Scale Fitting in the Psychometric Model of Judicial Decision Making

Abstract

The psychometric model, developed by Glendon Schubert, is a widely accepted way of conceptualizing judicial decision making. It entails the representation of justices' ideal points in a multidimensional space. A vector (representing the ordering of the justices derived from scalogram analysis) is projected through this space to aid in interpreting the underlying attitudinal dimensions of interagreement. Despite the apparent rigor of this model it contains certain indeterminacies, as no mathematical procedure has been used to determine the location of the scale vectors. Consequently ad hoc. hand procedures which may yield suboptimal solutions have been developed.This article provides solutions for the problem of fitting scale vectors to multidimensional configurations. These solutions are developed for the problem where the scale analogue is treated at the measurement level of interval or ordinal values. In order to demonstrate the advantages of using these procedures, we undertake a reanalysis of Schubert's results (for which he used ad hoc procedures). In one-half of the cases analyzed, the results reported by Schubert are suboptimal when evaluated in terms of the coefficient he sought to maximize. The mathematical solutions and the associated computer programs give greater rigor to the psychometric model and provide a response to the criticism that scale fitting in the psychometric model is subjective and inconsistent with the premises of the behavioral approach to judicial decision making.