Monday, December 19, 2005

Vetting papers takes time 

There was something vaguely familiar about this news release from UCLA that everyone is talking about. Yup, I had read this before. (See also this.)

And the reaction at that time was quite negative from some people. For example, consider this exchange between Geoff Nunberg and the authors Milyo and Groseclose on the study. It is easy for people to make mistakes on what the study does because it contains not one but two estimates of media bias, and the second one is quite complex -- the authors point out in their response to Nunberg that it "describes a maximum-likelihood estimation technique and it notes a set of random variables that follow a Weibull distribution." People want to skip that part of the analysis because it requires some real education in statistics, and instead focus on criticism of the press from the paper's first, admittedly "back-of-the-envelope" calculation. This reminds me of the discussion of criticisms of Keynesianism -- the critiques are better directed at Keynesians than at Keynes, since the former often do a bad job at explaining him.

Thus it's worth us understanding that the paper now has been reviewed by peers and is being published in a top-30 professional economics journal. It has taken a year to vet the results and be sure they meet standards. I do not know if QJE requires the authors to provide data for replication of their results, but one can certainly ask the authors.

This looks like the latest version of the paper, coming from Milyo's homepage. I read one that was an earlier draft, and I don't have time to do the comparisons yet.

Categories: ,