Stopping the Revolving Door: What we know—and still need to learn—about reducing recidivism in the United States

Milwaukee PRI interventionMore than three of every four individuals released from U.S. prisons are re-arrested within five years. Given the sharp negative effects that incarceration has on individuals and their communities, as well as the often staggering expense of jails and prisons to governments and taxpayers, addressing this “revolving door” has become a top priority for many policymakers.

Reentry interventions—programs designed to help people readjust to society following their release from jail and prison—are one promising strategy. Former inmates often face complex challenges after release and experience an increased risk of homelessness, unemployment, addiction, and trauma. Difficulty accessing housing, jobs, and treatment services puts this population at a high risk of reoffending, creating a self-perpetuating cycle.

Many employers are unwilling to hire individuals with criminal records, making it particularly challenging for former inmates to find gainful employment. Researchers at the National Institute of Justice found that 60 to 75 percent of recently incarcerated individuals were unemployed one year after release, and when they do find employment, former inmates can expect to earn 40 percent less on average than they did before going to jail. Low employment levels and low wages in the formal sector could drive formerly incarcerated individuals back to illegal activity, so helping former inmates find jobs could be a key strategy to reducing recidivism.

Individuals involved with the criminal justice system also frequently experience a disruption in healthcare. 70 to 90 percent of the 10 million people released from jail or prison each year are uninsured. While a lack of insurance can have serious ramifications for the general population, the impact is particularly acute for recently released inmates, who experience mental illness, substance use disorders, infectious disease, and chronic health conditions at a rate that is seven times higher than the general population.

Lack of access to medical care can increase risk of recidivism among those with complex medical needs. This is particularly true for those with substance use disorders. A randomized control trial involving formerly incarcerated men with opioid use disorders found that nearly three-quarters relapsed within 3-months of release, even if they received counseling. A study conducted in Missouri found that individuals who were convicted of drug crimes and/or who have a history of addiction had a 62 to 82 percent estimated probability of re-arrest within 48 months.

The research community has done an extensive amount of work documenting and exploring the challenges faced by formerly incarcerated individuals and the resulting effects on prison populations when those individuals reoffend. However, as policymakers face tough choices about which reentry programs to support with limited funding available, we don’t know enough about which approaches work, which work best, and why.

The Milwaukee Safe Streets Prisoner Release Initiative (PRI)

At J-PAL North America, we’re supporting randomized evaluations to rigorously test prisoner reintegration policies and programs so policymakers can direct their efforts in the most effective way possible. Looking at some past examples of research is helping to guide our thinking on what still needs to be tested. In one study, researchers Anthony Braga (Harvard University), Philip Cook (Duke University), Songman Kang (Hanyang University), Jens Ludwig (University of Chicago), and Mallory O’Brien (Medical College of Wisconsin) evaluated the impact of the Milwaukee Safe Streets Prisoner Release Initiative (PRI) on improving employment prospects and reducing recidivism. The PRI provides intensive, comprehensive services to inmates in Milwaukee, Wisconsin both before and after release with the aim of improving their chance of finding a job and staying out of jail.

From 2009 to 2011, several J-PAL affiliates worked with state, local, and federal stakeholders to design and implement a randomized evaluation of the PRI. Male prisoners aged 35 or younger were eligible for the randomized controlled trial if they had a history of violence or gang involvement and were scheduled to be released in Milwaukee with at least six months of community supervision. Researchers randomly assigned 236 inmates to either a treatment group, in which they automatically received extensive PRI services, or a control group in which they were free to sign-up for PRI programs but received no additional encouragement to do so and in some cases may have been placed on a waiting list if they did.

In the six months prior to release, inmates in the treatment group met with a social worker and were offered guaranteed access to a community employment program, vocational and soft-skills training, remedial education, the chance to participate in restorative justice circles (in which inmates met with victims to discuss the suffering caused by the crime), and treatment for drug and alcohol abuse. Inmates in the treatment group overwhelmingly took advantage of these services. Approximately 62 percent received treatment for addiction, the majority participated in restorative justice circles, and 54 percent participated in the community employment program. Inmates in the treatment group were also expected to participate in a twelve- to sixteen-week long cognitive therapy program to address behavioral issues that could lead to re-arrest. Approximately 72 percent participated in at least one session. In the 30 days before release, a coordinated-care team composed of social workers, job coaches, and other counselors worked with each inmate in the treatment group to address housing, transportation, and employment needs.

