A U G U S T   1 9 9 9

UTTING this idea to the test, Joseph Brown and I asked, How can the usual detrimental effect of stereotype threat on the standardized-test performance of these students be reduced? By strengthening students' expectations and confidence, or by strengthening their trust that they are not at risk of being judged on the basis of stereotypes? In the ensuing experiment we strengthened or weakened participants' confidence in their verbal skills, by arranging for them to have either an impressive success or an impressive failure on a test of verbal skills, just before they took the same difficult verbal test we had used in our earlier research. When the second test was presented as a test of ability, the boosting or weakening of confidence in their verbal skills had no effect on performance: black participants performed less well than equally skilled white participants. What does this say about the commonsense idea that black students' academic problems are rooted in lack of self-confidence?

What did raise the level of black students' performance to that of equally qualified whites was reducing stereotype threat -- in this case by explicitly presenting the test as racially fair. When this was done, blacks performed at the same high level as whites even if their self-confidence had been weakened by a prior failure.

These results suggest something that I think has not been made clear elsewhere: when strong black students sit down to take a difficult standardized test, the extra apprehension they feel in comparison with whites is less about their own ability than it is about having to perform on a test and in a situation that may be primed to treat them stereotypically. We discovered the extent of this apprehension when we tried to develop procedures that would make our black participants see the test as "race-fair." It wasn't easy. African-Americans have endured so much bad press about test scores for so long that, in our experience, they are instinctively wary about the tests' fairness. We were able to convince them that our test was race-fair only when we implied that the research generating the test had been done by blacks. When they felt trust, they performed well regardless of whether we had weakened their self-confidence beforehand. And when they didn't feel trust, no amount of bolstering of self-confidence helped.

Policies for helping black students rest in significant part on assumptions about their psychology. As noted, they are typically assumed to lack confidence, which spawns a policy of confidence-building. This may be useful for students at the academic rearguard of the group. But the psychology of the academic vanguard appears different -- underperformance appears to be rooted less in self-doubt than in social mistrust.

Education policy relevant to non-Asian minorities might fruitfully shift its focus toward fostering racial trust in the schooling situation -- at least among students who come to school with good skills and high expectations. But how should this be done? Without particulars this conclusion can fade into banality, suggesting, as Alan Ryan has wryly put it in Liberal Anxieties and Liberal Education, that these students "will hardly be able to work at all unless everyone else exercises the utmost sensitivity to [their] anxieties." Sensitivity is nice, but it is an awful lot to expect, and even then, would it instill trust?

That is exactly what Geoffrey Cohen, Lee Ross, and I wondered as we took up the question of how a teacher or a mentor could give critical feedback across the "racial divide" and have that feedback be trusted. We reasoned that an answer to this question might yield insights about how to instill trust more broadly in the schooling environment. Cohen's hunch was that niceness alone wouldn't be enough. But the first question had to be whether there was in fact a racial divide between teachers and students, especially in the elite college environment in which we worked.

We set up a simple experiment. Cohen asked black and white Stanford students one at a time to write essays about their favorite teachers, for possible publication in a journal on teaching. They were asked to return several days later for feedback on their essays. Before each student left the first writing session, Cohen put a Polaroid snapshot of the student on top of his or her essay. His ostensible purpose was to publish the picture if the essay was published. His real purpose was to let the essay writers know that the evaluator of their writing would be aware of their race. When they returned days later, they were given constructive but critical feedback. We looked at whether different ways of giving this feedback engendered different degrees of trust in it.

We found that neither straight feedback nor feedback preceded by the "niceness" of a cushioning statement ("There were many good things about your essay") was trusted by black students. They saw these criticisms as probably biased, and they were less motivated than white students to improve their essays. White students took the criticism at face value -- even as an indication of interest in them. Black students, however, faced a different meaning: the "ambiguating" possibility that the criticism was motivated by negative stereotypes about their group as much as by the work itself. Herein lies the power of race to make one's world insecure -- quite apart from whatever actual discrimination one may experience.

But this experiment also revealed a way to be critical across the racial divide: tell the students that you are using high standards (this signals that the criticism reflects standards rather than race), and that your reading of their essays leads you to believe that they can meet those standards (this signals that you do not view them stereotypically). This shouldn't be faked. High standards, at least in a relative sense, should be an inherent part of teaching, and critical feedback should be given in the belief that the recipient can reach those standards. These things go without saying for many students. But they have to be made explicit for students under stereotype threat. The good news of this study is that when they aremade explicit, the students trust and respond to criticism. Black students who got this kind of feedback saw it as unbiased and were motivated to take their essays home and work on them even though this was not a class for credit. They were more motivated than any other group of students in the study -- as if this combination of high standards and assurance was like water on parched land, a much needed but seldom received balm.

HERE is, of course, another explanation for why black college students haven't fared well on predominantly white campuses: they aren't prepared for the competition. This has become an assumption of those who oppose affirmative action in college admissions. Racial preference, the argument goes, brings black students onto campuses where they simply aren't prepared to compete.

