How light or dark is Barack Obama's skin? Depends on your political stance...

In the early days of the last US elections, Hillary Clinton's campaign was accused of deliberately darkening Barack Obama's skin in a TV ad. The implication was that by highlighting Obama's "blackness", Clinton's camp was trying to exploit negative associations that voters might have with darker skin. But you don't need editing software to do that - a fascinating new study suggest that people literally change the way they see a mixed-race politician, depending on whether the candidate represents their own political views.

Liberal American students tend to think that lighter photos of Barack Obama are more typical of him, while conservatives think he's best represented by darker photos. You can see this effect even after adjusting for any racial prejudices, be they hidden or overt, and even with a person less famous than Obama. And regardless of political views, people who associated Obama with lightened photos were most likely to vote for him.


Eugene Caruso from the University of Chicago, who led the study, thinks that this effect is the result of two biases: the positive associations of white and lightness among some Western cultures; and the tendency to view people of the same group (political or otherwise) more favourably than those of another group. He says, "Group membership provides a lens through which people generate representations of reality."

Caruso asked 221 students about their political ideologies and then showed them three photos of Obama and three of John McCain. On the grounds that some photos can capture the "true essence" of a politician better than others, the students were asked to rate how well each photo represented each man. But unbeknownst to them, two of each set of pictures had been altered with Photoshop, so that the subject's skin tone was either lighter or darker.

When it came to McCain, the students' political leanings had no bearing on their choice of photos. For Obama, it was a different matter - liberal students were more likely to pick the lightened photo as the one that represented him best. Conservative students were more than twice as likely to associate him with the darkened photo. These biases were reflected in the students' votes. Whether liberal or conservative, the more people associated Obama with the lightened photo, the more they were likely to vote for him.

Of course, this effect could simply be down to racism - people who harbour prejudices against Blacks would be more likely to associate Obama with a darker photo and less likely to vote for him. But Caruso accounted for that - he repeated the experiment with 49 people a week before the last election and specifically evaluated the recruits' attitudes on race.

Each of them filled in a questionnaire called the Attitudes Toward Blacks scale, which asked them whether they agreed with statements such as "Generally, Blacks are not as smart as Whites." Obviously, people can lie on these questionnaires, so each volunteer also did an "implicit association test" (IAT), designed to reveal any hidden prejudices (try one here).

The same patterns emerged as before and this time, Caruso found that they remained even after adjusting for racial attitudes, both hidden and explicit. A week after the election, Caruso caught up with his recruits and confirmed that those who thought the lightened photos represented Obama were actually more likely to have voted for him. Those who linked him to the darkened photo were more likely to have voted for McCain. Amazingly, the photo effect turned out to be a better indicator of voting choice than the scores on either of the two prejudice tests.

Barack Obama's fame is perhaps a bit of a distractor since people can judge him on his policies, personality and more than just his skin colour. But Caruso found the same effect using a non-celebrity.  He asked 102 students about their views on important educational issues and then showed them three photos of a man allegedly running for a position in the US Department of Education. He told them that the mystery politician either agreed or disagreed with most of their stances.


But Caruso kept two important things from the recruits. First, the 'politician' was actually Jarome Iginla, a mixed-race ice hockey player whose father was a Black Nigerian and whose mother was a White American. None of the students twigged to this. Secondly, as before, some of the photos had been doctored so that Iginla's skin tone was either lighter or darker.

The same trend emerged. Caruso found that students who were told that the politician supported their views were more than twice as likely to pick the lightened photo as the most representative one. Those who thought that Iginla disagreed with them were more likely to associate him with the darkened photo.  And across the board, people who picked the lightened photo were most likely to vote for him. 

Across all three experiments, the way that American students literally see a mixed-race politician depends on whether they agree with his views. If they felt aligned with a candidate, they tended to mentally lighten his skin, and Caruso suggests that this might reflect subconscious associations of white with good, and black with bad. Sadly, the study provides no information about the ethnicity of the students involved - it would be very interesting to see if the same bias in perception applies to viewers who are themselves Black.

Reference: PNAS  doi:10.1073/pnas.0905362106

More on politics:


More like this

I know I don't blog about pure politics much, but it's the weekend, I'm too tired to do anything heavy-duty about medicine or science, and this depressed me. As much as I'd like to delude myself that things have changed, it turns out that they haven't changed nearly as much as I'd like to think, as…
It's called "social desirability bias". And the voting public suffers from it. It leads likely voters to "underestimate their own prejudices when talking to survey takers", says Dalton Conley in the Chronicle Review. We know we are supposed to treat all candidates the same, regardless of race…
Those of us who have been on the receiving end of racial abuse know all too well that words can hurt. But they're also the tip of the iceberg. According to a study of popular US television, we're exposed to the spectre of racial bias on a regular basis, all without a single word being uttered.…
Elections are weighing heavily on our minds. In three short months, America will see the race between Barack Obama and John McCain come to a head, while we in Britain will probably have a general election within the next few years. Some people, of course, will vote based on long-held loyalties to…

I find the Obama photos highly dubious in terms of the purported manipulation--the degree of proper control is quite poor. This if for two reasons.

First, the leftmost ("light") one has Obama in a casual blue shirt--almost a workman's uniform shirt--with the sleeves rolled up, and with green vegetation behind him. He is also close up in the photo. The rightmost ("dark") photo has him in a formal executive's suit, with a symbol of government behind him, and he is at a somewhat greater distance. In the light one, he is addressing a crowd; in the dark one, he is just standing there, with arms folded, the universal sign for "back off". For these reasons alone, I could imagine those favorable to Obama picking the left one over the right (and vice versa), even if the skin tone had been made to be exactly the same.

Second, the tone is not comparable, because in the leftmost one he is standing in bright sunlight, and in the rightmost one he is standing mostly in shadow. When an object appears to be illuminated, our judgments about its inherent objective luminosity is very different; this is the basis for several jaw-droppingly compelling optical illusions, like this class one. So, it may be the case that a medium brown Obama on the left in sunlight can be perceived as "really" being darker than a slightly darker medium brown Obama on the right in shadow!

cm - interesting point. however, it sounds like the trend was similar using the jerome iginla photos where there was a little less variation in dress & style.

the positive associations of white and lightness among some Western cultures

as opposed to how color blind asian and middle eastern cultures are? :-) this is likely a cultural universal. from what i know africans also have skin color prejudice.

as a person of approximately obama's complexion i would add that it seems photos of me vary a lot in perceived complexion based just on the key of the lighting. more so that someone who is very dark or very light would. perhaps there needs to be some work done in the granularity of human perception on this sort of thing.

Okay I get the photos I've stuck up are a bit misleading. A bit more explanation is required: For every photo, they produced lightened and darkened versions. From the paper:

"Using Adobe Photoshop CS3 (version 10.0) software, we created two alternative versions of each photograph: one in which his skin tone was artificially lightened, and one in which it was artificially darkened. We accomplished this by isolating any exposed areas of skin (i.e., the head and hands) and adjusting the brightness and contrast settings by 15% in the appropriate direction. All other aspects of each pose were identical except for the darkness of the candidateâs skin (see Fig. 1 for examples).

The particpants had to choose between photos from different sets - this makes sense, for if they showed all three photos of a set to the volunteers, they'd soon figure out the point of the experiment. The stats reflect the averages, so they should account for differences in pose, lighting etc.

I agree with you, cm. These photos were not neutral at all (i.e. controlled for other factors). If asked about 'true essence' I'd likely have picked the one on the left for exactly the reasons you describe: light-heartedness, casual, unconstrained.

And unlike my south-of-the-border friends, I immediately recognized Jarome Iginla and know the second photo is most accurate to his actual appearance. In any case, the second photo lineup is also lacking controls for other factors, specifically, distance (the darkened photo being the furthest away).

Okay, Ed. I see your clarification (was posting at the same time as you). Thanks.

Interestingly when President Obama is Home in Hawaii, he likes to go body-surfing at Sandy Beach in the blazing Hawaiian sun. Barack doesn't care about skin tone. As for the Clinton Campaign's darkening the pictures it was probably an inexperienced campaign worker who mistakenly applied the darkening to Barack's picture instead of the originally intended and much needed destination: Hillary's cadaverous whiteness.

