An interesting quote I discovered about the old math SAT:

Although the normative reports for the PSAT and the SAT do not indicate the number of boys and girls earning the highest scores on these tests, Dorans and Livingston (1987) reported the number of very high scores earned by boys and girls on the SAT-Mathematics for all English-speaking examinees tested in June 1981 and May 1982. When the examinees from the two test administrations were combined, 96% of 99 scores of 800 (the highest possible scaled score), 90% of 433 scores in the 780-790 range, 81% of 1479 scores between 750 and 770, and 56% of 3,768 scores of 600 were earned by boys.

Thus, the degree of male overrepresentation was directly related to the level of SAT-Math performance. However,  the population of adolescents was not examined. Different percentages might be found for the subset of high-ability youths who did not take the SAT. But given the high correlation found between ability and SAT completion, it is doubtful that there were enough unrepresented youths to bias the reported percentages.


Source:  Feingold, A. (1988). Cognitive gender differences are disappearing. American Psychologist, 43(2), 95-103

If 96% of 800 math SAT people were male, it’s interesting to ask what an 800 math SAT score equated to on the IQ scale in the 1980s.  The New York Times states:

Out of about 1.5 million students who took the S.A.T. in 1982-83, 749 got a perfect score on the math section, according to the Educational Testing Service.

According to Ron Hoeflin, during the 1980s, roughly one third of American 18-year-olds had taken the SAT and virtually all of the top talent had, so a perfect math score was not just the 749 best out of the 1.5 million who took the test,  but the best out of all 4.5 million in that age group, and thus equates to the one in 6000 level, or math IQ 154+ (U.S. norms)

Although I think the math SAT measures only one kind of math IQ.  Women are perhaps better at more intuitive math like statistics.