A Stealth O'Connor or Souter After All?
SLevinson at law.utexas.edu
Thu Aug 11 19:39:33 PDT 2005
It occurred to me that "particular types of cases" does not at all equal "each and every case." Imagine that you are asked to predict how each particular justice will rule in a majority of cases dealing with, say, rights of criminal defendants, affirmative action, discretion of the military with regard to the war on terror, etc. I'd be very surprised if everyone on this list wouldn't be willing to bet a great deal of money on the accuracy of their predictions, assuming, of course, that the N for each group is more than one or two (which would therefore have the capacity for the "surprising vote"). And, as Sam has suggested, I'm even more confident that practicing lawyers in the particular subject areas would do even better. But I wouldn't bet the ranch (assuming I had one) on any given case, which supports Judge Boggs' point.
From: Danny_J_Boggs at ca6.uscourts.gov [mailto:Danny_J_Boggs at ca6.uscourts.gov]
Sent: Thu 8/11/2005 2:58 PM
To: Sanford Levinson
Cc: conlawprof at lists.ucla.edu; conlawprof-bounces at lists.ucla.edu
Subject: RE: A Stealth O'Connor or Souter After All?
Everyone on this list can predict quite easily how every one of the current justices will "rule in particular types of cases"
With respect to the above quotation from Prof. Levinson, you may wish to examine the recent "Supreme Court forecasting project" at Washington U Law School, where 3 "learned professors familiar with the field" (which included our esteemed moderator as one of the experts on the overall panel) -- and a computer program -- attempted to predict each Justice's vote on each case of the October 2002 term, before oral argument.
My recollection from following their website was that the overall success rate was only about 75% -- that the computer and the experts did roughly equally well -- and that there was not a single case where all three experts and the computer got every Justice correct.
(After typing the above, I checked and found that some of this has been published at 104 Columbia Law Review 1150 -- They show the computer doing somewhat better, but their listing of "selected major cases" confirms that in none of those cases did any one of the three human experts get the vote of each Justice correct).
Apparently they did the exercise again for the 2003 term, but I did not follow the results that year.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Conlawprof