The relationship between confidence and accuracy in clinical assessment of psychiatric patients' potential for violence.

Abstract

The authors studied the relationship between confidence and accuracy in clinical assessments of psychiatric patients' short-term risk of violence. At the time of entry to the hospital, physicians (N = 78) estimated the probability that each of 317 patients would physically attack other people during the first week of psychiatric hospitalization. The clinicians also indicated the degree of confidence they had in their estimates of violence potential. Nurses rated the occurrence of inpatient physical assaults with the Overt Aggression Scale. The results showed that when clinicians had a high degree of confidence, their evaluations of risk of violence were strongly associated with whether or not patients became violent. At moderate levels of confidence, clinicians' risk estimates had a lower, but still substantial relationship with the later occurrence of violence. However, when clinicians had low confidence, their assessments of potential for violence had little relationship to whether or not the patients became violent. The findings suggest that the level of confidence that clinicians have in their evaluations is an important moderator of the predictive validity of their assessments of patients' potential for violence.