Baby’s Got Blue Eyes

While taking care of some administrative details, I stumbled across a student’s senior research project from a couple of years ago that I thought was pretty fascinating. The student was an African-American female, who had brown eyes but often wore blue contact lenses. She worked as a waitress at a local restaurant, and decided to do a field experiment to test how people responded to her eye color(s). To avoid a wholly subjective evaluation she decided to focus on tip percentages, comparing nights she went natural vs. nights she wore her contacts. The results were intriguing–when displaying her natural brown eye color, her tip rate averaged less than 15%, while when she had blue eyes it averaged over 17%, and only night did it fall below 15%.

She also recorded comments that she received. Keep in mind her clientele was predominantly white, but it’s not as if black folks are at all unusual in our town (only about 4.5% of the population is black, but it’s a small town so it’s not like they’re cloistered far away from the eyes of us white folks). The two comments that caught me as just being crass, while attempting to be complimentary, were:

  • “You are the prettiest black girl I have ever seen.”
  • “You have pretty eyes girl, all the black guys must want you.”

And she noted,

When told that the blue eyes were contacts approximately one half of all guests seemed disappointed while the other half were careful to mention, “You’re still beautiful regardless”.

The student took this well, commenting that she understood they were trying to be complimentary, but noted that she felt, “an underlying sense of dissatisfaction, with hidden language suggesting that I did not quite measure up.” Of course that’s not very objective data about what the customers were thinking, but it’s pretty objective data about this particular African-American female’s subjective response.

Some racism is subtler than other racism, and some is wholly unintended and unconscious. Keep that in mind next time you tip your waitress.

About J@m3z Aitch

J@m3z Aitch is a two-bit college professor who'd rather be canoeing.
This entry was posted in Research Blogging. Bookmark the permalink.

21 Responses to Baby’s Got Blue Eyes

  1. Interesting stuff.

    I’m curious: did that student have any data about the percentage tips received by other waiters / waitresses, and those workers’ race, gender, and eye color?

  2. AMW says:

    Did she have enough data to compare within-race responses? For instance, was the bump in tips only among whites, or did black patrons also tip more when she had the contacts in?

    I ask for two reasons. First, I have read that blue eyes are a pretty recent human trait, maybe only a couple tens of thousands of years old. For a trait that recent to have spread as broadly as it has suggests that it is a beneficial one (at least in certain environments), and it could be that the average human (of any race) has some preference for blue eyes. Second, I know that within black communities there is often color distinction, with lighter skin considered more desirable than darker skin. It may be that blacks favor blue eyes for similar reasons that they do lighter skin.

  3. AMW says:

    P.s. I think the concept of the paper is fascinating. Do you think she would consent to having you post it or at least e-mail me a copy?

  4. Lance says:

    I agree with your statement,

    “Of course that’s not very objective data about what the customers were thinking, but it’s pretty objective data about this particular African-American female’s subjective response.”

    I would bet (a 25% tip) that she went in to her experiment thinking that “white” folks would treat her better if she had blue eyes.

    But let’s think about that for a minute. Do you suppose that “white” people dissaggregate physical characteristics to the point that a person that was otherwise identifiable as “African American” would be more acceptable just because they possessed one non-typical trait?

    Here’s a little “experiment” my wife’s been “conducting”. She is East African, (Ethiopian) and has mid-to-light brown skin, medium wide nose and lips and curly but not “kinky” hair. (Her descriptions not mine.) Once you have been to Ethiopia and/or have been around “Habesha” people it is clear that she is (physically speaking) a fairly typical Ethiopian.

    When she encounters African Americans they are often unsure of her heritage due to her slightly ambiguous “racial” traits and her Amharic accent. She reports that a large majority of these folks treat her much better when they find out she is “African”. Suddenly she’s an African American.

    Also when most African Americans find out she is married to a “white” guy their reactions range from surprise to open disgust. “You married a what?” has been a disturbingly common refrain.

    Sometimes this is expressed in my presence.

    Now my point isn’t that “white” people are less racist than “black” people. I think it is true however that African Americans are more likely to strongly identify with a person based on perceived “blackness” than are “white” people with people on perceived “whiteness”.

