I've been looking at some of Andrew Gelman's critiques of Kanazawa's studies of gender differentials in birth. Kanazawa argues that violent and tall men are more likely to have sons. Beautiful people are more likely to have daughters. Things like that.
Gelman's critique is on the stats. The idea is that the difference in the gender between beautiful and unbeautiful people's first child is indistinguishable from noise. The effect has to be greater than the normal margin of statistical noise to be significant. If the children of beautiful parents are only slightly more likely to be female, then who cares any way?
It seems to me, you have to look at this on several levels. First, you have to have a story, a theory of why beautiful people would have more daughters. The idea is that beauty is a more important attribute for women, so it would be a more desirable trait to pass on to your female offspring than your male. In my view this is your standard evolutionary psychology bullshit. In other words, there is no actual evidence that this story has any validity, but it conforms to the ideology of certain evolutionary psychologists, which Gelman explain as a "schoolyard" mentality: "the idea that, because of evolution, all people are equivalent to all other people, except that all boys are different from all girls."
Next, you have to have a biological mechanism for translating the advantage of female offspring into the reproductive process. In other words, how do the reproductive systems of the parents know they are supposed to produce female offspring? Beauty is not a single genetic feature, like red hair, but an amalgam of multiple biological and social factors, including access to healthy food, dentistry, or gymnasia. Does the beautiful woman's fallopian tubes know how to select which sperm get to enter? How exactly does this work, anyway? Does the pretty boy dad produce more x chromosome sperm than the tall, violent dad does?
Then, you would have to have the solid data to make a significant result. It seems to me, though, that the weakness of the initial story, and the lack of a clear biological mechanism, is at least as serious as the lack of statistical significance.
Put another way: normally, science should work the other way around. We should observe something strange and then try to explain it. Suppose it was true that pretty people produced way more daughters than average folk, so much so that everyone knew this to be the case. Then a scientist could come along and try to figure out why, both the theory (the story) and the mechanism, the way it actually happens biologically. Ideological science starts the other way: you have a story, and then you try to find empirical confirmation of it after the fact.