Why do boys so often outperform girls at the top end of mathematics achievement? This question perplexes academics, is the source of much consternation in hiring, and has caused no end of trouble for those (like Larry Summers) who have waded incautiously into the debate. As en educator, I want to understand if we as a society are not doing enough to help girls enter math and science. But I’m not convinced if someone tells me that girls don’t get as many math PhD’s as boys, because it doesn’t address the root cause. I need to see something about society’s impact on those numbers.
First, some background. Yes, in school, boys do test higher in math than girls on average, but the difference is usually very small. If you don’t like relying on tests, girls tend to have higher marks in school. The key thing to understand is the score distribution: more boys score at a high level, but more boys also score at a low level, bringing the average back to about the same score as girls. Hence, men are overrepresented as top achievers, but also as bottom achievers. (There is a provocative argument to be made that there are evolutionary reasons for men to be greater risk-takers and to have greater variation in their performance.)
Nevertheless, there is considerable evidence for strong cultural factors that impact girls’ performance. For example, here’s some evidence for factors that may affect girls in the aggregate (but which, fortunately, are also overcome by many girls):
- There is clear evidence of stereotype threat. For example, if the College Board asked for test takers’ gender after the AP calculus exam, rather than before, some 5000 more girls would pass it nationwide. Surely this stereotype danger affects not just test-taking in the moment, but also impacts students in each classroom, each day, little-by-little. (Full disclosure: there are also some doubts about the impact of stereotype threat.)
- Besides, in general, women don’t do as well on tests, and yet tests define a large part of our success and advancement in such subjects.
- Even when sexism is not overt, we all have mental biases of which we’re not aware. Just as double-blind auditions significantly boosted women’s admission to symphony orchestras (i.e. having potential candidates audition behind a screen), the same is true of peer review for journal articles. Again, this effect must surely play out again and again throughout school, with many little encouragements or discouragements building upon one-another.
- The clearest evidence of social factors affecting girls’ performance might be this study that shows that top-performing girls tend to consistently come from just a few high schools, while top-performing boys come from many different high schools. In theory, this means that those high schools have figured out how to more fully develop girls’ abilities.
- Another strong example is this study in which the math anxiety of elementary school teachers (almost all women) strongly affected girls’ math performance and perception of gender roles, but had no impact on boys.
- Another, similar study that might find causation is this recent one in which it was found that American parents talk to boys age 20-27 months about numbers much more often than they do to girls.
- Women who are exposed to romantic cues report less interest in mathematics, perhaps a reflection of what society views as “feminine” or desirable in women.
- This study demonstrates several countries in which girls and boys have the same variance in performance, and also notes that there are more girls on serious IMO teams when measures of gender equity are higher.
- Another cute example comes from a study that gave spatial puzzles to two tribes, one patrilineal, one matrilineal. The gender gap disappeared in the matrilineal society. (Although one has to ask, why was it merely eliminated and not reversed?)
If we search for a core issue, the popular perception of girls’ math ability seems to be a major factor. Perception of girls’ math ability can explain many of the above bullet points: it is likely the core cause of stereotype threat, it may affect parents’ conversations with children, and it likely impacts the kind of role model that an elementary teacher might be. If we could just address this one issue, if we could telepathically make it clear to each person in the world that women can be tremendous achievers in math and science, then we might eliminate the gender gap or significantly close it.
Now, I want to be clear: the jury is still out. Political correctness dictates that it’s much easier to publish a study that explains the gender gap through social rather than biological factors. The preponderance of evidence points to clear social factors that influence girls’ performance, and those social factors all seem tied to our view of girls’ ability to achieve, but there’s no way to judge what would change if we could correct for social factors. Women might choose other careers for very legitimate reasons. There might, yes, be biological differences. We don’t truly know.
However, this lack of evidence should make no difference to our policies! If it turns out that the impact of social factors is small, then not very much is lost by making a concerted effort to change the cultural perception of women’s achievement in math and science. On the other hand, if it turns out that the impact of social factors is large, then we gain tremendous value from repairing cultural perception of women in math and science. Moreover, it seems likely that the impact of social factors is large.
We should get rid of the cultural factors that prevent girls from making good on their ability, or we should at least strengthen the paths for girls to succeed independent of cultural factors. Which begs the question: how can we effectively do these things?
Note: Post has been updated to add the Wisconsin study in which some countries have the same variance in performance for both boys and girls. Post has also been updated to add the meta analysis that calls stereotype threat into question.
Consider Learning 
It is also worth noting that while the MC student body is roughly 1/3 girls, you find at most 10% girls in most classes ranking 3+ chilis. I don’t think MC somehow strengthens the gender gap, though this is often difficult to see. So what is it that makes this gap still appear?
This is a reflection of the applicant pool. In general, in admissions, we admit the same overall percentage of boys and girls, and the general cutoff for “you can handle camp and therefore should be admitted” is the same. However, we have a lot more boys above that cutoff than girls.
In other words, the differentials in mathematical background and sophistication exist before camp. It’s my hope that many girls, who may previously not have been exposed to quality mathematics/good role models, will become substantially stronger by virtue of the experience!
” Just as double-blind auditions significantly boosted women’s admission to symphony orchestras ”
I was annoyed recently by an interview article with, I believe, the manager of the Berlin Philharmonic (published maybe in a Lufthansa inflight magazine?): in one part of the interview, the manager proudly discusses the audition process where the auditioner has to perform in front of a selection of current orchestra members, and in another part of the interview the manager dismisses a question about the skewed gender ratio of the orchestra by stating that they just take the best talent. I would really have thought that the double-blind audition study would have made the rounds among major orchestras…
As an educator, I am most interested in showing that girls are not the stereotypical math-haters nor are outclassed by boys in the math/science realm. You say, “there is considerable evidence for strong cultural factors that impact girls’ performance. For example, here’s some evidence for factors that may affect girls in the aggregate…” but you do not give the sources of the evidence. Can’t really make an argument without them.
Sorry, I’m not sure that I understand your question! The sources for the evidence are linked in the bullet points.