Saturday, April 06, 2013

"Hypergamy"

One of my pet peeves is the abuse of the term 'hypergamy', which seems to be spreading like a parasite. A recent offender is James Taranto in the WSJ:

Why? People who go to Princeton or another elite university have a revealed preference for education and intellectual development. Because the female sex drive is hypergamous, Princeton women, as Patton asserts, will be attracted to men who are their cognitive or educational equals or superiors. Beyond the confines of the Ivy League such men certainly exist, but we are far less thick on the ground.

There are actually two senses of the term 'hypergamy' that are in use and whose conflation is making the problem worse. One of these senses is anthropological, and it is reasonably well-defined, useful, and established by evidence. Hypergamy in this sense is about marriage practices in the context of a society, and is the same thing we usually are talking about loosely and colloquially when we say that someone 'married up'. It does not have much to do with sex at all; many of the most clearly hypergamous societies are societies in which arranged marriages are standard: the usual hypergamous norm in such cases consists of parents marrying their daughters to the sons of families that have a higher social status than their own. You will notice immediately that it indicates a normative custom that is part of the structure of a society; it is the society that determines what counts as higher status and lower status, it is the social expectations and institutions that constitute its normative character. It is a genuine causal explanation; by saying that a society is hypergamous, one is saying that there is a fairly well-defined way of determining status in that society and that marriages tend to be status-asymmetric. In general by convention, we assume (where there is no other indication) that 'hypergamy' means that women marry into higher-status families, and 'hypogamy' means that women marry into lower-status families. (There are societies structured either way; and, for that matter, there are societies that where one group's norm will be hypergamous and another's hypogamous.) Societies structured on hypergamy for women tend to be somewhat more common than the reverse. The account of hypergamy is consistent, however, with the possibility of marriage tending to be hypergamous in both directions. All this requires is some standard or expectation by which one may compare the two directions; and this is, in fact, an extremely common situation, since many societies apportion status to men and women in different ways, so the expectation can be for a woman to end up marrying a man of higher status, as defined for men, and of a woman to end up marrying a woman of higher status, as defined for women. For instance, to take a toy example, a society may be structured in such a way that wealth is status for a man and breeding or education for a woman, and the marriage customs tend to drive men to try to marry women better-educated than they are while women try to marry men wealthier than they are. Hypergamy is the norm for both men and women in this society.

It does seem to be true that societies that are hypergamous for women are much more common than societies that are hypogamous; and that a large portion of societies tend to have stronger hypergamous expectations for women than they do for men. Hypergamy in the anthropological sense is capable of extremely diverse manifestations; there is no one way in which people are hypergamous, because (1) marriage customs vary from society to society; and (2) what counts as higher and lower status has to be defined relative to a given society. This ends up being extremely important when this sense is confused with the second sense, which comes from evolutionary psychology.

I said above that the anthropological sense was reasonably well-defined, useful for explanation, and established by evidence. It is also the literal sense of the term. The e-psych sense, on the other hand, is a metaphor that has never been fixed properly, is highly speculative, and whose explanatory usefulness is highly controversial. It applies, as Taranto's example above indicates, to sex, and thus is detachable from any particular custom. It would have to be biological in nature, it could not be relativized to a society, and would have to be descriptive of unconscious biological tendencies resulting from sexual selection rather than widely recognized social norms, many of which can be formed and disrupted on time scales much smaller than could possibly be useful for an explanation appealing to sexual selection. But this means immediately that we seem to be in the realm of occult causes, unless we can give a clear, precise definition of the 'hyper'. But this is not really forthcoming; the 'higher status mate' is defined in terms of any number of things like 'physical attractiveness', 'health' and 'intelligence', the last of which can be signaled by any of any infinite number of different cues, not always consistent with each other.

Without precision -- which it never really gets -- the e-psych sense runs the obvious danger of collapsing into a tautology. To say that women are hypergamous, for instance, is not obviously different from saying that women tend to mate with the kind of men that women tend to mate with, unless we give some precise meaning to "kind of men" (i.e., status) that is not selected out by seeing what kinds of men women like to mate with. This can be done on a very small scale, point by point. The best and least problematic work of this sort is also the kind of research that returns results everyone knows: both men and women, for instance, tend to prefer mates who are symmetrical in shape and facial form; minor asymmetries like scars, however, tend to have no effect if they do not change the overall symmetry of the face and frame; and although I don't know if there has been any study, one could well imagine a study showing that men, or women, or both, showed a preference for mates with a more symmetrical human form than themselves, and thus a "hypergamous preference" etc. But notice that we are here relativizing, too, not talking about a general hypergamy concept; we're just identifying a feature that is statistically connected to sexual partnering. This is as far as you can get at present on science that is not obviously bad: not hypergamy, but a loose collection of tendencies that could be labeled, if one wanted, with the metaphor "hypergamy". Likewise, the fact that we're really just dealing with statistical associations means that any attempt to use these hypergamy claims as explanations of anything (rather than, say, byproducts or effects) is highly problematic. This is all quite clear if you actually read Buss and the literature critiquing his work.

When someone says, "The female sex drive is hypergamous," they cannot possibly be using it in the well-established anthropological sense. But the way in which Taranto uses it as an explanation requires that it have the kind of causal structure that is possessed by the anthropological sense, and it is (at most optimistic assessment) unclear and controversial whether the evolutionary-psychological sense even has a genuine causal structure at all, and its status as rising above tautology (the female sex drive tends toward things that females find more, rather than less, sexually attractive) is unclear. Now, tautologies are not necessarily useless in explanation, but they don't themselves explain anything; at most they provide limiting constraints. The actual structure of Taranto's explanation, without some precise account of sexual hypergamy, which doesn't exist, is really:

Women at elite schools show, by being at elite schools, that they think education and intellectual development important. Because women tend to have sex with the kind of men women tend to have sexual preferences for, women at elite schools will tend to be attracted to men who are their cognitive equals or superiors.

Saying that this is a complete non sequitur does not quite convey how moronic it is. The revealed preference of women at elite schools is for education at elite schools; it has no necessary connection with their sexual preferences, and if it did, it would have to be established on empirical evidence, not assumed. Likewise, in the reverse direction, we have no way, as long as hypergamy is left undefined, of saying that the kind of revealed preference involved in attending an elite school is itself the kind of thing that could link up with the biological tendencies that would have to make up "the female sex drive". Cognitive equality and superiority is also left undefined, and without a precise definition, there is no way to say how it links up to any supposed female hypergamy. This is a non-explanation masquerading as an explanation.

Incidentally, I can't get over the "we" in the last sentence. First person is completely gratuitous here, so it means that Taranto is going out of his way to classify himself as the "cognitive or educational equal or superior" of the women he's talking about, which is (1) so irrelevant even if true that one wonders what sort of person would even say such a thing; (2) certainly not true on the educational side, since Taranto does not have a top-tier education; and (3) seems a somewhat hasty assumption on the cognitive side without letting us know how cognitive equality, much less superiority, is determined (and in context it could not just be any kind of cognitive scale, but one that could have an effect on the female sex drive!).