Studying the science of racism
Shortlisted for the Unwin Award and the British Book Awards’s ‘Discover’ Book of the Year and longlisted for the Jhalak Prize, The Science of Racism by the Runnymede Trust’s former head of research Keon West uses clear scientific research to expose what we know about racism, exactly how we know it, and what we can do about it. This edited extract throws a spotlight on the problems with the unconscious bias narrative.
A perpetually vexing aspect of the study of racism is the apparent lack of any actual racists.
Racism is real, significant, empirically detectable and almost everywhere. But if that’s true, then why can’t we tell where it’s coming from? Who are the people doing all these racist things, and why can’t we identify them?
This is no small problem, and for a time it was a matter of genuine scientific concern. If you enter any company or speak to any search committee and ask them if they discriminate against ethnic minorities in their hiring practices, they will all (or nearly all) tell you that they don’t. And yet, if you then send them a few hundred (or thousand) identically qualified CVs, they will predictably and reliably select the White CVs more often than the Black ones.
Similarly, if you ask any teacher or educator if they discriminate against ethnic minority students, they’ll say that they don’t. And yet, when you hand them a hundred or a thousand pieces of equivalent work, they will rate the work by Black students as worse than the work by White students.
If you ask any medical professional if she or he discriminates against ethnic minorities, you’re likely to get a stern dressing down and a request to leave the premises immediately. However, when you run the numbers, you’ll see that the ethnic minority appointments are scheduled to occur later than the ones for White people and that the ethnic minorities are less likely to be offered crucial life-saving treatments.
And if you ask any three-year-old or four-year-old if they prefer White people to Black people... well, a lot of them will just say yes, because kids are like that and most of them haven’t learned about the finer points of social norms.
But I think we can all accept that the three- and four-year-olds aren’t the ones responsible for all this racism in hiring, policing, customer service, healthcare and essentially everything else. So what’s going on?
‘The problem goes even deeper’
Frustratingly, the problem goes even deeper. Even if you ask people anonymously, even if you ensure that they know there will be no negative consequences for revealing their true thoughts or feelings, you’ll still find that almost nobody thinks of themselves as a racist person.
This is not a claim I make idly. In 2019, West (that’s me) and Eaton published interesting findings on just this topic. For our study, we asked 148 people to judge how racially egalitarian they were compared to both people in the wider society and to other people in the room with them at the time. We gave them a rating scale of 0 to 99, such that ‘0 would indicate that you are at the very bottom... or more racist than almost everyone else, 50 would indicate that you were “exactly average”, and 99 would indicate that you were at the very top, or less racist than almost everyone else.’
If the participants were rating themselves accurately, we should have found that about half of them gave themselves scores that were lower than 50 (indicating that they were more racist than average) and half of them should have given themselves scores that were higher than 50 (indicating that they were less racist than average).
This, however, was not at all what we found. Instead, we found that everyone gave themselves scores that were higher than 50, regardless of the context in which they were comparing themselves. That means that even when they were just considering how they ranked compared to the other people in the room with them on that day, every single person thought that they were less racist than the average.
You don’t have to be much of a mathematician to understand that this is not how averages work.
‘Implicit bias scores have proven their usefulness in the scientific community’
How can we possibly square these two findings – a world in which there is widespread, significant, detectable racism in almost everything, but simultaneously a world in which hardly anyone, anywhere admits, even to themselves, that they think, feel, or do racist things? This is the problem that research on implicit bias was perfectly poised to solve…
In 2007, Rudman and Ashmore found that Implicit Association Test scores predicted participants’ self-reported use of racial slurs, their social exclusion of ethnic minorities, their likelihood of causing ethnic minorities physical harm, and their desire to cut the budgets of ethnic minority organisations. In each case, the implicit scores predicted these behaviours even after statistically controlling for the participants’ explicit racial attitudes.
