When I recently watched the video recording of the panel debate from the Women in Secularism conference (which I posted here), a comment Rebecca Watson made about how gender stereotyping affects performance in mathematics and science caught my interest. She referred to a study mentioned in the book Delusions of Gender. I got the reference from her, and managed to get hold of the paper* through my university access.
This study, published in 2008 in the Journal of Applied Developmental Psychology, issue 29 (further detail in the footnote) looked at how stereotypes affect women’s performance in college-level mathematics tests. Even among women in science and mathematics studies, the effect of gender-stereotyping is significant.
The research shows that when stereotypes are not activated, or if they are nullified by other cues in the environment, girls and women perform better. When negative stereotypes are activated and left unchecked, they trigger a number of disruptive psychological processes that undermine test performance.
The argument has been made that women are under-represented in the science and mathematics professions because they are less capable than men due to innate biological differences.
Some background numbers referenced in the study: Male children and teenagers perform better than females in standardized mathematics tests . Examples from the paper are: 35% vs. 30% for fourth graders, 30% vs. 27% for eighth graders and 19% vs. 14% for twelfth graders. The numbers indicate who perform at or above proficiency. Women earn about half the doctoral degrees awarded in sciences. The numbers are 29% for physics, 24% in mathematics and 19% in engineering. I am assuming these are US numbers, but I have not looked up the references.
Although the paper doesn’t exclude the possibility of there being biological reasons for these numbers, as is indicated in other research , it is very clear by a wide range of research papers that socialization is a major factor. Research into stereotype threat has focused on college-level mathematics  and not generally looked at when these effects emerge. Children aware of race/ethnic stereotypes, for instance, perform more according to these than children who are unaware of them. This does not necessarily imply the same is the case for sex-based stereotyping, but studies show there likely is a cumulative effect resulting in undermining a young adult’s career choices and plans. Studies have found that stereotyping undermines female performance in mathematics from as early as sixth grade.
This study included 174 students enrolled in a “fast-paced calculus course”. 100 of the student reported being male, 57 female. The remaining 17 students were excluded from the final analysis since they did not report their gender. Extra credits was also awarded to ensure the students took the test seriously. A form with a short text describing the test was handed out with the test itself. It contained a paragraph worded in a way that is known to trigger stereotype threat. A random selection of the tests had an additional paragraph describing the test as having not previously been able to show a difference in performance based on gender. Attempting thus to nullify any stereotype threat for the women taking the test. Earlier research has tested for stereotype threat by reinforcing the stereotype that male students perform better instead .
The results were evaluated for 155 of the participants, and in the non-threat group, women outperformed men by a grade 3.60 to 2.60 out of 4.00 with a p-value** of .02. In the stereotype threat group women and men performed relatively equal, 3.13 vs. 3.08. The difference within the non-threat group was even larger for the Anglo-American subset of the larger group, 4.44 vs. 2.70, p-value < .02.
The test also ended with a questionnaire asking the students about how confident they were in their ability to score well on the test. Male students reported a significantly higher level of confidence, p-value < .01
They also tested, as a control, if there were any difference in performance between male and female students in the course as a whole. The result had a p-value of > .9, which indicates there were no significant difference in course performance.
In conclusion the authors write:
This study revealed that the effects of stereotype threat are not limited to the typical woman’s performance on general mathematics tests. Rather, even women at the upper ends of the ability distribution in college who opt to enroll in the most difficult math courses can be vulnerable to the effects of negative stereotypes. Specifically, framing a difficult calculus test as diagnostic of ability suppressed women’s calculus performance. However, ensuring women that the same diagnostic test was free of gender-bias reduced stereotype threat and unleashed their mathematics potential.
As for why women outperformed men in this specific test, the article offers this hypothesis:
Why would women outperform their male counterparts in this situation? It is likely that because women are more apt to self-select out of math and science fields early in their educational careers […], the women in advanced mathematics classes, such as those in our sample, comprise the ost motivated and prepared female mathematics students. Men, on the other hand, may be more likely to pursue math and science careers even if they are less prepared academically to do so.
- National Science Foundation. (2006). Science and Engineering Indicators, 2006. Retrieved August 1, 2006 from http://www.nsf.gov/sbe/srs/seind06
- Halpern, D. F. (1992). Sex differences in cognitive abilities. Hillsdale, NJ: Lawrence Erlbaum.
- Steele, C. M., Spencer, S., & Aronson, J. (2002). Contending with images of one’s group: The psychology of stereotype and social identity threat. In M. Zanna (Ed.), Advances in Experimental Social Psychology, 34 (pp. 379−440). San Diego: Academic Press.
- Spencer, S. J., Steele, C. M., & Quinn, D. (1999). Stereotype threat and women’s math performance. Journal of Experimental Social Psychology, 35, 4−28.
In statistics the p-value refers to how probable the null-hypothesis is to be contained in the result. The null-hypothesis here being that there is no gender difference. A p-value less than .05 is a good result in this type of studies. The null-hypothesis is rejected in all these results.