UVa study: Implicit bias difficult to overcome

Despite their best intentions, most people still favor their own racial and religious groups, according to an analysis of online implicit association test data conducted by researchers at the University of Virginia.

Aside from simple in-group preference, people tended to show favorable attitudes toward whites over blacks, and Christianity over Islam, according to the study, led by psychology doctoral candidate Jordan R. Axt.

Psychology professor Brian A. Nosek worked with Axt and graduate student Charles R. Ebersole to analyze more than 200,000 sets of results from the Brief Implicit Association Test. The online, anonymous test is designed to find participants’ unconscious attitudes about members of different racial, age and religious groups.

“It’s a way of really tapping into that fast action – that kind of gut reaction,” Ebersole said. “One of the key parts to measuring something that’s an implicit attitude is that it has to be a little bit indirect.”

According to the study, almost every racial group showed a preference for members of the same group. After that, Axt said, there was a hierarchy — after their own group, participants showed the most favorable attitude toward whites, followed by Asians, blacks and Hispanics (who were defined as a race for the purpose of the study).

A similar pattern formed with religions — after their own group, participants showed the most favorable attitude toward Christianity, then Judaism, Hinduism or Buddhism and Islam.

The Brief Implicit Association Test has received a lot of attention in recent years because it touches on lightning-rod issues of racial and religious division. Its proponents hail the test as a way to get at unconscious attitudes that may subtly influence behavior. But critics say it’s not clear what, if anything, these tests tell scientists about the human psyche.

The test is simple. After filling out a brief survey about their background and explicit attitudes about different groups, participants play an association game that requires instant responses. In the first two rounds, participants put positive and negative words into their appropriate categories, then match photos of people’s faces with the corresponding racial category.

Then the game gets tricky. Participants are given a mix of photos and words, and asked to put each into one of two categories. For example, a participant might be asked to press a particular button when she sees either an African-American face or a positive word, like “laughter” or “joy.” Then the categories are reversed.

Participants must respond instantly. The idea is that if a participant’s preferences are cross-wired (for example, the test-taker prefers whites but is asked to associate positive words with black faces), his or her response time will be slower, Axt said.

“We kind of inferred that when you’re faster in one block it’s more closely tied to your memory,” he said. “What we found is that there was a real clear consistent rule for how people would perform on this.”

But the test has its skeptics, including Gregory Mitchell, a professor in the UVa School of Law who also has a doctorate in psychology. Mitchell said it’s not yet clear whether reaction time actually tells psychologists anything about bias.

“The inherent problem is that associations between concepts may form for many different reasons,” Mitchell said. “Without linking the associations to other behaviors and thoughts, they remain ambiguous.”

If researchers conducted a similar test with sports logos, for example, participants may have positive associations with winning franchises like the New York Yankees or Dallas Cowboys, Mitchell said. This may not demonstrate favoritism as much as awareness of those teams’ popularity and success.

Other researchers, Mitchell said, hypothesize that participants are affected by apprehension at being perceived as biased or prejudiced.

Ebersole said response time studies have been predictive in many other areas — testing students on their implicit attitudes toward math and science, for example, is a good predictor of test scores. The results of the Brief Implicit Association Test also fall into patterns that researchers across disciplines have noticed for years, he said.

First, people favor their own group over others. Axt and Ebersole said they’re not sure what causes it. The favoritism could be instinctive or it could be the result of someone’s upbringing, Axt said.

“People prefer members of their own group,” Axt said. “Is that because they’re members of their own group or is it because that’s who they grew up with?”

The reason for the results isn’t clear, but Ebersole said it seems to be universally true. Forming a smaller group within the whole is normal human behavior, he said, and people form groups based on a variety of factors — which college they attended, for example.

Whatever the reason, Ebersole said, it’s generally advantageous for people to have positive feelings about the groups to which they belong.

“Our groups are an extension of ourselves,” Ebersole said. “So if I’m part of the group, and I want to feel good about me, it’s beneficial for me to feel good about my group.”

The results also support the perception that certain groups are seen more favorably than others, Axt said. Group affiliation aside, people typically had more favorable attitudes toward whites and Christians. The pattern was very clear and consistent, Axt said.

The implications for human behavior are less clear, according to the two researchers. Axt and Ebersole said they still aren’t sure how, exactly, these attitudes influence the way people conduct themselves, but it could point psychologists in the right direction.

“In navigating the social world, you have to make a lot of quick decisions,” Ebersole said. “[Implicit attitudes are] going to bleed into behaviors that you don’t really think about – your fast-acting decisions.”

Axt was careful to stress that the test is not designed to “catch” unconscious racism, but to give a more detailed picture of the psychological processes that help influence human behavior.

“It’s not like we’re getting at people’s real, true attitudes,” he said. “In our behavior, we are partially guided by our conscious beliefs and partially guided by our unconscious beliefs.”

Leave a Reply