Beyond prediction to explanation in risk assessment research. A comparison of two explanatory theories of criminality and recidivism.

Predictions of dangerousness, violence, and recidivism continue to play an ever-growing role in the legal system. These psycho-legal assessments determine a myriad of legal outcomes including, but not limited to, probation status (i.e., prediction of recidivism), civil commitment (i.e., prediction of danger to self or others), psychotherapist liability under a Tarasoff duty (i.e., prediction of danger to a specified victim), the imposition of the death penalty (i.e., prediction of a continued danger to society), and sentencing under the new sexual predator laws (i.e., prediction of danger to society and recidivism). Given the substantial inaccuracy of clinical predictions of recidivism, violence, and dangerousness and the extreme legal importance of these judgments, risk assessment research was undertaken during the 1980s to improve these predictions (Melton, Petrila, Poythress, & Slobogin, 1997; Monahan & Steadman, 1994). 1 Research determined that clinical predictive effectiveness was improved with specific populations under certain conditions. For example, empirical research found that clinicians were more accurate in predicting wellspecified behavioral outcomes over short time periods than they were at pre-