Friday, May 04, 2007

More on Race and the NBA

Yesterday, Mike summarized the findings of an academic article that found that white NBA referees call fouls more frequently on black players than on non-black players. (The finding works in reverse, but not nearly as strongly.) The NBA's defense against this academic paper, according to the Times article, is that they commissioned an analysis that uses a proprietary dataset which found no evidence of discrimination. The mere fact that that's what the NBA wanted their study to find doesn't automatically mean that the study was biased, but I'm skeptical.

The academic article was based on publicly available data, which limited them to counting fouls called in a game and matching that to the race of the refereeing crew. Thus, they correlated fouls called on black and non-black players by referee crews that are all-black, 2 blacks, one black, and no blacks. The evidence of race-based calls is, thus, indirect. It would be nice if unconflicted researchers could look at a data set that is based on individual refs and their calls. The NBA claims that they cannot release the data that they collect on each ref because of privacy concerns -- the refs could be personally identified from the data. This strikes me as nonsense. Researchers in tax policy, for example, can avail themselves of a process by which they are given access to data on taxpayers that are detailed enough to reveal the identity of the taxpayer (corporate or individual) so long as the researchers maintain taxpayer confidentiality in their use of the data and so long as the published results are sufficiently aggregated to conceal taxpayer identities.

Clearly, the NBA -- which has known about this paper for over a year -- could have done something like this on an ad hoc basis. Why wouldn't they? From the discussions on ESPN yesterday, my take is that everyone is simply afraid of the results. The concern, of course, is that the NBA does not want to be labeled a racist organization. Given that the racial composition of the referee corps is overwhelmingly white (68%), while the players are overwhelmingly black (83% of minutes were played by blacks), their concern is perhaps understandable. Players and league officials alike were quick to say that they've never even suspected racism among officials, which is good to know but explicitly not the authors' point.

One might ask why the NBA should not want to find out whether calls are race-based, if for no other reason than to reconsider their recruitment and hiring strategies for referees. According to the authors of the paper, after all, the differences in calls can amount to a two-game difference in won-lost record over the course of a season. Ask this year's LA Clippers if that difference matters. My guess is that the league's ignorance-is-bliss policy could simply be based on the belief that implicit bias is almost inherently impossible for them to correct, precisely because it is unconscious and automatic. If they cannot fix it, even in good faith, maybe they'd rather not have to admit that it exists.