"The Tests and the 'Brightest': How Fair Are the College Boards?", by James Fallows (February, 1980)
Do the SATs really discover the best and the brightest? Or do they chiefly identify the richest and the most expensively educated?

First, black students have better skills than the gap suggests. Most of the gap exists because the proportion of blacks with very high SAT scores is smaller than the corresponding proportions of whites and Asians. Thus when each group's scores are averaged, the black average will be lower than the white and Asian averages. This would be true even if the same admissions cut-off score were used for each group -- even if, for example, affirmative action were eliminated entirely. Why a smaller proportion of blacks have very high scores is, of course, a complex question with multiple answers, involving, among other things, the effects of race on educational access and experience as well as the processes dwelt on in this article. The point, though, is that blacks' test-score deficits are taken as a sign of underpreparation, whereas in fact virtually all black students on a given campus have tested skills within the same range as the tested skills of other students on the campus.

In any case, the skills and preparation measured by these tests also turn out not to be good determinants of college success. As the makers of the SAT themselves tell us, although this test is among the best of its kind, it measures only about 18 percent of the skills that influence first-year grades, and even less of what influences subsequent grades, graduation rates, and professional success.

Indulge a basketball analogy that my colleagues Jay Rosner and Lee Ross and I have developed. Suppose that you were obliged to select a basketball team on the basis of how many of ten free throws a player makes. You'd regret having to select players on the basis of a single criterion. You'd know that free-throw shooting involves only a few of the skills that go into basketball -- and, worse, you'd know that you'd never pick a Shaquille O'Neal.

You'd also wonder how to interpret a player's score. If he made ten out of ten or zero out of ten, you'd be fairly confident about making a judgment. But what about the kid who makes five, six, or seven? Middling scores like these could be influenced by many things other than underlying potential for free-throw shooting or basketball playing. How much practice was involved? Was the kid having a good or a bad day? Roughly the same is true, I suggest, for standardized-test scores. Are they inflated by middle-class advantages such as prep courses, private schools, and tours of European cathedrals? Are they deflated by race-linked experiences such as social segregation and being consistently assigned to the lower tracks in school?

In sum, black college students are not as underprepared in academic skills as their group score deficit is taken to suggest. The deficit can appear large, but it is not likely to be the sole cause of the troubles they have once they get on campus.

Showing the insufficiency of one cause, of course, does not prove the sufficiency of another. My colleagues and I believed that our laboratory experiments had brought to light an overlooked cause of poor college performance among non-Asian minorities: the threat to social trust brought about by the stereotypes of the larger society. But to know the real-life importance of this threat would require testing in situ, in the buzz of everyday life.

To this end Steven Spencer, Richard Nisbett, Kent Harber, Mary Hummel, and I undertook a program aimed at incoming first-year students at the University of Michigan. Like virtually all other institutions of higher learning, Michigan had evidence of black students' underachievement. Our mission was clear: to see if we could improve their achievement by focusing on their transition into college life.

We also wanted to see how little we could get away with -- that is, to develop a program that would succeed broadly without special efforts. The program (which started in 1991 and is ongoing) created a racially integrated "living and learning" community in a 250-student wing of a large dormitory. It focused students on academic work (through weekly "challenge" workshops), provided an outlet for discussing the personal side of college life (through weekly rap sessions), and affirmed the students' abilities (through, for example, reminding them that their admission was a vote of confidence). The program lasted just one semester, although most students remained in the dormitory wing for the rest of their first year.

Still, it worked: it gave black students a significant academic jump start. Those in the program (about 15 percent of the entering class) got better first-year grades than black students outside the program, even after controlling for differences between these groups in the skills with which they entered college. Equally important, the program greatly reduced underperformance: black students in the program got first-year grades almost as high as those of white students in the general Michigan population who entered with comparable test scores. This result signaled the achievement of an academic climate nearly as favorable to black students as to white students. And it was achieved through a concert of simple things that enabled black students to feel racially secure.

One tactic that worked surprisingly well was the weekly rap sessions -- black and white students talking to one another in an informal dormitory setting, over pizza, about the personal side of their new lives in college. Participation in these sessions reduced students' feelings of stereotype threat and improved grades. Why? Perhaps when members of one racial group hear members of another racial group express the same concerns they have, the concerns seem less racial. Students may also learn that racial and gender stereotypes are either less at play than they might have feared or don't reflect the worst-feared prejudicial intent. Talking at a personal level across group lines can thus build trust in the larger campus community. The racial segregation besetting most college campuses can block this experience, allowing mistrust to build where cross-group communication would discourage it.

Our research bears a practical message: even though the stereotypes held by the larger society may be difficult to change, it is possible to create niches in which negative stereotypes are not felt to apply. In specific classrooms, within specific programs, even in the climate of entire schools, it is possible to weaken a group's sense of being threatened by negative stereotypes, to allow its members a trust that would otherwise be difficult to sustain. Thus when schools try to decide how important black-white test-score gaps are in determining the fate of black students on their campuses, they should keep something in mind: for the greatest portion of black students -- those with strong academic identities -- the degree of racial trust they feel in their campus life, rather than a few ticks on a standardized test, may be the key to their success.