Wednesday, December 01, 2004


Steve Sailer has an article up at the moment which is getting a lot of attention. He shows an extraordinarily high correlation between birth-rate and voting for GWB. States with high birth-rates were almost all "red" and states with low birthrates were almost all "blue".

I am afraid that I have to issue a warning about what statisticians call "ecological" correlations, however -- and that's nothing to do with Greenies, surprisingly. Ecological correlations are correlations based on grouped data and grouping people only on the basis of the state they live in is very coarse grouping indeed. Such correlations are not comparable to correlations between individuals, allow no direct inferences about correlations among individuals and are commonly higher than correlations betweeen individuals. I say a bit more about them in the course of one of my academic articles here. So the correlations are a little less startling than Steve seems to think.

With all that statistician's caution out of the way, however, my best guess is that the results reflect failure to have children at all rather than family size per se. I think average birthrate is low in the blue states not necessarily because families are smaller there but because families with children are fewer. Lots of intellectual ladies never have children at all. I should know. I married two such women. My son comes from a third marriage to an intelligent but non-intellectual woman. And the low birthrate among highly educated people has long been a subject of much comment and heartburn anyway.

Why highly educated people tend Left is a subject I cover at some length here.


No comments: