in medical research. A high degree of myelinization in the brain has, for instance, recently been found to be helpful. So it should be no surprise that head size and IQ correlate only moderately -- at around .40, implying only 16% of shared variance between the two variables. Head size is, in other words, only one of many physical influences on IQ. So a correlation as low as .40 allows for a large number of exceptions to the rule. Somebody with a small head but (say) good myelinization can still be quite smart.

So what does Nisbett do in response to the careful survey evidence summarized by Rushton & Jensen? He quotes a whole lot of exceptions to the rule as if that proved something! He answers overall generalizations with particular instances. He puts up an argument by example and such arguments can support anything. They are certainly no basis for generalizations. I suspect that Nisbett needs to do a course in the philosophy of science some time.

No comments: