The Visible Hand

Let’s say you were in the market for an iPod and wanted to find a bargain, so you searched in a local online market like Craigslist to find one.? Would it matter to you whether, in the photograph of the unopened iPod, the person holding the iPod (all you can see is their hand and wrist) was black or white?? What if the hand holding the iPod had a visible tattoo?

I suspect that most people would say that the skin color of the iPod holder wouldn’t matter to them.? More people likely would say the tattoo might keep them from responding to the ad.

Economists have never liked to rely on what people say, however.? We believe that actions speak louder than words.? And actions certainly do speak loudly in some new research carried out by economists Jennifer Doleac and Luke Stein.? Over the course of a year, they placed hundreds of ads in local online markets, randomly altering whether the hand holding an iPod for sale was black, white, or white with a big tattoo.? Here is what they found:

Black sellers do worse than white sellers on a variety of market outcome measures: they receive 13% fewer responses and 17% fewer offers. These effects are strongest in the Northeast, and are similar in magnitude to those associated with the display of a wrist tattoo. Conditional on receiving at least one offer, black sellers also receive 2-4% lower offers, despite the selfselected-and presumably less biased-pool of buyers. In addition, buyers corresponding with black sellers exhibit lower trust: they are 17% less likely to include their name in e-mails, 44% less likely to accept delivery by mail, and 56% more likely to express concern about making a long-distance payment. We find evidence that black sellers suffer particularly poor outcomes in thin markets; it appears that discrimination may not “survive” in the presence of significant competition among buyers. Furthermore, black sellers do worst in the most racially isolated markets and markets with high property crime rates, suggesting a role for statistical discrimination in explaining the disparity.

So what can you conclude from this study?? The clearest result is that if you want to sell something online, whether you are black or white, find someone white to put in the picture.? I suppose you could say that advertisers figured this out long ago, and actually go one step further, making sure the white person is also a good looking blond woman.

It is much harder, in this sort of setting, to figure out why buyers treat black and white sellers differently. As the authors note, there are two leading theories of discrimination: animus and statistical discrimination.? By animus, economists mean that buyers don’t want to buy from a black seller, even if the outcome of the transaction will be identical.? Buyers wouldn’t like black sellers, even if black sellers provided exactly the same quality as white sellers.? With statistical discrimination, on the other hand, the black hand is serving as a proxy for some sort of negative: a higher likelihood of being ripped off, a good more likely to have been stolen, or maybe a seller who lives very far away so that it will be too much trouble to meet in person to do the deal.

The most impressive part of this paper by Doleac and Stein is their attempt to distinguish between these two competing explanations: animus versus statistical discrimination.? How do they do it?? One thing they do is to vary the quality of the advertisement.? If the ad is really high quality, the authors conjecture, maybe that provides a signal that could trump the statistical discrimination motive for not buying from the black seller.? It turns out that ad quality does not matter much for the racial outcomes, but possibly this is because the quality difference across the ads isn’t great enough to matter. The authors also explore the impact of living in an area with more or less concentrated markets, and also across places with high and low property crime.? Black sellers do especially bad in high crime cities, which the authors interpret as evidence that it is statistical discrimination at work.

I really like this research a lot.? It is an example of what economists call a “natural field experiment,” which has the best of what lab experiments have to offer (true randomization) but with the realism that comes from observing people in actual markets, and with the research subjects unaware they are being analyzed.

Eileen M. Wyatt

This study had no control group. Most CL "for sale" ads include no image of the seller at all.

That doesn't excuse prejudice, but it does make me wonder how the "white hand" ads would have done in relation to "normal" CL ads with no seller image.


Does this type of research have to meet any ethical guidelines? Should the research subjects know that they are participating in a research program?

Do they wind up actually selling the ipods?


Even intuitively you can tell that people would make buying decisions depending on the looks of the person selling it. Its for nothing that you see attractive looking people sitting at the receptions of super costly consultancies

Shaun G

One way to determine whether it's animus or statistical discrimination at work is to find out the races of the potential buyers.

In other words: Do blacks discriminate against blacks, or is it just whites who discriminate against blacks?

If the former, it points to statistical discrimination; if that latter, it points to animus.


Statistical discrimination in this case is bunk. There is no data that, once controlling for systemic variables, demonstrates there is any reason to conclude what the writer here suggested. Racism is racism, whether it's based on emotion (animus) or seemingly based on "facts" (statistical discrimination). Even if it could be proven that a black seller is more likely to be offering stolen goods or some other type of shadiness, that gives you no information about the specific individual you are dealing with. The tone here implies that racially profiling is potentially legitimate. And it never is.

Once upon a time

Eileen, you don't need a 'neutral' control group when you're not looking for an absolute value. Directly comparing black vs white will tell you which is higher or lower than the other, even if you don't know whether that makes it "higher than neutral" or "lower than neutral".

Steven, you missed another potentially confounding factor: The existence of online scams based in Africa. It's not just that the iPad may have been stolen locally.

Craigslist has a huge problem with African people (not African-American -- people currently living on the African continent) trying to scam people. It only takes one round of attempting to rent a house or buy a computer from an African scam artist to make you wary of any long-distance transaction that *might* be another African scam. These scams are why they spend so much effort to encourage face-to-face sales.

If you're looking at an image of a black person, the odds that the poster is in Africa instead of a fellow American have gone up.

(There are about 30 Africans with black skin for every American with black skin, although only about 4% of those 30 have internet access... which works out to about 50-50 odds that a randomly selected internet user with black skin is in Africa.)

This means, necessarily, that the odds of a poster with dark skin being a scam artist are higher than the odds of a poster with white skin. This is another negative externality of the crime in Africa.



Furthermore, think of the larger implications of this study. If we accept the findings to be true (and I would not be surprised if they were) and do nothing to change such appalling behavior, what is the inevitable result? People, both black, white, and everything in between, will exploit these biases. Criminals or people with nefarious intentions will put up glitzy ads with pretty white women. Even well-intentioned people will drive up the cost of their product by having the "right" hand be in the photo. These are the hidden costs of racism. So, conceding that the race of the hand in the photo is having an impact, what do we do about it? How about we work to remove these biases, whatever their cause is, instead of accepting and, therefore, legitimizing them?


I've noticed this for many years. I'm a freelance graphic designer with darker skin and black wavy hair in North Carolina. I was always hesitant to put my picture on my website for fear of some underlying prejudice in people. I noticed early on that when speaking and dealing with clients solely over the phone and via email I had much success with getting the job and being paid top dollar. There were cases where the phone conversations were promising. They loved my work and what I said. Then they wanted to meet in person. In every case, when this happened, I never heard from them after the meeting.

Then came Social Media. I lost three clients during the boom of Facebook. Two of which were well-paying, job-consistent clients. I never met these clients in person, then they requested my friendship. After I accepted their request, the worked seemed to dry up. Now It's much harder for me to get work. Add to the fact that I'm a former Collegiate linebacker standing at 6'2 and 230 pounds. I just don't fit the image of a graphic designer in most peoples minds.



I suspect this means people prefer to do business with people more like themselves. Trust is a big issue with Craigslist, and it is easier to trust someone who looks more like you. Also probably a perception that blacks in general are less trustworthy. I wonder what happens if the people in the ad is Asian?

Eric M. Jones

I think it is important to know whether or not the hand was male or female, too. Black men might get more animus discrimination than black women.


Add this to the long line of evidence that human's just like light skin better than dark. Could there be any scientific/evolutionary reason for this?

For what it's worth, I always have women pose in pictures of everything I sell on craigslist. I think it makes me look more trustworthy.

Ian Kemmish

In a similar vein to Shaun G's comment, as soon as I read the statistic about black sellers receiving 2-4% lower offers, I immediately wondered whether they were simply receiving offers from a pool of prospects which included more black people who (is it fair to assume?) would have had less money for discretionary spending, and whose lower offers would have dragged the average down. In turn, the disposable income of the prospect might well also influence what you've labelled "trust". Without statistics on the breakdown on the prospects in each case, it's hard to draw conclusions from the average offer price alone.

A thought for a follow-up experiment: a white hand in all the ads, but in half of them, an Anglo-Saxon contact name, and in half of them a Slavic name?

Eileen Wyatt

Once upon a time, my point is that showing any seller is so unusual that the coloration of the seller may become far more salient than it would be in ordinary transactions with persons of that color or tattoo status.

So what you'd be measuring would not be ordinary buyer attitudes toward darker or lighter persons they meet in the course of the day, but buyer attitudes when they are consciously evaluating skin color.

It's not unusual for psychological studies to make some idea salient and then discover that subjects' subsequent performance is affected by the salient attribute.

Additionally, since a CL ad with a seller image is so unusual, many ordinary CL users may self-select away from responding to that ad.


@BSK - You ask why we even distinguish between Stats vs. Animus. If it's statistical, you don't have a racial problem: you have a 'afraid they'll be screwed' problem. Take out the screw, and there's no problem. If it's based in animus, however, then you've got a bigger problem. That might only be fixed with a black man saving their kid from a burning building.

Know your enemy. That's Sun Tzu for you.


@Once upon a time: While Africa definitely has its share of scam artists (everyone's heard from a Nigerian prince), are they disproportionate to the number of similar people in Eastern Europe?


One of the most interesting aspects of this article hasn't been addressed yet. "These effects are strongest in the Northeast..." If you're going to associate this study with racism, then it seems like someone should address the regional differences that go directly against what many people assume is common knowledge. That Southerners are ignorant hicks who propagate racism while educated people from the Northeast have evolved beyond that.


@ BSK (#7): What makes you think this isn't how advertising is designed now? Take a look at the racial makeup of TV and print ads.



If there is an ad for iPod with a blue board that says: 20% chance of getting screwed, and a green board that says: 50% chance of getting screwed. Does choose the blue board make me animus-discriminate or statistical-discriminate? Or it doesn't matter and just make me a colorist?

Bob W.

So let's say the hypothesis is simply "white people are intimidated by black people (and tattooed white people." Where does that fit?

It's a pretty parsimonious explanation as well.


"@BSK - You ask why we even distinguish between Stats vs. Animus. If it's statistical, you don't have a racial problem: you have a 'afraid they'll be screwed' problem. Take out the screw, and there's no problem. If it's based in animus, however, then you've got a bigger problem. That might only be fixed with a black man saving their kid from a burning building.

Know your enemy. That's Sun Tzu for you."

Fail. First off, there is no data here demonstrating one is more likely to be screwed by a darker skinned person than a lighter skinned person. Stereotypes does not equal stats. My contention is that animus is present regardless, because the assumption is that a darker-skinned person is more likely to screw than a lighter-skinned person, something that simply isn't backed up by an legitimate stats.


If there is an ad for iPod with a blue board that says: 20% chance of getting screwed, and a green board that says: 50% chance of getting screwed. Does choose the blue board make me animus-discriminate or statistical-discriminate? Or it doesn't matter and just make me a colorist?
- YX "

Fail. No such boards exists. A proper analogy would be the two boards but no difference between the two and one person simply assuming a difference exists.

The idea that there is a statistical discrimination being demonstrated here begs the questions that there is a statistical backing to that discrimination. There isn't. The writer didn't even offer one. Instead, they played on the statistics relating to crime rates and race which are so fundamentally flawed as to call into question anyone who would rely upon them. Furthermore, the implication is that people are consciously sitting there and thinking, "I have an X% chance of getting screwed by the white guy and a Y% chance of getting screwed by the black guy." Racism is not that academic. Instead, subconscious, underlying biases are the problem here. People have been socially conditioned to equate white with good and black with bad. To have these biases does not make one a bad person, as they are often unaware they have them and are the result of living in a white supremist culture. That being said, to act as if the problem lies at the feat of black people, either here or in the African continent, is fundamentally flawed. There is ample evidence to demonstrate that white people are just as undeserving of trust online as PoC's, yet that does not fit the social narrative so it gets ignored. If I came up with a stat that showed white people were more likely to scam, would you be comfortable saying that these numbers would be acceptable if they went the other way? Probably not. This is just another example in a long line of them demonstrating the inherent prejudice that still permeates our society and how privilege/oppression are as real now as they ever were.