Bias tests

[venaja]

I took one of those associations tests where you are bombarded with images of faces and things which supposedly reveal your unconscious biases. I think it is a complete load of crock. Yes, people do have associations, probably even strong associations, but it doesn't automatically point towards racism, which is the implication. People recognize

Comments

[descant]

The problem with these sorts of tests is that it's really easy to interpret the results in a way that has no or little relation to what the test is really saying. The fault of the test isn't because it thinks it says more than it does. People are just designing the wrong tests.

Then there is the entire implication that race is purely social structure anyway, but that's another rant/story.

[venaja]

The problem is that people draw conclusions that are not at all substantiated by the test. Instead, they take the results and extrapolate from them, which is dangerous and unscientific.

You could say, "this person associates European with American." That is the only thing the test tells you. But to add, "therefore, this person thinks Europeans are more American/Asians are more foreign" is completely wrong. However, people have a tendency to believe it because bias and racism are exactly what they are looking for.

If you flip it around and say, "this person associates Asian with American. Therefore, this person thinks Europeans are more foreign", it begins to sound ridiculous. They completely overlook much more mundane reasons. In my case, for example, I may be more familiar with Asian faces and American images. I would be more inclined to press those buttons than I would for European faces or foreign images. It says nothing about how I feel about Europeans or foreign things, but just what I'm more used to.

[descant]

In a more direct way, you've stated what I may, or may not, have implied above. If one wants to investigate racism through association, then one would have to design a test that minimizes bias, and that is very hard to do.

The reason why unbiased tests are difficult is just as you have said: You associate two traits from exposure and experience. A similar association test done at Harvard [and repeated in learning psychology classes everywhere] reveals that if you prime someone to associate positive traits with patriotism, and patriotism with a certain ethnicity, then the person will more readily associate positive traits with that ethnicity.

Does that mean you favor one ethnicity more than another? No. You were taught by the test to associate one group of words or traits to another, or a set of images, that represent that ethnicity. The implication of that, however, is repeated exposure could cause racism, but such a test is no indicator of present racism.

Hell, you know this already. I don't need to lecture.

[venaja]

Yet you lecture in a most delightful way. I would be proud to have you as my teacher, Professor Ariose.

[madkiwi]

Don't tell anyone I said this because I don't want to be called an intolerant fascist right winger.

Problem is most of those "tests" were devised by liberal sociology professors (I know, that's redundant).

[venaja]

Whatever their leanings, it's obviously they're looking to draw only one conclusion, what with all the statements about how "the majority favor whites over non-whites" and how "many people are surprised at their results" scattered around the site.