Measuring unconcious bias

[Purple]

Which compensations do you deem "wild" and "over," exactly?

Wild - All of them. As instead of being ordered, calculated and in general systematic they seem to me to be wild stabs in the dark.
Over - Most of them. As typically when you are stabbing in the dark instead of approaching a problem systemically any effort you make tends to be in the wrong direction.

[Simon_Jester]

So are you just... taking for granted that any attempt to be non-discriminatory is a priori excessive?

[Purple]

Just the ones I see simon, just the ones I see.

There is nothing that says these things can't be done right in theory.

[Esquire]

Give some examples, then. Tell us what you see people doing and why you think it's excessive.

[Dragon Angel]

I took the Race IAT and ... I'm just confused. Some of the questions on the initial survey I felt were too simplistic, as in I had answers that went beyond the agreement-disagreement scale, and I don't know if I truly answered them as best as I could've. On the facial-verbal association test, I tended to lose concentration sometimes (thanks ADD), and when the categories were flipped and I had to reassociate everything again as fast as possible again, that's when my brain started to give up.

Speaking of the categories, I had a confused and ... somewhat disturbed reaction immediately when I saw "African-American OR Bad" grouped together, and it took me a minute to really figure out what the purpose behind it was. Take of that what you will.

I'm not sure how to resolve these with the actual result I got.

[Esquire]

Since this thread came back up to the top of the forum - you managed to find some of the major meta-methodological problems with qualitative research. A nonempirical* test is designed by some social scientist A to measure some concept B in some subject C; A may design a bad test, B may not be a valid concept, and C's concept of B may actually be a separate-but-related concept B', further threatening the accuracy of test results. Here, all three of those are almost certainly true. Purely as an example, a KKK Grand Dragon might take the test and not register as a racist if he were simply good at remembering which side of the screen to click. The theoretical underpinnings of implicit-bias research suggest that this won't be true on population-average, but... well, look at my signature. There are serious flaws in the whole conceptual superstructure surrounding the measurement of racism (or any other -ism or -phobia) which I'd be happy to have an excuse to think and talk about further.

More simply, I wouldn't worry about it. There are way too many confounding factors present in this sort of test for it to tell you anything about yourself with enough certainty that you ought to change your behavior solely because of the results.

*This is true for empirical tests as well, but more obviously and therefore less problematically.

[Crown]

Apparently I have "a strong automatic preference for African American compared to European American".