Skip to content

IAT and unconscious bias

April 8, 2010

Just took something called an IAT: Implicit Association Test. The one I took tests associations between “male” and “science” and “female” with “liberal arts” vs. “male” with “liberal arts” and “female” with “science. I, like 28% of online respondents, tested as having a moderate association of male with science.

IAT type tests get press because they tell us about things we don’t even know exist. They test biases we don’t carry consciously. There’s one that tests white/black with good/bad. Another tests views on disability and another on sexual orientation. (Incidentally I just took the sexual orientation one and test for a strong association of straight with good, again agreeing with 28% of the respondents.) The way they work is you’re asked to sort words into two columns. For instance, the gender science IAT has the following categories: Male, Female, Liberal Arts, and Science. They pair them up, say Female and Liberal Arts on the left and Male and Science on the right. You’re asked to sort words like “chemistry” and “daughter” into either the right or left columns, as fast as possible. Then they switch the pairing and have you do it again. If you’re significantly slower in one pairing, then that shows some type of bias. (they have a lot of careful controls, including randomizing the order so that the aggregate results are more trustworthy.)

Consistently, biases show up. Both women and men associate men with science and with having a career, while putting women with liberal arts and being a parent. White, blacks, and Asians show biases that whites are good and blacks are bad. I haven’t looked through the results carefully for other tests, but I think we can all guess what they show. We all carry around biases that we don’t want to recognize. And I think that’s got to change.

I’ve taken a race IAT before. I hated it. As I was taking it I could tell that it was harder to associate Black with Good than it was to associate White with Good. I hated knowing that this test was telling me that. At the time, I wasn’t wiling to admit that I might harbor some internal racism. I was too caught up in “racists are terrible people who all are going to HELL!!!!!!” I couldn’t look at that in myself and admit it. I didn’t consider that I am a product of a racist culture. I didn’t consider that I might be racist not because I’m a terrible human being, but because I live in a culture full of racial stereotypes. I wasn’t willing to admit my own racism because I didn’t consider that I could be a good person and still have such bias.

Thing is, not being willing to admit my own bias meant that I wasn’t attempting to lessen my bias. Denying made sure that it continued to exist. Admitting that everyone has them, though, freed me from the guilt. I’m no longer only a bit better than a slave owner for having racial bias. I’m allowed to be a good well meaning person, just one with some issues that need work. This means that I can actively work to lessen my own bias, by educating myself and being aware of my own subconscious.

Why am I rambling on about this? Because I have had many conversations with men here in physical sciences who close down when I say there might be barriers to women in science. I understand, because I used to do the same thing when people said racism might be important. I would shut down, because I felt accused. I’ve found that saying words like “sexism” or “feminism” cause people to clam up, even though I’m almost never accusing the guy I’m talking to of being sexist. What I’m trying to get across is that we ALL have this bias. Nearly everyone has some subconscious bias about women and science. That includes women. That, as evidenced above, includes me. I’m just as bad as the next person. But we’re not going to fix it if we keep denying it’s there. I have a pile of studies showing the effects of subconscious bias on women in various career situations. This is a real negative effect on women, and it’s not going to go away if we keep denying it. We ALL have subconscious bias, and I for one don’t blame anyone for having them.

If you’re interested in taking an IAT, Harvard has a website devoted to them. There’s a lot of them there, as well as more information about methodology. It’s an interesting sort of quiz, and I always end up laughing at how much harder it is to do the set that shows my bias.

4 Comments leave one →
  1. Julie permalink
    April 9, 2010 3:49 pm

    I took the disability one, the religion one, and the women/work one. I got “too many errors” to be scored on the women/work one, but I could tell that my brain has an implicit association with men and work – putting family-related terms in the man category was the hardest test set I did.

    With the religion one, I’m not sure how much familiarity accounts for being able to hit terms faster; also, I wonder about how similarities between Judeo-Christian-Islamic belief systems skews results. (I kept trying to associate “Abraham” with “Islam,” for example, and “Jesus” with “Jew.”)

    Interestingly enough, the disability test gives me a moderate automatic preference for disabled people. I might go back and play with the site later, but I’m tested-out for now.

  2. April 9, 2010 8:27 pm

    Huh interesting. I haven’t taken any of those actually, thought I’d be curious about all of them. I wonder why you tested so positive for disabled people. What did you get for the religious one?

  3. Julie permalink
    April 10, 2010 6:43 pm

    I got Christianity>Islam=Judaism>Hinduism for the religious one, hence the familiarity hypothesis. For the disability one, one possible theory is that my brain thinks, “Hey, I am a disabled person. I am awesome. Therefore, disabled people = awesome.”

    OTOH, given that people don’t normally do this – as evinced by the fact you cited above that minorities also show the unconscious bias of white = good – I don’t really know how well that would hold up overall. But it’s the best guess I can come up with, and I’d be interested to see subset data on how members of minority groups skew when taking these tests (i.e. are disabled people more likely to have an automatic preference for disability? Is that outcome changed by how much disability has impacted their lives? Or, to use another test, would LGBT people who have experienced more discrimination be more likely to show straight preference than those who haven’t experienced as many orientation-related difficulties?) Just some thoughts; I wonder if anyone’s actually tried to break down the data that way.

  4. April 11, 2010 1:54 pm

    They had tests like these at my museum. I did the male/female science/liberal arts one and came up with no association. I was neutral, which I was pretty excited about, seeing as I did not know what the test was even trying to do before I took it. Most people I was with ended up having a bias, though.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: