Stephen KrashenReading Today, vol 23 (2) Dec 2005/Jan 2006, p. 19
Freakonomics, by Steven Levitt and Stephen Dubner, is a very good book, but it contains a mistake, one that can cause a lot of harm. The Freakonomics authors claim that parents' reading aloud to their children does not predict the children's academic achievement.
This statement is based on data reported in a journal paper by Levitt and another colleague, Roland Fryer (Fryer and Levitt, 2004), an analysis of data from the Early Childhood Longitudinal Study. Levitt and Dubner concluded that frequency of being read to by parents was not a statistically significant predictor of scores on tests given to kindergarten students when they began school.
There is, however, a good reason why Levitt and Fryer got these results: Nearly all parents in the sample said they read to their children quite a bit. On a scale of 1 to 4, where 1 = read to "not at all" and 4 = read to "every day," parents of black children reported an average of 3, and parents of white children reported at average of 3.4. This produces a ceiling effect, with scores bunched near the top.
With data like this it is impossible to determine if those read to more did better than those read to less because so many children in their sample were read to a lot. (Of course, with self-reported data it is hard to be sure how much reading aloud really took place.)