Analysis of Differential Prediction of Law School Performance by Race/Ethnicity Based on 2011–2014 Entering Law School Classes (TR 17-01)
The Law School Admission Council (LSAC) has carried out annual predictive validity studies, also called LSAT Correlation Studies, since the Law School Admission Test (LSAT) was first administered. These studies are geared toward evaluating and ensuring the effectiveness and validity of LSAT scores for use in the law school admission process. In conjunction with these predictive validity studies, LSAC also conducts differential validity and differential prediction studies on the LSAT to ensure that the test is fair across racial/ethnic subgroups. The purpose of this report is to summarize the results of the 2012–2015 LSAT Correlation Studies, which are based on the 2011–2014 entering law school classes of participating schools, in a differential validity framework. The results presented serve to document and support the validity of LSAT scores for use in the law school admission process.
This study examined results for three racial/ethnic minority subgroups and the racial/ethnic nonminority (White) subgroup. Data were analyzed from 148 law schools, each of which over the 4-year study period enrolled 10 or more first-year students who identified themselves as Asian, Black, or Hispanic and 10 or more first-year students who identified themselves as White. Validity coefficients and prediction equations using LSAT score with first-year average (FYA), undergraduate grade point average (UGPA) with FYA, and the combination of both LSAT score and UGPA with FYA were calculated across subgroups and evaluated.
Results of analyses indicate that the validity coefficients calculated for each racial/ethnic subgroup were very similar to each other and that FYA tended to be, on average, slightly overpredicted for all three of the racial/ethnic minority subgroups studied in this report. The combination of both LSAT score and UGPA as predictors provided the least amount of overprediction for all of the racial/ethnic minority subgroups compared to the use of either predictor alone. Overall, results do not suggest that the use of LSAT score alone or the combination of LSAT score and UGPA contributes to unfair admission decisions for the racial/ethnic subgroups studied here.
At least two caveats should be remembered when evaluating the results of this study. First, only differences in average predicted performance were analyzed. That is, the performance of individuals within a subgroup whose FYAs are overpredicted on average may still be underpredicted, and vice versa. Second, differential prediction is only one aspect of an overall construct validity evaluation. Other aspects of validity should also be considered when deciding whether the use of any test scores is valid.