Allport's Legacy and the Situational Press of Stereotypes

Journal of Social Issues, Fall, 1999 by David M. Marx, Joseph L. Brown, Claude M. Steele

There are a number of points to emphasize about this reasoning. First, stereotype threat is a situational threat, not dependent on any internalized belief in the stereotype, and is potentially experienced by anyone who is a part of a group for which some negative stereotype exists. Second, stereotype threat may be more severe for students who are strongly identified with schooling than for less school-identified students. Students who report high levels of confidence and strong performance expectancies and are strongly school-identified should be especially disrupted by the prospect of being limited by stereotype-based treatment and judgment. Students who are less school-identified, the theory implies, should not be as threatened by this prospect since they care less about school, and because they are not strongly motivated to perform, may be little disrupted by stereotype threat.

Initial Research on Stereotype Threat

We have argued that stereotype threat is a situational pressure that can interfere with intellectual performance. An important implication of this reasoning is that if this situational pressure is removed, then the performance of negatively stereotyped group members should rise. C. M. Steele and Aronson (1995) tested this prediction. They reasoned that if a difficult standardized test was presented as diagnostic of verbal ability, the possibility of stereotype-based judgment would become salient to Black student test takers, since the negative stereotype about Blacks focuses on intellectual ability, but that this possibility would not become salient to Black test takers when the test was presented as nondiagnostic of ability. After adjusting for differences in skill levels (i.e., verbal SAT scores), it was found that Black Stanford University students solved fewer verbal problems than White students when the test was presented as diagnostic of ability. But they performed just as well as Whites when the same test was presented as nondiagnostic of ability. The same pattern of results was replicated in several subsequent experiments reported by C. M. Steele and Aronson (1995).

Similar studies have investigated the performance effects of negative stereotypes about women's math ability. Of particular importance were the experiments conducted by Spencer, Steele, and Quinn (1999). In these studies they recruited male and female students who were highly skilled in, and identified with, mathematics. The students then took a very difficult math exam composed of items from the GRE (Graduate Record Exam) subject test in math. They reasoned that the stereotype would be salient for female students taking such a test, but not on a difficult literature test (students were also selected for being strongly identified with their verbal skills). The results were consistent with their expectations. Women under-performed relative to equally qualified men on the math exam, but performed just as well as men on the literature exam. In a second experiment, women also performed just as well as men when the test was easier, presumably because an easier test made less relevant the possibility of stereotype -based judgments of their performance. Finally, Spencer and his colleagues showed that if the stereotype of gender inferiority was directly induced, such as by telling the students that the difficult math test they were about to take was one which had shown gender differences in the past, women, as expected, performed worse than equally qualified men. Yet--and this was the critical test of their hypothesis--when the same math test was presented as showing no gender differences in the past, that is, women test takers were not at risk of confirming the sex-limited stereotype about their math ability, they performed just as well as equally skilled men.