By OftenWrongTed (not verified) on 23 Nov 2009 #permalink

Also Razib, I wasn't implying that racial prejudice is unique to Western cultures. But I didn't think that linking white to goodness specifically would extend to non-Caucasian cultures. Any evidence that they do?

I've taken that IAT test, and it made my head spin. They switched the "associations" back and forth so much while expecting a fast reaction that I made mistakes mostly because my mind was trying to develop a pattern for speed that was constantly interrupted. I don't know how much stock I'd put into it.

I agree it would have been great to know the ethnicity of the students who participated. Interesting none-the-less. Thanks for the post.

There is no science in the IAT. It is just pushing buttons. How do you know the results are immutable? (e.g. constant across a person - as opposed to across just generated due to the events of that day?) And isn't the racism, or purported racism of the participants a basis for the test in the first place? It wasn't DESIGNED to measure people's LACK of racism - but rather to catch them being racist. There's a lot of experimenter bias here.

If you ask me, it's the kind of science that was designed by someone who stands to gain if they can talk about racism. It has no credibility in my mind. (PS: I scored, "no difference" in preference - so if you want to go calling me a racist, it just goes to show your test is crap.)

Ed, thanks for clearing that up. Sounds like they had their Good Experimental Design helmets on after all.

That IAT may valid for a population, because it randomizes the order of the association tests - associating bad/european before or after bad/african.

For individuals I'm not sure how accurate results are when you're first trained to associate good/european, and then have to retrain yourself to bad/european, as I was. I will be interested to see how I do in a years time should I remember to retake the test, knowing how the test works, and hopefully getting a reversed order.

For me it showed an expected moderate preference for european - I had indicated as much - unsurprising, as I'm a white male in a northern european country with a thing for redheaded girls. African males just don't excite me...

But I didn't think that linking white to goodness specifically would extend to non-Caucasian cultures. Any evidence that they do?

east asian cultures are racist against darker-skinned people in a way qualitatively different from the way they view whites. or at least that's what my relatives in japan and china have said when working there (these individuals were in white collar positions, so it wasn't just class). arabs also look down are south asians as dark skinned and ugly. as for south asians themselves, bollywood says it all. telling a south asian that they're dark-skinned won't be viewed as an objective assessment, it'll be seen as an insult. lighter skinned skinned punjabis viewed the darker skinned bengalis as racially inferior, and mass rapes during the 1971 civil war of bengali women by punjabi soldiers was justified in part on race improvement (i.e., "improving the stock of the black bengalis").

here's a blog post from james fallows in the atlantic,…
I have experienced this on a number of occasions. But after living in China for a while I realized that what we would consider racism in the West is simply a deeply ingrained cultural characteristic of mainland Chinese people. White skin (the Chinese like to consider themselves white) and or being a Han (the dominant ethnic group) means a person is good. Dark skin or not being Han means a person is inferior (and more likely to be a bad guy/a thief/incompetent etc.). It does not equal KKK style hatred. It does not even mean a Han Chinese wouldn't be friends with a person from India or Africa. It simply means that if a person is non-white or a member of certain Chinese minorities, they simply are to be considered less smart, less competent and less trustworthy than the average white person or Han. [Ed note: This accords with my observation, with the caveat that I have observed this all as a middle aged white guy. Early discussion of Obama in China fit this pattern -- but changed after he took office.]

anyway, the chinese are simply the most salient example right now because of their rise internationally. i've met individuals from the highlands of ethiopia who look down on darker-skinned ethiopians from the lowlands, the highlanders apparently consider themselves "yellow" or "red" in color as opposed to "black."

The ruling classes are almost universally associated with paler skin. Dark skin = works outdoors, pale skin = loiters indoors.

But things change. In the (post-)industrial age pale-skinned Europeans can afford spending their vacations in sunshine, and now a good sun tan is seen as a sign of wealth. Pale skin = works like a slave, dark skin = lots of free time.

By Lassi Hippeläinen (not verified) on 24 Nov 2009 #permalink

There is one possible confound: "lighter" images may simply make general features more prominent. In the "darkened" pictures that you show above, it is actually more difficult to see what the person looks like. Perhaps this effect is stronger for darker-skinned people? Could this explain some of the difference in the results?

