The Last Mile: Evaluating a Coding Program in Prisons

How to cite

Braga, Anthony; Tahamont, Sarah; Chalfin, Aaron. 2026. "The Last Mile: Evaluating a Coding Program in Prisons." AEA RCT Registry. AEARCTR-0018791. https://doi.org/10.1257/rct.18791-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Anthony Braga* , Sarah Tahamont , Aaron Chalfin

* Primary investigator

Registry ID
AEARCTR-0018791
DOI
https://doi.org/10.1257/rct.18791-1.0
First registered
Last updated
Status
In development
Trial dates
2022-03-01 – 2032-08-31
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/18791

Abstract

Despite the breadth of available programming and the theoretical foundation for the promise of vocational training, prior literature provides scant evidence about the effect of vocational training in prisons (Doleac 2023). One potential explanation is that the vocational programs on offer in U.S. prisons are not developing the kinds of labor market skills that will lead to jobs with attractive wages and the possibility of career advancement. To address this, we are evaluating The Last Mile's (TLM), a program that provides vocational training in prison in web development and audio/video production along with reentry services upon release. While individual outcomes vary, TLM graduates work at leading tech firms like Asana, Slack, and Meta. Using a randomized control trial, with a target sample of 1200 individuals assigned across treatment and control groups, this research will study both the effect of being offered TLM and of participating in TLM on in-prison misconduct, post-release employment, and recidivism. We believe this research will provide a rare, rigorous test of the offer of credibly lucrative labor market skills and will provide evidence not only for policymakers to facilitate reentry but on the credibility of a broader set of theories of desistance from crime.

Experimental design

This is a multi-state, multi-site RCT, TLM operates in 12 facilities across six U.S. states (with planned expansion). The target population for the study is all eligible TLM applicants.

TLM invites eligible individuals to participate on the basis of a fair lottery as lottery capacity allows. Lotteries are conducted at the facility-cohort. When a seat becomes available, the two eligible applicants whose expected release date is closest to the expected end of course date will advance to the lottery. Through this lottery, one of the two applicants will be randomly selected to participate in TLM. All individuals who enter the lottery will be part of the study. Those who win the lottery will be invited to participate in TLM and thus constitute the treatment group. Those who do not win the lottery will not be invited to participate in TLM and thus constitute the control group. Program enrollment, and thus the administration of the lottery, will be conducted by TLM. The research team will observe the lottery infrastructure on an ongoing basis to empirically assess the progress of the lotteries and ensure fidelity to the randomization procedures.

The study will compare average outcomes for the treatment group to those of the control group in both intent-to-treat (ITT) and treatment on the treated (TOT) framework.

Design details

Randomization method
TLM runs the lottery using a computerized lottery assignment system. The system operates in two parts: 1) eligibility screening and 2) lottery assignment. For each facility cohort, TLM screens applicants for eligibility. When it is time to fill the classroom seats, TLM runs eligible applicants through the lottery system.
Randomization unit
Randomization is pairwise by individual eligible TLM applicant within facility cohorts.
Clusters
The treatment is individually randomized, but the treatment is clustered by TLM classroom at the facility cohort level. The number of facility cohorts is unknown at this time. TLM assigns classroom seats by fair lottery, so the number of clusters will be the number of facility-cohorts that are required to reach the target sample of individuals. TLM classrooms vary in size.
Observations
1200 individuals
Arms
Randomization is pairwise so the target is 600 offered a seat in TLM and 600 control.
Primary outcomes
Recidivism and Employment

Cited by

No citation data available for this registration yet.

Keywords