The Science of How We See Obama's Skin Color


When it comes to the policies and politics of Barack Obama, it's no secret that liberals and conservatives don't see eye to eye. But according to behavioral sciencist Eugene Caruso of the University of Chicago's Booth School of Business, these differences in perspective may literally be a difference in perception. In a new study, Caruso and colleagues Emily Balcetis of New York University and Nicole Mead of Tillberg University asked a group of undergraduates which of a series of photographs of both Obama--some of them secretly lightened and darkened--best represented who he is as a person. The results were striking: while self-described liberals tended to pick the digitally lightened photos of the president, self-described conservative students more frequently picked the darkened images. The more you agree with a politician, in other words, the lighter his skin tone seems; the less you agree, the darker it becomes. To discuss how political affinities influence perception--and how politicians and the press could take advantage of these findings--NEWSWEEK's Andrew Romano spoke to Caruso. Excerpts:

Essentially we were interested in whether political party influences how people literally see the world, and how they may see different depictions of candidates as representative of who they really are. So to test this we gathered up a bunch of photos of Barack Obama and digitally altered them to create a version where his skin tone appeared a bit lighter and a version where his skin tone was a bit darker than it appeared in the original photograph. And then we just showed people several different photos and asked them to rate each one on how much they represented who he really is. What we found was that participants who told us that they had a liberal political orientation rated the lightened photographs as more representative of Obama than the darkened photographs, whereas participants who told us they had a more conservative ideology rated the darkened photographs as more representative of Obama than the lightened ones.

It’s a little bit hard to quantify the difference because they were just rating on a 7-point scale of representativeness. So to make it a bit more concrete we looked, for each participant, at which photo they rated as the most representative. They gave us three different ratings—say 1, 4 and 6—and we picked the photo that they gave the highest number to. From there we saw that liberals were about five times as likely to rate a lightened version of Obama as the most representative compared to a darkened version, whereas conservatives were about twice as likely to rate a darkened version as most representative compared to the lightened version.

That’s a great question. What we did was essentially take three different photos with three different poses, and created for each photo a lightened and a darkened version. And then we randomly selected the combination of pose and skin tone that we showed each participant.

Right. We were experimentally able to isolate the effect of skin tone because some people saw a lightened version of pose #1 and others saw a darkened version of pose #1—and independent of the pose the lightened versions seemed most representative to liberals and the darkened most representative to conservatives.

A little bit. Some of my research deals with how people who have different views on a subject are able to try to understand the views of someone on the other side, and the general finding is that people aren’t particularly good at really coming to understand the perspective of someone with whom they disagree. Beyond that, though, I got interested in this notion of whether our beliefs can actually affect the way we see the world—of whether they can actually affect our perception of objects or people in our environment. And it turns out they can.

Partisanship can affect all sorts of beliefs. It’s not surprising that a liberal and a conservative who read the same health care bill would come to very different conclusions about its merits. But I think our work is more akin to having a liberal and conservative look at the exact same physical copy of a bill sitting on the desk in front of them and disagreeing over how thick it is. That is, even something that we feel we should be able to see similarly, like a person’s racial identity or physical characteristics, can be influenced by our desire to see that person favorably or unfavorably.

There’s a long history in Western society of associating lightness with good and darkness with bad. Throughout history, throughout literature, et cetera. And we know now that these associations sometimes apply to the color of a person’s skin, and in addition to associating goodness with white, there’s some recent research in implicit attitudes suggesting that at an unconscious level people have a strong tendency to associate America with white. Which means that liberals, who are going to think that Obama is generally good and generally American, may have these subtle associations linking him to the concept of white, which is reflected in their representativeness ratings. The opposite would be true of conservatives.

That’s a great question. One of the things we’re trying to do now is experimentally try to tease those two options apart. Basically, what we have in our current paper, the one that’s out now, is correlational studies of Obama where we don’t really know what comes first or what’s causing what. The first study in the paper tries to address part of what you’re asking. If we get people to think about a novel candidate and simply manipulate whether they agree with a candidate or not, we can show that people who think this novel biracial candidate agrees with them later report that the lightened photos are more representative of him, suggesting that if you agree with someone then you may come to see him as lighter. From that we can speculate, exactly as you have, about the reverse path—and that is, seeing images of someone when his or her skin tone looks darker may cause people to like that person less than seeing images of that person with lighter skin tone.

