Friday, February 7, 2014

The Implicit Association Test: Racial Bias on Cruise Control

There is nothing more painful to me at this stage in my life than to walk down the street and hear footsteps and start thinking about robbery. Then look around and see somebody White and feel relieved.
                                                                                      Jesse Jackson

This is partly a housekeeping post. Several times, I've referred, or wanted top refer, to the Implicit Association Test (IAT), but didn't want to take the time to explain it. In the future, I can direct the reader to this post before continuing.

Before you go on, please go to the Project Implicit website and take the Race IAT.

You've probably figured out what happened. The first couple of tasks—classifying faces as black or white and words as good or bad—were pretty easy. This was largely a warmup to familiarize you with the procedure. But it became more difficult when you had to respond to two conceptual dimensions at a time. On some trial blocks, you were to press one key if either the face was White or the word was good, and the other key if either the face was Black or the word was bad. There were also some trials in which you were to press one key if either the face was White or the word was bad, and the other key if either the face was Black or the word was good.

The computer measured your reaction times. Trials on which you made an error or took an unusually long time were discarded. If you responded more quickly during the White-good/Black-bad trials than the White-bad/Black-good trials, the feedback stated that you had an automatic preference for White people compared to Black people. If you responded more quickly during the White-bad/Black-good trials, you had an automatic preference for Black people. The words "slight," "moderate" and "strong" indicate the degree of preference. An automatic preference for White people means you implicitly associate Whiteness with goodness and/or Blackness with badness. The more closely two concepts are associated, the easier it is to respond to them as a single unit. These implicit associations are completely automatic and are almost certainly based on past activation of the concepts through social interaction and media consumption.


The majority of Americans have an automatic preference for White people—70%, as opposed to only 12% who have an automatic preference for Black people. The responses of African-Americans are less predictable than those of Whites. About 40% prefer White people, 40% prefer Blacks and 20% have no preference. This suggests a conflict between the messages African-Americans get from their home and neighborhood, and the from mass media, where they encounter primarily negative stereotypes of Black people. The reponses of Asians and Hispanics are similar to those of Whites.

The results of the IAT suggest that we may not be willing to report our racial attitudes honestly, or that we may not be aware of our own racial attitudes. Many people respond with shock and disbelief when they get their results. I've heard many rationalizations from students. But all the obvious contaminants such as position preferences and order effects are corrected for in the data. My own automatic preference for White people is hard to take given my lifelong attraction to Black culture, but it pales by comparison to Jesse Jackson's self-reported discomfort.

Measures of racial attitudes fall into two broad categories—explicit or self-report measures, in which people are asked their attitudes directly, and implicit measures like the IAT that are designed to circumvent our defenses. There are two types of explicit measures of prejudice. Old-fashioned racism refers to agreement with statements that are obviously racist, such as the belief that Blacks are less intelligent than Whites. Because expressing blatantly racist attitudes is socially unacceptable, most people score low on measures of old-fashioned racism, although it does predict membership in the Republican Party. Modern or symbolic racism (sometimes called racial resentment) refers to prejudice that is revealed in subtle, indirect ways, through political beliefs such as the claim that Blacks receive more favored treatment than they deserve. These beliefs are correlated both with endorsement of social policies harmful to Blacks and negative emotional reactions to Black people. In effect, modern racism takes advantage of the fact that claiming to be a political conservative gives people plausible deniability of their racist attitudes. Implicit and explicit racial attitudes are positively related, but not strongly (r = +.35, approximately).

Many people wonder wonder if having an automatic preference for White people means that they are prejudiced. The fact that we usually define prejudice as a conscious response to a social situation implies that prejudice and implicit bias are not the same. On the other hand, IAT scores are not random. There are dozens of studies relating IAT scores to discriminatory behavior in social perception and decision making, and one analysis of the available studies finds that implicit measures are more strongly predictive of racial discrimination than explicit measures.

Can you have an automatic preference for White people and still be free of prejudice and discrimination? In terms of the central metaphor of this blog, the IAT represents System 1 thought, occurring without effort or conscious awareness. We encounter a Black person and stereotyped thoughts just “happen.” But the conscious, effortful, deliberate thought of System 2 can override our automatic behavioral tendencies. We can choose to disregard the stereotyped information that automatically came to mind. One way of understanding prejudice is to note that White people low in prejudice recognize their automatic biases and try to consciously control them, while White people high in prejudice either are unaware of their automatic biases or see no reason to disregard them.

If you would like to read a thoughtful and accessible article about the IAT, check out this one by Mahzarin Banaji.

You may also be interested in reading:


No comments:

Post a Comment

Comments are always welcome.