Sex and STEM: Stubborn Facts and Stubborn Ideologies

Sex and STEM: Stubborn Facts and Stubborn Ideologies

David C. Geary and Gijsbert Stoet
David C. Geary and Gijsbert Stoet
15 min read

Many academics in the modern world seem obsessed with the sex difference in engagement with science, technology, mathematics, and engineering (STEM) fields. Or rather they are obsessed with the fact that there are more men than women in some of these fields. There is particular concern about the lack of women in prestigious STEM fields, such as Ph.D.-level faculty positions, but surprisingly there is no concern about the under-representation of women in lower-level technical jobs, such as car mechanics or plumbing.

The concerned academics have been especially effective in convincing others, or at least intimidating them, into accepting their preferred interpretations regarding the source of these sex differences (as illustrated in the Google memo debate). These interpretations are not surprising and they include sexism, stereotype threat, and more recently implicit bias and microaggression. Each of these ideas has gained traction in the mainstream media and in many academic circles but their scientific foundations are shaky. In this essay, we’ll provide some background on the STEM controversy and consider multiple factors that might contribute to these sex differences.

The U.S. National Science Foundation reports that women are awarded 57 percent of undergraduate STEM degrees, but with substantial differences across fields. Women earn the majority of degrees in the life and social sciences but less than 20 percent of the degrees in computer science and engineering, sex differences that have held steady for several decades. The STEM debate is primarily about sex differences in educational and later occupational choices in inorganic fields, those focused on understanding non-living things. These differences are socially important because these tend to be prestigious occupations, and practically important because the different numbers of men and women in these fields contribute, in part, to the sex difference in earnings.1

Melba Roy at NASA, 1964

At the core of the obsession is the zeitgeist that there should be gender equity – equal outcomes – for anything of monetary or social value. The combination of an extreme agenda among some feminists and a stubborn sex difference has created a cottage industry focused on rectifying this ‘injustice.’ The federal governments in the U.S., U.K., and other Western nations have devoted hundreds of millions of dollars to interventions to close the gap. Some of the activities funded by these initiatives make sense and are possibly helpful in some ways, such as programs to increase interest in mathematics or programming among girls. Other programs, such as developing mentoring programs exclusively for women who are junior faculty in science and engineering in university settings (e.g., U.K. Athena Swan’s programs) are ethically problematic because they assume men do not need that level of support. From an evidence-based perspective, the most questionable and perhaps the most favored of these interventions are focused on stereotype threat, implicit bias, and microaggression.

Stereotype threat occurs when one is confronted with tasks or situations that trigger negative stereotypes (e.g., that “women are not as proficient at math as men”), that in turn result in a preoccupation about performing in a way that confirms the stereotype.2 Critically, the preoccupation is said to undermine actual performance, even when there is no factual basis to the stereotype. Implicit bias is a related concept and involves an unconscious association between group membership (e.g., sex) and stereotypical positive or negative attributes that can also, in theory, result in prejudicial behavior towards individuals within that group.3 Microaggressions are subtle behaviors (e.g., facial expressions) or statements that are not explicitly hostile but are nevertheless interpreted by the receiver as conveying contempt, stereotypical attitudes, or other negative beliefs.

Proponents of these theories and their activist followers believe that some significant proportion of the sex differences in STEM fields – but curiously only those in which men outnumber women – are thought to be caused by pervasive negative stereotypes about women’s abilities in these fields that in turn undermine their performance. Their argument is that in school and in the workplace, women in these fields are subjected to microaggressions by teachers and colleagues that seep from their unconscious belief in these same stereotypes. The result is the creation of unsupportive and even subtly hostile classrooms and work environments. These types of explanations fit hand-in-glove with the narrative of some feminist scholars; that the sex difference is largely due to oppressive social and cultural factors that undermine women’s pursuit of degrees and occupations in STEM fields.4

These concepts have been embraced by the mass media and beyond, and include accusations made in the New York Times that the wording of several SAT items trigger stereotype threat and undermine girls’ performance on the mathematics section of the test, and the publication of self-help books to purge one’s own unconscious biases.5 On the face of it, there is nothing wrong with an academic and mass media focus on these topics. The real issues concern the magnitude of these effects on women’s STEM participation and the foregone opportunities of not focusing on other factors that might have an even stronger impact on their participation.

