Defenders of affirmative action must work hard to explain away a serious problem: the tendency for the students admitted due to preferences to do relatively poorly in their coursework. When the class average in a calculus course is 85 but the average among the students who were preferentially admitted is 65, people start asking the question, "Wouldn't they have been better off at another school?"
An ingenious attempt to answer that question is the theory called "stereotype threat." One of its most famous exponents, Professor Claude Steele, recently spoke about it at Princeton. The argument is that students from "underrepresented minority groups" don't do as well as they really can because when taking exams they are gripped with fears that if they don't do well, that will make "their group" look bad, confirming the harmful stereotypes that "society" supposedly holds about them.
This theory is implausible. It asks us to believe that a minority student, while taking an exam, can focus well enough to get some questions right (let's say, 65 percent of the calculus questions) but some of the time he gets swept away with stereotype worries.
There is no reason to believe that occurs. College students are mature enough to stay focused on the academic work at hand. One who has learned the calculus material is perfectly able to do the problems without his mind fluttering off into a distraction over social stereotypes. Of course, when a student confronts a problem he doesn't know how to approach, he is apt to get nervous, but the nervousness is not over how his score might affect society, but how missing it will affect him. And that is true without regard to the student's race.
This is another of the many instances where we ought to apply Occam's Razor: the simpler explanation is more probably correct. When a student gets something wrong on an exam, it is more likely that he missed it because he didn't know it than that he did know it, but was so overcome with worries about society's stereotypes about his racial or ethnic group that he got it wrong.