I took an advanced Psychology class last semester focusing on Intelligence and the LSAT's predictive power with respect to 1L grades is roughly equivalent to the predictive power of the ACT/SAT for first-year undergraduate grades, with a correlation between the two in the range of 0.35-0.40, which is by no means great.
The reason, at least theoretically, that law schools continue to value the LSAT so highly is that while the predictive power is low, it is the best predictive tool available to them, unfortunate as that is for some of us (including me).
Just to let you know, a .35 to .40 correlation is pretty much as high as you'll ever see for any type of correlation test. (with regards to pshychology at least) I've also got to tell you that I'm highly skeptical about where you received your information about the LSAT's predictiveness. Did you do some type of meta-analysis? If you did, I would like to see it.
The LSAT isn't really predictive at all when you think that every school takes their own selective range of scores during admissions. Then they still end up ranking their students after first year, and there are people with high LSAT scores that end up near the bottom of the class, and there are people with low LSAT scores that end up near the top of the class. My point is that the predictiveness of the LSAT as an "indicator of first year success" is arbitrary at best.
If anything the LSAT only goes so far as to determine exactly what school an applicant can get into. I imagine there could be some other type of test made, but as I'm sure everyone that is actually in lawschool has noticed, lawyers (which includes a significant amount of people who have a hand in the admissions process) are a very slow evolving bunch. Further, the LSAC has a relative monopoly on the system, and it will be hard to get them to do anything to change their positions on the LSAT when they are surely making a very large profit on their current system. (where's the incentive for them?)
Bottom line is, the LSAT is just something that an admissions commitee can use to distinguish someone who has similar grades, from similar schools, with similar personal backgrounds from each other.
(well that is unless you're a minority....and please no one take offense..... but facts ARE facts, and we all know that if you're a minority, then the LSAT doesn't really matter as much....lets be honest)
Well that's just my two cents....any other thoughts?