We’ve actually just recently completed a new study that’s not in the current paper that looks at this question. We had people read about this new biracial candidate in the Department of Education, and for some participants we had them read this candidate’s biography with an unaltered picture accompanying the biography, while for some participants we had them read the biography with a picture of the candidate that had been lightened or darkened. Then we had them tell us how they felt about the six issues facing the Department of Education, and everyone was told the same thing—which was that this guy agrees with you on three of the six issues on the table, so it’s unclear really whether you like him or not. Then we asked them to tell us how much they supported him and how likely they’d be to vote for him if given the chance. And somewhat remarkably, the participants who’d seen a darkened photo just a few minutes earlier reported that they were less likely to vote for the candidate than those who’d seen the lightened photo.

I think our findings help explain the ways in which people may try to influence the level of support for, say, a biracial candidate. People have and may continue to strategically expose the public to images that alter certain characteristics of a person in the media spotlight. It reminds us of the Time magazine cover where an illustration had darkened an image of OJ Simpson following his arrest in 1994. Hillary Clinton’s campaign was actually accused of doing the exact same thing in the primary when it ran a television ad with a video of Obama during one of the debates in which the entire ad was artificially darkened. Although we didn’t find any direct evidence of this in our data, it’s possible that news directors may be susceptible to same sort of biases as our participants, without even really being aware of it, such that liberal and conservative media outlets may differ in the types of images of Obama that they tend to select and depict.

I wouldn’t advocate that people strategically try to manipulate things, but certainly political campaigns and ideologically-driven media outlets will always try to show their candidates in the best possible light.

Right. It’s the same as scrutinizing haircuts and clothing to make people as appealing as possible to the voters. With the Clinton ad, the goal was to try to make Obama appear more ominous.

Absolutely. It’s a short leap from “dark” to different, and from “different” to “dangerous.”

We expect to be objective, but if we can’t even agree on a picture sitting before us, it suggests that there are still deeper challenges to overcome before we can truly understand the perspective of someone we disagree with.

UPDATE: Reader jblackwell88 raises an excellent point. He or she writes: "The study showed liberals to show FIVE times the color BIAS and conservatives only TWICE the color bias. Not liking Obama because he's dark is the SAME as liking him because he's light.  How this interviewer can contort a study that shows color bias in EVERYONE to be a positive thing for the liberals is beyond me." I disagree with the part about me contorting the study; as I mentioned in one of my questions, MSNBC could potentially show lightened pictures of Obama that would reinforce or amplify the existing biases of its audience. Caruso himself explained how liberals may link Obama to "white" and "good" and "American" regardless of the efficacy of his policies. But jblackwell88 is right that I didn't ask Caruso to compare the SIZE of liberal and conservative biases here--i.e., the fact that liberals were five times more likely to find a lightened Obama representative and conservatives only twice as likely to pick a darkened Obama. It's a really interesting question. I'll put it to him today and get you an answer ASAP.

UPDATE II: Here's what Caruso has to say about jblackwell88's question.

What do you make of the fact that liberals were five times more likely to pick a lightened photo of Obama and conservatives only twice as likely to pick a darkened one? Does this mean that when it comes to Obama, liberals are more biased—i.e., that their perception, albeit positive, is even more divorced from reality?

Yes, there’s 5x and 2x difference in one of the studies using photographs of Obama (Study 3), suggesting a bigger difference for liberals than conservatives, but in the other study using Obama (Study 2) we actually found a bigger difference for conservatives (2.5x) than for liberals (1.3x). [Ed: Study 3 "examined whether perceived representativeness of the photographs were related to reported voting behavior in the 2008 Presidential election"; Study 2 "measured political party affiliation and examined its relationship to the perceived skin tone."] As a result, we can’t draw any strong conclusions about whether liberals or conservatives are ‘more biased’ because the numbers are not entirely consistent across the studies.