Let’s start with the magnitude of stereotype threat on girls’ and women’s mathematics achievement.6 Given the prominence of the topic and the resources devoted to it, we carried out the first meta-analysis (i.e., statistical aggregation of experimental results across many studies) of the effect of stereotype threat on sex differences in mathematics performance.7 We reasoned that if stereotype threat had a substantive effect on girls’ and women’s mathematics performance then the most basic experimental manipulation of the effect should replicate across studies. It replicated in about half of the studies that used the same and most basic experimental design. And of the half that replicated, half of these used a questionable statistical approach. The summary of the other half did not show a stereotype threat effect. Thus, if you accept the questionable statistical approach, you may still argue that a small stereotype threat exists.

In a related analysis, Flore and Wichert found a similar overall effect, but when they corrected for publication bias – the tendency for positive but not negative results to be published – the effect essentially disappeared.8 Because studies that do not find an effect tend not to get published, this means that even when there is evidence for a small stereotype threat effect in some reports, the real-world impact could be close to zero. Currently, a large replication effort is being carried out, and we are optimistic that this will be a significant step towards finally determining whether or not stereotype threat can undermine girl’s and women’s performance in mathematics, and if so to what extent.

It should be noted, though, that the largest study to date included nearly 1000 children (9-14 years old) and found no effects.9 This latter study is of particular interest, because it included adolescents, whereas most other stereotype threat studies were carried out with university students. If stereotype threat discourages girls from pursuing math-intensive STEM coursework and careers, its effect should be evident in high-school. The fact that a large and well-designed study could not find any effect, in our opinion, suggests either the effect does not exist or it is unmeasurably small.

In any case, the existing evidence indicates that stereotype threat has received outsized attention from educational policy makers and opinion makers. Thus, the considerable efforts at addressing this ‘problem’ will almost certainly have little if any effect on girls’ and women’s participation in inorganic STEM fields.

We suspect the same is true for implicit bias. For a variety of cultural and legal reasons, the level of explicit sexism has dropped considerably over the years in most school and work environments. But, as noted, girls’ interest and women’s participation in inorganic STEM fields has remained stubbornly low.4 So, there is now a fork in the road. Down one path is the conclusion that explicit sexism is no longer keeping girls and women away from these fields and so something other than sexism or bias must going on, as we’ll discuss below. The other path maintains the conceptual grasp on sexism but switches focus to an ‘unconscious’ and subtle form of sexism that results from implicit bias and its behavioral companion, microaggression.10

It seems that most academics and the general public have wandered down this second path. Indeed, implicit bias has achieved a cult-like status in some academic circles and in the wider culture. There are now on-line tests to assess one’s implicit bias in a number of areas, including sex differences in work and family. We are not doubting that people have all sorts of implicit beliefs that may or may not be accurate. At issue here is whether or not we can rigorously and accurately assess these biases, and whether or not the strength of any such biases is sufficient to explain the sex differences in STEM fields. The assessment of implicit bias is often done using the implicit associations test, whereby the strength of people’s associations between sex and certain attributes, such as work or science, is assessed by a series of categorization tasks. The difference between the speed of categorizing certain attributes (e.g., scientist, engineer) to one sex or the other is taken as an index of implicit bias. Nosek and colleagues found that people are generally quicker to associate men with science and women with literature, which is taken as an implicit bias against women in science, although they do concede that their results may reflect people’s knowledge of actual occupational sex differences.11

Although it doesn’t typically reach the ears of the general public, there is vigorous debate within the scientific community regarding what exactly is being measured by implicit bias tests, and whether they actually influence behavior.12, 13 Even if the tests are measuring bias, the influence on actual behavior is small at best, although proponents argue that these small effects add up over time.14 The ways in which implicit attitudes are thought to influence real-world outcomes include promoting stereotype threat and microaggressions.15 As we noted above for stereotype threat, there are serious concerns about the ability to accurately measure microaggressions, whether or not they are related to implicit bias at all, and – if it is indeed a valid concept – whether or not ‘victims’ suffer long-term consequences.16 These issues have not stopped the development of yet another rent-seeking industry to put a stop to this ‘aggression’ on college campuses, in the workplace, and in daily life.

