Predicting Court Cases Quantitatively

This article illustrates and systematically compares three methods for quantitatively predicting case outcomes. The three methods are correlation, regression, and discriminant analysis, all of which involve standard social science research techniques. Two prior articles have generated requests for a study dealing with the problems involved in handling a larger number of cases and predictive variables. The present article is also designed to provide such a study. It does not presuppose that the reader has read the earlier articles, although such a reading might help to clarify further some of the points made here. The cases used to illustrate the methods consist of 149 civil liberties cases decided by the United States Supreme Court from 1956 through 1960. The list of cases was obtained from a series of articles written by Sidney Ulmer and Glendon Schubert. Technical aspects have been eliminated from the body of this article, leaving a simple explanation that should be sufficient to enable the non-technical reader to employ the methods in his legal research.