On Jan. 14, 2005, Harvard University President Lawrence Summers unwittingly brought the simmering debate about women’s representation in science careers to a full boil. In a keynote speech at a conference on diversity, Summers hypothesized that the shortage of women in certain disciplines could be explained by innate differences in mathematical ability. “There is relatively clear evidence that whatever the difference in means—which can be debated—there is a difference in the standard deviation and variability of a male and female population,” he said. Thus, even if the average abilities of men and women were the same, there would be more men than women at the elite levels of mathematical ability—and also, though Summers didn’t say this, at the lowest levels as well.

The mass media—and, surprisingly, many academics—completely missed Summers’ point about variability. For example, in the Los Angeles Times, David Gelernter, a computer scientist at Yale and occasional conservative commentator, wrote: “[Summers] suggested that, on average, maybe women are less good than men at science….” Well, no, he didn’t. But in the public debate, that is how his statement was interpreted.

A study published in July of this year by Janet Hyde, a psychologist at the University of Wisconsin, partially vindicated Summer. Hyde and her colleagues compared the scores of girls and boys in grades two through 11 on the state mathematics tests mandated by the No Child Left Behind Act (NCLB). They found no meaningful differences in the average performance of boys and girls. But the variability of boys’ scores was 11 to 21 percent greater at all grade levels. Consequently, boys were indeed overrepresented in the top percentile, by a 2:1 ratio over girls.

Does this mean that Summers was right after all? There are many reasons not to jump to this conclusion. First, though the difference in variability is real (on this test), it is not necessarily innate. In Minnesota, for instance, the 2:1 ratio of boys to girls in the top percentile held only for white students. For Asian American students, the proportion was 0.9 to 1. That is, girls outnumbered boys in the top percentile. It is difficult to imagine an innate difference in math ability that would be present in whites but not in Asian Americans.

Second, even this apparently positive finding fails to explain the paucity of women in some disciplines. “If a particular specialty required mathematical skills at the 99th percentile,” writes Hyde, “we would expect 67 percent men in the occupation and 33 percent women. Yet today, for example, Ph.D. programs in engineering average only about 15 percent women.”

Third, it is doubtful that the NCLB tests measure skills that are actually required to succeed at science. Many studies have shown that standardized tests perform poorly as predictors of accomplishment in college and graduate school, let alone later in life. Hyde’s paper sharply criticizes the NCLB tests for concentrating on simpler “recall questions” and completely avoiding “strategic thinking” and “extended thinking” questions. “Complex problem-solving is crucial for advanced work in STEM [science, technology, engineering, and math] careers,” she writes.

Finally, do so-called differences in mathematical ability matter at all? “The national debate on the issue … has so far missed the central point: scientists are made, not born,” wrote economists Michael Cox and Richard Alm in The New York Times, at the height of the Summers brouhaha. At every level of the scientific enterprise, from grade school through grad school and beyond, our society is failing to make as many women scientists as it could—perhaps because we are too mesmerized by the idea that scientists have to be born.

—D.M.