Sunday, May 25, 2008

Biased, Sure, but WHY?

There is an interesting article over at Cognitive Daily about covert racism. It's a summary of some studies done by John Dovido (with colleagues) designed to demonstrate that, while overt racial bias has declined into near-insignificance since the 1930s (of course it still exists on the fringes, however), covert racial bias remains stubbornly hard to identify and ameliorate.

The results discussed work like this. Some white students were given hypothetical pairs of job applicants of varying levels of ability (by pair), one of the pair being shown by a photo to be black, the other white. When looking at a pair of CVs for highly qualified applicants, no evidence of bias was found: white students are, statistically speaking, roughly equally likely to hire either candidate. However, when the two candidates are nothing to write home about, white students are more likely to choose the white applicant. Why is that?

As evidence of racial bias, it's quite convincing, of course. But I think it is important to identify the basis and mechanism of the bias before committing ourselves to any solutions. In this case, I strongly suspect that the operative bias is what I might style "affirmative action bias."

In short, "affirmative action bias" is just what it sounds like: the assumption that a minority applicant is more likely to have been given help along the road to where he is than a white applicant - that is, that a smaller percentage of his on-paper "qualifications," if I may speak in this way, are accounted for by effort and ability than would be the case for a comparable white applicant. It's an employer's way of compensating for the playing field having been artificially tilted toward the black applicant.

So, the idea is that if you're in the top percentile no amount of legs up could've gotten you there. Even if you were helped along the way by an affirmative action scholarship (or a rich uncle with Good Ol' Boy Network connections, to be fair), your presence in the top tier shows that you have the requisite amount of ability or drive, and there is no need for an employer to compensate for anything, hence the lack of racial bias. If, however, you're in the average tier and black, your qualifications become more suspect. The employer tends to assume that while you look just as good on paper, a certain number of your internships might just have been awarded on the basis of the company's need to show racial balance for PR reasons rather than because you were really the man for the job. Or that your grades might be what they are not because you're a great student, but because someone gave you enough race-based scholarships that you were able to avoid working a part-time job and thus had more time on your hands in college.

Naturally the response will be that white kids have certain systemic advantages that compensate for these affirmative action helping hands - and that may or may not be true. It's an empirical question that I have never seen solved to my satisfaction, truth be told. But absent the empirical evidence in front of him, a white employer is, it seems to me, likely to fall back on his own gut feelings about the situation, and his own gut feelings are probably this: he knows a LOT of poor white people who never had anything handed to them, and comparatively few connected rich white kids. "White privilege," to the extent that it exists, benefits only a comparatively small percentage of whites. Affirmative action, however, is freely available to all blacks simply for the point of fact of having been born black. So the employer's gut instinct is likely to tell him that while there are indeed a handful of white snot-noses who will be worthless to his organization (having been handed their qualifications through relatives' connections) in his applicant pool, there are a comparatively greater number of blacks who may have padded resumes through affirmative action programs and are worth slightly less in real economic terms than their paper qualifications let on. To be sure, the "penalty" for hiring one of the well-connected white applicants is much more expensive than hiring one of the affirmative-actioned black applicants. This is so because for most affirmative action programs you still do have to meet some minimal qualifications, which is obviously not the case if you're playing "white privilege" connections. But the law of averages nevertheless tells the employer that hiring the white guy in the general case will compensate him for these duds since he assumes that the white applicant of the pair is generally more likely to have gotten where he is by effort and ability.

I'm not sure that this counts as "racism." Certainly it's "racial bias," but it's a rational kind of bias. Whether or not it's actually supported by the facts, it's a reasonable conclusion to draw, from the employer's point of view, based on the evidence that his life experiences have given him; fighting it with more affirmative action is therefore both unfair and unlikely to address the root issue in any case. This highlights the importance, I think, of going beyond simple matters of identifying racial bias to trying to understand where it comes from. If it is, in fact, a malevolent racism that causes white people to screen applicants in this way, then affirmative action programs become more appropriate. But if it's affirmative action programs themselves that are the root of the bias, then it's time to start talking about eliminating them and replacing them with more applicant-netural schemes. Things like the NFL's practice of putting restrictions on the racial makeup of the hiring pool rather than the hiring outcome spring to mind. The rule here is that at least one applicant for a coaching position has to be a minority. There is no requirement to hire or even give preference to this person, just to give him a fair interview. Apparently, it has been effective at "diversifying" the coaching rolls. On the college level, programs in which state universities agree to accept all applicants from their state within a certain grade range have been successful "diversifiers." And, naturally, there's the option no one talks about: handing college admissions screeners applications with all personal information replaced by an identifier number so that they have no notions at all what the background of the applicant might be. In any case, the point is that racial bias is not an "across-the-board" phenomenon. There are many potential flavors, each requiring its own type of solution. Narratives focused exclusively on "racism" are not always very helpful, and experiments that document bias without probing it aren't much better.