After their release, ex-offenders in the treatment group were eligible for substance-use treatment and the services through the community employment program. Employers were incentivized to hire individuals in the treatment group through subsidies, such as tax-credit certification. Those within the treatment group also took part in follow-up meetings with their coordinated-care team and underwent enhanced supervision from parole agents.

Results from the randomized evaluations found some positive indications within the first year of release. The PRI program increased the likelihood of some employment from 55.5 to 80 percent. 81 percent of the treatment group reported earnings in their first year out of prison, compared to 59 percent of the control group. Those within the treatment group experienced a reduction in the likelihood of re-arrest as compared to the control group (63 versus 72 percent) but there was no discernable difference in the rate of re-incarceration.

However, the PRI’s success was limited—while the program increased the likelihood of former inmates earning an income at all, the treatment group still earned very low wages. Median earnings were approximately $2,700 among the treatment group, leaving many in poverty. The intensive services offered in this intervention did not improve the financial situation of formerly incarcerated individuals to a point of reliable economic self-sufficiency. Their income was not enough to lift a family of two out of poverty. Furthermore, the decrease in the likelihood of reimprisonment for those in the treatment group, as compared to those in the control group, was not statistically significant. This suggests that these intensive services are not cost-effective due to the minimal impact on recidivism.

Through this evaluation, the researchers also aimed to explore a possible causal link between employment and recidivism that has been suggested by previous non-randomized studies. For example, researchers at the Urban Institute conducted a longitudinal study in which they interviewed 740 formerly incarcerated men. Researchers found that formerly incarcerated individuals who were able to secure jobs were less likely to be re-incarcerated 8 to 12 months after release. This reduction in recidivism was even more pronounced for those receiving higher wages. Depending on wages rates, employed formerly incarcerated individuals experienced an 8 to 16 percent likelihood of being re-arrested, depending on wages, compared to 23 percent for those who were not employed. Those receiving the highest wages were more likely to stay out of the criminal justice system. However, since there were so many facets of PRI, it is unclear whether employment deterred recidivism or if other services offered through the program played a bigger role.

A Different Approach to Recidivism: Behavioral Interventions

A cognitive behavioral therapy (CBT) program used in the Cook County Juvenile Temporary Detention Center (JTDC) in Chicago, IL offers a potentially more scalable model for policymakers to adopt in new contexts. The therapy sessions, studied by Sara Heller (University of Pennsylvania), Anuj K. Shah (University of Chicago), Jonathan Guryan (Northwestern University), Jens Ludwig (University of Chicago), Sendhil Mullainathan (Harvard University), and Harold A. Pollack (University of Chicago), generated small but significant and sustained reductions in recidivism at a much lower cost than the PRI.

From 2009 to 2011, the JTDC leadership randomly assigned nearly 2,700 high-risk young men admitted to JTDC to either treatment or control centers. Twice a day, trained detention staff led group CBT sessions for members of the treatment group, which were targeted at reducing impulsive and harmful behaviors that can lead to criminality. This intervention took place in some residential units within JTDC, but not others. This CBT curriculum—developed by Dr. Bernie Glos and his associates at the DuPage County Illinois Juvenile Detention Center—was created by surveying other CBT programs and adapting them based on behavioral science research. This allowed for an extensive understanding of the key elements necessary for successful implementation of CBT in a juvenile detention center.

This CBT program helped high-risk youth avoid being readmitted to JTDC after their release. Juveniles who complied with the treatment were 13 percentage points less likely to be readmitted to JTDC than their control counterparts within two months after release, and 16 percentage points less likely after 18 months (equivalent to a 21 percent reduction in readmission rates to JTDC).

While both the CBT program at JTDC and the PRI intervention were successful by some measures, the PRI intervention cost about $5,000 per participant, whereas the JTDC program cost approximately $60 each. Intensive programs like the PRI may need to have a greater and more sustained impact on recidivism to justify the high cost. On the other hand, investing in cognitive behavioral therapy programs like the one used at JTDC can produce returns ranging from 5-to-1 to 30-to-1 in averted recidivism costs.

Governments, service providers, and researchers are continuing to develop innovative projects aimed at finding scalable, effective solutions to close our criminal justice system’s revolving door. With support from J-PAL North America made possible by the Robert Wood Johnson Foundation, Jennifer Doleac and Benjamin Castleman of the University of Virginia are testing a tablet-based reentry module in two county jails that aims to strengthen inmates’ transition back into society. Through this module, inmates create a personalized transition plan prior to release, and after leaving jail receive support and reminders to encourage them to adhere to their plan. If effective, this highly-scalable, technology-based and behavioral-science informed strategy may be a promising tool for successful reentry reforms.

