are limited, generally because of inadequate program implementation and because of failure to fully account for self-selection bias in evaluations.

These findings have major implications for the design of services and interventions, for both parole systems and other community-based programs. A first-time parolee who committed a violent crime needs a different program than a drug-using repeat parolee who committed a drug crime. However, concentrating supervision and services in the first days and weeks out of prison is likely to have most effects on desistance.


The evidence on the rates of death and crime commission within the first weeks of a person’s release argue strongly for a redirection of postrelease program and service provision efforts to those first days and weeks after release. We recommend that parole authorities and administrators of both in-prison and postrelease programs redesign their activities and redirect their resources to provide major support to parolees and other releasees at the time of release. Such programs may take many different forms, including: intensive and detailed prerelease and postrelease counseling; immediate enrollment in drug treatment programs, intense parole supervision; assistance in finding work; short-term halfway houses; mentors who are available at the moment of release; and assistance in obtaining identification, clothes, and other immediate needs. The key is that a person should not leave prison without an immediately available person and plan for postrelease life.

We also recommend that longer term assistance for parolees include cognitive-behavioral treatment approaches.


Future research on parole and desistance needs to be more methodologically rigorous. In program design, random assignment should be used when possible; when it is not possible, more attention is needed to the selection of comparison groups and to the use of appropriate statistical techniques to account for differences between program participants and comparison groups. In program implementation, more attention is needed to ensure fidelity to program principles and procedures. In program evaluation, more attention is needed to avoid the possibility of self-selection bias.

Rigorous research is needed to explain gaps between the research findings on what influences desistance and evaluation findings of program effects. Also needed are improvements in the conceptualization and design of program content based on research findings on desistance. Research is

The National Academies of Sciences, Engineering, and Medicine
500 Fifth St. N.W. | Washington, D.C. 20001

Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement