Interesting Research Alert: Girls and Math

There is a theory explaining why there are so few women in the highest levels of math and science which is very comforting to scientists (see Larry Summers) looking to justify the persistent gap which remains many decades after the 'women's lib' movement.  Its quite simple.  You posit that on average there is no difference between the ability to do math and science of men and women (the means are the same), however, men vary more in their ability so there are more really smart men than really smart women.  Let's look at the figure below and take the X axis to be math ability, with 0 being average and the Y axis to be the probability for a person to have X ability in math. This theory says that, contrary to what people thought in the dark ages (pre women's lib), women's ability in math is not described by the green curve, and men's ability is described by the blue curve.  Rather, they believe that women's ability is described by the blue curve and men's ability is described by the red curve.   You see that the red curve is higher than the blue curve for values of X greater than .7, i.e. there are more men than women who are .7 in ability greater than the average, while the mean value of both distributions are 0. 

image from wikipedia

Some people say that both men and women are described by the blue curve for values of X which are less than 1, but for values greater than 1, the men's curve is enhanced so that the 'tails' of the distribution are higher for men.

Some new research from economists at MIT suggests that this simple theory does not hold.  Part of the motivation for the theory I outlined above is that not only are there very few women in the highest levels of academic math and science, there are also very few girls performing at the highest math levels in high school.  The researchers, as described in this article, focused on those top math student by looking at the results from the past 60 years of American Mathematics Competitions (AMC) contest, a math exam taken by some 125,000 "exceptional high school students".  They found that there were 4 boys for every 1 girl in the top 94% and in the top 99.9% that gap increased to 12 to 1.  However, suspecting that environment rather than innate ability might explain these results, they looked at the distribution by school of the highest performing students, those competing in the International Math Olympiad.  They found that while the boys were distributed over approximately 200 high schools, the girls came from just 20 schools.  In their words, "the gender gap at the very highest levels is in part due to extreme selection effects". 

I've skimmed through the paper, and it presents a more complicated picture than what is outlined in the article.  For example, while they found variation among high schools in the AMC sample, it was barely statistically significant, which might lessen the claim for environmental bias.  In any case, they say they do not want to draw conclusions, although they make broad but non committal statements about girls susceptibility to peer pressure and say their results aren't inconsistent with the 'men vary more in their ability' argument.  It's a murky field but what it can be said that there are persistent gender differences at the highest levels of math but that clearly there are some environmental factors at work.