There was much controversy when Freakonomics said that reading at home doesn’t help test scores. Now the Freakonomics blog points out that the much-cited stat about kids and exposure to words before school is, well, sketchy (A Data Pool of One). The number comes from a 1990 book whose author used her child as the sole point of comparison between rich and poor kids being read to. (There were only 24 low-income children in the study pool all together.) If a stat seems too good to be true, it probably is. Not that the UPK mafia won’t keep using it.