I've done a few of the tests at Project Implicit, and each time the results have been inconclusive or I seem to show no preference. I've been skeptical of their tests [but haven't taken the time to deconstruct them], seeing as how I'm pretty much ambidextrous, but most of the population, being pretty much right-handed, also seem to get measured as having implicit biases. So, what I'd like to see is their results broken down by handedness of the participants.
However, like I say, until recently I didn't care enough to do something about it. Then I saw this:
Now, I don't know Nicholas Kristof from Eve's house cat, but I would have bet he's one of the least racist persons on the face of the planet. So I decided to take a closer look at this 'test.'
It's another one where I thought handedness might figure into the result, so after going through the test the first time as instructed --- right hand shoots, left hand holsters the gun --- I went through it a couple more times: shooting all 100 the next time through, and not-shooting all 100 the third time through. Both of those two trials, instead of reacting as quickly as possible, I did two things: looked only at the hands [no faces], and waited until I was sure what the object was before hitting the key.
All 3 times I had roughly the same results as Kristof, but after studying the images a little more closely, I'm convinced that the designers of that test need to re-evaluate their images. It looks to me like [and I saw a similar comment in that huge thread on Kristof's blog]
the images of blacks holding guns are readily apparent as such and the images of whites holding not-guns are just as readily apparent, but the images of blacks holding not-guns and whites holding guns are more difficult to decipher.
I'd like to see somebody do a rigorous test of that possible confounding variable before labeling another huge chunk of us as implicitly racist.