If you work at a university or live in a cosmoplitan city, you'll probably have heard the claim that we're all unconsciously racist to some degree. Unlike a lot of the things people say about sexism and racism in such environments, this one's actually based on research. The only problem is that the research now looks flawed.
The research focuses on the so-called Implicit Association Test, or IAT. This test shows you various faces and words and asks you to click the mouse if you see (say) a black face or a positive word, or a white face or a negative word. Then it switches them around, asking you to click if you see a white face or a positive term, or a black face or a positive term. If you're slower at the first of these tasks than the second, then you're said to exhibit a positive bias for white faces.
That's one part of the research, which claims to establish that we all have unconscious bias. But that only raises the question of whether that unconscious bias actually gets translated into prejudiced behaviour. To try and demonstrate a link, researchers have taken a further step, which involves various sorts of experiments involving inter-racial interactions using people who've previously done the IAT. If people display prejudiced behaviours, they can then link the unconscious biases implied by the IAT with actual racist or sexist actions.
Critics of the IAT have pointed out problems with both stages of this research paradigm. The main problem with the IAT itself is that there's very little consistency in people's results if they take the test multiple times. Of course, with most tests you'll get slightly different results each time you take it, but if there's too much variance then it suggests the test isn't really measuring anything. (Imagine a scale that said you weighed 200 pounds one minute and then 3 pounds then next. You'd think it was broken.)
Variance in test results is, unsurprisingly, a common issue in science. Because of this, scientists have developed a scale which measures how much results vary between the different times you take a particular test. On this scale, 1 is perfection (the test always give the same results for a particular person) and 0 is disaster (the variance is huge). 0.8 is seen as an acceptable score for publications in the social sciences. How does the IAT score? Somewhere between 0.3 and 0.6. Not only is it not a very reliable test - it's well below the conventional bar for respectable research.
Let's just imagine that enormous problem didn't exist, though. If we take certain people's IAT scores, do they predict racist behaviours? Again, scientists are used to looking at a spectrum of explanation. Some things predict behaviours perfectly, and others not at all. What percentage of behaviours do IAT scores explain? According to one meta-analysis (a combination of many different studies), it's something like 2 or 3%. (Another meta-analysis, which confirmed the powerlessness of IAT scores to predict prejudiced behaviour, was published just this year.)
The IAT is currently being used by police forces, schools, and other institutions. It's being pushed especially hard on university campuses. Often the test is followed by some sort of 'cultural awareness training,' which aims to correct the unconscious biases the IAT has supposedly identified.
Of course, the weakness of the research doesn't necessarily mean that we're not unconsciously racist and sexist. But given the unreliability of the IAT and the flimsiness of its link with actual racism, it's fair to say that there's no good evidence that unconscious prejudice of this sort exists, and even less evidence that it's doing any harm.
The scientists who've been pushing the claim that we're all unconscious racists are welcome to try to find more evidence that we are. But until they do, we should resist or ignore attempts to introduce compulsory IAT testing and compulsory 'cultural awareness' programs. Such endeavours may often be well-intentioned, but it's pretty clear now that they're based on flawed research.
is an English settler in New Zealand, a writer and an anti-intellectual.