Affirmative action - cutting the knot of misinformation

Affirmative action continues to inspire fierce debates and repeated litigation. It is also frequently mischaracterized. In recent years there have been repeated challenges to the right of US universities to use race as a factor in their admissions policies. This has become an imbroglio, muddying John F. Kennedy’s original ambition to address racial disparity.

As the issue of the legality of affirmative action in college admissions wends its way through American courts, it is worth a look at the non-legalistic background. Discussions of affirmative action are often framed as a question of balancing a desire to use neutral and fair criteria in admissions, or hiring in the case of businesses, against a desire to achieve broad objectives such as diversity or rectifying historical wrongs. But this framing is misleading; admissions criteria are, inevitably, neither “neutral” nor “fair”.

Harvard, for instance, evaluates applicants on four dimensions: personal qualities, extracurricular activities, athletic abilities, and academic achievement. The subjective, non-neutral nature of “personal qualities” is obvious; the same is true of judgements on the merits of extracurricular activities. Consider the more objective, fair-seeming criterion of academic achievement. Grade point averages are not readily comparable across high schools, so that admissions committees make idiosyncratic adjustments. Partly to compensate, universities also use national standardized tests, such as the SATs.

But, while all students take the same SAT test, the test is hardly fair. Parents who can afford it, hire tutors for their children; businesses, including Advantage Testing and Princeton Review, offer costly courses whose sole purpose is to raise students' scores on standardized tests. As a result, these tests are biased, in a statistical sense, against students who cannot afford extra help.

There are statistical methods to check for such biases, but it is not a purely statistical matter. Rather, there is a confounding incentive effect. A student from a disadvantaged background is hard-pressed to compete with a more privileged student. Hence, the student has less reason to put in extra studying effort to try for a better college placement, resulting in a lower performance. This feedback loop makes it difficult to check for statistical bias, as I show in my paper “Color-Blind is not Color-Neutral: Testing Differences and Affirmative Action” Journal of Law, Economics, and Organizations. An affirmative action program can help rectify incentives.

Other factors also cloud the notion of neutral and fair assessments. Studies have shown that stereotype threat may cause minorities, and women, to underperform on diagnostic tests, relative to their true abilities; interviewers are subject to implicit and explicit biases. Several investigations have found that whites evaluate the performance of non-whites more harshly than they evaluate the performance of whites.

On the employment front, one audit study supplied Black males and white males with similar resumes and training, before sending them out to interview for advertised job openings. Despite their similarities, the Black candidates were offered jobs three times less often than the white candidates. This discrepancy can be interpreted in several ways. One interpretation is that while employers used one hiring standard for all applicants, the interviewers subjectively evaluated African-American applicants as having performed worse on the interviews than white ones. Put differently, African-Americans received lower marks than equally qualified whites on the interview test.

Just as academic scores do not properly capture all that a university values, employment criteria unavoidably reflect factors unrelated to job ability. As an example, following a bias suit in 1985, the New York City Police Department agreed to promote minority police officers with low scores on the sergeants exam. This settlement was reached after city officials determined that they could not establish that the exam was job related. Thus, even absent bias, rankings based on tests and other markers, are less informative than they are often portrayed.

Affirmative action is misunderstood. Appropriately implemented, it counters flaws in admissions and hiring processes, while providing better incentives to the disadvantaged. As a conceptual framework, it supports a national racial justice agenda that helps fight centuries-old discrimination -  in law and custom - which manifestly restricts the economic, political, and educational opportunities available to minorities. The continued survival of affirmative action policies remains important for the dismantling of institutional practices that limit the progress of marginalised groups.