With a renewed wave of political attention and practitioner innovation in this space, there is so much more we can—and must—learn about how best to support formerly incarcerated individuals to make our communities safer and reduce the burden of our criminal justice system. At J-PAL North America, we’re always excited to hear from prospective partners interested in expanding the evidence base on reentry policy. Reach out to J-PAL North America’s crime sector manager, Ben Struhl (, with your own ideas for evaluations.


Experts Gather at MIT to Explore New Research in Education Technology


Technology is developing at a breathtaking pace, and it’s fundamentally changing the way teachers, policymakers, and researchers think about education. On March 31st. J-PAL North America hosted a conference at MIT to discuss the role of research and evidence in education technology, bringing together a diverse group of leaders across academia, education companies, education practice and administration, and philanthropy to share their experiences implementing and evaluating technology both in and out of the classroom. Throughout the conference, speakers and participants advocated for rigorous evaluation to advance our understanding of how technology can help students—regardless of income level—learn.

Technology—an Opportunity, a Challenge, and the Need for Research

Quentin Palfrey (Executive Director of J-PAL North America) and Phil Oreopoulos (Professor of Economics and Public Policy at the University of Toronto and J-PAL Education Co-Chair) discussed the transformative promise of education technology and some of its most exciting uses, including approaches to personalize learning and scale instruction to learners across different contexts. However, they warned that rapid advances in education technology create the risk of leaving those without access behind, exacerbating already stark inequalities between affluent and low-income students—a public policy problem known as the “digital divide.”

“Emerging fields like machine learning, big data, and artificial intelligence will likely compound the influence of technology even further, increasing the range of tools that ed-tech can draw on and speeding up cycles of learning and adjustment…[but] these technologies are arising in a context of persistent inequality. Despite expanding access, the digital divide remains very real and very big. If ed-tech practitioners and researchers don’t pay close attention to equity of access and tailoring programs to the needs of those at the lower end of the income spectrum, there’s a risk that the growing influence of technology will aggravate the educational inequalities that already exist.” – Phil Oreopoulos

Oreopoulos set the stage with a review of the current evidence on what works, what works best, and why in education technology, drawing on over 90 studies across economics, education, and social psychology. Technology-assisted personalized learning programs emerged as an especially effective approach from the review, which stems from an upcoming education technology literature review. However, many open questions remain about how to leverage technology to help disadvantaged learners, which technologies are the most cost-effective, and why successful approaches work.

Kumar Garg, a former White House advisor who spearheaded President Obama’s efforts to improve STEM education, underscored the tremendous need for investment in education research to help us answer these questions. In 2015, only 0.4 percent of the federal education budget was spent on research, compared to 6.3 percent in health and 12.3 percent in defense. By increasing investments in rigorous research, we can better understand how to use technology to truly transform education, Garg stated.

Not a Silver Bullet for Education

Despite the excitement around education technology, a consistent theme throughout the conference was how technology alone will not serve as a panacea. Rather, it’s best used as a complement to good pedagogy.

“Technology is not a silver bullet, but education is,” said Karen Cator, the CEO of Digital Promise and former Director of the Office of Educational Technology for the U.S. Department of Education. During her keynote address, Cator highlighted the need to produce and use evidence to understand how we can make the most of technology both within and outside of the classroom. She went on to discuss educational equity, technology, and the profound impact of education on social justice and economic development.

Dr. Ken Eastwood, superintendent of the Middletown City School District in New York, shared his personal experience with innovative approaches to improving high-poverty schools in his home district. In his experience, “pedagogy and the art of teaching trumps technology every time,” and emphasizing complementary professional development is key to optimizing technology in the classroom.

Working at the Intersection of Policy, Research, and Philanthropy

Alongside practitioners and researchers, the conference featured philanthropic leaders like Emary Aronson, the interim Chief Program Officer of the Robin Hood Foundation. Aronson spoke as part of a panel focused on improving access to education in the 21st century. “Technology enables access to information, and access to information is a poverty issue,” Aronson said of the Foundation’s role in the education technology space.

Speakers also addressed the challenge of translating research into policy action. Tom Kane (Harvard Graduate School of Education), a leading education scholar, discussed how he aims to keep the research process and results localized and timely in order for evidence to be actionable. Former U.S. Chief Technology Office Aneesh Chopra and former White House advisor, R. David Edelman shared their perspective on how research can impact large-scale federal policies.

