Stereotype Threat Among Schoolgirls in Quasi-Ordinary Classroom
Circumstances
Pascal Huguet
Centre National de la Recherche Scientifique and Aix-Marseille
University
Isabelle Re´gner
Centre National de la Recherche Scientifique and University
Toulouse-Le-Mirail
There is ample evidence today in the stereotype threat literature that women and girls are influenced by
gender-stereotyped expectations on standardized math tests. Despite its high relevance to education, this
phenomenon has not received much attention in school settings. The present studies offer the 1st evidence
to date indicating that middle school girls exhibit a performance deficit in quasi-ordinary classroom
circumstances when they are simply led to believe that the task at hand measures mathematical skills.
This deficit occurred in girls working alone or in mixed-gender groups (i.e., presence of regular
classmates) but not in same-gender groups (i.e., presence of only same-gender classmates). Compared
with the mixed-gender groups, the same-gender groups were also associated for girls in the stereotype
threat condition with greater accessibility of positive role models (i.e., female classmates who excel in
math), at the expense of both stereotypic in-group and out-group members (i.e., low-math-achievement
girls and high-math-achievement boys). Finally, the greater accessibility of positive role models mediated
the impact of the activated stereotype on girls’ performance, exactly as one would expect from C. M.
Steele’s (1997) stereotype threat theory. Taken together, these findings clearly show that reducing
stereotype threat in the classroom is a crucial challenge for both scientists and teachers.
Keywords: stereotype threat, schoolgirls, classroom, visual memory, ROCF task
Women all over the world have participated (as mathematicians,
physicists, astronomers, and so on) in unraveling the secrets of
nature. Yet they remain underrepresented in mathematically inten-
sive disciplines and careers such as the natural and physical
sciences and engineering. In Western Europe in 2003, for example,
35.7% of university graduates in mathematics, natural and physical
sciences, and computer science were women (averaged over the 15
member states hereafter referred to as EU-15; see European Re-
search Council, 2003). Similarly, among engineering graduates,
women were in the minority (20.2%) across the EU-15. Women
outnumbered men only in the field of education and were more or
less equal in number of graduates to men in fields such as the
humanities and the arts, and health and social services. Women
were underrepresented in science-related activities and careers as
well, whether among senior university staff (34%), as members of
scientific boards (27%), and within the business sector (15%).
Finally, women applicants from EU-15 were slightly but consis-
tently less successful than men in receiving funding for their
scientific projects. The fate of women scientists in the United
States is roughly the same (see National Science Foundation,
2004). In 2001, they accounted for more than half of all graduate
students in science fields like psychology (74%), biology (54%),
and the social sciences (52%), but they were in the minority in
mathematics (35%), physical sciences (30%), computer science
(30%), and engineering (20%). Women made up about 26% of
employed science or engineering doctorate holders in 2001. Thus,
men still predominate in scientific institutions both in the United
States and abroad.
Mathematics has been identified as the critical filter that pre-
vents women from gaining access to the hard sciences and related
occupations (Sells, 1973). In line with this, the existence of gender
differences favoring men on standardized math tests (especially
word problems and geometry items from the math portion of the
SAT) has frequently been a topic of debate.
GENDER DIFFERENCES ON STANDARDIZED
MATH TESTS: FROM BIOLOGY TO
SOCIALIZATION
Relying on very large samples (thousands) of intellectually
talented adolescents (12- to 14-year-olds), Benbow and Stanley
(1980) reported that boys outperformed girls by about half a
standard deviation (0.40) on the math portion of the SAT. Al-
though they attributed this difference to “superior male mathemat-
ical ability, which may in turn be related to greater male ability in
spatial tasks” (p. 1264), they also noted that this superiority was
“probably an expression of a combination of both endogenous and
exogenous variables” (ibid). Benbow and Stanley (1983) later
concluded that boys predominate in the highest ranges of mathe-
matical reasoning ability before they enter adolescence (with a
male-to-female ratio of 13:1 in the group that scored above 700).
Pascal Huguet, Cognitive Psychology Laboratory, UMR CNRS 6146,
Centre National de la Recherche Scientifique; and Aix-Marseille Univer-
sity, Marseille, France. Isabelle Re´gner, Social Psychology Laboratory,
UMR CNRS 5551, Centre National de la Recherche Scientifique; and
University Toulouse-Le-Mirail, Toulouse, France.
This research was funded by Grant JC6082 from the French Ministry of
Higher Education to both authors.
Correspondence concerning this article should be addressed to Pascal
Huguet, Universite´ Aix-Marseille 1, 3 Place Victor Hugo, Case D, 13331
Marseille Cedex 3, France. E-mail: huguet@up.univ-mrs.fr
Journal of Educational Psychology Copyright 2007 by the American Psychological Association
2007, Vol. 99, No. 3, 545–560 0022-0663/07/$12.00 DOI: 10.1037/0022-0663.99.3.545
545

Although they again acknowledged that the “reasons for this sex
difference are unclear” (p. 1031), Benbow and Stanley (1983)
suggested sexual differentiation of the brain as a possible candi-
date (relying on Guy & McEwen, 1980; see also Benbow, 1988;
Benbow & Benbow, 1984).
1
Eccles and Jacobs (1986) challenged this conclusion. Gender
differences on standardized math tests, they claimed, are mostly
rooted in stereotypic gender-role beliefs that parents (and teachers)
communicate to their children (and students) on a daily basis,
resulting in differential expectations, confidence, and attitudes
toward math in boys and girls (see also Eccles, Adler, & Meece,
1984; Eccles, Jacobs, & Harold, 1990; Jacobs & Eccles, 1992).
