You are using an outdated browser.
Please upgrade your browser
and improve your visit to our site.
Skip Navigation

Men, Women, And Test Scores

Dear Reader,


We here at The Stash are delighted to introduce Zubin Jelveh, who'll be joining me beginning today to help make sense of the financial crisis and other matters economic. Zubin's most recent blogging gig was at Portfolio, which is how I came across his terrific work. He's now in graduate school studying quantitative methods--a set of skills he'll be bringing to bear on the news of the day.


Noam Scheiber

Back in 2003 2005, ex-Treasury Secretary Larry Summers made the infamous claim that genetic differences between men and women might* partly explain the gap we see between the number of male and female super-geniuses in sciences.

He relied on research (similar to this) which showed that the distribution of scores on apptitude tests was wider for men than women. That means that even though there might be no difference between average test scores, men tend to score both better, and worse, than women at the top and bottom of the distribution. But two recent pieces of research cast some doubt on this genetic and statistical argument.

First, this working paper by economists Scott Carrell and Marianne Page of UC Davis and James West of the United States Airforce Academy finds that the gender gap is sharply reduced when a female student has a female professor:

...the effect of female professors on female students is largest among students with high math ability. In particular, we find that among students in the upper quartile of the SAT math distribution, being assigned to a female professor eliminates the gender gap in introductory course grades and science majors. We also find that professor gender has minimal effects on male students’ outcomes.

And second, a study published last year in Science found that the different distributions we see between male and female test scores is not a universal phenomenon. Researchers examined data from a standardized test given to 15-year-old students in OECD countries and found four places (Indonesia, United Kingdom, Iceland, and Thailand) where there was no gender gap at the high end of the distribution for math scores. In fact, in Iceland and Thailand girls outperformed boys.

These results suggest that gender bias and sexism could play a bigger role in academic achievement than we probably think. And looking ahead, the real story might be the gender gap that's opening up in reading, where the same Science study found that in nearly every OECD country, girls outscore boys, often by a substantial amount.

--Zubin Jelveh

* As many of you pointed out in the comments, Summers did not explicitly say that genetic differences caused some of the observed differences we see in male and female achievement, but that they could be a possible cause. I changed the wording in the post to reflect this.