When the brouhaha over the Psychology Today “Why Black Women Are Less Physically Attractive Than Other Women” article broke, I wrote a quick blogpost pointing out some of author Satoshi Kanazawa’s most ludicrous, obvious mistakes. But now someone with a bit more competency has gone back to look at the actual data Kanazawa used, and discovered that the problems with his “study” go much deeper.
Much, much deeper.
Basically, Kanazawa completely misrepresented the data. His source material just flatly doesn’t say what he says it says.
Here’s the deal. Kanazawa drew his conclusions on the relative attractiveness of black women from the “Add Health” study, a long-term survey of American adolescents. He claimed that the study showed — proved — that black women were less attractive than women of other races. But that’s not the case.
The attractiveness “data” is itself suspect, for one thing. It consists of the subjective judgments of interviewers who were asked to rate their interviewees’ appearance. There’s no effort in the numbers to control for the interviewers’ (unstated) ethnicity, no protocol for their judgments, no reason to believe that their conclusions are in any way representative. It’s just their opinion, and different interviewers reached dramatically different conclusions about the same interviewees’ attractiveness.
Let me underscore that last bit. According to a review of the original data, most of the difference in attractiveness between individuals in the study can be explained by different interviewers “grading” the same interviewee differently.
But it gets worse.
This study is, as I noted above, a study of American adolescents, tracked through early adulthood. And though Kanazawa portrayed his article as a study of the attractiveness of adults, the samples he used included children as young as twelve. He based the majority of his conclusions on data on the youngest two groups, who had an average age of just sixteen.
Still with me? It gets even worse.
Kanazawa admitted that the supposed difference in attractiveness was less in “Wave III” than in “Wave I” and “Wave II,” though he actively concealed the fact that Waves I and II weren’t adults at all. (He labeled the relevant charts “Wave I: Men,” “Wave II: Men,” “Wave I: Women,” and “Wave II: Women,” even though the vast majority of those subjects were teenagers and pre-teens.)
What he didn’t admit was that there’s a Wave IV.
Wave IV, it turns out, is the only wave composed entirely of adults. And an analysis of the Wave IV data shows that it doesn’t support Kanazawa’s thesis.
In Wave IV there is no difference between the perceived attractiveness of the black women and that of the other ethnic groups examined.
And again, I want to underscore something. Wave IV is composed of the same interviewees as the previous waves. So what the data really shows is that some (presumptively white) interviewers thought that the black adolescent girls in the study were a little less cute than the white, Asian, or Native American girls.
But when interviewers went back and spoke to the same women as adults, that “attractiveness gap” disappeared. Completely.
This isn’t just shoddy statistics. This isn’t just crap reporting. This isn’t just incompetence. It’s scholarly malfeasance.