Racial biases in voters are more prominent in those who lack strong party preferences, according to a study conducted by UW psychology professor Anthony Greenwald.
Approximately 30,000 people participated in an online survey Greenwald conducted that measured both self-reported and implicit racial attitudes. Implicit attitudes were tested through the Implicit Association Test, which assesses people on preferences they may not realize they have.
Greenwald collected survey responses July through September from the participants, 8,600 of which are eligible voters. Now he is in the process of following up with respondents to see which candidate they actually vote for in November.
“I was especially interested in [the study] because after the election in 2008, a lot of people were saying that it proved that … if a black man could be elected president, then we don’t need to worry about race bias,” said Sianna Ziegler, a UW graduate student who collaborated with Greenwald on the study. “Studying unconscious race attitudes is one way to show that that’s not really the case.”
The survey also measured respondents’ knowledge of candidate platforms, general political knowledge, their own political opinions, their perception of candidates, and the candidate for whom they voted in 2008.
So far, results show that 25 percent of respondents lack strong party preferences. Among this group, 10 percent of voters’ candidate choice was influenced by racial attitudes.
On the other hand, 25 percent of respondents had strong party affiliations and appeared to have already determined which candidate they will vote for come election time. Among these voters, only 2.4 percent of voters’ candidate choice was influenced by racial attitudes.
This data supports arguments that unpolarized voters are significantly influenced by racial attitudes.
Greenwald said although the amount of moderates showing racial bias is significant, it is not substantial enough to determine the election outcome, because voters in this group are less likely to vote.
“That’s probably the reason why the level of race biases we are observing is not causing national polls to show that there is a preference for Romney,” Greenwald said. “Although Obama faces what amounts to a severe handicap due to racial bias among Americans, he appears to be doing well enough to overcome that.”
Since moderate voters rely less on political preferences, these voters look for other factors to determine their vote.
“If you are on the margin of not knowing who to vote for, the influences of racial bias — which are not huge but are definitely noticeable — can easily tip you one way or another,” Greenwald said. “If you have very strong beliefs, the same magnitude of racial biases will not push you away from the candidate you prefer on the basis of your political ideology.”
Although the survey gives insight about racial bias, it is not a representative survey due to the demographics of the respondents, Greenwald said.
Racially, the survey pool looks much like the electorate. Eighty percent of respondents were white, and approximately 8 or 9 percent were African-American. However, a greater percentage of women participated in the survey than there are in the electorate.
Furthermore, results show an overrepresentation in highly educated voters — college students and those with advanced degrees.
As Greenwald continues surveying, he hopes to find a way to make the survey more representative.
Greenwald is exploring weighting methods used by pollsters who don’t get a random sample. These methods give heavier weight to those who are generally under-selected.
“In our case, those who are over-selected are liberal and highly educated,” Greenwald said. “[To solve this], we can do this type of weighting too. One of the things we are exploring is how to use weighting methods on our drop-in sample.”
Reach reporter Ann Huynh at firstname.lastname@example.org. Twitter: @AnnThuy
Please read our Comment policy.