Bayesian Models Predicting the Return to Prison

• Activity Number: 32
• Type: Contributed
• Date/Time: Sunday, August 1, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Gail Blattenberger+ and Richard Fowles* and John Krantz
• Companies: The University of Utah and The University of Utah and The University of Utah
• Keywords: Bayesian ; Model selection ; Model averaging ; Recidivism

Abstract

The United States has the highest prison incarceration rate among the developed countries in the world. States' budgets devoted to departments of corrections have grown substantially with an ever-increasing prison population. About half of all persons in the United States who are released from prison return within three years. This high recidivism rate exacerbates States' public budget problems. Three Bayesian statistical methods and a rich Utah data set that covers release and return over a period of three years are utilized in this research. We explore criminological, sociological, and economic factors to predict parolees return to prison. Comparisons of the predictive performance of models based on Bayesian Classification and Regression Trees, Bayesian Model Averaging, and Extreme Bounds Analysis using reasonable classes of priors are designed to provide useful policy guides.