Written by Andrew Neff

Edited by Jonathan James

June 2019

Closet Racism:

Why Words and Action Don’t Align

racism expeiment

We’ve all seen casual racism go viral. Barbeque Becky, Golfcart Gayle, policemen escorting brown bodies out of coffee shops. The strange pattern that emerges is that in most cases, the alleged racists keep saying that race has nothing to do with it. Hypothetically, assuming these acts were racially motivated, why not deny it? Why not accomplish a racist goal without the social cost?


But, I mean, at least some of these people are definitely racist. We see it in our behaviors. Hiring practices and wages remain unequal despite few hiring managers admitting to a male or white or hetero preference. And so, if we want to probe racial attitudes, as enlightened individuals, we tend not to rely on self-report. Instead, we recognize that actions speak louder than words.


In science, we’ve begun formalizing these “attitude tests”, and we’re now pretty scientifically sure, people score differently on implicit and explicit tests of racism. But the findings haven’t always been so black and white. A group of researchers were curious, when it comes to the scientific evaluation of our attitudes, just why is it that implicit and explicit tests diverge?


On the surface, it’s pretty apparent; one task is asking whether you want to explicitly acknowledge that you’re a racist, the other is asking whether your actions are consistent with racism. They’re just different things. But at the same time, both our actions and words have something to do with our attitudes. And according to one piece of research, there may be a subtler explanation for why the two aren’t aligning.


Please, help us continue bringing you more content.

You can help us by sharing this post on twitter,

or providing a monthly contribution on patreon.

Thank you for reading!


The Experiment

If I wanted to understand your explicit attitudes towards my writing, I could just ask you whether you liked my essay. If I were curious about implicit attitudes, I might look at the number of claps (not that I’m asking for them…) or twitter shares (@Neuroscience_fu). Telling me how you feel, and sharing something on twitter, superficially, are totally different things. The question the authors wanted to ask was whether the discrepancy in racial attitudes between implicit and explicit tests could be chalked up to superficial features of the tests. By administering three different racism tests, scientists devised a clever approach to tease apart action/word discrepancies from superficial test characteristics.

racism test

Implicit Test: Subjects were first asked to rate how much they liked an abstract symbol, like a chinese character - which is a weird question, but that’s what they did. But just before that, subjects were shown a picture of a face, a white face or a black face. So first you see the face, then you see the symbol, then you rate the symbol. It’s an interesting approach, because when we talk about racism, we’re for sure talking about how the presence of black faces makes abstract symbols less likeable. Definitely what we’re talking about. Not the belief that black people are more aggressive or less intelligent, but that they are ruining our appreciation of abstract art - this test nails it. But, sorry, continuing on.

Explicit Test 1: Subjects were presented with the same task, black face → abstract symbol → rate for likeability. But in this case, subjects are asked to rate the likeability of the person, not the abstract symbol. Since the implicit and explicit tests were (basically) the same, they can be directly compared without the risk of superficial differences in test-type.


Explicit Test 2: Subjects were asked, pretty much, just, “Are you a racist?“


When comparing the two very different looking tests of racism, implicit and explicit racial attitudes are not well correlated. But when they used almost the exact same test, there was a pretty good correlation between implicit and explicit attitudes. To the scientists, what that means is that test structure, rather than implicitness, is driving the differences between the two test types.


People lie to each other, and people lie to themselves. Society puts a cost on certain types of speech, and therefore discourages it. Acknowledging whether or not you’re a racist is about so much more than an honest introspective dive into your implicit racial attitudes. My guess is that explicit endorsements of racism don’t really track a hypothetical “racial attitude”, insofar as that exists, because knowing the nature of our mind is a difficult proposition. And in the rare instances that we may really know ourselves, honesty is only one of many goals in a conversation. But nonetheless, this study, and a subsequent replication, found a pretty strong correlation between the implicit and explicit test. People who don’t like black faces also prefer their abstract symbols not to be preceded by black faces. A ridiculous test? Completely disconnected from what we actually mean when we talk about racism? Yes. But the data is the data. Maybe we know ourselves better, and are more honest than we seem.



  1. Payne, B. K. Burkley, M. A. Stokes, M. B. (2008). Why Do Implicit and Explicit Attitude Tests Diverge? The Role of Structural Fit. Journal of Personality and Social Psychology, 94, 16–31

  2. https://osf.io/79y8g/



Skeptical Science

With a

Humanistic Touch

A Psychology and Neuroscience Blog