I know of only one case where the "whiter" group is looked down upon: the Pashtuns/Pathans/Afghanis of South Asia. They are the palest of all South Asian groups, yet they are widely regarded as uncouth dimwits by their (slightly darker) neighbours in Pakistan (esp. the Punjabis).

Curiously, Pathans even have a reputation for habitual homosexuality. Apparently, using deviation from the sexual norm as a slur seems just as universal as colour-based prejudice.

So are people who think he's _really_ black unenlightened or are they simply being kept in the dark?!

I think your last sentence is really important: "Sadly, the study provides no information about the ethnicity of the students involved..."

I suspect most of those surveyed were white. The subconscious thought process might not be "white = good" but "white = like me." If they associate themselves with the candidate's political views, they think of themselves as similar to the candidate, and then it's easy to see that they might associate the candidate with their own race.

#16 ... That may have a good bit to do with it.

It's hard to darken McCain enough to make his features hard to discern.

By Tsu Dho Nimh (not verified) on 24 Nov 2009 #permalink

I guess I'm atypical but the darker photos just look badly exposed to me. (Maybe because I have photo experience.)

I think it might be wise to re-do this experiment with well-lit photographs, to control for some odd variables that likely affect pictures of brown people more or differently than lighter colored people.

Without stopping to think about it, the three pictures of Obama look to me like the same guy with about the same skin tone, just under different lighting conditions, and the darkest one seems backlit and underexposing the subject. (Note the dark eye sockets and not being able to see his eyes, at least in the lo res pic above.) It's a crap picture, and people who like Obama might have just have a higher standard of what a good picture of him would consist of, without that being directly about skin tone.

The hockey player pix are a bit different. The left one looks washed out, like the picture was underexposed and they tried to fix it but the darkest shadow information just wasn't there.
All three photographs are badly lit---too much light from the sides/back and not enough from the front, but that shows up most in the darkened one---the central face goes really dark.

I'd recommend doing this experiment over with properly lit pictures, specifically standard 3-point lighting. (Strong key light from in front of the subject, toward one side, and above, plus a weaker fill light from the front and off to the other side, plus a bit of top lighting to highlight boundaries, etc. and make the picture pop, i.e., have depth and crispness.)

I suspect you can find several pictures of each politician with that kind of lighting, because it's the single most common default lighting arrangement when photographers have a choice. (And approximations are the second most common kind, e.g, using fill flash when there's no fill light.)

I suspect most of those surveyed were white. The subconscious thought process might not be "white = good" but "white = like me." If they associate themselves with the candidate's political views, they think of themselves as similar to the candidate, and then it's easy to see that they might associate the candidate with their own race.

IAT tests have long shown that blacks are racist against blacks and women are sexist against women. so controlling for race probably won't eliminate that issue. are fat people any less prejudiced against fat in terms of aesthetics? perhaps somewhat, but not much.

I know of only one case where the "whiter" group is looked down upon: the Pashtuns/Pathans/Afghanis of South Asia. They are the palest of all South Asian groups, yet they are widely regarded as uncouth dimwits by their (slightly darker) neighbours in Pakistan (esp. the Punjabis).

many south indians view north indians as better looking (lighter, more european features) but less intelligent (south indian brahmins used to staff the indian civil service positions all across northern india). when i was talking about colorism, i was being narrow in terms of phenotype as to the evaluation. my personal experience is that bengalis do feel inferior to punjabis in physical appearance because we are darker and smaller (i have known bengalis who pride themselves on not looking bengali, and these are invariably the punjabi looking ones, not the east asian looking ones), but in no way do bengalis feel culturally inferior. on the contrary.

as i have noted elsewhere, in the chinese imperial examinations the southerners tended to do better than northerners. to the point where there were quotas on the southerners to maintain regional balance. but to my knowledge the chinese perceive white skin as more beautiful than darker skin.

I thought they all represented him pretty well, but one seemed overexposed and on another his face was shadowed. They didn't change his colour. Ditto the second example. Who would be fooled by that? Don't their minds compensate by what the rest of the picture is like?

