Forensic Psychology
Class: Risk Assessment
March 20, 2012
Harris, G. T., Rice, M. E., & Quinsey, V. L. (1993). Violent recidivism of mentally disordered offenders:
The development of a statistical prediction instrument. Criminal Justice and Behavior, 20, 315-335.
Purpose
•Develop an actuarial risk assessment instrument for the prediction of violent recidivism
Participants
•618 male offenders/patients
•332 admitted for treatment to maximum security psychiatric institution
•286 admitted only for brief pretrial psychiatric assessment
Procedure
•All predictor variables were coded retrospectively from institutional files
Inter-rater Reliability
•Three coders
•20 randomly chosen participants
•Reliability criterion was set at .70
•For the variables that met this criterion, mean correlation was .90 and mean kappa was .83
Violent Recidivism
•Post-release offences coded from official files
•Any new charge (or equivalent) for a Any new charge (or equivalent) for a violent offences
•Violent offences included all assaults and sexual assaults, armed robbery, forcible confinement,
threatening, and pointing a firearm” (p. 322)
•Mean time at risk = 81.5 months (SD = 60.6)
•Violent recidivism rate = 31% (191/618)
Analysis
•Separate stepwise discriminant analysis for each set of variables
–Childhood history Childhood history
– Adult adjustment
– Index offences
– Assessment result
•These discriminant analysis were conducted separately for each of five groups of participants
–Entire Entire sample (N = 618) sample (N = 618)
– Treated participants (n = 332)
– Assessed participants (n = 286)
– Randomly selected half of the entire sample
–Remaining half of the entire sample
•Only variables selected by at least three of the five discriminant analysis of the different
participant groups were different participant groups were entered into the final discriminant
analysis