When then Harvard University president Lawrence Summers suggested in 2005 that innate differences between men and women may account for the lack of women in top science and engineering positions (and subsequently resigned), he was referring to the greater male variability hypothesis. Women, it holds, are on average as mathematically competent as men, but there is a greater innate spread in math ability among men. In other words, a higher proportion of men stumble mathematically, but an equally high proportion excel because of something in the way male brains develop. This supposedly explained why boys tend to dominate math competitions and why men far outnumber women in elite university math departments. Since then, scientists have put the variability hypothesis to the test, and it comes up short.

In the most ambitious study so far, mathematics professor Jonathan Kane of the University of Wisconsin–Whitewater and oncology professor Janet Mertz of the University of Wisconsin–Madison analyzed data on math performance from 52 countries, including scores from elite competitions such as the International Mathematical Olympiad. In particular, they examined variance—roughly, how spread out scores are. Two patterns emerged, they report in a paper in the January issue of the Notices of the American Mathematical Society. The first is that males’ and females’ variance is essentially equal in some countries. The other is that the ratio of males’ to females’ variance differs greatly from one country to another. These ranged from 0.91 to 1.52 (where a ratio of 1 means the two sexes’ variance is equal, and a number greater than 1 means males’ scores were more spread out than women’s).

The finding that males’ variance exceeds females’ in some countries but is less than females’ in others and that both range “all over the place suggests it can’t be biologically innate, unless you want to say that human genetics is different in different countries,” Mertz argues. “The vast majority of the differences between male and female performance must reflect social and cultural factors.”

Such as? One clue comes from the finding that a widely used measure of a nation’s gender equality, called the Global Gender Gap Index, correlates with the ratio of boys versus girls scoring in the top 5 percent on an international math competition called PISA. In some countries, such as the Czech Republic, the boys’ and girls’ distribution of math scores were nearly identical. Another clue that gender differences in math performance are not innate comes from the shrinking gender gap. In the U.S., the ratio of boys to girls scoring above 700 on the math SAT fell from 13:1 in the 1970s to 3:1 in the 1990s.

Psychology professor Stephen Ceci of Cornell University calls the new analysis “a very important argument” in the debate over the sources of sex differences in math careers. But, he adds, the findings do not mean that biology plays no role. Just because diet affects human height, for instance, does not mean “that nature is unimportant.” Now that the greater male variability hypothesis has fallen short, nature is not looking as important as scientists once thought.