Posts from Josh Wright & David Hoffman. Here is a taste of Wright’s post:
- [A]n inherent suspicion of empirical work I think lies beneath much of the conclusion that the work is of low quality. Professor Bainbridge gets at some of that in this post when he writes that empirical work will “always be suspect — and incomplete in my book.” That seems wrong to me. Suspect and incomplete need not be the same thing. And too often the responses I hear from legal scholars to empirical work that actually has legal relevance and is well done is reflexively dismissive: “but you can’t possibly control for everything can you?” Sometimes there will be a short quip about “omitted variable bias” or something like that. In the panel data setting, I’ve often sympathized with the econometrician who tries to explain exactly what state fixed effects are controlling to an audience to whom the answer sounds like he is selling them a used car without an engine. I suspect when Leiter complains of the discussion that follows empirical papers as always involving “the same tedious pattern of wondering how one controls for this-or-that variable, with the presenter showing, cleverly, how s/he already controlled for it, or admitting that s/he didn’t, so that this is an issue for future work, etc.” — that there might be some meaningful information in there. It, in fact, matters a great deal whether something is properly controlled for, whether the error structure is appropriate.