Is Satoshi Kanazawa the Glenn Beck of Pseudoscience?
The man who claimed to prove that black women are less attractive distorted scientific facts to make his case. Sound familiar?
It's only been a few weeks or so since controversial evolutionary psychologist Satoshi Kanazawa posted the blog heard round the world -- his blatantly discriminatory study, "Why Are Black Women Rated Less Physically Attractive Than Other Women?" In a nearly unanimous verdict, Kanazawa's peers -- his fellow science bloggers and psychologists -- have found him guilty of perpetrating pseudoscience, embarrassing the profession of evolutionary biology and offending multitudes of black women and men.
Actually, black women are just Kanazawa's latest target. His page on the London School of Economics' site shows that he's also denigrated the poor and asserted that racism in the United States is a myth, among other claims.
But what's really amazing about this work is how far Kanazawa appears to have stretched the rules of statistical analysis, while ignoring much of what we know about biology and culture, to produce a hurtful, racist assessment of black women. Originally posted on Psychology Today's blog, and now found here, Kanazawa's work is a startling example of the degree to which racism is alive and well in some corners of academia.
A group of concerned scientists has taken to the Web to voice their concerns over Kanazawa's misuse of the data from the National Longitudinal Study of Adolescent Health (Add Health) to draw a conclusion so odious that many wonder why he would even ask such a question. But few have discussed the degree to which he discounted much of what we know about culture and race in America and how it would inform any random survey of attractiveness and black women.
As an African-American psychologist, T. Joel Wade, chair of the department of psychology at Bucknell University, can certainly understand the racial impact of Kanazawa's work. "You have to factor in culture to do a study like this, but Kanazawa does not address culture here at all."
Specifically, Kanazawa's inaccurate interpretation of information from a highly reputable federal survey renders his conclusion about black women virtually meaningless, suggests Ernest Davenport, a professor of psychology at the University of Minnesota and an expert in statistical analysis and measurement. "While the author [Kanazawa] extols the use of factor analysis in this study -- a method for eliminating all random measurement errors -- he forgets the 'garbage in, garbage out' rule. No statistical analysis can overcome a bad theory and bad data.
"In terms of statistics, Kanazawa gives no clue to how he resolved many issues with the data he gathered," Davenport continues. "Add Health uses oversampling of certain subgroups [a common practice in studies of multicultural populations], and any analysis of the data should be weighted to compensate. Kanazawa makes no mention of weights. He also refers to 'statistical differences' without specifying how he handled aspects of the data that would be affected by cluster sampling [distortions caused by the tendency of people in clusters to hold similar opinions]."