I knew Chinese friends very much,and my friend,girl,was dating a Korean guy.Her family,Chinese,was absolutely so against this;he was considered inferior.(she married Chinese.)Young Chinese do date white people,but their families are very against it.(that does not always work.)Now,this gets complicated;my dad was born in China,raised there till he was 14,but he was white.Most of his friends were Chinese(not all.)He viewed them as equal,and even superior.When a young Chinese man was interested in me,(he was a friend of my dad)my dad totally approved.

To this day,I still like Oriental men,they appear attractive to me,and when I would go out in public with my roommate(who was half Japanese)white people would stare at us together(even in a big city)and I was very annoyed because that is so rude to stare.Ditto when I went out with a Philipino friend,male.My sister had numerous Chinese girl friends,and family friends.My mother was very bigoted that I had this Japanese roommate,but I was,"Duh,mom,dad was practically Chinese!So what?"African-American men,I was not attracted to;and I did not like their personalities.(however,I often have african american girlfriends.)

I'm unusual;my upbringing was very different,and I do tend to like Asians a lot.(I have a Chinese doctor I really like.)It was my family that influenced me,not the society.Even my mother did not discourage me in that,and she's pretty "white-bread."I think Pres.Obama is just as white,as he is African-American.I just do not like Democrats anymore.I've had some very bad experiences with black men,and that influenced me,too.So,how can you categorize me?

The definition of racism is confusing here. To me, seems like
both those who picked lighter photos and darker photos are
racist - both seem to consider skin color important one way
or the other. Weren't there people, non-liberal and
non-conservative I guess, who picked the unaltered photo?

Three thoughts.

1) If the people who agreed with Obama and tended to vote for him thought his skin was lighter than it is... AND if they were white people, isn't that racism just as much (and perhaps less realized) than those who thought he was darker because they disagreed?

2) Paul W. (#20) is absolutely right. Plus, anyone who has fiddled with photographs in any photo program can recognize artificially lightened and darkened photos. I'm not a professional photographer, but I'm a decent amateur and the most difficult shots I've taken were of my pale-skinned daughter, her dark-skinned husband, and their in-between baby.

3) I've noticed that in the south, very light-skinned white people with pale blue eyes and natural white-blonde hair are considered 'dumber' than if their eyes are darker blue and their hair has a more golden/brown tint.

A couple more photo tips for anybody who might do this sort of experiment, or just anybody who wants to take good pictures:

1. Don't take straight-on pictures of people's faces most of the time, especially if you don't have good lighting.

Straight-on pictures are much more ambiguous in terms of how to infer shape from shading, because the face is basically symmetrical---the left half of the picture doesn't give you much different information that the right half.

An off-axis picture (say, 30 degrees off axis) much more clearly and vividly reveals the actual 3D shape of the head and face, because you can see one half-head from a perspective where it's turned partly toward you, and the other turned partly away. That makes the the shape of the head unambiguious in a way that he viewer's brain instantly infers the right head/face shape. (The visual cortex is hella good at this, especially with regard to faces but you've got to give it some perspective disparity to work with.)

2) Don't use a typical flash on the camera as your main light if you can avoid it. It makes the lighting weird in a way that the brain does not like at all because you almost never see things lit that way in a pleasant environment. (Even car headlights aren't that close to the line of sight.) This mis-cues the brain and makes it likely to infer the wrong 3D shape. (Things generally look elongated toward the camera, because the shading is so dramatic as you go far around the sides of objects. Things toward the front part of a curved surface look flattened, though. Bleah.)

If the ambient light level is enough to get by without flash, but your camera has a "fill flash" setting, it's usually a good idea to use it. (It's the only thing a flash mounted close to the line of sight is good at.) The fill flash is a weak flash, which doesn't affect the brighter parts of the image noticeably, but lightens the darker parts a lot. (That's because of the logarithmic scale of intensity that our eyes use---a little bit of light changes the shadows by a big ratio, and the highlights by a tiny one, because the darkest parts of the image are many, many times darker than the lightest, in terms of raw quantity of light.

