Was Larry Summers right?

You may have seen write-ups reverberating around the Internet about a recent study, in which the writers conclude that girls and boys have the same average math test scores, and therefore the long-standing assumption that boys are better at math than girls is false. However, Alex Tabarrok at Marginal Revolution digs a little deeper and finds that the story hasn't been very accurately portrayed. While average scores are similar, the variance among scores for boys and girls is very different: namely, boys are much more likely to be either very good at math or very bad at math than girls. So, at the upper end of the achievement scale, you're likely to see many more boys than girls. Which is exactly what you see in university-level math and science programs. Looks like Larry Summers might have been onto something, after all.