I just counted the number of men and women grad students in the Berkeley Math department. Not being able to gender non-European names, I count 120 men and 25 women. That’s a 1 to 5 ratio.
According to the graph here, if IQs in that department are at or above 175 IQ (assuming IQ is a good proxy for math ability or at least the difference between the male and female distributions of math ability is similar to the difference in distribution of IQ), then variation in math ability can explain the underrepresentedness ((Get your prefix here! Suffix here! Get your hot, fresh -fixes!)) of women in that department.
Well, according to the Berkeley math admissions page, applicants score above the 80th percentile in the math subject test. Assuming this test is only taken by math majors and that the average attendee is smarter than the average applicant, we can assume Berkeley math grad students are probably in the 90th percentile of math grad students (another way to get this percentile would be to assume all the top students go to Berkeley, its the top math program, and estimate the total number of math grad students… but I’m too lazy).
Ok. So, using the distribution of GRE scores on this page (pdf) and using this table to convert GRE to IQ, this means the average IQ at Berkeley Math is 146.92. Rounding to significant digits, this would suggest Berkeley students are 2 standard deviations too dumb to justify the gender imbalance.
Damn sexists… or there’s more problems with this exercise than I can name.