This paper extends the analysis begun in Arnold (2007), The Dynamic Predictive Validity of the Level of Service Inventory-Revised. The combination of stability and change in criminal behavior over the life course represents one of criminology's most enduring puzzles. This study examines how the criminal offender's risk of recidivism changes over time and provides guidance for criminology, forensic psychology, and criminal justice. Using 3,190 probationer assessments collected from a Central Minnesota county, a multidisciplinary approach leads to improvements in the methods of dynamic risk assessment, and provides an analysis of the stability and dynamism present in offender risk levels, both between and within individuals. Methods are demonstrated for improving measurement accuracy in test-retest environments and creating measurement criteria to determine the change in risk score needed to predict a change in recidivism. Results include comparisons of between group and within group changes in risk levels over time, indications of how the rank order levels of risk change over time, measurements of the temporal stability of risk, and an indication that changes in risk are typically nonlinear rather than linear in nature. An analysis of the results suggests methods for improving the risk assessment process, precautions needed when measuring results from treatment programs, ways to improve desistance research, and future directions for theory. This is a working paper. Comments and feedback are welcome. Please contact the author via ORCID: 0000-0003-3537-3774
Thomas K. Arnold (Tue,) studied this question.