We suspect that concepts like stereotype threat, implicit bias, and microaggression have gained traction because they fit the inequalities-equals-oppression narrative.17 In many cases, explicit oppression is hard to find in classrooms and the workplace and, thus, the resort to explanations from unconscious bias and fleeting behaviors (microaggressions) that continually ‘assault’ and undermine the ‘victim.’ In this case, the victims are girls’ and women’s aspirations towards and performance in certain STEM fields, especially engineering, computer science, and the physical sciences. The logical response to this narrative is the development of interventions to reduce stereotype threat, implicit bias, and microaggressions. But, what if these factors have much smaller effects on girls and women than proponents argue?  The associated time and resources devoted to addressing these problems will have little or no long-term effect on girls’ interest in or women’s participation in inorganic STEM fields.

If not implicit and subtle oppression, then what’s really going on?

We’ve recently found that countries renowned for gender equality show some of the largest sex differences in interest in and pursuit of STEM degrees, which is not only inconsistent with an oppression narrative, it is positive evidence against it.18 Consider that Finland excels in gender equality, its adolescent girls outperform boys in science, and it ranks near the top in European educational performance.19 With these high levels of educational performance and overall gender equality, Finland is poised to close the sex differences gap in STEM. Yet, Finland has one of the world’s largest sex differences in college degrees in STEM fields. Norway and Sweden, also leading in gender equality rankings, are not far behind. This is only the tip of the iceberg, as this general pattern of increasing sex differences with national increases in gender equality is found throughout the world.20

The recent uptick in interest in concepts such as stereotype threat, implicit bias, and microaggression may be a reaction to the low female STEM participation in highly developed nations. At one time, there were substantive social and educational impediments to women’s participation in these (and other) fields, but as explicit sexism and restricted educational opportunities faded into history, the sex differences (e.g., fewer women than men physicists) attributed to them should have faded as well. Some of them have even reversed, such that more women than men attend and graduate from college and women may now have structural advantages (e.g., hiring practices) in STEM fields.21 Even with these changes, many other sex differences remain or have become larger over time. The latter are serious problems for anyone with strong beliefs about purely or largely social influences on sex differences; if the obvious social causes have been addressed, then there must be other, more subtle oppressive factors afoot. This is where stereotype threat, implicit bias, microaggression and related concepts enter the oppression narrative.

We believe that with economic development and advances in human rights, including gender equality, people are better able to pursue their individual interests and in doing so more basic sex differences are more fully expressed.22 The differences in STEM are related in part to student’s personal and occupational interests and relative academic strengths. Sex differences in occupational interests are large, well-documented, and reflect a more basic sex difference in interest in things versus people.23 Men prefer occupations that involve working with things (e.g., engineering, mechanics) and abstract ideas (e.g., scientific theory) and women prefer working with and directly contributing to the wellbeing of others (e.g., physician, teacher). The sex difference in interest in people extends to a more general interest in living things, which would explain why women who are interested in science are much more likely to pursue a career in biology or veterinary medicine than computer science.24

Programs designed to steer women into inorganic STEM fields would in effect steer these same women away from the life sciences. Such programs would, in our opinion, only be justifiable if women are not provided a fair opportunity to pursue inorganic STEM fields (for which there is no good evidence). The main argument from gender activists is that inorganic STEM fields are a better choice for women either because these jobs lead to higher incomes or that there is a labor market demand for them. Both arguments are fundamentally capitalist and dehumanizing in the sense that considerations of personal interest are overridden by considerations of societal demand. This is ironic, given that the agenda arguing for more women in STEM seems most popular among left-leaning people.

