« Violent crime soaring in Venezuela | Main | Guns in DC taxis »

More on Lott's lawsuit

Empirical Legal Studies notes a problem Dr. Lott may face in the defamation suit he's filed.

Lott's complaint argues that his critic Levitt said that other statisticians had been unable to replicate Lott's results, and that this is false because (1) in the social sciences, "replicate" means to re-run an analysis, using the same methods and data and (2) every statistician who has done this with Lott's work has replicated his results. Further, the Levitt claim was defamatory because, with this definition of "replicate," failure to replicate means the original study involved "cooking the books."

ELS argues that different social scientists use "replicate" in different and broader ways than this, and in fact the term appears to have no fixed meaning, or at least no one fixed meaning. (In many fields of science, it means to run the original method on a different data body, a different set of lab rats, so a failure to replicate may mean that the original study's conclusions were honestly reported but a product of chance or of its limitations).

That's the big risk with libel suits -- you lose on some narrow point, but the loss is interpreted as proof that the defamatory statement was proven as true.