The glib response to stats on blacks and police is to cite so-called “black crime” or “black criminality.” But this depends on a major analytical error. Yes, blacks are overrepresented in arrest and conviction rates. At the same time, “criminal blacks” are a tiny, unrepresentative fraction of all black Americans. If you walked into a group of 1,000 randomly selected blacks, the vast majority—upward of 998—would never have had anything to do with violent crime. To generalize from the two is to confuse the specific (how blacks are represented among criminals) with the general (how criminals are represented among blacks). Statisticians call this a “base rate error,” and you should try to avoid it. [Emphasis added]
SlateStats: they’re the opposite of HateStats.