Barbara Kay: Actually, it turns out that you may be less racist than you’ve been led to believe

What Kay misses is the usefulness of the IAT for people to become more mindful of their implicit biases, and, in so doing, be more aware of their “thinking fast” mode to use Kahneman’s phrase.

It is not automatic that being more mindful or aware changes behaviour but it can play a significant role (and yes, the benefits can be overstated). Having implicit biases does not necessarily mean acting on them.

Kay did not mention whether or not she took the test. Given her biases evident in her columns, it would be interesting to know whether she took the IAT and what, if anything, she learned.

I certainly found it useful, revealing and most important, discomforting as I became more aware of the gap between my policy mind and views, and what was under the surface.

Anyone can take the test on the Project Implicit Website, hosted by Harvard U. By October 2015, more than 17 million individuals had completed it (with presumably 90-95 per cent of them then self-identifying as racist). Liberal observers love the IAT. New York Times columnist Nicholas Kristof wrote in 2015, “It’s sobering to discover that whatever you believe intellectually, you’re biased about race, gender, age or disability.” Kristof’s tone is more complacent than sober, though. For progressives, the more widespread bias can be demonstrated to be, the more justifiable institutional and state intrusions into people’s minds become.

Banaji and Greenwald have themselves made far-reaching claims for the test: the “automatic White preference expressed on the Race IAT is now established as signaling discriminatory behavior. It predicts discriminatory behavior even among research participants who earnestly (and, we believe, honestly) espouse egalitarian beliefs. …. Among research participants who describe themselves as racially egalitarian, the Race IAT has been shown, reliably and repeatedly, to predict discriminatory behavior that was observed in the research.”

Problem is, none of this can be authenticated. According to Singal, a great deal of scholarly work that takes the shine off the researchers’ claims has been ignored by the media. The IAT is not verifiable and correlates weakly with actual lived outcomes. Meta-analyses cannot examine whether IAT scores predict discriminatory behaviour accurately enough for real-world application. An individual can score high for bias on the IAT and never act in a biased manner. He can take the test twice and get two wildly different scores. After almost two decades, the researchers have never posted test-retest reliability of commonly used IATs in publication.

It’s a wonder the IAT has a shred of credibility left. In 2015 Greenwald and Banaji responded to a critic that the psychometric issues with race and ethnicity IATS “render them problematic to use to classify persons as likely to engage in discrimination,” and that “attempts to diagnostically use such measures for individuals risk undesirably high rates of erroneous classifications.” Greenwald acknowledged to Singal that “no one has yet undertaken a study of the race IAT’s test-related reliability.” In other words, the IAT is a useless tool for measuring implicit bias.