In any event, on top of differences in career preferences, there are important and largely overlooked sex differences in relative strengths in reading, mathematics, and science.25 Students who are relatively better in reading-related areas (e.g., literature) than they are in science or mathematics, independent of their absolute level of performance are more likely to pursue college degrees in the humanities and enter non-science occupations. The reverse is true for students who are relatively better in science and mathematics than literature.26 This is where the results from Finland and elsewhere make sense. Although Finnish girls perform as well or better than Finnish boys in science, the gap is even larger in reading. The result is that more Finnish girls have relative advantages in reading than science. Most adolescent boys in contrast are relatively better at science or mathematics than reading, independent of their absolute level of performance. Individuals with this academic profile are likely to enter STEM areas, either as research scientists or technicians, and there are more boys than girls with this pattern throughout the world.25

We found that the sex difference in academic strengths explains part of the gap between the proportion of adolescent girls who have the absolute level of science and mathematical competencies needed to pursue a STEM degree in college and the proportion that actually obtain such a degree. The other contributing factor is their stated interest in and enjoyment of science. At the same time, there was still a gap between the number of women capable of obtaining a STEM degree and those that actually completed one. There was no sex difference in the math and science abilities needed to pursue a STEM degree. This leaves some room for stereotype threat, implicit bias, or related factors, but their relative contribution (assuming they exist) would be small.

A better route to increasing women’s participation in STEM might be to focus on the substantive numbers of girls with relatively higher science or mathematics than reading achievement; 24 percent of Finnish girls, for instance. These girls have the academic profile that is common in boys that pursue STEM-related careers but fewer of these girls than boys actually pursue them. It seems to us that interventions focused on this group of girls (e.g., individual mentoring) holds much more promise for increasing the number of women in inorganic STEM professions than do currently vogue interventions that focus on purging the wider society of stereotypes, implicit bias, and microaggressions.

Adapted from a chapter for D. Allen and B. Howell, editors, Groupthink in Science: Greed, Pathological Altruism, Ideology, Competition, and Culture. New York: Springer.



