We Dug Into Data to Disprove a Myth About Women in STEM

The argument used to be that women were simply biologically less capable. Now it’s that they’re less interested. Both are wrong.

Three women in white lab coats work in front of a whiteboard in a lab.
Getty Images Plus

In 2018, a man-bites-dog claim appeared in the journal Psychological Science: In countries with less gender equality, like Algeria and United Arab Emirates, women were more likely to get higher education degrees in science, technology, engineering, and math than they were in more gender-equal countries like Norway and Finland. The authors, psychologists Gijsbert Stoet and David Geary, called this the “gender-equality paradox” in STEM. The counterintuitive finding brought headlines like the Atlantic’s dreary “The More Gender Equality, the Fewer Women in STEM.”

The stakes here are high. If men and women really have inherently different career preferences, then any attempt to get more women into STEM fields would be misguided and likely to fail. That was certainly how men’s rights activists such as Jordan Peterson and conservative think tanks such as the American Enterprise Institute, who have cited the theory, interpreted the results.

Curious about Stoet and Geary’s findings, the GenderSci Lab, an interdisciplinary group of scientists and gender studies scholars at Harvard (to which we both belong), tried to replicate them. But their numbers didn’t add up. For example, in Poland, 43.63 percent of STEM graduates are women, which would place it fifth for representation of women in STEM out of the 45 countries included in Stoet and Geary’s analysis. Yet Stoet and Geary reported a value of 26.9 percent, ranking Poland 20th. Why?

We discovered that Stoet and Geary weren’t looking at “women’s share of STEM degrees,” as they had claimed, at all. After we sent over some of our results that contradicted Stoet and Geary’s, the journal editors investigated, and Stoet and Geary revealed they’d actually used their own undisclosed measure of women’s representation in STEM. The case of Algeria highlights a central point of contention. In Algeria, 53 percent of STEM graduates are women. Still, only 9 percent of women college graduates choose a degree in STEM, compared with 13 percent of men. Stoet and Geary had claimed that they were reporting the 53 percent number, but they were actually focusing on the statistic that men were receiving degrees in STEM at a higher rate.

After internal review, Psychological Science required extensive corrections to the published study. In their revised paper, Stoet and Geary maintain that there is a gender paradox in STEM but clarify that it relates to their more obscure measure, which they termed the “propensity” of women and men to attain a higher degree in STEM. This week, the same journal published a peer-reviewed paper from our group citing conceptual and empirical problems with the “gender-equality paradox in STEM” hypothesis. The paper shows that the negative association between gender equality and women’s STEM achievement does not persist when the measures of gender equality and achievement change.

The focus on “propensities” is a trend in the broader conversation about the role of social and biological factors in women’s and men’s STEM achievement. Twenty years ago, biological hard-liners, including Geary, argued that women are biologically less capable in the STEM fields. Now that women are outnumbering men in many scientific and medical fields, outperforming males on many standardized STEM tests, and receiving larger numbers of higher degrees of all sorts than men worldwide, hard-liners such as Steven Pinker and Charles Murray have moved away from such assertions in favor of the claim that women are biologically wired to find less joy in STEM or are simply less interested in it.

Stoet and Geary claim that their formula captures how much women naturally do or don’t prefer STEM, which is why they prefer it to the simpler “number of women in STEM.” Aside from the fact that their formula obscures and minimizes leaps and bounds in numbers of women’s university degrees in STEM across all nations over the past several decades, this is a problem for two reasons. First, you can’t infer women’s preferences about STEM as individuals from the behavior of women as a group defined coarsely by national borders. Second, they claim that women’s behavior is not only an expression of their preferences, but that these preferences are innate and biological. Even if the patterns Stoet and Geary report do reflect preferences, that doesn’t tell us anything about whence those preferences come.

Is there a paradoxical relationship between women’s achievement in STEM and gender equality? It depends on which countries you include and how you measure gender equality. For the small sample of 45 countries examined by Stoet and Geary, women’s STEM achievement does negatively correlate with the World Economic Forum’s Global Gender Gap Index, or GGGI, which ranks countries based on markers of gender parity. But correlations between women’s STEM degrees and nation-level gender equality don’t stand up when parts of the equation—how we measure women’s STEM achievement and how we measure gender equality—are changed. As we report this week in Psychological Science, when we tested the correlation between women’s STEM degrees and Stoet and Geary’s own 2019 proposed alternative gender-equality index, a separate index that includes basic measures of well-being such as health, education, and life satisfaction, there was no paradox to be found.

In any case, all nation-level gender equality measures are highly imperfect for understanding the drivers of gender equality. Stoet and Geary naively adopt the GGGI as a social science measure of gender equality, but it was not designed for that purpose, and it should not be used as a measure of gender empowerment or attitudes about gender. For example, Rwanda ranked sixth in the world on the 2015 GGGI due to high representation of women in economic and political life. This outstanding representation stemmed from Rwanda’s post-genocide sex ratio imbalance, not a campaign to increase women’s empowerment. Similarly, a negative correlation between STEM degrees and GGGI rank in a particular country—say, Luxembourg—tells us nothing about whether gender equality is causally related to STEM achievement or a product of other factors such as coeducational opportunities on offer for higher education in STEM in that locality.

Women in science have been pushing back the tide of claims about women’s lack of interest and ability in STEM for decades. The work continues today. As we see it, the so-called gender equality paradox is a new entry in an old playbook of arguing that biological sex differences, not social inequalities, drive the gender disparities we see in areas such as STEM. But a little digging shows that the paradox is the product not of innate sex differences in STEM interest, but the use of contrived measures and selective data to tell a particular story.