Academic

I recently had a look at Richard Posner's and David Landes's new statistical analysis of judicial ideology.  Basically, the paper:

1: Corrects the liberal-conservative ratings of the well-known Songer and Spaeth databases on judicial voting in the Supreme Court and courts of appeals in ways that I can't believe passed muster up to this point.  A real service, and it sets up the potential excitement of competing versions of the database being used by different authors, etc, etc.  A taste:

Commercial speech cases (301) had been coded so that a vote for the broadest interpretation of First Amendment protection was liberal. We changed this to ?other? because businesses typically assert claims of commercial speech. ... We made a similar change for all votes in categories 710?713, which cover copyrights, patents, trademarks, trade secrets, and personal intellectual property. All votes for the plaintiffs in such cases had been coded liberal. Yet an intellectual property case often is brought by a large firm against a small firm or an individual, as when a giant pharmaceutical company sues the manufacturer of a generic drug or a large record company sues a file-sharing college student to make an example of him.

2: Runs straightforward, but still interesting, statistical analyses on a number of usual metrics on the corrected data.  Landes and Posner find some interesting stuff that was new to me (polarized voting works differently in Supreme Court and courts of appeals, liberal appellate judges tend to have "conservative" voting records in criminal and employment discrimination matters, and so on).

3: But perhaps most of note, pronounces the "politics is a particularly good predictor of judicial voting" established by their analysis.  Since one of the authors is a sitting federal judge, both know how to slice doctrinal salami, and neither has done a lot of statistical work in the past (as far as I know, and I don't know Landes's work well), this is a conclusion from a striking quarter - one that I can't imagine, say, Harry Edwards or even Stephen Breyer drawing.  I predict a big splash in law school faculty offices - and possibly also in the popular media.  And, though I liked the paper, I also predict some exasperation by political scientists who think they did this work already, only using complex Bayesian analyses that epitomize accuracy, if not readability (said political scientists are going to need to explain why it doesn't matter that they based their analyses on data that coded broad protection of IP as "liberal" and strict construction of commercial speech as "conservative," though).

Here's Brian Tamanaha with another early analysis.  Net-net, you should definitely download this while it is hot.