Violence Risk Assessment: An Experiment in Improvement (VIDEO in link) 1


For many years, forensic psychiatrists and psychologists have been tasked with assessing individuals’ risk of violence. There are three fundamental approaches to violence risk assessment: using unaided clinical judgment, using statistical models, and using structured professional judgment. Unaided clinical judgment, the oldest approach, is basically when an evaluator forms an opinion on the individual’s risk of violence based on data gathered from an unstructured clinical interview. Statistical, or actuarial, methods, use groupings of risk factors that have previously been associated with violent behavior to generate a violence risk estimate. Structured professional judgment, the most recently developed method, guides evaluators in a structured approach to use known risk factors to estimate an individual’s violence risk without utilizing statistical methods. Each of these approaches could be used in combination.

Along with a colleague, I sought to determine whether violence risk estimates improved when clinicians used their professional judgment to adjust actuarial violence risk estimates. Most professionals who endorse actuarial assessment of violence risk recommend combining clinical judgment with actuarial risk to generate a final risk estimate. The developers of the Classification of Violence Risk (COVR), for instance, advise that the administering clinician develop a final risk estimate that includes consideration of information not addressed within the instrument. However, at least one scholar argues that “actuarial methods are too good and clinical judgment is too poor to risk contaminating the former with the latter.”

To our knowledge, no study had examined whether using clinical judgment to adjust actuarial violence risk estimates would more accurately predict violence risk than actuarial assessment alone. To answer this question, we prospectively followed almost 900 patients starting upon admission to a state forensic hospital, almost 600 of which were administered the COVR. We documented violent acts (classified as either impulsive, psychotic, or predatory) committed by 20 weeks post-admission or time of discharge, whichever came first. Two violence risk estimates were generated for the group: one by using the COVR alone, and the other by using clinical judgment to adjust the COVR’s actuarial risk estimate. Our data indicate that adding clinical judgment to actuarial assessment more accurately predicts overall aggression and impulsive aggression than actuarial assessment alone. It did not more accurately predict psychotic or predatory aggression. Our findings were presented at the American Academy of Psychiatry and the Law (AAPL) conference in 2014, and we plan on publishing our results. 

Here is a news video of me explaining the study at the AAPL conference.


Leave a Comment

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

One thought on “Violence Risk Assessment: An Experiment in Improvement (VIDEO in link)