Examining the Predictive Validity of the Public Safety Assessment

The purpose of this study was to examine the predictive validity of the Public Safety Assessment (PSA), an actuarial pretrial assessment instrument, administered to 15,931 individuals in Volusia County, Florida, between 2016 and 2017. A series of logistic regression models analyzed the influence of the PSA's risk scores for Failure to Appear (FTA) and New Criminal Activity (NCA), as well as gender, race, and the length of time spent in pretrial custody on incidents of failure to appear and new pretrial arrest. The findings suggest that while both the FTA and NCA scales predicted pretrial failure fairly well, the variation explained by the models suggest that there is much that we do not understand about predicting pretrial failure to appear and new pretrial arrest, indicating the need for further research and refinement of pretrial assessment instruments.