Sentence Length and Recidivism Among Prisoners Released Across 30 States in 2005

Abstract

Longer prison sentences correlate with lower recidivism. However, it is difficult to disentangle individual effects related to aging and prior criminal histories from state practices and policies that impact the likelihood of reoffending. This study examines a large, administrative data set containing the criminal histories of prisoners released in 2005 to investigate this problem. We used Cox models with a shared frailty framework to account for the latent, state-level correlations that have a multiplicative effect on the individual likelihood of rearrest. These models reveal that the state effects account for considerable variation in rearrests, while also clarifying the roles of sentence length and disparate criminal histories on recidivism.