Date of Award


Document Type


Degree Name

Doctor of Psychology (PsyD)


Graduate Department of Clinical Psychology

First Advisor

Kathleen Gathercoal, PhD

Second Advisor

Mary Peterson, PhD

Third Advisor

Diomaris Safi, PsyD


Violence risk assessment has evolved significantly over the past 50 years. The newest generation of risk assessment tools is intended not only to assess risk but also to inform treatment. The Short-Term Assessment of Risk and Treatability, i.e., the START, is an example of a 4thgeneration tool that identifies risk and protective factors, highlighting the dynamic variables that may decrease risk. As treatment for mentally ill offenders shifts from inpatient hospital environments to community-based treatment settings, a focus on dynamic risk factors is crucial for maintaining safety for patients and the community. This study aimed to identify the predictive validity of START measures completed at time of discharge on participants who were as adjudicated Guilty Except for Insanity (GEI) and transitioning to conditional release under the Oregon Psychiatric Securities Review Board. Using archival data, collected from PSRB files, 31 adjudicated GEI participants on conditional release in the year 2013 were found to have completed START results and records of positive, negative, and revocation outcomes. The static factors of Length of Time PREDICTIVE VALIDITY OF START IN COMMUNITY iv in years from Date of Judgment to date of Conditional Release and Length of Sentence were not predictive of positive, negative, nor revocation outcomes. Dynamic factors of Strengths total scores and Vulnerabilities total scores were predictive of positive modifications, and revocation outcomes, respectively. Specifically, the ROC area Under the Curve (AUC value), used to predict the utility of forensic assessment tools, indicated that Strengths total scores had an AUC = 0.85 (good) when predicting positive modifications and the Vulnerabilities total scores had an AUC = .70 (fair) for negative modifications and AUC = .84 (good) for revocations. These results should be interpreted with some caution because of the limited sample size, but suggest that the dynamic factors of the START have predictive validity. These findings are discussed with regard to the practicality of assessing the START on a regular basis and directions for future research.