The American Enterprise Institute hosted an event last week titled “Why Aren’t There More Female Scientists?” that discussed whether the underrepresentation of women in engineering and math fields is the result of gender bias or alternatively because of differences in academic interests and cognitive strengths.
Recently released data by the College Board for math scores on the 2009 SAT exam (data available here and here) shed some light on the controversy because the test results suggest that there are significant gender differences in mathematical abilities as demonstrated by performance on standardized tests. The College Board data reveal that high school girls have math preparation that is equivalent to high school boys, measured by: a) the average number of years of math study (3.8 years for girls vs. 3.9 years for boys), b) the highest level math taken (50 percent of both boys and girls take math through calculus in high school), and c) both males and females have average GPAs of 3.14 for their high school math classes.
And on several other general measures, females in 2009 might have actually been better prepared for the SAT math exam because: a) 117 high school girls took advanced placement honors math courses for every 100 boys, b) girls outnumbered boys in both the top 10 percent of their high school class and the next 10 percent, and c) there were 150 girls with GPAs of A+ for every 100 boys and 156 girls for every 100 boys with a GPA of A.
And yet despite equal or better academic preparation, girls underperform boys on the SAT math test. These differences are significant and they persist over time, suggesting that some gender differences exist for certain cognitive abilities. For example, the mean math SAT test score for males in 2009 was 534 compared to the mean score for females of 499, for a difference of 35 points that is comparable to the 30+ point male-female gap that has persisted at least since the early 1970s (see chart below).
The next chart displays the ratio of male to female test-takers for the 2009 SAT math test by 10-point increments for scores between 200 (lowest) and 800 (highest, perfect score). For perfect scores of 800, males (6,928) outnumbered females (3,124) by a ratio of 2.22 to 1, and therefore 69 percent of the high school students who got perfect math scores were males vs. 31 percent for females.The graph further shows that high school boys outnumbered girls at all of the 23 SAT math test scores between 580 and 800 in 10-point intervals. There is also a clear trend that male students increasingly outnumber girls at higher levels of test performance, as the male-female ratio rises from 1.0 at test scores of 580 (equal number of boys and girls) to a ratio of 2.2 at the highest test score of 800 (222 boys per 100 girls with perfect scores).
If we are trying to explain the underrepresentation of females in top science, math, and engineering positions in industry and research universities, and if we realistically assume that those high performers are likely to have very high scores on the SAT math test, the explanation of the gender gap in science seems pretty clear: males are over-represented by a factor of 2:1 for math SAT test scores above 750 (98th percentile). Unless and until there is something close to gender parity for math test scores on standardized exams like the SAT, especially for test scores at the high end, women will likely continue to be underrepresented in engineering and math fields as a natural and expected outcome of some gender differences in certain mathematical aptitudes.