February 05, 2007
Measuring Bias
Ace of Spades found an interesting article at Slate over the weekend about a series of online bias tests called the Implicit Association Test, or IAT, with the most popular one being the race IAT, to judges your biases (prejudices) based upon timed responses to paired words and images.
It works like this:
In the test's most popular version, the Race IAT, subjects are shown a computer screen and asked to match positive words (love, wonderful, peace) or negative words (evil, horrible, failure) with faces of African-Americans or whites. Their responses are timed. If you tend to associate African-Americans with "bad" concepts, it will take you longer to group black faces with "good" concepts because you perceive them as incompatible. If you're consistently quicker at connecting positive words with whites and slower at connecting positive words with blacks—or quicker at connecting negative words with blacks and slower at connecting negative words with whites—you have an implicit bias for white faces over those of African-Americans. In other words, the time it takes you to pair the faces and words yields an empirical measure of your attitudes. (Click here for a more detailed description of the test.)The elegance of Banaji's test is that it doesn't let you lie. What's being measured is merely the speed of each response. You might hate the idea of having a bias against African-Americans, but if it takes you significantly longer to group black faces with good concepts, there's no way you can hide it. You can't pretend to connect words and images faster any more than a sprinter can pretend to run faster. And you won't significantly change your score if you deliberately try to slow down your white = good and black = bad pairings.
Banaji, now a social psychologist at Harvard, has found that 88 percent of the white subjects who take her test show some bias against blacks. The majority of all subjects also test anti-gay, anti-elderly, and anti-Arab Muslim. Many people also exhibit bias against their own group: About half of blacks test anti-black; 36 percent of Arab Muslims test anti-Arab Muslim; and 38 percent of gays show an automatic preference for heterosexuals.
Sounds interesting, no? A test that won't let you lie, even to yourself. I'd like to see how CY readers fare on this, so if you have the time (about 10-15 minutes), take the test at this link (follow the link on this page to the "Race IAT."
Then post your age, race, and state where you grew up, along with your results in the comments. I've already taken it, and I'll post a screen capture of my results page tonight.
Also tell me if your results surprised you, or if they were about what you thought they would be.
As promised, a screen capture of my results for the Race IAT:
The results mark me as part of a fairly rare group, part of the 12-percent of whites that have no bias against blacks, and part of a smaller sub-group of whites that is actually biased in favor of blacks.
I'm sure this probably surprises some people, and I must admit that I am a bit amused that, statistically speaking, I'm probably less biased than quite a few of the liberal bloggers and their commenters that pitch hysterical hissy fits because of the name of this blog.
I'd invite them to also take this specific Race IAT and post their results, but I have a sneaking suspicion that many wouldn't like their results known.
Age=48
Race=white
State=OH
Results:
Your data suggest a strong automatic preference for European American compared to African American.
I wasn't surprised at all.
Posted by: buckeyefan at February 5, 2007 09:46 AMAge = 40
Race = white
State = Michigan
Little to no automatic preference between African American or European American.
Posted by: Mikey NTH at February 5, 2007 10:13 AMAge=51
Race=White
State=Virginia
Results:
Your data suggest a strong automatic preference for European American compared to African American.
Results about what I expected, considering I'm a redneck cracker honky.
Posted by: SicSemperTyrannus at February 5, 2007 10:13 AMAge = 66
State = Washington
Race = white
Little to no automatic preference between African American or European American.
I did have some suspicions about the devisors of the test itself. Portraying concepts as simply "good" or "bad" - and then running the same exercise on photos of strangers - seems a damn crude measurement of whatever they're after.
And considering that just about every social science department in the country is dominated by the groupthink that votes for racial preferences and lives by the religion of PC - which could largely be described as the political movement to disestablish or villainize the white male (formerly known as WASPS, before the PC purveyors wised up, separated out the women, and used murkier terminology to cultivate their funding Foundations) - one would think that said devisors would be happy to come up with an oh-so-neutral test that 'proved' that them honkies is racists after all.
