Rossegger Astrid, Laubacher Arja, Moskvitin Konstantin, Villmar Thomas, Palermo George B, Endrass Jérôme
Criminal Justice System, Canton of Zurich, Switzerland.
Int J Offender Ther Comp Criminol. 2011 Aug;55(5):716-31. doi: 10.1177/0306624X09360662. Epub 2010 Feb 11.
Research in the area of predicting recidivism has produced several well-validated standardized risk assessment instruments. The question arises, which instruments best serve which purposes? The objective of this study was to evaluate and compare several actuarial and dynamic risk assessment instruments as to their predictive accuracy and their usefulness in forensic practice. The sample consisted of 109 violent and sex offenders who had been released from prison in Switzerland between 1994 and 1999, and for whom the Psychopathy Checklist-Revised (PCL-R); Historical, Clinical, Risk Management-20; Level of Service Inventory-Revised; Violence Risk Appraisal Guide (VRAG); and the Swiss assessment instrument FOTRES were scored. Using bivariate logistic regression analyses, all instruments were able to discriminate between recidivists and nonrecidivists. The receiver operating characteristic analyses yielded area under the curve values between 0.70 (VRAG) and 0.84 (PCL-R). Furthermore, it was shown that solely examining AUC values does not suffice to determine usefulness. A comprehensive evaluation of an instrument's usefulness for forensic practice should also look at qualitative criteria such as area of application, specificity of risk assessed, and inclusion of dynamic items among others.
在预测累犯的研究领域已经产生了几种经过充分验证的标准化风险评估工具。问题来了，哪些工具最适合哪些目的呢？本研究的目的是评估和比较几种精算和动态风险评估工具在预测准确性以及在法医实践中的有用性方面的表现。样本包括1994年至1999年间从瑞士监狱获释的109名暴力和性犯罪者，对他们进行了《修订版心理变态检查表》（PCL-R）、《历史、临床、风险管理-20》、《修订版服务水平量表》、《暴力风险评估指南》（VRAG）以及瑞士评估工具FOTRES的评分。通过二元逻辑回归分析，所有工具都能够区分累犯和非累犯。接受者操作特征分析得出曲线下面积值在0.70（VRAG）至0.84（PCL-R）之间。此外，研究表明仅检查AUC值不足以确定工具的有用性。对一种工具在法医实践中的有用性进行全面评估还应考虑定性标准，如应用领域、评估风险的特异性以及是否包含动态项目等。