    If an African America refers to fellow African Americans as their “people” it is openly accepted and even encouraged by society as a whole. If I refer to “white” folks as “our people” I sound like I’m in the Aryan Nation and even my own family and close associates would frown upon hearing it, let alone society as a whole.

    Now I am not unaware of the history of racial bigotry and alienation that lead to the need for African Americans to bond together as a “people” to fight racism or just get by.

    But I would hope, at some point, that all people could be encouraged to abandon this type of “racial” self-segregation.

    I am a who not a “what”.

  5. James Hanley says:

    Pierre–Unfortunately, no. It would be neat to see someone with the resources to do a fuller research project play around with this.

    AMW–She didn’t record the tips by race/ethnicity of customer, but I think she commented once that she got favorable responses to her blue eyes from black customers, too. One of the things she and I talked about in general conversation about the project was the green eyes that some “black” people of mixed ancestry have (“black” in square quotes because it tends to occur, I think, in people who greater than half white, but of course socially we still identify them as black, I guess because we can’t do math well). If you’ve ever seen such a person, it is a stunning color, a bright emerald green that really pops out. Her “blue” eyes had a similar effect, and I wonder if at some level the responses weren’t so much racially based as just a response to the perceived brightness of her eyes when she had her blue contacts in.

    Lance–She did go into the experiment thinking she would get better responses from white customers when she had blue eyes, because she felt like she had experienced that, and wanted to get more solid data to see if her subjective feelings had some substance. (The great methodological weakness, of course, is that she knew what her eye color was on any given day, so we can’t absolutely rule out that she acted differently on those days. Having had her wait on me a few times when my family ate out, and having observed her working, I think probably not because she’s a genuinely warm and friendly person, the type people normally respond well to. But there’s no way to really know.)

    As to your question about disaggregating features to rate more favorably based on just one atypical one, I think it’s possible here for two reasons. One is that I think we’ve become just post-racial enough that for most white people innate discomfort with black people isn’t at the level they’ll hate them no matter what, but at a level where they just need some reassurance that this person isn’t “too black.” The second is that this is a feature that’s exceptionally noticeable–as I noted in reply to AMW, when she put in the blue contacts her eye color really popped out and was hard to miss. An aquiline nose might not have been noticeable enough, perhaps, to have as much effect. (And there’s an experimental opportunity–find a black waitress with an aquiline nose, and fit her with a prosthetic “typically black” nose, and see what happens with tips.)

    That’s speculative, of course, and I don’t pretend it’s any more than that.

    As for people’s response to your wife about her marriage, are you sure they’re not reacting badly to you being a math teacher?

  6. Dr X says:

    Interesting, but dot a double blind (sorry), so we don’t know if her behavior subtly changed with her changed lenses. Not a small problem in an experiment like this.

    But it would be tough to do a double-blind. You could use clear and blue lenses put in by someone else, but the customers would comment when she’s in the blue condition.

  7. Dr X says:

    not a double blind.

  8. James Hanley says:

    Dr. X,

    Yes, an unfortunate difficulty. It’s worth noting that the student also put pictures of herself on a website called, I think,, where people rate the “hotness” of people in pictures. She did the same pose, with the same clothes, one with blue eyes and one with brown. She received a higher average rating with blue eyes. So that got rid of the possible factor of her response, but a) we don’t know anything about the respondents, and b) she didn’t give enough information on response frequencies to test for significance in the ratings difference. I think the most interesting part of that finding is that it tended in the same direction as her tip findings.

  9. AMW says:

    I’m working on another project with a hard deadline right now, so I’m wasting time that I do not have (as professors typically do). But after reading this post I’ve been thinking about how you might do a larger study to account for things as properly as possible, because I’m seriously considering attempting it myself. Here are some ideas I’ve had to improve the design:

    1. Do the experiment at a mall restaurant in a large city. As James mentioned, he went to that particular restaurant many times, and saw the same waitress. That means a lot of her customers are probably regulars who know her, and the data points probably aren’t independent. Running the experiment in a heavily trafficked restaurant that serves a large population should minimize the number of repeat customers in the data set.

