The students had no idea of the real purpose of the study they had volunteered for--it is, after all, standard operating procedure in psychology to keep subjects in the dark on that little point. (If volunteers know they're being studied for, say, whether they will help a blind child cross a busy street, it tends to skew their behavior.) So when 40 black and 40 white Princeton undergraduates volunteered to play mini-golf, the psychologists dissembled a bit. This is a test of "natural ability," Jeff Stone and his colleagues informed some of the kids. This is a test of "the ability to think strategically," they told others. Then the students--nongolfers all--played the course, one at a time. Among those told the test measured natural ability, black students scored, on average, more than four strokes better than whites. In the group told the test gauged strategic savvy, the white kids scored four strokes better, the researchers reported last year. "When people are reminded of a negative stereotype about themselves--'white men can't jump' or 'black men can't think'--it can adversely affect performance," says Stone, now at the University of Arizona.

Another group of students, 46 Asian-American female undergrads at Harvard, thought they were taking a tough, 12-question math test. Before one group attacked the advanced algebra, they answered written questions emphasizing ethnicity ("How many generations of your family have lived in America?"). Another group's questionnaire subtly reminded them of their gender ("Do you live on a co-ed or single-sex dorm floor?"). Women who took the math test after being reminded of their Asian heritage--and thus, it seems, the stereotype that Asians excel at math--scored highest, getting 54 percent right. The women whose questionnaire implicitly reminded them of the stereotype that, for girls, "math is hard," as Barbie infamously said, scored lowest, answering 43 percent correctly.

The power of stereotypes, scientists had long figured, lay in their ability to change the behavior of the person holding the stereotype. If you think women are ninnies ruled by hormonal swings, you don't name them CEO; if you think gays are pedophiles, you don't tap them to lead your Boy Scout troop. But five years ago Stanford University psychologist Claude Steele showed something else: it is the targets of a stereotype whose behavior is most powerfully affected by it. A stereotype that pervades the culture the way "ditzy blondes" and "forgetful seniors" do makes people painfully aware of how society views them--so painfully aware, in fact, that knowledge of the stereotype can affect how well they do on intellectual and other tasks. Now, with half a decade of additional research under their belts, psychologists are discovering the power of stereotypes not only over blacks, but over women, members of ethnic minorities and the elderly, too. And the research is shedding light on such enduring mysteries as why black kids, even those from middle-class families and good schools, often score lower than white kids on standardized tests.

In their seminal 1995 study, Steele and Joshua Aronson, now at New York Uni- versity, focused on how the threat posed by stereotypes affects African-Americans. They reasoned that whenever black students take on an intellectual task, like an SAT, they face the prospect of confirming widely held suspicions about their brainpower. This threat, the psychologists suspected, might interfere with performance. To test this hunch, Steele and Aronson gave 44 Stanford undergrads questions from the verbal part of the tough Graduate Record Exam. One group was asked, right before the test, to indicate their year in school, age, major and other information. The other group answered all that, as well as one final question: what is your race? The results were sobering. "Just listing their race undermined the black students' performance," says Steele, making them score significantly worse than blacks who did not note their race, and significantly worse than all whites. …