National Review’s Kevin Williamson Comes Out Against Daughters, Misunderstands Science

Kevin D. Williamson, Deputy Managing Editor of the National Review, has taken a bold anti-daughters stance in a recent post arguing that Mitt Romney’s sons make him a stud, and studs should be president. SCIENCE!

Williamson writes:

It is a curious scientific fact (explained in evolutionary biology by the Trivers-Willard hypothesis — Willard, notice) that high-status animals tend to have more male offspring than female offspring, which holds true across many species, from red deer to mink to Homo sap. The offspring of rich families are statistically biased in favor of sons — the children of the general population are 51 percent male and 49 percent female, but the children of the Forbes billionaire list are 60 percent male. Have a gander at that Romney family picture: five sons, zero daughters. Romney has 18 grandchildren, and they exceed a 2:1 ratio of grandsons to granddaughters (13:5). When they go to church at their summer-vacation home, the Romney clan makes up a third of the congregation. He is basically a tribal chieftain.

Professor Obama? Two daughters. May as well give the guy a cardigan. And fallopian tubes.

Fallopian tubes and a cardigan? Ewwww. Why not just replace the Constitution with a Talbots catalog?

Clearly the only ways this could possibly backfire would be if any of the following are true:

  • The National Review is a conservative publication, and meanwhile conservatives in the US are fighting a PR war because they’re perceived by many as espousing virulent, mean-spirited sexism;
  • There is a non-trivial number of men who actually love their daughters and are neither disappointed they exist nor convinced that it means they’re low-status males;
  • There is a non-trivial number of human beings who are unrepentant daughters and/or fallopian tube havers and/or cardigan wearers;
  • There was a Republican president in recent memory who had two daughters;
  • There are social conservatives in other quarters trying to be seen as the ones who care about the value of daughters.

One can only assume that NRO folks ran these scenarios and determined none of them were problems worthy of their attention. (Mmkay. Noted.)  Unfortunately, Williamson also seems to have misunderstood the Trivers-Willard hypothesis.

The Trivers-Willard Hypothesis is an if/then statement. If the variance in male fitness is much higher than the variance in female fitness, then then mothers in good condition should invest in male offspring.

In other words, you have to assume a situation where any male offspring will someday either “cash in big” and produce a whole lot of offspring himself, or else be shut out, but female offspring will probably go on to crank out a predictable number of their own offspring every year.

In such a situation, mothers in good condition can afford to take the risk on male offspring, because they’re playing with the “house money” that nature has given them. Meanwhile, mothers in poor condition should favor female offspring, because in this scenario it’s daughters who are the “sure thing.”

Notice the common theme of “mothers”? That’s because the Trivers-Willard Hypothesis predicts something about (class? hands? anyone?) mothers. (Who often have fallopian tubes, and sometimes even have cardigans.)

The hypothesis does seem to work where it has been studied, for example, in red deer. The important thing here, though, is that it has certain conditions under which it is supposed to work, and if those conditions are not satisfied, there is no reason to even bring it up. Without the “if”, there is no “then.”

So in baboons, where males compete fiercely for mates, and some cash in and some are shut out, we can look for Trivers-Willard. Unfortunately for Williamson, we find exactly the opposite: mothers in good condition actually favor females. Because for baboons, social behavior has a stronger effect on the parenting than does the evolutionary bias predicted by Trivers-Willard. In this case, males transfer into another social group at puberty, so a mother cannot really affect her son’s reproduction. But a mother in good condition can indeed have an effect on her daughter’s fitness, because they are going to stay together. So the social system overwhelms the small bias predicted by Trivers-Willard.

And when it comes to humans, a man tends to have about as many children as another man, and about as many children as a woman has. Stories of sultans with 900 children notwithstanding, humans are generally characterized by a much lower variance in male reproductive fitness than other closely related species, because of the institution of marriage. Once again, without the “if” statement, there is no “then” statement.

Those who have tried to demonstrate the existence of Trivers-Willard in humans generally don’t even try to demonstrate it with actual sex ratios (which are notoriously noisy sets of data), but with measures of parental investment, such as “time spent with sons vs. daughters”. What they come up with is that there is, wonder of wonders, a lot of gendered cultural variation in parenting. (No! You think?)

Since Trivers-Willard doesn’t even hold up in baboons, who at least meet the boundary conditions of Trivers-Willard, there is no reason to think that it ought to apply rigorously to humans, much less that it actually does.

And anyway, when did Republicans become such cheerleaders for evolution?