Like Prince Harry, I never knew I had unconscious bias until it was pointed out to me, but now it has been I know I will have to do something about it. Except that in my case that ‘something’ is not to moan to Oprah Winfrey about members of my family speculating on the colour of my baby’s skin. It is to dig a little deeper and ask: do I really have an inner Ku Klux Klan that is controlling all I do and preventing me from becoming a good person?

I had heard of unconscious bias training on many occasions – not least when the then Cabinet Office minister Julia Lopez told the Commons that a government review of evidence had suggested it was ineffective and would therefore be phased out in the civil service. Actually, it seems to be taking a long time to be removed: more than a year after Lopez had made her statement, Civil Service Learning, which organises courses for civil servants, was still offering unconscious bias training. As for other organisations, it seems to be a booming industry. One company, for example, offers a one-day training course which ‘provides a non-judgmental approach aimed at understanding how unconscious bias operates in the workplace’. The cost is £795 per delegate face-to-face or £295 via Zoom.

Do I really have an inner Ku Klux Klan preventing me from becoming a good person?

Until last week I had never tried it myself, not least because I have never applied for a job in a public body or woke corporation. I then discovered that I could visit the Harvard University website and try for myself the Implicit Association Test (IAT), which lies at the heart of unconscious bias training. What’s more, I could do it for free, rather than having to pay £795.

This is how I got on. The IAT comes in many different flavours, promising to weed out unconscious bias on anything from disability to age, but the best-known is the one which deals with race. The test began by presenting me with a number of faces which I had to classify as either belonging to an African American (in which case I had to tap the ‘e’ key on my computer) or a European American (in which I had to press the ‘i’ key). If I got it wrong – and there was one face which befuddled me – I would be corrected. Then I was presented with a series of words like ‘friend’ and ‘disaster’ and had to identify them as ‘bad’ words (‘e’ key) or ‘good’ words (‘i’ key). Then it started to mix up the faces with the words, but asked me to do the same thing: press ‘e’ if I saw an African American face or a bad word; ‘i’ if I saw a European American face or a good word.

I paused at this point. Was it really trying to sow in my head the idea that African Americans are bad and European Americans are good? I wondered if that was what I was supposed to do; whether it was like Stanley Milgram’s experiment where volunteers were instructed by a man in a white coat to administer an apparent electric shock to a human being sitting behind a window. Was I going to be branded a racist if I went further?

But I reckoned I clearly wasn’t harming anyone, so I carried on. After I finished that section, the rules suddenly changed. I still had to press ‘e’ for a bad word and ‘i’ for a good word, but now I had to press ‘e’ for a European face and ‘i’ for an African face. Unsurprisingly, I found it a little confusing given I had got into the habit of doing it the other way round. Then it was all over and the computer delivered its verdict: I had taken much longer when black faces were on the same key as good words as I had when they were on the same key as bad words. Therefore I had a bias: I found it easier to associate black people with bad things. In fact, I had taken ‘much longer’ on the second part of the test. I really was right out there on the racist fringe.

It ought not to be hard to see the problem here. In the second half of the test I was having to counter what I had been doing in the first part. If I associated black people with bad things it was surely because that was the way my muscle memory had just been conditioned a few minutes earlier, not because of my inner bias. The objection has been raised before, indeed quite soon after the IAT was introduced in 1998. Yet the Harvard webpage rejects it. It turned out that while I had been given a version of the test with first associated black faces with bad things, some people – at random – got the test the other way around: where black faces were initially associated with good things. ‘The order in which you take the test can influence your results,’ it said, ‘but the effect is small.’

Most popular

Fraser Nelson

McMafia: inside the SNP’s secret state

So I took the test again, posing as someone else just in case Harvard’s mainframe spotted me as a returning customer. And this time I did indeed get the test the other way around. And guess what? This time I was taking longer when white people were associated with good things and black people bad things. In other words, I was no longer a white supremacist; rather I had a bias towards black people.

Ah, but was that because of the effect I described above – or was I subconsciously taking longer over the second part of the test because I didn’t want to be branded a racist? Either way, it undermines the test. Once you have done an IAT and you know how it works, you can game it so as to get the result you want. I then took the test a third time, this time counting to three each time before I pressed each key, to make sure I had enough time to counter any confusion and make sure I was getting it right. This time my result came back: ‘You were about equally fast at sorting “black people” with “bad” and “white people” as “good” and at sorting “white people” with “bad” and “black people” with “good”.’ Miraculously, I had suddenly lost all my prejudices. That is my advice if ever you are asked to undertake one of these tests: count to three before pressing any key and you will pass with flying colours (if I haven’t just unconsciously made an association between ‘colour’ and something good).

 In spite of these flaws, a vast industry has grown up around the IAT. It is nice business, at £795 a delegate. Some time the whole edifice of unconscious bias training will surely crumble and the IAT will be used with the same prefix as tends to be used with the Milgram experiment – ‘the now dis-credited’. But not before great fortunes have been made on the back of this junk science.  />