Analysis of Differential Prediction of Law School Performance by Racial/Ethnic Subgroups (TR-98-02)
by Lisa Anthony Stilwell, Lynda M. Reese, and Peter J. Pashley
This study was designed to address questions of differential prediction in the law school admission process for various racial/ethnic subgroups. Such research is essential for assuring that the admission process is fair to all subgroups in the applicant population. Differential prediction was evaluated by comparing the predicted and actual law school first-year average (FYA) for various racial/ethnic subgroups within individual law schools based on regression equations commonly used in the admission process.
The sample used in this study was drawn from 1993, 1994, and 1995 entering law school classes, using data that were available from the Law School Admission Council (LSAC)-sponsored Correlation Studies. Data from 160 law schools, each of which enrolled 10 or more first-year students who identified themselves as Asian American, Black, or Latino were analyzed and reported.
Least-squares regression analyses were carried out to predict FYA using Law School Admission Test (LSAT) scores alone, undergraduate grade-point average (UGPA) alone, and the best predictive linear combination of these two variables. Analyses were carried out separately for each law school included in the study, resulting in three regression lines for each law school. The regression analyses were carried out for the combined group of students included in the study (i.e., combined minority and nonminority subgroups).
The results reported here indicate that FYA tends to be slightly overpredicted for all minority subgroups analyzed when LSAT score, UGPA, or the combination of these two variables are used as the predictors. The most serious overprediction was observed for the use of UGPA alone. The results do not support the concern that the LSAT score or the traditional combination of LSAT score and UGPA result in unfair admission decisions for the racial/ethnic subgroups studied here.