Since their introduction almost twenty years ago, implicit association tests (IATs) have become fairly well known as a tool that attempts to measure a person’s unconscious or unrevealed biases. Such unconscious biases are sometimes explained as the cause for discrimination, stereotyping and inequality. Many large companies now have some sort of unconscious bias training program in place, and the number is expected to grow.
One way that many students often become aware of unconscious bias is through being directed to Harvard’s Project Implicit website, maybe through a class or other activity. There you can take an IAT on your own, which shows you images and words and has you quickly categorize them by pressing certain keys, and based on timing differences, suggests that you might unconsciously associate certain images with certain words. I had tried one of these a few years ago regarding Asians and being American versus being foreign, which told me that I moderately associate Asians with being American — apparently an uncommon result. I understand two East Asian languages to some degree though, so perhaps that’s not too surprising.
It’s time UW and the Madison community widen its update of racism
So I decided to try another one recently, the more famous race IAT, which shows white and black faces, along with positive and negative words. I expected this one to be fairly neutral based on my beliefs and interests. After going through several screens of the different combinations, I was presented with a results screen that stated “your data suggest a slight automatic preference for African Americans over European Americans.”
That is also uncommon. Further down on the page were the statistics: 68 percent of results showed a greater preference for white people, compared to 14 percent showing a greater preference for black people, with the rest showing no preference. That’s a rather large disparity, so surely this is evidence of a widespread problem that needs to be addressed, right? Well, not so fast.
Despite the marked results for a large population of test takers, IATs are actually fairly unreliable at the individual level, with statistical measures of test validity falling far short of the standard for psychological tests — in other words, your results are somewhat random. That’s bad news if you end up in a company or organization that conducts unconscious bias training and tries to give personalized training based on your results, or even if you just take them too seriously yourself. Furthermore, there is now some evidence that training people not to stereotype others might actually make the problem worse.
It also isn’t entirely clear what these tests measure. There is no good evidence that they actually predict discrimination or that systematic efforts to change implicit biases will affect people’s actual behavior. People are not automatons who just react to unconscious stimuli, after all. Other researchers have proposed alternative explanations for the difference in reaction times, such as empathy and familiarity.
Since we don’t know what they’re measuring, it’s too easy to project assumptions onto the statistics. If you’re looking for a way to confirm a preexisting belief that most of America is still deeply discriminatory at the very depths of our soul, these numbers and the fact that individuals can do little to influence them might seem compelling. IATs have now been in the public eye for so long that people have actually built careers around them, with unconscious bias training serving as a kind of modern day exorcism for unseen demons of racism and sexism. We should be more skeptical than that.
We also shouldn’t forget that unfair discrimination still occurs because of explicit biases. American society has come a long way since racial discrimination and limitations on women’s rights were written into law, but some people have misguided beliefs about others. There actually is evidence that explicit biases exist and affect people’s actions. Challenging those would likely be more productive than hunting down invisible demons.
Additionally, if you want to make an effort to improve yourself and treat others better, that’s an admirable goal. Just don’t trust a computer program to tell you that you unknowingly believe something so you can try to fix a problem that isn’t there.
Amanda Love ([email protected]) is a senior majoring in mathematics.
