Why We're Wrong About Affirmative Action: Stereotypes, Testing and the 'Soft Bigotry of Low Expectations'

Earlier this month a divided Fifth Circuit Court of Appeals upheld the University of Texas' right to use race amongst its criteria for undergraduate admissions, however limited that right may be. While the decision will be viewed as a small victory for supporters of race-based affirmative action, there is little reason to believe that the widely held claim that black and Latino students enter selective universities as comparatively inferior students will not cease to rear its ugly head. It is a pervasive stereotype that minority students must face from matriculation to graduation, a stigma with undoubted adverse psychological and economic consequence that follows them well beyond higher education and into the labor market.

This claim, reasserted in a study published in 2012 in the Journal of Labor Economics by two Duke faculty members and a graduate student (Peter Arcidiacono, Esteban Aucejo, and Kenneth Spenner), anchors on a combination of black students' relative standardized-test scores and the authors' prior beliefs about the operation of affirmative action in admissions. It also should be noted that Arcidiacono has been a member of the research team for Project SEAPHE, a collective led by one of the chief architects of the war against affirmative action waged in California, UCLA Law School professor Richard Sander.

In their study, which was undertaken at the highly selective Duke University, the authors express concern about the pattern of students switching from majors in the natural sciences, engineering, and economics, which they view as more demanding majors, to majors in the humanities and social sciences, ostensibly less demanding majors. For black students, they argue, the tendency to switch from the "harder" majors to the "softer" majors is fully explained by the black students' weaker academic backgrounds. Thus the presence of black students with weaker academic backgrounds is attributable, in their view, to affirmative action in the admissions process.

It is true that at Duke University, black students' Scholastic Aptitude Test (SAT) scores are, on average, significantly lower than those of white students. For the cohort of students they examined, students entering Duke in 2001 and 2002, the mean SAT score for the math and verbal sections for white students was 1416, and for black students it was 1275. That constitutes a 130-point gap, although Duke's black students' scores were well above the national average for all students taking the SAT (about 1030), and substantially above the national average for black students taking the SAT (about 860). But much like the vast array of so-called "education reformers," the authors incorrectly assume that the SAT is a useful indicator of racial differences in preparation, and that sorting of black students by major is independent of stereotypical beliefs, which are based in part on SAT scores, held by the faculty themselves.

Students at Duke protested the study when it first received public attention, identifying it as not only misguided but offensive. However right their grievance may be, there is a much larger and more substantive danger than hurt feelings: The study was used in the anti-affirmative-action amicus brief brought before the Supreme Court for the 2013 Fisher v. University of Texas case. The plaintiff's complaint in Fisher charged that affirmative action leads to institutions of higher education lowering their admissions standards for targeted minorities.

We believe there are at least two ways to view affirmative action. One perspective, our view, is that it serves as a set of positive anti-discrimination measures designed to include persons in preferred positions of society from which they would otherwise have been excluded, despite their qualifications and merit.

An alternative view -- the view that informs the authors' study -- treats affirmative action as a compensatory measure that de facto gives extra points to candidates who otherwise would not have met the admissions standards. From this perspective, affirmative action invariably leads to the admission or inclusion of less-qualified candidates. To the extent that faculty share the second view, they will be predisposed to believe that black students are inferior or less prepared than their non-black peers.

In comparison, we observe that there is no compelling research that supports the view that black employees in the wider workforce who get hired via affirmative action are inferior. Quite the contrary, comprehensive studies of affirmative action and hiring by researchers like Major Coleman and Harry Holzer and David Neumark find no significant difference in work performance between employees who entered jobs via affirmative action and those who entered via traditional hiring practices.

But what about black students and entry into selective universities? There is a major factor -- a factor conveniently missing from anti-affirmative-action legal briefs -- that can depress their test scores: stereotype threat. Stereotype threat constitutes the existence of widely held negative beliefs about the cognitive abilities of a particular group of people with real-life, everyday consequences. In the case of high-stakes, performance-based assessments, whether or not a person in the stigmatized group actually shares those beliefs, stereotype threat can adversely affect their test performance. Students consciously or subconsciously do not want to confirm the stereotype, thereby putting too much pressure on themselves while taking high-stakes tests; they then underperform relative to their full capabilities. A disheartening irony of stereotype threat is that the students from the stigmatized group that are most motivated to perform well on the test are the ones most susceptible to underperform on the test.

Experimental studies undertaken nearly 20 years ago by social psychologists Claude Steele and Joshua Aronson found that black students at Stanford University under the stereotype-threat condition had scores 13-percent lower than comparable black students in the no-threat condition. This is a nontrivial effect; a 13-percent reduction means that a student who might have scored 1200 under non-threat conditions would score only 1044 under threat conditions. Steele argues that stereotype threat is necessarily activated under high-stakes testing conditions like taking the SAT, a test that influences college admissions.

While at least some significant portion of the entry-test-score gap can be explained by stereotype threat, it can also operate pervasively after students enter college. Beliefs about black students' alleged inferiority create conditions where there is little reason for faculty members to try to "keep" black students in fields where, presumably, they don't really "belong." In fact, at the first sign of a poor grade, or even before, faculty members and course advisers may discourage black students from majoring in science or math fields. Indeed, information about the racial test-score gap at their respective institution may reinforce faculty members' expectation that black students are not as academically competent as their non-black peers.

But the depressant effect of stereotype threat can disguise or hide black students' academic competency. A team of researchers at the University of Waterloo and Stanford University, led by Christine Logel and Gregory Walton , demonstrated in a paper that appeared in Educational Psychology in 2012 that when black students have the opportunity to be in a stereotype-safe environment, they outperform white students with comparable test scores upon admission. That's correct: Black students outperformed white students! Indeed, Walton, Steven Spencer, and Sam Erman have made the case that because test scores underestimate black students' academic potential, affirmative action actually performs the task of including stronger students in selective institutions of higher education; thus they aptly describe affirmative action as "affirmative meritocracy."

Applying this to the case of Duke University, black students' "latent ability" also is suppressed by the impact of stereotype threat on their educational experiences both prior to and after entry to the university. The assumption that, relative to non-black Duke students, they are academically deficient is incorrect; the problem is that Duke, and most other selective universities, is deficient in providing them with a stereotype-safe environment. Thus the use of a study that ignores this reality in any judicial case involving affirmative action -- despite this month's small (and probably temporary) victory for "affirmative meritocracy," is at least brazenly irresponsible.

Unfortunately, there is still more to consider. A remarkable study by economists Talia Bar and Asaf Zussman on the phenomenon of "partisan grading" finds that faculty in the natural sciences tend to assign lower grades to black and Latino students even after accounting for their actual SAT scores -- scores already depressed in part due to stereotype threat. Thus, momentum builds to push black and Latino students out of the natural sciences, a momentum linked to faculty beliefs that affirmative action has brought an inferior student -- a black student -- into their classroom. Discouraged from science and math, black students often shift to the humanities or social sciences other than economics. Given the circumstances, this is not an unwise decision. Bar and Zussman find that, generally, black and Latino students in these majors are graded with more equanimity.

But the circumstances should not be a given. If the absence of black and Latino scholars in science, technology, engineering, math and economics is genuinely a matter for concern, it would seem that universities would place a special interest in nurturing students from underrepresented groups in these fields to complete the major and to pursue doctoral work. Rather than promoting a self-fulfilling prophecy of anticipated failure by black and Latino students, faculty in these fields, together with college administrative support staff, need to learn how to minimize stereotype threat and promote black and Latino students' success.