Predicting the Constitutionality of Punitive Damages: A Statistical Approach
Edward K. Cheng
Vanderbilt Law School
University of Toronto - Faculty of Law
June 28, 2009
The constitutional doctrine governing punitive damages captivates legal scholars and jurists in part because it is both complex and evolving. The unpredictability, however, presents difficulties for attorneys and their clients, who need greater certainty to make practical decisions about litigation and settlement. In this Essay, we offer a statistical approach for predicting court decisions on the constitutionality of punitives. As it turns out, basic logisitic regression methods with appropriate model selection can be quite effective, although we make further gains using a Bayesian hierarchical approach. Using a new dataset of cases challenging punitive damage constitutionality from 1989 to 2008, our hierarchical model can predict out-of-sample outcomes with 76-85 percent accuracy. These results suggest that while constitutionality may not be subject to a deterministic formula, it can be effectively modeled statistically. Beyond the punitive damages context, our work additionally offers a glimpse of the potential of statistical models for predicting a wide variety of legal questions.
Number of Pages in PDF File: 21
Keywords: punitive damages, constitutionality, statistical prediction, prediction, forecasting, BMW v. Gore, model, punitives
Date posted: June 30, 2009
© 2016 Social Science Electronic Publishing, Inc. All Rights Reserved.
This page was processed by apollobot1 in 0.219 seconds