Evaluation of the predictive validity of LS / CMI in the prediction of incidents in prison

The purpose of this study was to address the critical need to improve accuracy in risk assessment for prison misconducts and hence improve decision-making. With the increased prison crowding in the correctional facilities in Quebec, there is a critical need for the development of better instruments to assist clinicians in their task. Since February 2007, all offenders sentenced for a violent crime or to six month or more are now evaluated with the French-Canadian version of the LS/CMI. Although the LS/CMI is generally good at predicting recidivism, it was not specifically designed to predict prison misconducts. The present study investigates the potential predictive validity of the LS/CMI. The sample consisted of 934 male offenders sentenced for a violent crime or to 6 month or more in the Correctional Services of Quebec between February 2007 and December 2007. In general, results indicate that the LS/CMI can predict violent and non violent misconducts.