Todd Zywicki, over at The Volokh Conspiracy, helpfully restates a key element of the controversial Larry Summers Hypothesis as follows:

The distribution of natural endowments for math abilities for men show[s] the same mean but greater variance than math abilities for women. Therefore, men will be disproportionately represented at the tails of the distribution relative to women.

Summers offered this as possibly a part of the explanation for the statistical over-representation of men in the very upper end of the distribution of math users (namely, among professional scientists and engineers). Todd notes that greater variance also has an implication less flattering to males: “there are likely to be more men in society than women with unusually poor and below-average math skills."

Washington Post columnist Robert J. Samuelson now points to evidence that males dominate the lower tail of other distributions:

Compared with girls, more boys take drugs for attention-deficit disorder (80 percent of users are boys); more are held back (8.3 percent vs. 5.2 percent among 5- to 12-year-olds); more are high school dropouts (12.2 percent vs. 9.3 percent among 16- to 24-year-olds); and the gap between boys and girls on reading tests is widening.