Yeah, I noticed the same thing (see my followup post). The differences in how responses between blacks and whites distribute is pretty interesting, so it’s disappointing that the author didn’t discuss this more. It suggests that white kids are less apt to be “color blind” when it comes to their perceptions, since their responses tended to be more polarized. Blacks kids seem more adept at seeing the good and bad in everyone.
My daughter, who’s 7, treats skin color like a distinguishing feature of a person, i.e. “Mary has green eyes, dark brown hair, and brown skin.” “I have brown hair, hazel eyes, and pink skin.” She knows that all of these features are dependent on where a person’s ancestors came from. I was shocked the first time I heard her describe someone like that, because I myself was taught that it was rude to “notice” someone’s skin color out loud. After thinking about it I realize I was taught wrong and her way is much more natural and right.
We tend to talk about different cultures and religions out of context of skin color. We have talked about segregation and prejudice of people that look different than us (we’re white, but I’m darker than she is), but she realizes that it’s really dumb to treat someone differently just because they look different.
I don’t think we’ve really used the word “race” when we’ve talked.
If the hypothesis you’re testing is that black and white kids show no racial biases wrt to the assignment of negative and positive traits, how would you use race-neutral questions to answer this question?
We have questions that we suspect are answered in a race-biased way. We compare the potentially biased answers to answers we know are not race-biased. If there is a statistical difference between the suspect answers and the known non-biased answers, then we can conclude the suspect answers are biased.
That’s the basic outline of a control group in a scientific study. Just like when we test a drug, we compare the drug’s results to the results of a placebo. If there’s a statistical difference, then we know the drug has an effect.
Without a control group, there’s nothing to compare to. In this study, for all we know, the children will show race “bias” in questions like “show me the hassenpfeffer”. It’s possible the study’s conclusion that bias exists will stay the same, but we don’t know unless they actually test for it.
Can you give some examples to illustrate what you are talking about? It’s still not clicking for me. Please note: Designing studies is how I make my livelihood.
To get away from race for minute, supposed the hypothesis is “men and women think blondes are more attractive than non-blondes”.
So we ask men and women a bunch of questions about people with a range of hair colors. “Who is the sexiest in this picture?” “Who is the frumpiest?” "Who would you be the most likely to show off to your friends, if you were dating them?“Who will have the best looking babies?” Questions like this.
Consistently, men and women pick the blondes as the sexiest and prettiest, etc. So after all of this why would it be incorrect to conclude that men and women find blondes more attractive? What other explanations could you come up with account for this observation, and what “control” questions would be needed to make our conclusions stronger?
Clinical trials use different methods than behavioral and epidemiological studies. A “control group” is hard if not impossible to come by in a setting that can not be controlled artificially. It’s not legitimate to criticize this study because it wasn’t performed like a clinical trial or laboratory experiment.
You can compare two groups with each other and get meaningful results without one of them necessarily being the “control group”. It’s done all the time.
That’s what my “show me the hassenpfeffer” example is trying to illustrate. The study could ask questions like that and see what responses they get. There’s no reason to expect that the answers will be connected with race in any way (and they can ask multiple neutral/nonsense questions to be sure about that–truly neutral questions should have statistically similar responses). The answers will be as “random” as the children can provide and they will answer “I don’t know” at some rate.
Then, we look at the responses to “show me the pretty one” and if they are statistically different from the responses of the neutral questions (by showing a higher correlation and higher response rate), we can conclusively say there is a bias in the response.
My scientific background is physics, so these sorts of studies are completely out of my field, but I do know how to ask basic scientific questions. The first question that comes to my mind when give a result of “X is different (larger/faster/more biased/etc)” is “different (larger/faster/more biased/etc) from what?”.
Well, you’re making my point with the part in bold. If your question about hassenfeffers does not return significantly similar responses, what should you conclude about this question? I and others would conclude that this question does not elicit race-neutral responses because racially biased reasoning might be at work (“Hassenfeffers sounds German and Germans are white, ergo the white one is a hassfeffer.”) What may seem like a nonsensical question may not elicit nonsensical results.
It’s not necessary to do all this if you simply compare responses to “show me the pretty one” and “show me the ugly one”.
The study results isn’t just saying “X is different”. It’s saying “White kids are more likely to assign negative features to black people than they are white ones.” It’s a very straightfoward finding that doesn’t require any controls or assumptions about race-neutral questions.
My background is in epidemiology. Nothing is wrong with study design based on what I see, and it’s pretty consistent with other work described in the peer-reviewed literature.
Thank you. I was inspired after opening up a fresh box of crayons and huffing them.
I’m a social scientist. I’m pretty happy with the study design as well. Heck, I’m in a field that generally uses secondary data rather than experimental data like this, so I think this stuff is spectacularly clean!
So now that we see children are racist little monsters, how do you talk to them about race?