Assessment of risk for violent recidivism through multivariate Bayesian classification.

Abstract

Bayesian reasoning has already been applied in the area of assessing recidivism risk. Based on single predictors for re-offending, various authors have pointed out that Bayesian analysis was suited to the problem because the base rate of recidivism could be accounted for in terms of a prior probability. The present paper extends this argument towards the multivariate case. The result is a case-specific probabilistic assessment that allows judges and juries to reach informed decisions. The present paper illustrates the method through the combination of offender's age with data from a structured professional risk assessment instrument, the Psychopathy Checklist-Revised (PCL-R), for a sample of N = 393 German convicts. The combination of these two criteria emerged as optimal from all available subsets of predictors (including the History Clinical Risk-20 and its components). The effect size for the Bayesian combination measure with regard to violent offense recidivism was large and significantly higher than the predictive value for each of its constituents. The study design was retrospective, average time at risk was 6.5 years.

Normative Data for the Psychopathy Checklist–Revised in German-Speaking Countries

“…So far, there is little support for the notion of normal distribution properties or even distribution symmetry for PCL-R/PCL:SV data from assessments within the German-speaking domain. Nuhn-Naber and Rehder (2005) as well as Mokros, Stadtland, Osterheider, and Nedopil (2010) reported bimodal sample distributions. Dahle (2005) found a unimodal distribution, yet the distribution was nonnormal with a slight left skew.…”

Bayes and Base Rates: What Is an Informative Prior for Actuarial Violence Risk Assessment?

“…Nevertheless, both assume nature comprises objective realities and both are expected, incrementally, to achieve the same accurate representation of those realities by gathering data. Indeed, in many practical contexts, it appears the two approaches yield very similar answers and conclusions (Beauregard & Mieczkowski, ; Berk, Campbell, Klap, & Western, ; Jensen, McShane, & Wyner, ; Liu, Chen, Yen, & Chen, ; Mokros, Stadtland, Osterheider, & Nedopil, ; Newcombe et al, ; Oleson, ; Sebastiani, Solovieff, & Sun, ). Where they differ is in the logic underlying conclusions drawn from those data.…”

Final-Stage Diversion

“…The current study suggests further research using risk scores and other variables to identify felony probationers at great risk of failure. However, we recommend the use of multiple risk measures (e.g., actuarial and structured) with an examination of their relative usefulness (see, e.g., Mokros, Stadtland, Osterheider, & Nedopil, 2010). At present, the 30-day trial period appears to have merit in testing participants’ investment in treatment and determining whether sufficient resources are available.…”