Saturday, April 22, 2006

Why Proper Controls are Important

Steve Sailer mentioned this study in a recent post, with some very insightful, if obvious (but that is what Sailer does, points out obvious things that most people seem to ignore).

Essentially, the article says that putting black people on a jury trying a black man accused of a crime, but without terribly good evidence, makes the white people on the jury more likely to vote to acquit, and thus helps black defendants get a fairer trial. (If he had just found that putting blacks on the jury means that the black jurors vote in favor of a black defendant, that would hardly be news).

I can't help but notice that the ways the juries voted are not broken down by all-white vs. 1/3 black juries:

In this study, one all-white jury reached a unanimous guilty verdict, 16 groups reached a unanimous not guilty verdict, and 12 groups did not reach a verdict within the one-hour time limit.

Yes, but other than the guilty verdict, which were all-white and which diverse?

In any case, as Sailer pointed out, this is a very poor study because it doesn't contain anywhere near enough controls. For one thing, there should have been all-black juries as well, and also, I would think, juries with two whites and four blacks. For another, there should have been cases where the black guy was clearly guilty, to see whether the diverse juries were actually mroe fair, or whether black people just don't like to vote to convict black people. Finally, there should have been white men accused as well (the victim's races, whatever they were, ought to be reversed as well so that the cases are equivalent), both obviously guilty and with not terribly great evidence, to see whether the diverse juries are fairer than all-white juries, or are just fairer toward black people (and likewise, whether diverse juries are fairer than all-black juries).

My gut feeling is that what we are actually seeing here is that people are more willing to give the benefit of the doubt to people of their own race (which would mean more likely to vote to acquit defendants of their own race or to believe victims of their own race). But I'd really like to see a study that tries to determine that.

That is all.

No comments: