[Note as of February 2011: The most recent score distributions by sex are in the report Adjustment of the protonorms to norms conversion]
This report focuses on the performance by sex over seven tests that received substantial numbers of submissions: Long T.F.G., C.I.T., Final Test, Q.M.C. #3, Numbers, Short T.F.G. and S.T.H. Combined it concerns 732 test submissions (656 male and 76 female).
For each test, the average I.Q. per sex is computed, after which the weighted averages over the seven tests are (S.D.=15):
Combined average is 140.2, difference is 11 points. Compared to the Male-Female difference at I.Q. 100 (estimated at 4 points by Richard Lynn) this means an increase of the difference of 7 points over a distance of 2.68 general population standard deviations.
The difference between the male and female population S.D.s therefore appears to be 7/2.68 or about 2.6 I.Q. points. In other words, the male mean and S.D. are 102 and 16.3 while the female mean and S.D. are 98 and 13.7. These numbers roughly match an estimate I made earlier, based on my observation that the 98th centile is exceeded by about twice more males than females (Mensa membership) while the 99.9th centile is exceeded by 15 times more males than females (Glia Society membership).
The male/female ratio at 99.9 is in fact predicted to be 15.2 by the means and S.D.s given in the previous paragraph, so the match is not too bad. Future reports may yield even better estimates.