City Research Online

The prediction of risk level and violent offending using models from classical and bounded rationality - Volume 2

Broad, G. (2004). The prediction of risk level and violent offending using models from classical and bounded rationality - Volume 2. (Unpublished Doctoral thesis, City, University of London)

Abstract

Cues Consultant Psychiatrists use to estimate the probability of future violence were obtained through postal survey. The ten most popular were taken as a basis for data collection. One hundred patient risk assessment reports done by Kent Forensic Psychiatry Service staff were checked in a retrospective file study for these cues. Note was also made of the report writer’s risk conclusion at the time (i.e. low or high), and a search carried out on the Police National Computer to determine if the patients concerned had subsequently offended violently or not. There were 46% assessed as high risk, and 54% considered low risk. Over the follow up period (29 - 76 months) 12 had offended violently while 88 had not. Assessors’ decisions could be predicted at a better than chance rate by using one cue in a non-compensatory model (the matching heuristic) or two cues in a compensatory model such as the logistic regression approach. These two showed little change in accuracy by comparison with a logistic regression model using all ten cues (i.e. 62%, 60% and 60% respectively). There was greater variability when attempting to forecast outcome (i.e. 72%, 88% and 88% respectively). Implications for clinical practice will be discussed.

Publication Type: Thesis (Doctoral)
Subjects: B Philosophy. Psychology. Religion > BF Psychology
Departments: School of Health & Psychological Sciences > Psychology
School of Health & Psychological Sciences > School of Health & Psychological Sciences Doctoral Theses
Doctoral Theses
[thumbnail of Broad thesis 2004 Vol 2 PDF-A.pdf]
Preview
Text - Accepted Version
Download (6MB) | Preview

Export

Add to AnyAdd to TwitterAdd to FacebookAdd to LinkedinAdd to PinterestAdd to Email

Downloads

Downloads per month over past year

View more statistics

Actions (login required)

Admin Login Admin Login