Women are naturally worse at math and spatial reasoning than men due to biological differences.

In December 1980, the journal Science published a study by Camilla Benbow and Julian Stanley of Johns Hopkins University that produced headlines across the country. Benbow and Stanley had analyzed the math SAT scores of roughly ten thousand academically gifted seventh-graders and found that boys outscored girls by a significant margin, and that the ratio of boys to girls among the highest scorers was thirteen to one. Their article did not claim to have identified a cause. But its title, Sex Differences in Mathematical Ability: Fact or Artifact?, and the way it was received delivered a clear implication. Time, Newsweek, and wire services across the country reported the finding as evidence that boys' math advantage was biologically fixed.

The belief the headlines reinforced had roots running back decades. By the middle of the twentieth century it was standard pedagogical wisdom that girls excelled at reading and verbal tasks while boys were the natural mathematicians. The explanation given in textbooks and teacher training programs was straightforwardly biological: male brains were structured for spatial and quantitative reasoning in ways female brains were not. Girls who struggled in math were told, in effect, that they were built that way. Girls who excelled were treated as exceptions who proved the rule.

What the Benbow and Stanley study could not account for, and what critics pointed out immediately, was everything that preceded the test. Boys and girls did not arrive at seventh grade having had identical mathematical experiences. Boys were more likely to have been encouraged to pursue math by parents and teachers. Boys' mathematical errors were more likely to be treated as worth correcting; girls' as worth accepting. The SAT scores measured an outcome that had been shaped by years of differential preparation, expectation, and feedback. Benbow and Stanley had no way to disentangle nature from that accumulated history.

The stereotype threat research that emerged in the 1990s made the mechanism visible. Steven Spencer, Claude Steele, and Diane Quinn ran a series of experiments at the University of Michigan in which men and women with equivalent math preparation took identical tests. When the test was described as one that typically produced gender differences, women performed significantly worse than men. When it was described as one on which men and women performed equally, the gap disappeared. The only thing that changed between the two conditions was a sentence at the start of the exam. That sentence, activating awareness of the stereotype, was enough to measurably impair performance among people who were otherwise equally prepared.

If the biology argument was correct, if women's math underperformance reflected structural differences in the brain, stereotype threat could not eliminate the gap with a paragraph of instructions. The gap's sensitivity to social context was itself evidence against a biological explanation.

The cross-national evidence told the same story from a different angle. If male math advantage were biologically fixed, it should appear consistently across cultures. It did not. Girls in Iceland and Sweden regularly outperformed boys in mathematics; performance patterns tracked cultural attitudes toward gender equality more closely than they tracked any biological variable. The gap was not universal. It was contextual.

Janet Shibley Hyde and colleagues published a definitive analysis in Science in 2008, drawing on state standardized math tests taken by approximately seven million students across ten states. The overall gender difference was essentially zero, d = 0.05, a trivial effect size. Among the highest performers, the thirteen-to-one ratio Benbow and Stanley had reported in 1980 had shrunk to roughly three to one, a shift too fast by orders of magnitude to be explained by any biological process. The belief in innate female mathematical inferiority had been doing the damage it was always invoked to explain.