Consistent with this, Eccles et al. (1990) showed that parents’
beliefs about their daughter’s versus son’s math abilities were
related subsequently to their child’s math self-efficacy, identifica-
tion with math, and math performance.
Hyde, Fennema, and Lamon (1990) found that gender differ-
ences favoring boys on standardized math tests did not emerge
until the high school years. Their meta-analytic findings (100
studies, 254 independent effect sizes) also helped clarify the over-
all picture. For example, girls were superior to boys in arithmetic,
and there were no significant gender differences in the understand-
ing of mathematical concepts per se. Likewise, using meta-analytic
findings from the 1970s as a baseline, Hyde et al. showed that the
magnitude of gender differences varied over time (see also Fein-
gold, 1988), which is hard to accommodate with any biological
account (for a similar argument regarding spatial math ability, see
Halpern, 1992). As reported by Hyde et al., however, gender
differences favoring men grew larger with increasingly selective
samples and were the largest for highly selective samples and for
samples of gifted persons (no significant gender differences were
found among adults or children taken from the general population).
Hedges and Nowell (1995) showed that the male advantage
among the higher ability samples was not an artifact caused by
biased sample selection and reported a higher proportion of boys at
both extremes of the distribution.
2
More evident among the better
high school students, the male advantage is in fact minimal on easy
test items but increases as the items become more difficult, even
when gender differences in variability are controlled (Penner,
2003). Combined with the fact that spatial aptitude (i.e., mental
rotation of three-dimensional objects) mediates gender differences
in math (Casey, Nuttall, & Pezaris, 1997; see also Benbow, 1988),
this observation suggests that the gender differences in math
ability, if any, only show up on difficult material.
STEREOTYPE THREAT AS AN ALTERNATIVE
EXPLANATION TO BIOLOGY AND
SOCIALIZATION
However, it is precisely in this situation (i.e., when they are
faced with difficult math tests) that women who are good at math
may feel threatened by the possibility that their performance will
confirm—to others, to themselves, or both—the negative stereo-
type about their gender’s math abilities (Spencer, Steele, & Quinn,
1999; Steele, 1997). This threat unfortunately leads to poorer
performance and thus produces the expected negative outcome.
In Spencer et al.’s (1999) studies, for example, women with high
math ability performed less well (than equally qualified men) on
difficult math tests both when they were told that the test produced
gender differences and when that information was not given, but
performed as well as men when told that no gender differences had
been found (for similar findings, see Johns, Schmader, & Martens,
2005; Keller & Dauenheimer, 2003; O’Brien & Crandall, 2003;
Quinn & Spencer, 2001; Sekaquaptewa & Thompson, 2003; for a
review, see Ben-Zeev, Duncan, & Forbes, 2005). The very fact that
falsifying the gender stereotype about math not only reduced the
male advantage but eliminated it altogether runs counter to any
biological account of gender differences in this domain. In Cadinu,
Maass, Rosabianca, and Kiesner’s (2005) study, women who were
told that the math test they were about to take produced gender
differences engaged in negative math-related thoughts (e.g.,
“These exercises are too difficult for me”) that were associated
with poorer performance (compared to women who were told that
no gender differences had been found). The negative thoughts
mediated the impact of the activated stereotype on math perfor-
mance, exactly as one would expect from Steele’s (1997) stereo-
type threat theory (STT; see also Steele & Aronson, 1995).
According to STT, women and minority-group members are
expected to experience additional tension—over and above that
associated, for most people, with taking difficult tests—because
they are preoccupied by fears of confirming a negative stereotype.
From a purely cognitive point of view, such a preoccupation might
reduce working memory capacity, which is critical to performing
well on complex intellectual tasks (Kane et al., 2004). Schmader
and Johns (2003) showed that stereotype threat was indeed asso-
ciated with a lower working memory capacity (compared with a
no-threat condition), which in turn led to lower math performance
in women.
The fact that gender differences favoring men on standardized
math tests are confined to the higher ability samples is also clearly
consistent with STT. As noted by Steele (1997), susceptibility to
stereotype threat derives not from internal doubts about one’s
ability based on one’s history of failure and/or the internalization
of the stereotype under the influence of socialization, but from
one’s identification with the critical domain and the resulting
concern about being stereotyped in that domain. To the extent that
women who excel in math identify strongly with this domain—in
the sense that they perceive math as self-relevant—stereotype
threat is expected to be especially prominent in women from
higher ability samples. One may argue that the persistent presence
of women in these selective samples is inconsistent with STT (as
well as with biological accounts). However, if STT is supported,
then women who excel in math would do even better in stereotype-
free environments. Having to face stereotype threat might, over
1
According to Benbow (1988), at least three physiological factors are
relevant to understanding gender differences in mathematical reasoning
ability, namely (a) left-handedness and (b) symptomatic atopic disease
(allergies), which may be related to bihemispheric representation of cog-
nitive functions or to the impact of prenatal testosterone exposure; and (c)
myopia. As noted by Benbow, “these physiological factors, especially
prenatal testosterone exposure, lend credence to the view that sex differ-
ences in extremely high mathematical reasoning ability may be, in part,
physiologically determined” (p. 180).
2
The National Science Foundation (2006) recently released a report
essentially declaring that the gender gap on standardized math tests has
disappeared, but the related statistics did not focus specifically on highly
selective samples.
546
HUGUET AND RE
´
GNER