g, factor analysis

Cosma Shalizi's article (it is not a post) on g and factor analysis can be praised but it can not be praised enough. I particularly like the bit about factor analysis, being that I work at a consulting firm with a bunch of I/O Psychologists. When not doing regular data analysis they develop employment tests. I know little enough about this arcana other than the couple of statistical procedures I've found useful for my work, so I'm grateful for Cosma's explanation.

I have happily worked at a testing company for over six months without developing any opinions about g as a platonically really real thing. We try to stay away from g loaded tests because we're trying to create tests that discriminate as little as possible, and there are plenty of ways to predict job performance without g (though I gather it is a bit harder).

I can recall several different economics articles that use IQ scores as quick and dirty proxies for human capital, a practice which entails no commitment to any dangerous or mistaken views about human psychology. Which is not to say that most of economics is not shot through with mistaken views of human psychology, but most of these are made out of either laziness or a desire for tractability.