Get ready for a brand new defense of affirmative action that you’ve never heard before: preferences are necessary to assure selection by merit. How can that be? Simple. Just rework Claude Steele’s theory of stereotype threat–that minorities do less well on tests than their abilities warrant out of fear that their performance will confirm negative stereotypes about their race or ethnicity. Greg Walton, an assistant professor of psychology at Stanford, and several co-authors have just dressed it up in a new study currently in press at the journal Social Issues and Policy Review. They plan to include their findings in an amicus brief supporting affirmative action in Fisher v. University of Texas.
According to a celebration of their work just published in Stanford News Today,
the authors believe they have come up with a gold-plated defense of
racial, ethnic, and even gender preferences that does not rely on the
promotion of diversity. The conflict between diversity and merit, Walton
et al. believe, is bogus. “Our argument is that you need affirmative
action to make meritocratic decisions – to get the best candidates.”
They see stereotype threat as the silver bullet that slays the dragon of the false dichotomy between merit and diversity.
“When people perform in standard school settings, they are often aware of negative stereotypes about their group,” Walton says. “Those stereotypes act like a psychological headwind – they cause people to perform worse. If you base your evaluation of candidates just on performance in settings that are biased, you end up discriminating….”
As a consequence, Walton says, “Grades and test scores assessed in standard school settings underestimate the intellectual ability of students from negatively stereotyped groups and their potential to perform well in future settings….”
This argument is developed at considerable length in the soon-to-be-published study, ambitiously titled Affirmative Meritocracy, which is currently available on Walton’s web site here. It may well be the most extensive compendium of stereotype threat research available, and scholars who have been critical of the theory–especially those involved with Fisher–will want to consider it carefully. As a defense of affirmative action, however, its central argument–“that, despite a lower SAT score, a Black student who took the SAT under the burden of stereotype threat could be more qualified for college and likely to perform better in college than a White student”–is quite familiar.
Why Not ‘Blind Grading’?
Claude Steele, the father of stereotype-threat theory and now dean of the school of education at Stanford, provided expert testimony in support of racial preference in the University of Michigan cases saying much the same thing. I criticized his argument at the time in a blog post on “Claude Steele, ‘Stereotype Threat,’ And Racial Preference,” noting that, since the theory holds that blacks don’t do as well on tests where their graders are aware of racial differences in performance on tests, it would seem to follow that insofar as stereotype threat is the problem race-blind admissions would be the solution.
All tests, it seemed to me, should be given and graded the way that “blind grading” is customarily done in law schools, with the grader unaware of the identity of the student whose blue book is being read. (I have commented before on the oddity of so many law professors who so avidly oppose colorblindness for everyone else practicing it themselves at the heart of their profession.) Steele, of course, recommended the opposite, increasing rather than eliminating color consciousness — discounting test results for blacks, thus reinforcing the notion (or confirming the stereotype) that blacks don’t do well on tests.
One often-noted problem with Walton’s stereotype threat theory that admissions tests “underestimate the subsequent level of performance of one group relative to that of another” is, bluntly, that they don’t. As Russell Nieli noted here on Minding The Campus in an excellent September 2010 essay summarizing much research, “blacks with the same entering SAT scores as whites and Asians earn substantially lower grades over their college careers and wind up with substantially lower class rankings.”
An Ingenious Solution
Moreover, If stereotype threat depresses minorities’ standardized test performance, as one study not cited by Walton states, “it would be expected to lead to underprediction, because affected students would perform better in college than their (depressed) test scores would indicate.” Actually, however, the SAT overpredicts for minorities. Their performance in college is lower than the test predicts, which requires the hypothesis “that stereotype threat has more effect on college grades than on admissions test performance, which seems contrary to Steele and Aronson’s (citation omitted) implication that standardized testing situations are particularly evocative of stereotype threat.” Moreover, as Nieli noted, “the problem of underperformance has been shown to extend far beyond the kinds of majors where test-taking is a central part of a student’s overall evaluation.”
Walton et al. offer an ingenious solution to this problem: performance in college itself is measured in classrooms that are polluted by achievement-depressing stereotypes. “Psychological threat,” they claim, citing Steele, “is the norm in academic environments.” Minority underperformance, they argue, can be mitigated or eliminated if institutions “create environments that minimize stereotype-related psychological threats.”
They support this argument by citing experiments conducted in “nonthreatening [laboratory] environments” in which “stereotyped students … perform better than non-stereotyped students with the same level of prior performance.” In laboratory conditions where “threat” has been reduced, they conclude, “stereotyped students do better than expected based on their prior performance. That prior performance was itself polluted by stereotype threat. It underestimated stereotyped students’ ability and potential in an environment without threat.”
In short, as presented by Walton et al., stereotype threat posits that objective tests (they don’t challenge the contents of the SAT etc.) underestimate the true ability and potential of minorities to perform in college, but they then claim that it also explains the undisputed if troubling fact of minority underperformance in college grades and graduation rates. As a theory that cannot be falsified by evidence from the world of higher education that actually exists and will no doubt continue to exist, it is like a rare, exotic plant that can thrive only in the hothouse of a purified academic laboratory.
Since it is far beyond the power of even the Supreme Court to require educational institutions to cleanse their “learning environments” of all traces of threatening stereotypes (entirely apart from the question of whether a stereotype that is accurate — blacks as a group do less well on standardized tests than other groups — can actually be labeled a stereotype), the Court should not be persuaded by the theory of stereotype threat to renew the license it granted in Grutter to continue discriminating on the basis of race and ethnicity.
Steele gave a public lecture at Illinois several years back. It had the flavor of a rival meeting, not a scholarly lecture. He put up a bunch of slides to prove his point, but each one lasted only a few seconds. I had the impression that things were not as they seemed, but you could not make heads or tails of the data in just a few seconds. This is not the way scholarly presentations usually work. Meanwhile, he kept on reassuring the sympathetic crowd that all differences in academic outcome resulted from stereotype threat. Nobody objected and the crowd ate it up.
John already mentioned that the SAT over predicts minority performance.
The new data coming out of Duke have some meaningful information on the impact of stereotype threat.
The authors not only checked the minorities’ performance relative to their incoming academic data in STEM courses, they also looked for legacies as they probably are not likely to get in on merit alone.
This minority vs legacy method showed that both groups were essentially the same in their grades and the rate of dropping out of STEM course when co-related with incoming academic data.
If the legacies also suffer stereotype threat, then stereotype threat can be noted.
It seems more likely that stereotype threat is not significant (the Duke Pres was happy that there was not a difference between minorities and legacy with same academic credentials).
How do we know that some Asian students don’t also suffer from “stereotype threat” — worrying that if they don’t do exceptionally well, they’ll be seen as “losers” who let “their group” down? Getting a B is sometimes called “the Asian F.”
It seems to me that this whole business of assuming that a student “really has more ability” than he or she demonstrates on standardized tests because psychological stresses mysteriously get in the way of answering some questions correctly is just groundless speculation.