Weekend Stubble

It's gone unnoticed in any print media or blog yet, but this week's new issue of New Scientist -- behind a subscription wall, so I will quote at some length here -- has a short piece by Jim Giles that puts the statistical crunch on Henry Louis Gates:

Next April, the prominent Harvard historian will publish In Search of Our Roots: How 19 extraordinary African Americans reclaimed their past. Gates has studied the ancestry of some of the country's most successful black individuals, including Winfrey, a daytime TV host. He found that 15 had grandparents who owned land, even though just 25 per cent of African Americans had obtained property by 1920. This "astonishing pattern", as Gates described it recently in The New York Times, suggests that promoting home ownership can alleviate contemporary poverty....

So what do[es this book] tell us about the roots of success? From a scientific point of view, almost nothing.

The problem lies in two fundamental mistakes.... Let's assume that he really has found an unusually high rate of landowning grandparents among successful African Americans (it is not clear that he has, given that the odds of one grandparent among four having owned land are actually quite high). What about land ownership among the grandparents of less successful African Americans? To suggest that property is important, Gates needs to show that a lack of this legacy contributes to normal folk being, well, normal. In epidemiological terms, he needs to run a "case-control study". But, as Harvard political scientist Andrew Eggers puts it, book sales probably wouldn't be helped by the inclusion of "a few chapters about thoroughly unfamous people"....

An epidemiologist would employ simple mathematics to correct for this problem.... "Most of the time these books are just making intuitions," concludes Gary King, a social scientist also at Harvard. "They're like astrology."

Ouuuuch.

Well, that'll make for some awkward dining in the Harvard Faculty Club.

To be fair to Gates, it's not clear from the article whether their criticism is based on reading an ARC of the book, or on his Times piece. But fans of, ahem, intuitive statistical scholarship will also recall Richard Posner's "evidence" a few years back of a decline in public intellectuals, which was then brilliantly dissected by Caleb Crain in The Nation...