Comments on: Malcolm Gladwell on Juries and Race

By: pacatrue

pacatrue — Tue, 07 Oct 2008 21:07:55 +0000

I actually read up a bit in the FAQ and one thing they address in there is something like “if it keeps giving me the red X, does that mean this test is not for me?” The eventual answer they give is yes and that's what I was seeing. The test seems to revolve around reaction time measurements of some sort. They must be looking for either a facilitation or interference effect due to unconscious assumptions. However, I see no point in measuring RTs of things I think are wrong. If I think the pic looks happy and it's forcing me to say it's unhappy over and over, you aren't going to learn much of interest from whether I say it's unhappy faster with black faces than I do with white faces. The task the dependent measure is based on has been compromised. The reason is that the time it takes me to remember “oh, I'm supposed to say this guy looks unhappy” and click appropriately is easily greater than the interference or facilitation effect they are trying to measure.

By: GreenDreams

GreenDreams — Tue, 07 Oct 2008 15:48:33 +0000

pacatrue, try again. The point is that you can't trick your reflexes, and if you honestly try to make the association as quickly as possible it shows how you are hardwired. But it's telling that you say “forcing me to make a decision in a way I didn't want to.” Isn't that the point? You want to make the other association. We all do. Not even blacks can quickly associate good with African American as easily as good with white. It's disturbing. And not even women can “see” women in business, management and leadership as easily as they can associate men with those roles.

By: superdestroyer

superdestroyer — Tue, 07 Oct 2008 10:16:08 +0000

Is the author really implying that blacks and whites commit crimes at the same rate and that everything in the justice system is due to racial bias. Maybe the author should go back and look at the racial differences in being a crime victim.

Blacks are much more likely to be crime victims. It that also due to racial bias? If I remember, blacks are six times more likely to be the victim of a homocide than whites. I wonder how the explain that away or is it more likely that they just ignore data that does not support their position.

By: pacatrue

pacatrue — Tue, 07 Oct 2008 09:38:17 +0000

I started taking that test and quite halfway through. They seemed to be forcing me to make a decision in a way I didn't want to. I ended up killing it and writing a long methodological essay to them in the feedback area. In short, I don't know if I believe anything from that methodology at all.

By: JWindish

JWindish — Tue, 07 Oct 2008 04:33:09 +0000

I had not heard of project implicit before. Very interesting. At the risk of being to revealing, my results from their "featured task: U.S. elections 2008:

"Your data suggests a strong automatic preference for Barack Obama over John McCain

Your data suggests a moderately stronger automatic association between Christian and John McCain, compared to Barack Obama

Your data suggests a slightly stronger automatic association between American and Barack Obama, compared to John McCain"

By: GreenDreams

GreenDreams — Tue, 07 Oct 2008 03:57:51 +0000

Thanks for the post, Joe. After reading Blink, I started directing people here to the Harvard implicit associations test, to demonstrate to themselves that we have racial and gender biases that we cannot easily overcome. It's why we need programs, such as affirmative action, to overcome that bias. Here it is again. http://www.implicit.harvard.edu

By: PattonGuy

PattonGuy — Tue, 07 Oct 2008 03:40:03 +0000

I wonder how constitutional that would be…but it's an interesting thought. I'm not sure, though, how much the analogy is applicable. The comparison of music auditions with criminal trials seems a bit of a faulty analogy to me. With a trial, despite all of the efforts to prevent the jury members from knowing anything about the case besides what they hear at the trial itself, people will know at least some facts. This would be especially bad in prominent cases, such as, to use a local (Georgia) example, the Brian Nichols case. That story has been covered to hell and gone, and everyone who was selected for the jury knows his race. I suppose if you could somehow anonymize the defendant- but that might preclude character witnesses…I'd go on, but I don't actually know anything about law and I have a paper to work on. It's an interesting idea, though.