Women, Math, and Stereotype Threat

By Diane M. Quinn

Diane Quinn is an assistant professor of psychology at the University of Connecticut-Storrs. She
received her B.A. from the University of Virginia and her Ph.D. from the University of Michigan
(in 1999). STATUS editors first became aware of Dr. Quinn’s study via the June 10, 2001
Washington Post article, “Mind Over Math” by Richard Morin. A copy of this article can be
found in the American Astronomical Society’s (AAS) December 2002 newsletter.

January 2003

This article appeared in the Fall 2002 issue of The Gazette, the newsletter of the American Physical Society’s Committee on the Status of Women in Physics (APS/CSWP). It has been reprinted with permission.

For many years, social scientists have tried to explain the gender gap on standardized mathematical
tests. Explanations have ranged from biologically based to developmentally
socialized. For example, researchers have examined differences in brain formation and exposure
to neonatal hormones, as well as whether girls are less likely to be encouraged to experiment
with math and science outside of the classroom. I am not wholly disputing these or other related
possibilities; however, I would like to suggest that, when examining why the best and brightest
of women underperform on math tests or drop out of math related fields, the subtle effects of
cultural stereotypes have been largely overlooked.

Few would argue that the American culture abounds with stereotypes. When I ask students
in my undergraduate psychology classes to name stereotypes, they can spout ten to twenty stereotypes
with ease. One stereotype that we all know is that boys/men are better at math and science
domains, whereas girls/women are better at English and reading domains. These stereotypical
beliefs are transmitted throughout the culture via mass media, books, parents, peers, and teachers.

How might these negative stereotypes account for a gap between men and women on
tests of mathematical ability? My colleagues Steve Spencer, Claude Steele, and I believe the
answer lies in the interaction between cultural stereotypes and the test-taking situation, what
we call a “stereotype-threat” situation. Stereotype threat occurs when a person is in a
situation in which a negative stereotype about that person (or that person’s group) could be
applied to the person and used to judge the person’s behavior. In the case of gender and math,
imagine a boy and girl sitting down to take the SAT for the first time. They have equivalent
math experience. Taking the SAT is a tense, sometimes frustrating experience for both of
them. However, as the girl is taking the test she has an extra worry to contend with that the boy
does not: a stereotype that she, as a girl, has inferior math skills. As she experiences frustration
and difficulty with the problems, she has the burden of knowing that her difficulty could
be judged as proof of the veracity of the stereotype. The boy has none of these doubts
or thoughts to interrupt his performance. It is important to note that in this situation neither
the girl nor the boy has to believe that the stereotype is true. Stereotype threat is not an
explanation based on internalized inferiorization. Just the knowledge of the stereotype itself is
enough to affect performance in the situation. How do we know this occurs?

My colleagues and I have tested the stereotype- threat hypothesis in a series of studies. In
all of our experiments we bring university men and women matched for equivalent math backgrounds
and interest into the laboratory. In the first of these studies we simply gave participants
an easy or difficult math test. We found that women only performed worse than men on the
difficult math test. To demonstrate that it was the threat of the stereotype that caused this
underperformance, we gave a second group of men and women the same difficult math test. In
order to make stereotypes about math explicit, half of the participants were told that the test
had shown gender differences in the past. In order to eliminate a stereotype-based interpretation
of the situation, the other half of the participants were told that the test had been shown to be
gender-fair — that men and women performed equally on this test. In line with our predictions,
when the stereotype was not applicable to the situation, when men and women were simply
told that they were taking a gender-fair test, men and women performed equally on the test.
When told that the exact same test had shown gender differences in the past, women scored
lower on the test than men. Just a simple change in the situation — a different line in the
instructions — changed an outcome that many believed intractable. Notably, and perhaps more
ominously, we have also conducted studies where we have a condition in which we do not
mention gender at all — we simply describe the math test as a standardized test. In this situation,
women also score lower on the test than men, suggesting that standardized mathematical
testing situations are implicitly stereotype-threat situations. Follow-up research in our own and
other laboratories has replicated these findings and explicated some of the boundaries of
stereotype threat. Stereotype threat occurs most strongly for women who are highly identified
with math and are taking a test that is pushing the limit of their skills. When a test is easy or the
women no longer care about how they perform on the test, changing the stereotype relevance of
the situation is unlikely to affect performance.

We have found some provocative clues to how stereotype threat works to undermine
women’s performance. Stereotype-threat situations lead to both increased feelings of
anxiety and more cognitive activation of female stereotypes. Both anxiety and stereotype
activation have been linked to worse performance. When we look at what women and men are
actually doing when working on the difficult test, we found that women and men primarily
used the same strategies to solve the problems; however, women in stereotype-threat situations
were less likely to think of any way to solve a problem. That is, women were more likely to
“blank out” or “choke” on a problem when they were in a stereotype-threat condition. Thus
research results so far point to the following scenario: when women with a strong interest
and identification with math are in a situation in which their math skills could be negatively
judged, their performance is undermined by the cognitive activation of gender stereotypes
combined with some feelings of stress or anxiety.

Although more research is needed to fully delineate the stereotype-threat process, we do
know that women are not alone in being affected by negative stereotypes. Research on stereotype
threat has demonstrated its effect on African- Americans and Latinos in intellectual situations,
on the elderly in memory testing situation, and even on white men in sports situations.

What can be done about a cultural stereotype? Some might argue that if the stereotype is
“out there” in the culture, there is nothing that can be done to stop its effects. However, we are
not so pessimistic. In our studies we make very simple changes — for example, adding a line in
the instructions communicating that a test is gender-fair or non-diagnostic — that have a
dramatic effect. If girls and women encounter fewer situations in which they experience stereotype
threat, their increasing performance may one day break the ugly cycle of the stereotype
leading to poor performance and the poor performance in turn feeding the stereotype.

Back to January 2003 Contents

Back to STATUS Table of Contents