Identifying Criminogenic Needs

• In Event: Applications of Analytics to Community Corrections in Washington, DC

Abstract

Dynamic criminogenic needs—one of the three core inputs into the RNR model—are typically quantified using some type of an unsupervised learning algorithm. Since need levels are rarely observed directly, needs scales are constructed using theoretically defined and measurable manifestations. Once constructed, tests are often undertaken to ensure the latent needs are indeed related with risk. This paper uses a confirmatory factor analysis to assess the adequacy of available manifestations in assessing offender needs and then applies a mixed model approach—combining the available manifestations and observed misconduct outcomes—to measure criminogenic needs. Findings and practical implications are discussed.

Author

• Avinash Bhati, Maxarth, LLC