Since this question catalyzed the swift firing of former Harvard President (and current Obama adviser) Larry Summers, I've found boys' and girls' math achievement a fraught subject at dinner parties and hang-outs. So I was intrigued to see that Freakonomics economist Steven Levitt teamed with another well-known academic (from Harvard, no less) to tackle the question. While they characterize their own attempts at explaining the gender gap a "failure," they manage to debunk other theories, such as less study time for girls, lower moral support from their parents, and male-biased tests.
I think these kind of studies are invaluable. My understanding is that there is something resembling a scientific consensus that there are major differences between male and female brains. I'm not terribly offended by the finding that girls are falling behind boys in math, just as I'm not offended by the finding that boys lag behind girls in verbal skills and concentration, or graduation rates.
The vale of these male-female comparison studies -- especially in education -- isn't to use scientific observations to justify and cement the differences (eg. "Look, women will never have men's spacial capacity, so let's just stop teaching them geometry") but rather to discover the ingredients of male-female differences so that we can find more effective ways of teaching challenging material. If the Levitt/Fryer study is a "failure," I still hope it's a catalyst.
[Via Real Time Economics]