Additional speakers from academia and education companies discussed diverse strategies to embed rigorous evaluation in the rollout of new education programs—such as former First Lady Michelle Obama’s Reach Higher Initiative—to better understand how real-world policies affect student outcomes. Building off the lessons from the conference, J-PAL North America looks forward to catalyzing new research and promoting evidence-based policymaking in the education technology space.

Watch videos of each speaker here and check out the conference agenda

What does it take to launch and implement a randomized evaluation?


Randomized evaluations can be challenging to launch and implement in practice. Before designing a randomized evaluation, it’s important to understand the administrative steps needed to get your project off the ground. In the US, there is a complex web of rules and regulations for research. Decisions made in the early stages of an evaluation may have lasting implications on issues like data access and use, and the ability to publish research results. We developed a checklist, Administrative Steps for Launching a Randomized Evaluation in the United States, to lead researchers through the key tasks they must tackle to launch a successful evaluation.

We also created a guide to Real-World Challenges to Randomization and Their Solutions to help researchers and implementing partners develop evaluation designs that fit their program’s context. Using real examples from ongoing and completed randomized evaluations, we describe multiple research designs that accommodate existing programs, mitigate foreseeable implementation challenges, and demonstrate the flexibility of randomized evaluations across contexts.

With these resources, we aim to make it easier for researchers, policymakers, and practitioners to design randomized evaluations and produce rigorous evidence to inform their priority policy questions in the fight against poverty. For more information, please contact Rohit Naimpally at

J-PAL affiliate Michael Greenstone testifies to Congress on the social cost of carbon


In February, J-PAL North America affiliate and co-chair of J-PAL’s Environment & Energy sector Michael Greenstone (University of Chicago and EPIC) testified on the social cost of carbon before the U.S. House Committee on Science, Space and Technology, Subcommittee on Environment, and Subcommittee on Oversight. The social cost of carbon is a measure that helps policymakers calculate the monetary costs and benefits of environmental regulations to more accurately assess the social impacts of federal policies.

In his February testimony, Greenstone noted, “Ultimately, society needs to balance the costs to our economy of mitigating climate change today with climate damages. Wishing that we did not face this tradeoff will not make it go away.” Greenstone co-led development of the social cost of carbon measure during his time as a chief economist for the White House Council of Economic Advisors. Last week the U.S. government withdrew use of the measure, which had been in effect since 2008. Read Greenstone’s testimony and the government decision, and check out randomized evaluations by J-PAL North America affiliates on effective energy conservation approaches.

“Acting Wife”: Marriage Market Incentives and Labor Market Investments

Do single women avoid career-enhancing actions because they could signal personality traits, such as ambition, that are sometimes considered undesirable in the marriage market?

J-PAL affiliate Leonardo Bursztyn (University of Chicago), Thomas Fujiwara (Princeton), and J-PAL affiliate Amanda Pallais (Harvard) took on this question through two field experiments in an elite MBA program in the United States. In the first experiment, newly-admitted MBA students filled out a questionnaire on job preferences and personality traits for the career center to use for first-year summer internship placement. The first-year summer internship is one of the most important steps for post-MBA employment, which meant the questionnaire had relatively high stakes for these students. While all students knew the questionnaires would be seen by the career office, some randomly-selected students also thought their classmates would see their answers. This treatment was subtle: half of the students were told “your” answers would be discussed in their career classes, while half were told that “anonymized” answers would be discussed.

Single women changed their answers decidedly in response to this one altered word. When they were told their classmates would see their answers, single women reported wanting to earn $18,000 lower per year and work four fewer hours per week, and being willing to travel half as frequently. They also reported less professional ambition and tendency for leadership. In contrast, single and non-single women answered similarly when they believed their classmates would not see their responses, and there were only very small differences between women’s and men’s responses when they believed they would be anonymized. In short, the bulk of the gender gap was driven by single women’s expected observability and not by differences in private, which may better reflect their “true” preferences.

To test whether single women’s behavior changed when they believed their choices would be seen by men as opposed to women, the researchers ran a second experiment. In the second experiment, students were asked to choose between hypothetical jobs—career-focused positions (e.g. higher-paying jobs with longer hours and jobs with more travel but better prospects for promotion) versus less career-focused jobs (e.g. lower-paying jobs with shorter hours and jobs with less travel but worse prospects for promotion)—before discussing their choices in their career class small groups. The research team randomly varied the discussion groups by gender composition in which some groups included only women while some included men. When their answers would be shared with male peers—especially single ones—single women were much less likely to select career-focused jobs.