1 Del Río, C., & Alonso-Villar, O. (2015). The evolution of occupational segregation in the United States, 1940–2010: Gains and losses of gender–race/ethnicity groups. Demography52(3), 967-988.
2 Spencer, S. J., Steele, C. M., & Quinn, D. M. (1999). Stereotype threat and women’s math performance. Journal of Experimental Social Psychology, 35, 4–28.
3 Greenwald, A. G., McGhee, D. E., & Schwartz, J. L. (1998). Measuring individual differences in implicit cognition: the implicit association test. Journal of Personality and Social Psychology74, 1464;  Greenwald, A. G., Poehlman, T. A., Uhlmann, E. L., & Banaji, M. R. (2009). Understanding and using the Implicit Association Test: III. Meta-analysis of predictive validity. Journal of Personality and Social Psychology, 97, 17-41.
4 Hill, C., Corbett, C., & St Rose, A. (2010). Why so few? Women in science, technology, engineering, and mathematics. Washington, DC: American Association of University Women.
5 Thiederman, S. (2015). 3 keys to defeating unconscious bias. San Diego, CA: Cross-Cultural Communications.
6 Walton, G. M., Logel, C., Peach, J. M., Spencer, S. J., & Zanna, M. P. (2015). Two brief interventions to mitigate a “chilly climate” transform women’s experience, relationships, and achievement in engineering. Journal of Educational Psychology107, 468-485.
7 Stoet, G., & Geary, D. C. (2012). Can stereotype threat explain the sex gap in mathematics performance and achievement?  Review of General Psychology, 16, 93-102.
8 Flore, P. C., & Wicherts, J. M. (2015). Does stereotype threat influence performance of girls in stereotyped domains? A meta-analysis. Journal of School Psychology53, 25-44.
9 Ganley, C. M., Mingle, L. A., Ryan, A. M., Ryan, K., Vasilyeva, M., & Perry, M. (2013). An examination of stereotype threat effects on girls’ mathematics performance. Developmental psychology49, 1886-1897.
10 Basford, T. E., Offermann, L. R., & Behrend, T. S. (2014). Do you see what I see? Perceptions of gender microaggressions in the workplace. Psychology of Women Quarterly38, 340-349.
11 Nosek, B. A., Banaji, M. R., & Greenwald, A. G. (2002). Harvesting implicit group attitudes and beliefs from a demonstration web site. Group Dynamics: Theory, Research, and Practice6, 101-115.
12 Greenwald, A. G., Banaji, M. R., & Nosek, B. A. (2015). Statistically small effects of the Implicit Association Test can have societally large effects. Journal of Personality and Social psychology108, 553-561;  Oswald, F. L., Mitchell, G., Blanton, H., Jaccard, J., & Tetlock, P. E. (2013). Predicting ethnic and racial discrimination: A meta-analysis of IAT criterion studies. Journal of Personality and Social Psychology105, 171-192;  Rothermund, K., & Wentura, D. (2004). Underlying processes in the implicit association test: Dissociating salience from associations. Journal of Experimental Psychology: General133, 139-165.
13 Blanton, H., Jaccard, J., Klick, J., Mellers, B., Mitchell, G., & Tetlock, P. E. (2009). Strong claims and weak evidence: Reassessing the predictive validity of the IAT. Journal of Applied Psychology94, 567.
14 Oswald, F. L., Mitchell, G., Blanton, H., Jaccard, J., & Tetlock, P. E. (2013). Predicting ethnic and racial discrimination: A meta-analysis of IAT criterion studies. Journal of Personality and Social Psychology105, 171-192;  Greenwald, A. G., Banaji, M. R., & Nosek, B. A. (2015). Statistically small effects of the Implicit Association Test can have societally large effects. Journal of Personality and Social psychology108, 553-561.
15 Miller, D. I., Eagly, A. H., & Linn, M. C. (2015). Women’s representation in science predicts national gender-science stereotypes: Evidence from 66 nations. Journal of Educational Psychology107, 631-644;  Sue, D. W. (2010). Microaggressions in everyday life: Race, gender, and sexual orientation. Hoboken, NJ: John Wiley & Sons.
16 Lilienfeld, S. O. (2017). Microaggressions: Strong claims, inadequate evidence Perspectives on Psychological Science12, 138-169.
17 Hicks, S. R. (2004). Explaining postmodernism: Skepticism and socialism from Rousseau to Foucault. Scholarly Publishing, Inc.
18 Stoet, G., & Geary, D. C. (2018). The gender equality paradox in STEM education. Psychological Science. Advanced online See also Pinker, S. (2008). The sexual paradox: Men, women and the real gender gap. New York, NY: Simon and Schuster.
19 World Economic Forum (2015). The Global Gender Gap Report 2015. Geneva, Switzerland: World Economic Forum; Programme for International Student Assessment, 2016;
20 This is not restricted to STEM fields. Many sex differences are larger in gender-equal countries. Lippa, R.A., Collaer, M.L, & Peters, M. (2010). Sex Differences in Mental Rotation and Line Angle Judgments Are Positively Associated with Gender Equality and Economic Development Across 53 Nations. Archives of Sexual Behavior, 39, 990-997.

21 Ceci, S. J., & Williams, W. M. (2015). Women have substantial advantage in STEM faculty hiring, except when competing against more-accomplished men. Frontiers in Psychology6. e1532; Williams, W. M., & Ceci, S. J. (2015). National hiring experiments reveal 2: 1 faculty preference for women on STEM tenure track. Proceedings of the National Academy of Sciences USA112, 5360-5365.
22 Geary, D. C. (2010). Male, female: The evolution of human sex differences (second ed)Washington, DC: American Psychological Association.
23 Su, R., Rounds. J., & Armstrong, P. I. (2009). Men and things, women and people. Psychological Bulletin, 135, 859-884.
24 Lofstedt, J. (2003). Gender and veterinary medicine. The Canadian Veterinary Journal44, 533-535.
25 Stoet, G., & Geary, D. C. (2015). Sex differences in academic achievement are not related to political, economic, or social equality. Intelligence, 48, 137-151.
26 Humphreys, L. G., Lubinski, D., & Yao, G. (1993). Utility of predicting group membership and the role of spatial visualization in becoming an engineer, physical scientist, or artist. Journal of Applied Psychology78, 250-261.


Diversity DebateScience / TechTop

David C. Geary and Gijsbert Stoet

David C. Geary is a Curators’ Distinguished Professor at the University of Missouri. Gijsbert Stoet is a Professor of psychology at the Department of Psychology, University of Essex.