Affirmative Action and the SAT

Racial diversity is a fascinating subject to me, especially that now (I think) the culture wars of the 90s have simmered down somewhat, and we may be able to discuss the issue with a degree of sanity. If only we could develop that capacity with regard to gender, or sexual orientation…

In any case, Sam Barr over at HPRgument has penned a thoughtful response to a recent Harvard Educational Review article on racial bias in the SAT, and its implications for affirmative action (a sympathetic Washington Post columnist’s take here.) Barr concludes that affirmative action “isn’t the privileging of ‘justice’ or ‘compensation’ over ‘merit.’ It’s part of the search for merit…” That phrase – “the search for merit” – has exactly the kind of ring to it that would befit an article title or journalist’s catchphrase. But as with all such phrases, it needs a bit of unpacking before it can do service in an argument.

One might initially imagine the SAT to be neutral with respect to issues of discrimination. It is an attempt to measure, impartially and with minimal uncertainty, the capacity of the test-taker to (for example) process information and use it to perform cognitive tasks. Affirmative action advocates can argue that certain groups are behind because of historic discrimination, and need opportunities to level the playing field – but they can’t say that there aren’t aggregate disparities in whatever it is the test measures. The HER piece, however, casts pretty fundamental doubt on the quality of the tests themselves (in this case, the SAT).

So how does this produce an argument for affirmative action, rather than just against testing? Well, Barr doesn’t make the leap from merit as evaluated by tests to merit as evaluated by, say, hiring committees, but he certainly could. His implicit point seems to be that whatever ways we come up with of evaluating people – tests or otherwise – will, on net, be biased against minority groups. At least, that’s the provocative point that I hope he’s making. And the implication of that would be the following: as long as we live in a “race-conscious society” (to use his words), we need to take race into account when evaluating people, since we can’t avoid bias. I wouldn’t use the word “race-conscious,” but rather just “racial.” That is, we can’t ignore race if it’s a real thing. Indeed, race matters – like institutions and ideas (which also, it turns out, have consequences). Affirmative action, then, is part of the search for real merit, beyond the limitations of our ability to measure it.

And in all of this, I totally support Barr. But I do want to clear a couple things up: most importantly, I think he improperly maligns his opponents’ views. Quoting himself from a month ago, he suggests that, “In order to justify assessments of “merit” where blacks and whites perform differently from one another, you have to assume that they perform differently because they actually are different and immutably so: because blacks are dumber, or less cut-out to be firefighters.” In other words, because the anti-affirmative-action position assumes that tests are impartial, it must therefore believe that certain minorities are intrinsically inferior.

That suggestion is silly. If you assume a fair test, you don’t conclude that black people are inherently less capable than white people, but that the specific black people who took the test are, on average, less capable than the specific white people who took the test. Note that this is a possible outcome whether or not the black and white populations at large have the same aggregate abilities. It’s also possible that circumstances of upbringing or history, rather than the test itself, have produced the difference in outcomes, and therefore (assuming a fair test, and statistical significance) the difference in ability. This explanation is at the core of the traditional liberal argument for affirmative action, which Barr eschews.

His stronger point, which I agree with, is that we should not start with the assumption that the test is fair, but with the assumption that racial groups are inherently equal in their capacities. I contend that the two assumptions are not mutually exclusive, but since I’m unwilling to defend the fairness of the test I won’t belabor myself.

The final point I want to make is this. Barr, quite correctly, asserts that we need to think about race. Colorblindness is, for the time being, unthinkable. But how do we think about race? How do we change our evaluations in light of the possibility that they have bias? The HER article’s methodology was to look at so-called “Differential Item Functioning,” or questions certain groups do significantly better on than other groups. Do we weight the scores of participants in each group toward questions favoring their group (as the author of the original 2003 study suggests)? But what if some individuals are not exposed to whatever factors have influenced their group – do we tailor each individual’s test to him or her? Clearly, that would defeat the purpose of testing. Should we, then, just give disadvantaged groups a boost? Given Barr’s vision of affirmative action as “the search for merit,” I think that’s a pretty crude tool. My personal preference would be to massively de-emphasize standardized testing as an evaluative instrument…but I really doubt that’s going to fly with the general population. So what do we do?

As fun as this discussion is, it hasn’t yet really begun.

Tags: Affirmative Action Sam Barr SAT