The structural coherence of clinically derived dynamic indicators of reoffending risk

• Published in: Criminal behaviour and mental health Vol. 14; no. 4; pp. 263 - 279
• Main Authors: Philipse, Martien W. G., Koeter, Maarten W. J., van den Brink, Wim, van der Staak, Cees P. F.
• Format: Journal Article
• Language: English
• Published: Chichester, UK John Wiley & Sons, Ltd 2004
• Subjects: Adult; Clinical indicators; Coherence; Crime - legislation & jurisprudence; Crime - prevention & control; Criminal Psychology - legislation & jurisprudence; Expert Testimony - legislation & jurisprudence; Female; Forensic psychiatry; Hospitals, Psychiatric; Humans; Male; Netherlands; Offenders; Personality Assessment - statistics & numerical data; Prisoners - psychology; Prisoners - statistics & numerical data; Psychometrics - statistics & numerical data; Recidivism; Reproducibility of Results; Risk assessment; Risk Assessment - legislation & jurisprudence; Risk Assessment - statistics & numerical data; Secondary Prevention; Statistics as Topic; Violence - legislation & jurisprudence; Violence - prevention & control; Violence - psychology; Netherlands
• ISSN: 0957-9664; 1471-2857
• DOI: 10.1002/cbm.593

Background Forensic psychiatrists aim to reduce the risk of reoffending through treatment. With few exceptions, research evidence tends to favour risk assessment aids reliant on fixed historical rather than clinical data, but transparency in the making of clinical judgements is lacking. Aims To evaluate further a clinically derived 47‐item dynamic risk assessment checklist; specifically to test first whether it has a meaningful dimensional structure and, second, the extent to which items and underlying dimensions are associated with a separate, direct clinical judgement of risk of reoffending. Methods Data sets from four previous studies on the reliability and validity of the Clinical Inventory of Dynamic Reoffending Risk Indicators (CIDRRI) were merged, yielding 370 cases. The resulting data set was analysed using principal axis (common) factor analysis, with orthogonal (varimax) rotation. In addition, receiver operating characteristic (ROC) curves were calculated. Results The model of best fit depended on treatment stage; for those in residential treatment it was a six‐factor model (responsibility, self‐reliance, antisocial/narcissistic traits, treatment compliance, goal attainment and avoidance); for those back in the community a five‐actor model in which, effectively, the compliance and goal attainment factors became one, provided a better fit. These dimensions and some individual scale items were associated with clinical judgement of risk of reoffending. Conclusions and clinical implications This study provides evidence that the CIDRRI is an adequate operationalization of underpinnings of clinical risk assessment, and that these underpinnings are part of meaningful higher‐order dimensions. The CIDRRI is a viable instrument for clinical use, taking only 15 minutes to complete and identifying dynamic risk factors, the validity of which needs to be further established. Copyright © 2004 Whurr Publishers Ltd.