Dangerousness: Assessing the risk of violent behavior.

Abstract

Results of discriminant analyses for identifying dangerous inpatients and prison inmates are presented. Analysis of a hospital sample (N = 100) yielded a discriminant function containing 5 variables, which was 85% accurate in classifying the sample. Analysis of a prison sample (N = 100) yielded a discriminant model with 6 variables, which was 72% accurate in classifying the sample. Stepwise discriminant analysis of the combined hospital and prison derivation sample (N = 200) yielded a discriminant function containing 8 variables, which was 75% accurate in classifying the sample as dangerous or nondangerous. It was concluded that the derived population-specific (i.e., hospital or prison) models constitute empirically valid measures of dangerousness for the populations studied.