Three empirical investigations of forensic decision-making were conducted: a study of 104 hearings by a forensic tribunal; an evaluation of which aspects of forensic patients' clinical presentation were empirical predictors of violence; and a survey of forensic clinicians to determine which factors they said they used to assess risk of violent recidivism and which they actually used. Results showed a significant correlation between actuarial risk and clinical advice to the tribunal, and a nonsignificant trend for patients higher in actuarial risk to receive more restrictive dispositions. Psychotic diagnoses and symptoms were not indicators of increased risk of violent recidivism. Clinicians endorsed some empirically valid indicators of risk, but also relied on some invalid indicators. There was also inconsistency between factors clinicians said they used and factors actually related to their hypothetical decision-making. An automated system is presented as an illustration of how the consistency and validity of forensic decisions could be enhanced.
2007 John Wiley & Sons, Ltd