Stilwel, et al wrote: This national summary of the LSAT correlation studies conducted in the years 2005 and 2006 lends continued support for the validity of LSAT scores in the law school admission process. Major findings from this study are summarized as follows:
- The combination of LSAT score and UGPA is a useful predictor of academic performance in the first year of law school. The average multiple correlation between FYA in law school and the combined predictors of LSAT score and UGPA is 0.46 for both 2005 and 2006 study years. This value is high, but slightly lower than multiple correlation coefficients reported for previous correlation study years. As has always been the case, these combined predictors continue to be superior to either predictor alone for predicting FYA.
- LSAT score alone continues to be a better predictor of law school performance than is UGPA alone. The median validity for LSAT score alone is 0.34 for 2005 and 0.33 for 2006, compared with mean validity values of 0.29 (2005) and 0.28 (2006) for UGPA alone.
- When schools are grouped by the correlation between LSAT score and UGPA, the validity coefficients increase when the correlations between the predictors increase. This relationship provides some indication of the impact of the restriction of range resulting from using only matriculated students on the estimates of validity, particularly in the presence of a compensatory admission model.
- A substantial amount of the variability in validity coefficients obtained among different law schools is directly attributable to the amount of variation in LSAT scores and UGPAs in the data used to estimate the validity.
- Cross-validation studies support the use of regression equations based on previous first-year classes to predict future performance of law school applicants.
-The restriction of range typically observed for LSAT score and UGPA was more pronounced for LSAT scores for the 2005 and 2006 correlation study years as compared to previous correlation study years, resulting in a reduction, on average, in the correlation between LSAT score and FYA. This further resulted in a reduction in the mean percentage of variance accounted for by LSAT score in the equation combining LSAT score and UGPA to predict FYA, and a corresponding increase in the mean proportion of variance accounted for by UGPA. However, in general, LSAT score continues to account for the greater proportion of the variance in this prediction equation.
The LSAT is a very good predictor of a student's grades in law school. Most people assume that studies would find an even high correlation if it weren't for the fact that students at a given law school already have a very clustered range of LSAT scores. The 25% 75% spread for most of the T14 is only 4 or 5 points. To be fair, the predictive validity of UGPA would probably also rise if the ranges were more extreme.
I'm not saying the LSAT is the end all be all. It's not a perfect correlation. Low scorers can excel and high scorers can fail. But to claim that the test is stupid and doesn't generally predict how well a student will perform in law school is refuted by very strong empirical evidence...