In 2008, other researchers (Greenwald, Tucker Smith, Sriram, Bar-Anan and Nosek) found that the Implicit Association Test could predict which Americans would vote for John McCain or Barack Obama in the US presidential election, even after their explicit attitudes were taken into account.
In 2021, Bell, Farr, Ofosu, Hehman and DeWall found that the Implicit Association Test predicted which hopeful parents were less willing to adopt a Black child. Indeed, their implicit bias scores were better predictors than their explicit bias scores.
I could go on, but hopefully the point has been made. As Greenwald and colleagues hoped back in 1998, implicit bias scores have proven their usefulness in the scientific community. They allow us to detect bias and to predict who is most likely to behave in a biased way, even if the person is ‘unwilling or unable’ to admit it.
So far, this all sounds really good. From what I’ve said, the scientific research has clearly demonstrated the existence of implicit bias, and has clearly shown that implicit bias scores can be useful for predicting actual discriminatory behaviour: medical discrimination, hiring discrimination, educational discrimination, political discrimination, familial discrimination, and so on.
So, if that’s all true, why do I keep hinting darkly at a problem with the unconscious bias narrative? Wasn’t Prince Harry, Duke of Sussex, correct to attribute his past behaviour and the current behaviour of some of his family members to unconscious bias?
‘The problem with the unconscious bias narrative’
To explain my concerns, I’d like to return to the interesting phrasing that Greenwald and colleagues used when describing implicit bias: bias that a person might be ‘unwilling or unable’ to tell you about. Why, exactly, might a person be ‘unwilling or unable’ to accurately report their own levels of racism?
There are many, many possible reasons, which I’ve condensed under three general headings. The first and most obvious of these is that a person may be perfectly aware of their levels of bias, but deliberately misrepresenting these levels to you in an attempt to deceive you. In layman’s terms, I believe this is often referred to as ‘lying’.
The second reason could be that a person is partially or imperfectly aware of their levels of bias, but employing a set of psychological mechanisms to protect themselves from recognising or admitting their biases. This second reason, which I’ll put under the general heading of ‘psychological trickery’, is, to me, the most interesting one...
For now, however, the third and final reason why someone might be unwilling or unable to tell you how biased they are could be that the person is completely unaware of their own level of bias, only discovering this after taking some kind of implicit bias test. This last reason is the only one that matches the ‘unconscious bias’ narrative that has become so popular in recent years.
This is why I am so pedantic about the distinction between ‘implicit bias’ (which the scientists like to talk about) and ‘unconscious bias’ (which the newspapers like to talk about). Research on implicit bias covers a multitude of different types of bias that individuals are ‘unwilling or unable’ to report explicitly and accurately.
From the very beginning, its goal has been to take types of bias that are difficult to detect and make them easier to detect. It is a set of tools that allows us to better predict who will pick the White CVs over the Black CVs, who will rate the White student’s work as better than the Black student’s work, or who will schedule the White patient’s appointment sooner than the Black patient’s appointment – even if the person in question is ‘unwilling or unable’ to admit to doing those things.
In contrast, the unconscious bias narrative sweeps away two of the important reasons why someone might be unwilling or unable to report their levels of bias. It ignores both simple deception and complex psychological trickery, leaving us with a narrative in which most or all people are entirely, and innocently, unaware of their own biases.
As such, the problem with the unconscious bias narrative is not that it is inaccurate. Unconscious bias does exist. It can be objectively detected and it can be used to predict certain kinds of biased behaviour. Rather, the problem with the unconscious bias narrative is that it ignores and downplays other real, damaging manifestations of bias. The unconscious bias narrative is restrictive and myopic.
And this myopia is dangerous.
The Science of Racism: Everything you need to know but probably don’t – yet by Keon West, head of reparations at the Joseph Rowntree Charitable trust, is out now.
The views and opinions expressed in this article are those of the author(s) and do not necessarily reflect those of the Runnymede Trust.
Join the fight for racial justice: support the Runnymede Trust’s work by making a donation.
