Because both the quality of the eggs and the quality of the sperm decrease with age, it is tempting to explain the declining likelihood of having a son among older parents potentially in terms of such quality of gametes (although I’m not aware of any argument that suggests that lower-quality gametes are more likely to produce girls). However, such explanations, even when correct, are proximate, not ultimate. They answer the question of how; they don’t answer the question of why. The lower quality of gametes, if it indeed lowers the probability of producing boys, is the mechanism that evolution employs to make sure that older parents are more likely to have daughters. But such a proximate mechanism does not explain why evolution “wanted” to make sure that older parents are more likely to have daughters, in other words, why it is adaptive for older parents to have daughters, not sons. That requires an ultimate evolutionary explanation.The adaptive explanation is given as a necessity. Yes, proximal and ultimate explanations aren't the same thing, but the point here is that there need not be an evolutionary reason. Being more likely to have girls when above 40 does not need to be adaptive, as assumed by Kanazawa. This very assumption, however, is the very livelihood of working evolutionary psychologists.
Here's Kanazawa's proposition:
Being orphaned young is bad both for boys and girls, but it’s much worse for boys than for girls.Sure, I'll bite. That is a great explanation right there, but in order to work with this hypothesis, it helps (though isn't strictly necessary) to further assume that there is enough variation for selection to act on. For example, it would without a doubt be great for humans if they had evolved flight to better escape predators back in the day, but also without a doubt selection couldn't select for flying humans because no children were ever born with this ability (or even anything that comes close).
In terms of genotype-phenotype maps (the relationship between a given genotype and the phenotype it codes for), it is not easy to reconfigure the genome so that humans are born with rudimentary wings. Some things are more likely to evolve than other, simply because they can occur at all - and wings aren't so likely (though not impossible). Genetic constraints make some things more likely to appear, and for selection to act on - or not. As with the observation that older parents are more likely to have daughters, which may stem from genetic constraints which make it less likely to make sperm with Y-chromosomes, for example. So, even if there is no advantage for older parents to have daughters, it could still be that way without it being selected for. Having more daughters could be a neutral trait.
But that's no fun. It does not make for a newsworthy story, so people in that business are more apt to assume that selection did it. But it really lacks scientific rigor.