"Through a statistical survey, he shows that success on the LSAT is a much better predictor of success on a timed, in-class test than it is on a take-home exam or a long written assignment. Both "blue book" exams and the LSAT assess not only ability but also speediness, a skill that psychological research has suggested is unrelated to intelligence. While Henderson asserts that a good lawyer must have some capacity to think quickly—if you snooze in court, you typically lose—he argues that tightly timed tests have little in common with the work done by practicing lawyers. Writing briefs, doing scholarly research, and trying cases are all activities that are typically completed over periods of days, not hours." http://www.lsacnet.org/lsac/publications/CAUTIONARYPolicies2003.pdf
I thought this was interesting. This is a publication produced by LSAC and distributed to schools which discusses how they should use the LSAT in admissions decisions.www.etax.byu.edu/Faculty_Pub/Thomas/LSAT%20study.pdf
After discussing the use of the LSAT and UGPA for predicting law school performance, the author says..."The predictive power of any of these measures is not strong, and only the most general patterns may be discerned
. Entering law students should not feel that their future academic success is either unduly limited or assured by the quality of their academic credentials."http://www.geocities.com/CapitolHill/2381/Lawschoolscase/lawschoolscasesummary.html
U.S. District Court Judge Bernard A. Friedman ruled that the University of Michigan Law School's should stop using the LSAT as an admission tool. "One such solution may be to relax, or even eliminate, reliance on the LSAT. The evidence presented at trial indicated that the LSAT predicts law school grades rather poorly (with a correlation of only 10-20%)
and that it does not predict success in the legal profession at all."
I know I've run across more articles, but these are the ones I've found in a short amount of time searching on the internet. I think there are cases for both sides, but there is certainly enough data to suggest that the LSAT is not always an accurate predictor of law school success. A 10-20% correlation is nothing to get excited over. Apparently, however, no better method exists, so schools continue to use the LSAT in admissions decisions.