    2. Recruit multiple wait staff at the restaurant to participate. This could do a lot for the study, including:
    a. Generate more data in a shorter time frame to avoid repeat interactions with customers.
    b. Generate data on a broader spectrum of race, gender and eye color in the wait staff pool. This could help you delineate between some possible causal explanations of the results. If people just like blue eyes, then white waitresses should earn more when they’re wearing brown rather than blue. If people like to tip “white” looking people, then there should be no effect among whites and a blue-eyed tipping bias for blacks/asians/hispanics/etc.
    c. Potentially get around some of the double-blind problems. Staff with all eye-colors will be wearing contacts to change their color. If just changing one’s eye color changes one’s expectations (positively) about how others will treat them, then blues changing to browns should cancel out the effect of browns changing to blues.

    3. Switch up clear and colored lenses so that whether a member of the staff is on a “control” day or “experimental” day, they’re going through the same routine.

    4. Take photos of each participant with each eye color. Gather data on how attractive third parties find them with either eye color. This could be done through something like or some more scientific approach.

    5. Have the wait staff make brief demographic notes on their patrons.

    Anybody have other ideas?

  10. James Hanley says:


    I think this would be very interesting. It’s possible you’d need to hit a couple of restaurants to recruit sufficient wait staff, but maybe not. The mall restaurant idea spurs one note of caution, perhaps, in that some are what a friend of mine calls “soft food” restaurants, those catering particularly to elderly patrons. You’d probably want to make sure you select a restaurant with a diverse clientele.

    Off the top of my head, I see the greatest difficulty, after recruiting wait staff to be your research assistants, as being directing these research assistants properly. Particularly on busy days they’re likely to be busy enough that keeping detailed records will be difficult for them. You’d need a clear and concise reporting form. The aggregate results of tips at the end of the day might not be a big deal, but the reporting of percentage tips (as my student did) for each ticket could be difficult to ensure they do consistently, and reporting on ethnicity of customers (if you’re interested in that) could be very hard because of a) reliance on research assistants’ subjective interpretations of ethnicity, b) the problem of multi-ethnic parties, and c) the difficulty of remembering which ticket went with which person of which ethnicity when multi-ethnic parties do separate checks. So I’m dubious about number 5.

    2, 3, and 4 sound good. I’ll try to give it some more thought, and see if I can add anything useful. I would be delighted to see a followup study on this. As I tell my students, once you know how to do good research, the rest is a matter of resources, primarily time and money. I can’t demand too much of them because they generally don’t have either of those available. But a study by someone who has more of those would be really neat; great behavioral economics stuff.

  11. Lance says:

    There is a possibility that the blue-eyed “black” person appearance made her more memorable and/or exotic and therefor increased the likelihood that customers would put more thought into tipping her rather than say a good server that wasn’t as memorable.

    James Hanley,

    Oh, and being a math teacher is one of the least objectionable things about me. I didn’t earn the nickname “Obnoxious Man” for nothing. I have a rather high measured (so they tell me) IQ but an extremely dodgy EQ. (Emotional Quotient). A disturbingly high proportion of even my closest friends say things like “I really hated you the first time I met you.”

    How do you respond to a comment like that? Uh, thanks? Apparently I have a somewhat “polarizing” personality.

    I seem to remember you blasting me a few times back at the original DIspatches.

  12. James Hanley says:

    Me? Never. I was probably the least polarizing person who ever posted at Dispatches.

  13. Matty says:

    Some racism is subtler than other racism, and some is wholly unintended and unconscious.

    This got me thinking, what prejudices do I have without even knowing it. Following on are their prejudices that are widespread but never, or almost never conscious? Could there be people who are effectively discriminated against by society without anyone in that society actually thinking they deserve such treatment?

  14. Lance says:

    Maybe its my low emotional intelligence quotient but I have less and less patience for discussions of race.

    I live with and spend a great deal of my time with people that could be characterized, by some, as belonging to various “races”. I see no useful purpose to continue using the term “race”. It is has no scientific meaning.

    There is only one useful metric when it comes to humans and that is the “individual”. If you take the idiotic idea of race to its extreme we are all races of “one”. (Except for perhaps identical twins.)

    Culture on the other hand is quite real and that is where the discussion should proceed.

    I believe it is nearly universally accepted that physical characteristics are not useful in making value judgments about people. So how about we all just drop the idea of “race” and move on to the social issues that have caused there to be cultural misunderstandings that continue to negatively influence some individuals.

