Regression of The Mean

Monday, 2 December 2002

I came across a newish term today, while I was browsing those regions of the blogosphere inhabited by people who believe in the scientific truth of Mr Sax Rohmer's classification of the human race into 3 basic types: wily orientals, such as the fiendish Doctor Fu Manchu and his devious daughter the lovely (and slightly oversexed) Fah Lo Suee; slightly less intelligent, but nonetheless decent whites, such as Nayland Smith, Doctor Petrie and Shan Renfrew; and the rest of the human race whose intelligence, in general, is inversely propotional to their degree of skin pigmentation. The (newish) term is Regression to the Mean and I say newish, because in my chequered academic career I did study a year of statistics (did pretty well too, if I say so myself) so I may have come across it before, even if all I remember of the section on correlation and linear regression these days is my lecturer's frequent reminders that correlation is not causation.

John Ray of Dissecting Leftism, offers this statistical artifact as an explanation for poor academic performance by black students as reported in the New York Times and commented on by Razib at Gene Expression. Dr Ray obviously knows his statistics, so I think that his suggestion that the observation of generalised poor academic performance in american black students is a merely an accidental quirk of statistics, rather than an indicator of underlying genetic differences in IQ, has some merit.