The effect of prison gang membership on recidivism

How does prison gang membership affect recidivism? In this paper, we use a unique dataset of all releasees from prisons operated by the Illinois Department of Corrections during the month of November 2000, which includes demographic information and data on gang participation. We attempt to control for confounding factors that are traditionally associated with both prison gang membership and rearrest. We develop a potential-outcomes framework and describe the conditions under which a counterfactual can be estimated when gang membership is not randomly assigned. We combine regression analysis with Coarsened Exact Matching, which has several advantages over the more popular propensity score matching, to estimate the effect of gang membership on recidivism. Prison gang membership results in a six percentage point increase in recidivism. Despite the strengths of the data, unobserved heterogeneity among inmates could still bias estimates. However, there are probably important subtleties to the gang participation decision such that experimental or quasi-experimental data are unlikely to increase our understanding of the relationship between gang-membership and post-release outcomes. We recommend incorporating ethnography with survey data collection, because ethnographers are able to document otherwise unobservable contextual information concerning the selection process which could be used to identify causal relationships.