FORECASTING CRIMINAL SENTENCING DECISIONS

This paper investigates the feasibility of developing bootstrapping mo dels of criminal sentencing decisions to predict sentence length and t he outcome of sentencing appeals. New Zealand Court of Appeal sexual o ffence sentencing decisions are modelled as a function of variables de scribing the character of the offender and the circumstances of the of fence. The model outperforms both an equal weights approach and a naiv e mean projection in predicting sentence lengths in a holdout sample. As applications and further validations of the model, the disparity be tween actual and predicted sentence is used to forecast the success of an appeal and to predict change in sentence, given a successful appea l. The results confirm that bootstrapping models have the potential to contribute to a field traditionally believed to favour solely judgeme ntal approaches.