Causal Identification, Imprisonment Effects and the Life-Course

• In Event: Imprisonment and Reoffending

Abstract

Estimating the effects of imprisonment on recidivism and other life-course outcomes remains a continuing causal identification challenge. Even in countries where widespread use of imprisonment is common, the imprisoned are still a highly selected population of individuals. This situation complicates the construction of a suitable counterfactual comparison group for the estimation of imprisonment’s effects on recidivism. In order to clarify the extent of this challenge, the present paper reconsiders the estimation of imprisonment effects through the lens of the life-course paradigm. Imprisonment, typically conceived of as an isolated treatment or experience, is reconceived as an instance of a potential life-course transition embedded within a larger life-course trajectory. Under this framework, the experience and timing of imprisonment become crucial components of an individual’s life-time cumulative contacts with the criminal justice system, which cannot be ignored when measuring recidivism or other effects of imprisonment. This reconceptualization generates two principal insights: (1) the importance of recognizing that all estimates of imprisonment effects are local average treatment effects within the life-course trajectories of the imprisoned and (2) the implausibility of using observational data and conventional statistical techniques to estimate the effects of imprisonment. The benefits of this approach to the estimation of imprisonment effects are demonstrated using data from a large urban jurisdiction in the United States.

Author

• Charles Loeffler, University of Pennsylvania