    Are there idiots that still make value judgments based on trivial physical differences? Yes.

    But over-analyzing every subtle subconscious reaction and then trying to determine if it is based on these perceived physical differences only reinforces the idea that these traits are indeed important.

    They are not.

    Laws exist to prevent discrimination based on these traits and there is also a great deal of social pressure to not act on the basis of these traits.

    Unfortunately there also exists social pressure in some cultural settings to continue to acknowledge these differences and act differently to people based on these traits.

    This must stop!

    You cannot end racial discrimination with more racial discrimination.

  15. Matty says:

    Lance actually largely agree with your post and am no fan of ‘race’ as a category. The strange thing is although I was race was the obvious referent of what I wrote the quote was more of a jumping off point than the point I was trying to make. Let me try again –

    I try to treat all individuals fairly, so do most people I know and that includes not judging on non-relevant characteristics. But do we succeed? is there some person or people I am being unfair to without realizing? I want to know not because I think there is intrinsic value in whatever trivial differences are the trigger but because I want to stop being unfair. It really isn’t about race for me if I unfairly judge maths teachers, or libertarians or people whose first name starts with L I want to find out so I can stop.

  16. Matty says:

    Gah, syntax fail I hope it is still readable

  17. AMW says:


    I see your point, but I think we are hardwired to take notice of differences in individual characteristics. Since “race” is essentially constructed by a suite of individual characteristics widely shared among a group of people, there may be no way to totally get around our attention to race.

    Things that men and women find physically attractive about one another often correlate strongly with fertility and/or the ability to raise children to reproductive age. It’s an evolutionarily useful bias in our make-up not to want to have sex with highly asymmetric people.

    Likewise, it’s highly evolutionarily useful to be altruistic towards and trusting of one’s close kin, because that promotes replication of one’s own genes. Similarly, it might be evolutionarily useful to be distrustful towards much more distantly related kin, as their genetic makeup is much less similar than one’s own. Individuals who share a similar suite of individual characteristics to my own are almost certainly closer kin than those who share a dissimilar suite of individual characteristics.

    It may be that racism, like human metabolism and appetites, is one of those things that made a lot of sense 10,000 years ago but is now outliving its usefulness.

    P.s., I’m not equating “made a lot of sense” with “was morally right.” Things can be evolutionarily useful but morally repugnant.

  18. Lance says:


    Good points. I also want to make sure that I don’t make illogical assumptions based on limited information. Also I am not claiming that by virtue of being married to an African woman that I am somehow immune to making these type of snap judgments.

    As a math teacher I was often glad when my brightest students would be African Americans. I felt that this would dispel the odious racial stereotype about relative aptitudes of the “races”. But I also realized that I wasn’t 100% sure that I wasn’t proving it to myself.

    Then I thought, what evidence would be required to make a “final decision” on the matter and I realized that there was no objective metric that could be applied to such a problem and that the very question was a non-sequitur that said more about my attitudes than the relative mathematical aptitude of individual students.

    Still it just frustrates me that so much energy and emotion is spent on this one subset of “limited information”.

  19. Lance says:


    Probably some evolutionary genetic component exists but I think that it is a learned behavior more than an instinctual one.

    I know that I have always been attracted to people that are different than me both physically and intellectually.

    I have walked into African villages where I was viewed almost like a space alien. After the initial reaction I am almost always received positively. (Of course they don’t know me that well.)

    In Ethiopia my wife and I occasionally get a dirty look or rude comment from passersby but on the whole we are warmly received even though I have taken so to speak, one of their women. So if suspicion and malice towards strangers is a genetic human trait the Ethiopian cultural tradition of kindness and deference to strangers (engeda) has overwhelmed the genetic component.

  20. AMW says:


    All good points, and let me just add that even though I suspect we have some hard-wired biases in this matter, that doesn’t mean I think we’re doomed to fall prey to them. Every time we practice birth control we’re thwarting our genes, as Dawkins says.

  21. Lance says:

    “Every time we practice birth control we’re thwarting our genes, as Dawkins says.”

    Hehe, kind’a puts a different spin on masturbation as well.

Comments are closed.