Forecasting Methods in Crime and Justice

Abstract

Researchers who study crime and justice have been forecasting a variety of outcomes for nearly a century. Perhaps the most well known are forecasts of behavior after release from prison. Other examples include forecasts of prison populations and time trends in crime. However, it is very difficult to determine how accurate such forecasts have been because few forecasts have been properly evaluated. In the hope of improving crime and justice forecasts in the future, this article reviews modern forecasting methods and their applications to crime and justice questions. Among the key challenges for researchers are how best to arrive at useful forecasting models, taking the costs of forecasting errors into account, and estimating nonlinear relationships.