The LA Times publishes Jon Wiener on Lott's lawsuit:
But Lott is not suing those who have said some of his pro-gun research was "invented," "faked" or "cooked." The lawsuit turns on the definition of "replicate," from the "Freakonomics" sentence about how other scholars have tried and failed to "replicate his results." Lott maintains "replicate" means "analyze the identical data in the way Lott did." Because nobody tried to do that, he argues, "Freakonomics" is wrong. Most people, however, understand "replicate" to mean something like "confirm."
Lott's reputation has indeed been "seriously damaged" by critics, but only because they have described many apparent holes in his dubious research and misleading citations. Blocking the sale of a book based on a literal interpretation of a single word would be outrageous.
Note that the story of Lott's missing survey has appeared in many major papers such as the Chicago Tribune and the New York Times as well as in Freakonomics, which has sold over a million copies. Supposedly eight or so Chicago students made many thousands of phone calls conducting a this survey for Lott. It's not something you would forget. But apparently none of them have heard of the affair and come forward. You'd think students who studied at Chicago would be particularly likely to read Levitt's book, since he is a professor there.
Via Kevin Drum.
Doesn't replicate mean that you repeat the entire experiment and obtain results that confirm the original hypothesis? It doesn't even mean get the "same" results -- just results consonant with the original -- but it generally does mean gathering ones own data. Recalculating the statistics on the original data means just that -- it's hardly replication, it's merely checking the arithmetic.