Most indoor pictures that most people take suck because of the first two issues, combined. They have people look right at the camera, which is hard on the brain due to lack of depth cues, then use a flash from the camera, which is outright misleading in terms of depth cues.

If you take pictures like that, the camera always lies, because you are basically lying to the camera both by omission (of perspective) and explicitly (by bogifying the lighting).

Some people may look better in pictures like that, but the large majority of people look worse. (My wife being a really vivid example---she's actually nice looking, but most people take ugly pictures of her. By following the above guidelines, I can consistently take nice-looking ones.)

Oh, by the way, the fill flash tip above is especially helpful when taking pictures of both light- and dark-skinned people. It lightens up the dark skin tones, especially the shadowy areas, more than the light skin tones.

(Don't worry about whether that's "realistic." Given the differences between normal viewing conditions and photo-viewing conditions, it's the least of your problems, and it usually makes things look more realistic overall. You want to get both the highlights and the shadows comfortably within the intensity range of your camera's sensor, or you'll get muddy shadows or blown-out highlights. If you're very worried about "realistic" skin tones, fiddle with the brightness and contrast in an image-processing program later. When shooting, just make sure you get a decent exposure.)

Paul W. -- thank you for the hints, especially about straight-on facing the camera. I have some photos that have turned out OK that way, but they are all with 100+ length lenses. Actually, I was far enough away that they probably were at least 10 degrees off straight on.

And most of them are of babies. Babies are different!

If I had the chance to re-stage the photos of my daughter's family, I would have had my daughter facing the natural light and my son-in-law getting the fill light from the reflector.

I put six layers of waxed paper over the built in flash on my camera years ago and that has definitely helped. I'm not dealing with redeye anymore, at least. But people are always asking what's wrong with my camera.

I do want relatively accurate skin-tones, but I'm more interested in getting well-defined facial features and capturing an expression.

Personally, I think all three photos of Obama above are bad, but the middle photo of Iginla is OK.

In commercial advertising, when an African-American couple is portrayed, the male invariably has darker skin than the female. In fact, Barak Obama is one of two men I can remember seeing in the media who has a wife who is darker skinned than he is. I have wondered for years why this is so. I have been told by my African-American friends that lighter skin is preferable, and that some people have "better," less "nappy" hair than others. If anyone can explain this, I would appreciate it.

Looks to me like all the photos are of the same skin tone, with more or less light being *reflected*-- I would end up assuming some of the photos are taken in harsh lighting, some with reflectors to get a more even tone, and some with a lot of shadowing.
However-- the darkest photos make their smiles look less genuine.
Perhaps people have gotten used to how photography effects perception of skin tone... but still judge smiles?

Not that I doubt racism is still very common in society, including subconscious racism. I just don't think, looking at the photos side by side, it looks like the skin color changes, only the lighting.

By Samantha Vimes (not verified) on 28 Nov 2009 #permalink

Now that I've read Ed's clarification in the comments.
Okay, so they did show different students lightened and darkened versions of all the photos? Cool, then unless the darkening process actually makes the smile look strange, the experiment is well designed.

And I thought the guy in the bottom photos looked like David Lister from Red Dwarf.

toto, I understand the K!ung-san are looked down on by their darker neighbors. The hunter-gatherer lifestyle is a big part of that, but they are distinctively different because of their light skin.

By Samantha Vimes (not verified) on 28 Nov 2009 #permalink

Samantha, you're right, he totally looks like Lister.

Although I agree with many of the posters that lighting etc can make a photo look overexposed/poorly lit/etc, I'm not sure that would explain the experimental results. If it was just the lighting and not the skin tone, I wouldn't expect such a strong correlation between political views and which photo was chosen. Why would conservatives think an underexposed photo is more representative (i.e., if they just interpret the right-hand photo as underexposed, why would they choose it in preference to the others, which they would presumably recognize as properly exposed)? It seems that if that were the case--if there were easily recognizable problems with the photos--then all of the respondents should have chosen the unaltered photo.

Also, I do agree with other posters that this test suggests racial bias in ALL respondents, not just the conservative ones. Associating positive traits with a person because of said person's race is still racism, even if it may have slightly more benign effects.