Formulating Generic Levels in Risk Assessment

• In Event: Advances in Risk Assessment and Prediction

Abstract

The potential of different risk assessment instruments being able to “talk” to each other would greatly assist in the communication of risk estimates among systems. After identifying the median score of the instrument, the second step is to calculate the cut-score between the second (II) and fourth (IV) risk levels, which used odds ratios of recidivism is used to identify the ratio that is associated with the expected reduction effect of treatment. Based on meta-analyses of treatment outcome studies, expected treatment change is r = .10, d = .20, or odds ratios of .70/1.43. The median scale score = “1.0” and the cut-points for the third (III) level are defined by the average effect of treatment (between odds ratio of .70, II/III boundary, and 1.43, the Level III/IV boundary). The third step calculates the cut-scores between the first and second levels and the fourth (IV) and fifth (V) levels. The predicted probabilities are derived from logistic regression models for the ≤ 5% provides the level I/II threshold and the predicted probability of ≥ 85% provides the level IV/V threshold. This system is tested among probationers (N = 24,936) and parolees (N = 36,303).

Author

• Daryl G. Kroner, Southern Illinois University