Posted by: Insufficiently Sensitive at February 5, 2007 11:08 AMAge = 44
Race = white
State = Illinois
Little to no automatic preference between African American or European American.
And I'm a hard-headed conservative.
Posted by: Steve at February 5, 2007 02:11 PMAge = 50
Race = White
State = NY
Little to no automatic preference between African American or European American.
-----
I think I noted strong conservative but I might have checked moderate c. I was disappointed there wasn't a libertarian option.
Took the religion test, too, but first and came up with the same classification. Too bad they don't have several of that category. I would have liked to have done a Shinto v the world one.
Hey, I see I am not the oldest fogy to frequent here ... hmm, maybe it's time for me to take the young v old test to see which group they consider me in.
Posted by: Dusty at February 5, 2007 04:42 PMIsufficiently Sensitive, somehow I think they might smoke your computer after you finish the Sociologist IAT, or the Psychologist IAT. Which, I suppose, is why they don't include those in the choices.
Posted by: Dusty at February 5, 2007 04:48 PMNothing. I do not associate anything with people, only with their actions. This tests is stupid. Almost everyone wants to associate with their own kind, and that does not, necessarily, show bias.
Posted by: Len at February 5, 2007 05:41 PMThis test, its uses, and its hidden surprises were discussed in detail in the book Blink by Malcolm Gladwell. Len's dismissal of the idea is not surprising, but he misses the whole point of the thing.
Posted by: Doc Washboard at February 6, 2007 09:08 AMCome on Doc, tell us the surprises. Don't make me go read.
Hey, CY, where's your results, you baiting hack of a blogger?! Skip the screen cap, just type the info.
Posted by: Dusty at February 6, 2007 10:29 AMDarn, did you post the results there before I wrote the last comment?
Posted by: Dusty at February 6, 2007 11:32 AMDarn, did you post the results there before I wrote the last comment?
About 8:45 AM.
Posted by: Confederate Yankee at February 6, 2007 12:06 PMMy bad. I apologize though I'll also note that I didn't notice the "Read More ..." that was created by your adding the results to your post. I was looking for an update to the number of comments.
Not to nit pick, but I think your "part of the 12-percent of whites that have no bias against blacks" should be "29%". It seems to me even the "Little" in "Little to no automatic preference between African American or European American" could swing either way but is not measureable to a standard level of confidence to note which direction.
BTW, since I don't comment often, I ought to note here that I think you're doing a heck of a job here, CY.
It seems to me it could have as much to do with an individuals language patterns and words you are accustomed to using or not. Timing delays could be related to comprehension.
Posted by: lonetown at February 7, 2007 07:04 AMIt seems to me it could have as much to do with an individuals language patterns and words you are accustomed to using or not. Timing delays could be related to comprehension.
lt, The words they use are exceedingly commmon.
Dusty, the researcher found 88% of whites had some bias against blacks. Math tells me that 12% of whites, therefore, do not have bias against blacks. You're basing your numbers, I think, off the web-only results in my screen cap, which are not broken down by the test-takers race.
While the results posted here by readers is of course self-selected and in no way scientifically valid, it seems that blog readers that posted their responses here as a group seem to have less bias that the general population at large.
Posted by: Confederate Yankee at February 7, 2007 07:20 AMGot ya, CY. A random coincidence of 12%'s between the two, helped by the fact you were in the 12%.
But the oddity, and the high percentage in the Slate article did cause me to look into it further. In their FAQ #9 they had this:
"Moreover, if the IAT result represented an ingroup preference exclusively, then Black Americans should show for their group the same level of automatic preference. We know that that is not the case. 50% of Black Americans show automatic Black preference, but the remaining half show an automatic White preference. We conclude from such data that the IAT preference is some combination of an automatic preference for one’s own, moderated by what one’s learns is regarded to be “good” in the larger culture."
It surprised me to see it reported as a 50-50 breakdown. It seems to me, this would mean the IAT Project harbors results that do not allow for the "Little or no" category. I wonder why that is and have asked them.
Posted by: Dusty at February 7, 2007 03:19 PM
