New Research Suggests Potential of Prison Furloughs, But Shadow of Willie Horton Still Looms

It remains the paradigmatic moment in the modern history of tough-on-crime politics. In  the summer of 1988, Michael Dukakis, the Democratic Governor of Massachusetts, seemed to be cruising toward a presidential election victory in November. Then, Republican operatives began to pummel him for a horrific failure in Massachusetts’s prison furlough program. This program offered short leaves for inmates to spend time at home, which was thought to help prepare them for their permanent release. The program had a good track record until an inmate named Willie Horton absconded during one of his releases and brutally assaulted a young couple. As the Horton story became more widely known nationally, Dukakis’s lead in the polls evaporated. His eventual loss seemed to confirm that politicians could no longer afford even a tangential association with policies or programs that were perceived to be soft on crime.

The Horton story reverberated for years across the whole field of criminal justice, but perhaps its most direct impact was a sharp constriction in prison furlough programs, which had previously been widely accepted and utilized by American corrections officials.

As furlough programs faded away, so, too, did research on their effectiveness. Although several older studies suggested that furloughs might help to reduce post-release recidivism, there has been a growing need for updated research.

A new paper by L. Maaike Helmus & Marguerite Ternes helps to fill the gap.  


Helmus and Ternes studied a sample of more than 27,000 Canadian prisoners who were released between 2005 and 2011. They found that twenty-two percent had been granted a “temporary absence” (TA). This included both escorted TAs (e.g., a trip to perform community service under the supervision of a correctional officer) and unescorted TAs (e.g., a weekend home with the family).

Helmus and Ternes sought to compare the post-release outcomes of those inmates who had TAs with those who did not. In making this comparison, they controlled for a range of well-recognized risk factors, such as criminal history. The results were consistent with earlier research in finding a positive association between TAs and post-release success.

More specifically, for inmates who had a TA, the odds of post-release unemployment were reduced by about one-third, while the odds of being returned to custody with a new offense were reduced by nearly twenty percent. The effects were stronger for those who had unescorted TAs than for those with only escorted TAs.

Helmus and Ternes also sought to investigate dosage effects, that is, whether and to what extent effects varied based on the number of TAs. Their data indicate that higher numbers of TAs are associated with better outcomes. This conclusion held for total TAs and escorted TAs, although the results for unescorted TAs did not achieve statistical significance. For instance, “offenders who participated in 60 or more TAs had substantially lower unemployment rates compared to offenders who participated in 20–30 TAs, who also had substantially lower rates of unemployment than offenders with only one to five TAs” (33).

The study suggests that TA/furlough programs may pay sizable dividends, but it is, of course, not conclusive. Despite the effort to take into account basic risk factors, one cannot rule out the possibility that there remain important differences between the TA and non-TA populations that were not controlled for. Ideally, future research would involve random assignment of inmates to TA and non-TA conditions. Additionally, it is possible that the Canadian context may somehow play an important role in producing the improved outcomes found by Helmus and Ternes; results might be quite different in the U.S.

If further research continues to yield positive results, will U.S. policymakers embrace the prison furlough concept again, or is the shadow of Willie Horton still too long?

The question draws attention to a fundamental political impediment to good correctional policymaking. When an inmate who is furloughed or otherwise out early commits a serious, victimizing offense, the program and anyone who supported it are apt to find themselves in the crosshairs of public criticism and political attack. On the other hand, when a furlough or early release program succeeds in reducing recidivism rates, the good news gets far less public attention. There may be fewer victims, but the crime reduction is a statistical abstraction; it is impossible to point to any specific person who was saved from victimization by the program. There is a seemingly unavoidable imbalance: failures generate compelling stories of horrific injury, while successes are unlikely to penetrate the public consciousness. Thus, even a demonstrably successful program that heads off dozens of victimizing offenses per year might be torpedoed by a single, isolated failure—with significant net costs to public safety.

The Helmus-Ternes paper is Temporary Absences from Prison in Canada Reduce Unemployment and Reoffending: Evidence for Dosage Effects from an Exploratory Study, 23 Psychology, Pub. Pol’y, & L. 23 (2017).