The Implicit Association Test

The Edge website recently put up a talk with Mahzarin Banaji and Anthony Greenwald, concerning the Implicit Association Test, aka IAT. In the interview they basically cover why and how the IAT was created, and the importance of the test.

Very quickly, the IAT is a social psychological research tool used to help investigators explore the various unconscious preferences and attitudes that affect our behavior. Specifically, it measures how strongly one automatically associates a concept or entity with an attribute as compared to how one associates another concept or entity with the very same attribute. So, in other words, would you be quicker to associate a flower (concept 1) with pleasantness (attribute) and an insect (concept 2) with unpleasantness (attribute) than the other way around? If so, then you, purportedly, have a stronger preference for flowers than insects.

Because the test works on split-second associations, you have no time to think about them. This means, ostensibly, that these attitudes are unconscious and inaccessible to one’s self-awareness, otherwise known as implicit associations. As well, because these attitudes are unconscious, they affect aspects of our behavior without us even realizing it, which gives way to enormous implications they have on social cognition and behavior. Being able to understand unconscious attitudes will give us great insight into all kinds of social issues concerning stereotypes, prejudices and biases. And because the IAT is able to quickly measure unconscious attitudes, it is a very useful tool for studying these social issues. Let’s face it . . . the IAT is hot right now, haha. It seems like everyone wants to do an IAT study. But the three main IAT researchers are Mahzarin Banaji, Anthony Greenwald (featured in the Edge interview) and Brian Nosek from the University of Virginia.

Yet the IAT is not without its detractors. In fact, there was a decent debate between the IAT camp and Hart Blanton and James Jaccard in 2006, which was published in the American Psychologist. Blanton and Jaccard came up with some very worthy criticisms of the IAT. Firstly, they state that the metric the test is measured on is arbitrary for measuring a psychological dimension, such as prejudice. We don’t really know what an IAT score means. An example they give deals with self-esteem. Imagine that you scored an 8 on a self-esteem scale. And scores can range from 0 to 10. A score of 8 is meaningless in diagnosing one’s level of self-esteem. We would have to know how that score of 8 relates to behaviors associated with self-esteem. In the same way, to gauge our implicit attitudes we have to link our IAT scores to observable behaviors relevant to automatic preferences, and this has not yet been done with IAT scores.

Well, anyways, if you are interested in implicit social cognition or bias, stereotypes and prejudice, then definitely check out the Banaji/Greenwald interview and, as well, I’ll list some citations for further reading you might enjoy. Lastly, the Edge website has a link to an IAT on preferences for the presidential candidates. So perhaps you’ll find that you hold an implicit preference for a candidate that differs from the candidate that you consciously prefer, if you decide to take it that is.

For further reading:

Greenwald, A., McGhee, D. & Schwartz, J. (1998). Measuring individual differences in implicit cognition: The implicit association test. Journal of Personality and Social Psychology, 74(6), 1464 – 1480.

Blanton, H. & Jaccard, J. (2006). Arbitrary metrics in psychology. American Psychologist, 61(1),
27 – 41.

As well, the same issue of American Psychologist (vol 61 num 1) contains a Greenwald reply and Blanton counter-reply.

Go check them out and enjoy!

another historical day for psychologists

So how's it going sports fans?

It seems that I have been neglecting my blog a little bit lately. I know, I know . . . it seems like I get into a good rhythm for a couple days, and then BOOM!, I'll hit a dry spell and won't write anything new for weeks. Oh, the time graduate study takes from me. Well, I do have a decent excuse . . . I've been doing a lot of traveling. Firstly, I spent some time in Albuquerque, NM for the 9th annual meeting of the Society for Personality and Social Psychology. And secondly, I'm interviewing for Ph.D. programs. Fun stuff. Interestingly, before February, I have never flown on a plane before (I guess a live a sheltered life, ha). Yet I flew on 10 different flights in the past week and a half!

Well, besides leaving me tired, my traveling has left me with several posts that I plan to write up soon. The current post, on the other hand, is about an event that occurred on February 17th, 1890. Give up? It's the birth date of R. A. Fisher!

You might be asking yourself, who is R. A. Fisher and why is he important? I'll tell you. Sir Ronald Aylmer Fisher was a British statistician and evolutionary geneticist, and one of the founders of the modern evolutionary synthesis. He basically showed quantitatively that "inherited traits were consistent with Mendelian principles." As well, he built the foundation for modern statistical theory and population genetics. To say the least, he was a bright guy.

A lot of Fisher's work concerned the variation of inherited traits among plants, so why is he important to psychologists? Well, like I stated before, he practically fathered modern statistical theory. He created advanced techniques that we still use today and wrote influential books on research design and analysis, including his first book, Statistical Methods for Research Workers (of which I actually found a used early edition and got it for free!). Most importantly, especially for social psychologists, he invented the statistical technique referred to as "analysis of variance," or simply, "ANOVA."

To put it simply, ANOVA is a procedure that deals with the differences between groups (specifically, two or more groups), rather than just describe the relationship between variables. This makes it a considerable advancement from the statistical technique of correlation. In correlational research, you can't really make statements about cause and effect. Whereas, in experimental research using ANOVA's, you are given more insight to do so, which is why it has become the most popular (and often abused) statistical procedure in psychology . . . especially social psychology! ANOVA gives us it's extra insight by examining the ratio of the observed variability BETWEEN groups (what we can account for) and the observed variability WITHIN each group (uniqueness that we can't account for). Or in even simpler terms that my thesis adviser, John Nezlek, would say, an ANOVA is the ratio of "what we know" over "what we don't know."

Here's an example to make it a little easier to understand:

Let's say that I have a drug and I think it makes people more aggressive. So I draw two random samples of people and I give one sample a dose of the drug, while the other sample gets a placebo (sugar pill). Then I measure how aggressively (perhaps how many times each person physically harms another) each person in each sample acts. To say that my drug causes aggression, one would have to say that the variation in aggressive behavior between the two groups (drug group and placebo group) is much larger than the variation within each group (do all placebo participants act similar? do all drug participants act similar?). So if the the drug group does act more aggressively than the placebo group, and each member of the drug groups acts similarly aggressive, then it's likely that my drug causes aggression.

Well, that's my short and simple description of Analysis of Variance. I know that my meager post does not give Fisher's brilliance the full justice that it deserves, but I try. So, even though I'm just a little late (my time says 3:18 am on Feb. 18th) . . . HAPPY BIRTHDAY Ronald Fisher!

Happy Darwin Day!

Almost too late! Happy Birthday Charles Darwin!

And here's a funny comic I found to go with the special occasion: