Oddly enough, I’ve received two unrelated emails attaching articles shooting down hypotheses of the notorious Satoshi Kanazawa: a paper by Kevin Denny in the Journal of Theoretical Biology:
Recently Kanazawa (2005) proposed a generalization of the Trivers–Willard hypothesis which states that parents who possess any heritable trait that increases male reproductive success at a greater rate than female reproductive success will have more male offspring. . . . This note shows that analysing the same data somewhat differently leads to very different conclusions.
and one by Vittorio Girotto and Katya Tentori in Mind & Society:
According to Kanazawa (Psychol Rev 111:512–523, 2004), general intelligence, which he considers as a synonym of abstract thinking, evolved specifically to allow our ancestors to deal with evolutionary novel problems while conferring no advantage in solving evolutionary familiar ones. We present a study whereby the results contradict Kanazawa’s hypothesis by demonstrating that performance on an evolutionary novel problem (an abstract reasoning task) predicts performance on an evolutionary familiar problem (a social reasoning task).
These, on top of other debunkings of this work by Volscho, Freese, and others, makes me think that Kanazawa is actually serving a useful role in the fields of biology and sociology by evoking such interesting rebuttals.
P.S. I probably should stop bringing this stuff up–it’s just that I got those two emails one right after the other. As David Weakliem and I discuss in our paper, Kanazawa’s work is not particularly interesting in itself except as an example of genuine statistical challenges that arise in the estimation of very small effects. Basically, the multiple comparisons problem in action, but with a twist in that Kanazawa has been successful enough at getting his ideas out there that he’s attracted debunkers. Presumably, there’s lots of stuff like this out there in the scientific literature that nobody even notices. In studying these problems, I’d like to think that I’m contributing to the search for better methods of estimating small effects, not simply making fun of the errors of non-statisticians.
It’s also interesting to me that biologists and economists seem to fall for this stuff, while sociologists and psychologists see the flaws right away. Presumably because sociologists and psychologists have lots of experience studying small effects in the context of individual variation.