Modeling Recidivism within the Study of Offender Reentry: Hierarchical Generalized Linear Models and Lagged Dependent Variable Models. (January 2017)

• Record Type: Journal Article
• Authors: Mowen, Thomas J.; Culhane, Scott E.
• Other Names: Helmus L. Maaike guest-editor.; Babchishin Kelly M. guest-editor.
• Is Part Of: Criminal justice and behavior. Volume 44:Number 1(2017:Jan.)
• Page Start: 85
• Page End: 102
• Publication Date: 2017-01
• Subjects: reentry -- criminal offending -- hierarchical generalized linear models -- lagged dependent variable models; Correctional psychology -- Periodicals; 364.6
• DOI: 10.1177/0093854816678647
• Languages: English
• ISSNs: 0093-8548

Abstract

Although there are multiple statistical approaches used in understanding reentry, there is little consensus on the benefits and limitations of some of the more popular techniques as they relate to each other. Here, two common methods, lagged dependent variable modeling and hierarchical generalized linear modeling, are contrasted. To examine how particular modeling strategies may lead to different understandings of recidivism within reentry, we use data from the Serious and Violent Offender Reentry Initiative (SVORI; N = 1, 697) to provide an example of the two statistical approaches and discuss the benefits and limitations of each strategy. While researchers will need to make important decisions about which strategy best addresses their research question, results of our analyses show that in dealing with reentry data across more than two waves, a hierarchical generalized linear model is often the preferred approach.