The Threat in the Air

Is fear of "stereotypes" really why blacks do poorly on tests?


People do badly on tests for different reasons. Some do badly because they're anxious or fear failure, others because they don't know how to answer the questions. How important are these factors? When it comes to the long-observed patterns of black-white differences on a range of academic measures, social science purports to know the answer. Journals and textbooks of psychology will tell you that the principal cause of black students' poor performance on tests is something called "stereotype threat." Black test-takers fall short because they're afraid that the results will be used to confirm negative views about their group's abilities. It follows that if some way can be found to dispel this "threat," group differences in scores will disappear.

Yet the belief that stereotype threat is the sole or even the chief cause of the differences is without foundation. In the main study in this area, done by Claude Steele and Joshua Aronson in 1995, black and white Stanford undergraduates were given a difficult test of verbal ability. The students of both races were divided into two groups. Half of both black and white students were told the test would be used to assess their groups' intellectual ability--the "stereotype threat condition." The black students who were tested under threat did worse than the black students who were not told this. White students performed the same, whether or not they were threatened.

The key to this study, and to its misuse, can be found in how the results were reported. The average incoming verbal SAT scores of the black Stanford students lagged about 40 points behind the white students in the experiment. In order to control for those academic disparities, the authors adjusted scores on the experimental tests to account for any background SAT score differences. Since the adjustment allowed them to compare students as if they were equally qualified, it's no surprise that black and white students were reported as achieving the same scores when the stereotype threat was removed.

But they did not in fact achieve the same scores. As noted by University of Minnesota psychologist Paul Sackett and his colleagues in the January issue of American Psychologist, the raw, unadjusted scores of African-American and white students in the Steele/Aronson paper actually "differed to about the degree that would be expected on the basis of differences in prior SAT scores." Although stereotype threat warnings widened the gap between black and white student scores somewhat, purging the threat did not close or even narrow the actual gap in scores on the experimental test.

The few subsequent studies by the original authors or other researchers either report similar modest findings, or show no racial stereotype threat effect at all. For example, a 2002 paper in the Journal of Applied Social Psychology showed that stereotype threat had no statistically significant effect on blacks' performance on a commonly used nonverbal test of cognitive ability. A 2003 paper in the journal Human Performance found no effect on undergraduates' performance on a job selection test. Finally, economics Nobel laureate James Heckman and his colleagues have noted that significant racial gaps in cognitive performance show up in preschool children, before low teacher expectations and fears about stereotyping would come into play.

Although the studies to date suggest that stereotype threat may sometimes depress minorities' academic performance, they do not show that stereotype threat is responsible for most--let alone all--of the test-score gap. They also provide no clue on a key question: In accounting for the gap, how important is stereotype threat relative to differences in students' true skill, knowledge and problem-solving ability? Messrs. Steele and Aronson could have designed their study to get at this question, at least for their Stanford subjects, by reporting raw scores on the experimental verbal test without adjusting for prior SATs. This would have shown a significant baseline gap between black and white students in the non-stereotype-threat condition and a modest increase in that gap (due to somewhat lower scores by blacks) under stereotype threat. Adjusting for prior SAT scores--which makes it look as if stereotype threat were the only cause of any performance differences between black and white students--adds no useful information.

It is also raises the question of whether stereotype threat has any effect on standardized test scores in the real world. For their decision to control for background SAT scores to make sense, Messrs. Steele and Aronson must assume the SAT is a baseline measure of "true" ability. Indeed, they state that "it is unlikely that stereotype threat had much differential effect on the SATs of our black and white participants." But then what about the almost 200-point difference between the SAT scores of blacks and whites in the population overall? Does stereotype threat affect those SAT scores or not? If not, how can stereotype threat explain SAT score gaps? The authors can't have it both ways.

Lack of evidence and grave methodological defects haven't prevented the stereotype threat industry from taking off. Distortions are now pervasive. According to a survey by Mr. Sackett and his colleagues, 10 of 11 references to the threat in scientific journals, more than half the descriptions in psychology textbooks and 14 of 16 discussions in the media incorrectly state that racial differences in academic performance disappear when stereotype threat is eliminated. In this vein, a recent New York Times article on stereotype threat and the racial test-score gap declared that "performance is psychological." A "Frontline" special falsely stated that blacks who believe that a standardized test was merely a research tool, rather than a gauge of their abilities, performed just as well as whites.

Article originally in the WSJ of Sunday, April 18, 2004 but now taken down (!)

Back to Table of contents for this site

Back to "Wicked Thoughts"