Although recent U.S. Supreme Court decisions regarding the death penalty (e.g., Atkins v. Virginia, 2002) have renewed interest in mental health issues, one topic that has not received much attention recently is the ongoing use of expert testimony to support claims that defendants represent a "continuing threat to society." In this article, we (a) review prior research relevant to determining the accuracy of clinical predictions that capital defendants will commit future acts of criminal violence; (b) summarize new data from current and former death row inmates in Texas that bolster the claim that such predictions are gross overestimates of risk; and (c) review extant research addressing the potential utility of various risk assessment instruments that increasingly are being used to reinforce clinical predictions in capital trials. Despite significant recent advances in the field of risk assessment, clinical assertions that a defendant is likely to commit future violent acts appear to be highly inaccurate and ethically questionable at best. Moreover, available research offers little support for the claim that the accuracy of these predictions will be appreciably improved by relying on more structured risk assessment measures that have some demonstrated predictive validity in other contexts.

The (Sorensen and Pilgrim, An actuarial risk assessment of violence posed by capital murder defendants. J Crim Law Criminol 90: [1251][1252][1253][1254][1255][1256][1257][1258][1259][1260][1261][1262][1263][1264][1265][1266][1267][1268][1269][1270] 2000) actuarial model was developed to predict institutional violence among lifesentenced murderers. However, despite its presentation at capital sentencing, the model has not been validated on death row inmates specifically. This study examined the association between Sorensen and Pilgrim model scores and five types of institutional violence (serious assaults, minor assaults, verbal assault/threats, prison order offenses, and non-violent infractions) among a sample of 155 individuals who had been incarcerated on death row in Texas. Results revealed that risk scores performed better for non-violent infractions than for serious assaults, calling into question the utility of this measure for capital sentencing evaluations.