Inside Information: Sorting, Screening, and Signaling Desistance in Correctional Settings

Research Summary: Correctional staff face an inference problem. They are legally required to support and recognize an incarcerated person's efforts at rehabilitation, but they cannot directly observe their true identity or the effort they are making to change. This paper applies the concepts of sorting, screening, and signaling from information economics to create an analytical framework to guide improved correctional practice. The paper first classifies the incarcerated population into three “types”: committed criminals, conformists, and desisters. Sorting at intake in the criminal justice system reliably separates out committed criminals using criminal history, but desisters are often pooled with conformists because intent to change is a dynamic, self-authored process that no risk-needs instrument, however refined, is designed to capture. Screening, whereby staff provide explicit rewards for behavioral compliance, fosters a safer environment, but likewise does not differentiate conformists from desisters, as desisters would comply even without incentives. Signaling solves this problem by shifting the initiative to the inmate, whereby desisters would be motivated to act without any guarantee of an explicit reward. Policy Implications: Continued investment in ever more refined risk-prediction tools misdiagnoses the problem: risk instruments detect risk, not intent to change. Parole boards and correctional institutions should shift weight at reclassification and release away from risk scores and toward screens and primarily signals, including microsignals such as consistent attendance at a work assignment or willingness to accept added supervision. Correctional institutions should expand access to voluntary rehabilitative opportunities without layering on extrinsic rewards. They should also avoid the temptation to make programs mandatory, which dilutes their value as a signal. This framework for solving the inference problem generalizes beyond incarceration settings to any situation where decision makers need to identify desisters among a group of people with criminal history records.