This supplements results from an observational survey where almost three-quarters of single female students reported avoiding activities they thought would help their career because they did not want to appear ambitious, assertive, or pushy, eschewing these activities at higher rates than did men and non-single women. Additionally, unmarried women participated much less in class than married women, despite performing similarly to married women on exams and problem sets.

Bursztyn, Fujiwara, and Pallais’ results help explain gender gaps in labor market outcomes and also highlight the importance of social norms—particularly what is expected from (or preferred in) a husband versus wife—in explaining gender gaps. Even in the 21st century, men prefer women who are less professionally ambitious than they are, researchers have found. Women make important schooling and career decisions while looking for a romantic partner, and their desire to succeed in dating or marriage markets may impact choices that have long-term effects on their careers.

Read the full paper here and listen to coverage of the study on NPR’s Morning Edition.

High-stakes Nudges: Providing social information can affect what job you choose and keep

A single sentence at the end of an email can affect whether or not someone accepts a job—and stays at it for years to come. Lucas Coffman (Harvard), Clayton Featherstone (University of Pennsylvania), and J-PAL affiliate Judd Kessler (University of Pennsylvania) found that providing social information—in this case, the job acceptance rate of the previous year’s admitted applicants—increased the likelihood that an individual would accept a teaching job with Teach for America. Those told about the previous year’s matriculation rate were not only more likely to accept the job, but also to complete training, start the job, and return for a second year. Based on this study’s results, Teach for America has started including a line about their historical matriculation rate in all admissions letters.

Teach for America, a non-profit organization that recruits recent college graduates and professionals to teach in under-resourced public schools, sends thousands of offer letters via email to job applicants each year. To test how the provision of social information would influence an applicant’s likelihood to accept a job, the researchers added one line to the end of randomly selected job offer letters: “Last year, more than 84 percent of admitted applicants made the decision to join the corps, and I sincerely hope you join them.”

Admitted applicants who received offer letters with this sentence were 1.8 percentage points more likely to accept the position. In other words, including this information persuaded 8.4 percent of admitted applicants who would not have joined Teach for America to do so. Furthermore, the researchers found that providing social information increased matriculation by 3 to 5 percentage points among three subgroups of individuals who were likely to be “on the fence” about joining Teach for America. This means that providing social information persuaded 12 to 14 percent of “on-the-fence” admits who would not have joined Teach for America to do so. In addition, teachers who received this information also returned for their second year of teaching at higher rates, which suggests that the social information did not convince those who were likely to drop out at a later point to join the teaching program. 

Why might one simple sentence convince individuals to change their decisions? According to the researchers, information on their peers’ choices may influence admitted applicants’ opinion of the value of the Teach for America experience. For example, the high acceptance rate may signal that the program is particularly effective at improving student outcomes, or that it looks especially good on resumes.

While social information has been found to influence low-stakes decisions (e.g. donating to charity or taking an environmentally friendly action), Coffman, Featherstone, and Kessler show for the first time that social information—such as information about the previous decisions of others—can affect high-stakes behavior and do so persistently. As policymakers seek out cheap, subtle interventions to shape behavior, this study shows that providing social information can be a potentially powerful option.

Read the full paper published in the American Economic Journal here.

Staff Profile: Elisabeth O’Toole


What do you do at J-PAL? I provide research support for the Nurse-Family Partnership study, coordinating between the research team and providers on study implementation and analysis. I have also contributed to the facilitation of several J-PAL conferences, including for the State & Local Innovation Initiative and the Health Care Delivery Initiative. Lastly, I have been assisting editing a research resources document regarding RCT challenges.

What drew you to want to work at J-PAL? I rightfully anticipated that J-PAL would be a perfect fit for me to put my economics background to good use. It is incredible that the research resources created and impact evaluations conducted in our department have the potential to impact domestic and even global policy.

What is your favorite place in the world that you have been? Lake Atitlan, Guatemala – I went on a zip-line tour, and despite being incredibly nervous, the views were incredible as we were flying over coffee fields and rain forests as we looked out on the lake.

If you could have dinner with one person, dead or alive, who would it be? I would love the chance to speak again with the departed Fr. Ted Hesburgh, CSC – a former president of my alma mater whose actions had meaningful global impact.

If you could buy one material thing, and money was not an issue, what would you buy? A Viking stove top/oven to bake delicious cookies.

Read Elisabeth’s bio on the J-PAL North America website.