Empirical, particularly quantitative empirical, scholarship is all the rage these days in law schools. (By the way, as a quantitative legal empiricist,that makes me really nervous. If there's one constant in legal academia, it's that things go in and out of style as fast in law schools as they do in Milan fashion shows.)
One thing that has been bothering me lately about this next phase, new wave, dance craze aspect of legal scholarship is the use of appellate cases as datapoints. It's tempting to think that one can code appellate decisions or judicial opinions pursuant to some neutral criteria, then look for trends, tease out inferences of causation, etc. Here's a note of caution: they're not i.i.d. They're probably not i.i.d. given X (whatever X is). Precedent matters. In our legal system, the fact that a previous appellate case (with a published opinion) was decided a certain way is a reason to decide a subsequent, facially similar appellate case the same way, even if the first decision might have been (arguably) wrong. Folks will argue over how much precedent matters; all I can tell say is that as a law clerk to an appellate judge, I participated in numerous conversations that resulted in the sentiment, "I might/would have decided the present case differently had Smith v. Jones not been on the books, but I see no grounds for departing from the reasoning of Smith v. Jones here." I.i.d. models, or analyses that assume non-interference among units, should be viewed with great caution in this setting.
Posted by James Greiner at March 20, 2007 4:40 PM