Which Prisoner Reentry Programs Work? Replicating and Extending Analyses of Three RCTs

twitter.com/‌jenniferdoleac/‌status/‌1181551102292107264

We replicate and extend three studies: one on a swift, certain, fair (SCF) program of graduated sanctions for drug-involved probationers; one on aftercare programs for recently released, drug-involved offenders; and one on a comprehensive reentry program for inmates in Minnesota. We find suggestive evidence that the SCF program reduced recidivism, but estimates are too imprecise to draw clear conclusions. Our reanalysis suggests that endogeneity bias in the original study affected the magnitude and sign of some coefficients, but not statistical significance (although this is because the study is substantially underpowered).
***
Doleac (2019a) reviews the literature on desistance from crime, including existing empirical evidence on the effects of various programs and policies on prisoner reentry outcomes. The above analyses contribute new evidence to relatively thin literatures in three areas: SCF programs, aftercare programs for those with substance-use disorders, and wrap-around services. A number of recent RCTs have attempted to replicate the initial success of the HOPE program in Hawaii. DYT was part of this batch of RCTs, and the authors of that evaluation concluded that DYT had no impact on participants. Combined with null effects from other RCTs of similar programs, this contributed to a sense that HOPE (and SCF more broadly) did not replicate in other contexts. Our results above suggest that this punchline may be misleading. The DYT experiment cannot rule out large beneficial effects of the program on participants, and in fact the point estimates suggest meaningful benefits.