Adventures in Risk: Predicting Violent and Sexual Recidivism in Sentencing Law
58 Pages Posted: 28 Mar 2014 Last revised: 31 Mar 2014
Date Written: March 27, 2014
Abstract
Risk has become a focal point of criminal justice policy. Officials draw upon the sciences for the best evidence to differentiate between offenders at high risk of being a future threat to society, for whom preventive incapacitation may be justifiable, and those at low risk, for whom diversion might alleviate the overuse of imprisonment. A recent turn in evidence-based practices is to borrow the newest technologies developed in the forensic mental health field to better classify offenders accordingly to their predicted likelihood of recidivism.
Actuarial risk assessment is considered the new frontier as a progressive sentencing reform, representing best practices in predicting recidivism risk. The actuarial turn is adjudged to offer probabilistic estimates of risk that are objective, reliable, transparent, and logical. Policy groups, state legislatures, judges, and probation offices actively promote the use of actuarial risk assessment, believing the empirically-derived tools effectively standardize sentencing practices, mitigate bias, and thereby increase the legal and moral standing of sentencing outcomes.
Actuarial prediction is promoted as founded upon scientific and empirical principals. This Article critically analyzes the predictive abilities of actuarial risk prediction tools utilizing statistical, empirical, and legal methods. A specific focus herein is the risk prediction of those criminals for whom fear is strongest: violent and sexual offenders. Several questions are of interest: Is widespread reliance on actuarial sentencing justified? Are actuarial risk results sufficiently relevant, valid, and reliable for sentencing law? Is actuarial evidence too prejudicial, confusing, and misleading to meet evidentiary standards in sentencing? The Article addresses proponents’ arguments that, regardless of any weaknesses, actuarial risk results should be admissible because they constitute merely one piece of evidence in a multi-faceted decision and that any flaws or errors in the evidence can be deduced through normal adversarial processes.
Keywords: sentencing, risk assessment, violence, recidivism, expert evidence, validity and reliability
JEL Classification: K14, K41, K42
Suggested Citation: Suggested Citation