Graduate Record Exam

Quantitative scores for engineering majors

32,810 men and 8,385 women who planned to or who did major in engineering took the GRE test in 1997, and they scored 697 and 683 respectively in the quantitative portion of the test.  The men who took this test represented .57% of the 5,768,000 men in college, and the women who took it represented 0.09% of the 9,330,000 women in college.  In order to make a fair comparison between the sexes, we would need to compare sample sizes which are proportional to college enrollment by sex, because the 5.7x larger sample size of men skews the results.   Comparing the top 0.1% of men, or 5,768 men, to the top 0.1% of women, or 9,330 women, is the only fair way to compare their scores.

The standard deviation for men was 88 points, which means that 16% of the men, or 5,250 of them, scored higher than 785, so the median score of the top 0.1% of men would have exceeded 800 had this not been the upper limit of the GRE test scores.

Conversely, the standard deviation for women was 93, which means that 16% of the women, or 1,342 of them, scored lower than 590.   Increasing the sample size for women by 11% would require the inclusion of 945 women who scored significantly lower than 590, which would reduce the median score of this sample to something less than 650.  So where the gap between the median scores of men and women on the quantitative portion of the GRE appears to be only 14 points, the real gap between representative sample sizes is greater than 150 points.

	See also ftp://etsis1.ets.org/pub/gre/250365.pdf
	Comments: