Work After Prison - One-Year Findings from the Transitional Jobs Reentry Demonstration, Redcross et al, 2010

• Aug. 9, 2016 • Topics: Prisoner Privileges, Release and Reentry

Download original document:
Brief thumbnail

This text is machine-read, and may contain errors. Check the original document to verify accuracy.

Work
after
Prison
One-Year Findings
from the
Transitional Jobs
Reentry Demonstration

Cindy Redcross
Dan Bloom
Erin Jacobs
Michelle Manno
Sara Muller-Ravett
Kristin Seefeldt
Jennifer Yahner
Alford A. Young, Jr.
Janine Zweig
October 2010

Electronic copy available at: http://ssrn.com/abstract=1786932

Work After Prison
One-Year Findings from the
Transitional Jobs Reentry Demonstration

Cindy Redcross
Dan Bloom
Erin Jacobs
Michelle Manno
Sara Muller-Ravett
Kristin Seefeldt
Jennifer Yahner
Alford A. Young, Jr.
Janine Zweig

October 2010

Funding for the Transitional Jobs Reentry Demonstration is provided by The Joyce Foundation.
Additional funding was provided by the JEHT Foundation and the U.S. Department of Labor. The
Urban Institute and the University of Michigan’s Gerald R. Ford School of Public Policy are research partners on the project. The National Transitional Jobs Network is providing technical assistance to the project.
Dissemination of MDRC publications is supported by the following funders that help finance
MDRC’s public policy outreach and expanding efforts to communicate the results and implications of our work to policymakers, practitioners, and others: The Ambrose Monell Foundation,
The Annie E. Casey Foundation, Carnegie Corporation of New York, The Kresge Foundation,
Sandler Foundation, and The Starr Foundation.
In addition, earnings from the MDRC Endowment help sustain our dissemination efforts. Contributors to the MDRC Endowment include Alcoa Foundation, The Ambrose Monell Foundation,
Anheuser-Busch Foundation, Bristol-Myers Squibb Foundation, Charles Stewart Mott Foundation, Ford Foundation, The George Gund Foundation, The Grable Foundation, The Lizabeth and
Frank Newman Charitable Foundation, The New York Times Company Foundation, Jan Nicholson, Paul H. O’Neill Charitable Foundation, John S. Reed, Sandler Foundation, and The Stupski
Family Fund, as well as other individual contributors.
The findings and conclusions in this report do not necessarily represent the official positions or
policies of the funders.

For information about MDRC and copies of our publications, see our Web site: www.mdrc.org.
Copyright © 2010 by MDRC.® All rights reserved.

Overview
More than 2 million people are incarcerated in the United States, and around 700,000 are released from
prison each year. Those who are released face daunting obstacles as they seek to reenter their communities, and rates of recidivism are high. Many experts believe that stable employment is critical to a
successful transition from prison to the community.
The Joyce Foundation’s Transitional Jobs Reentry Demonstration (TJRD), also funded by the JEHT
Foundation and the U.S. Department of Labor, is testing employment programs for former prisoners in
Chicago, Detroit, Milwaukee, and St. Paul, using a rigorous random assignment design. MDRC is
leading the evaluation, along with the Urban Institute and the University of Michigan. The project
focuses on transitional jobs (TJ) programs that provide temporary subsidized jobs, support services, and
job placement help. Transitional jobs are seen as a promising model for former prisoners and for other
disadvantaged groups.
In 2007-2008, more than 1,800 men who had recently been released from prison were assigned, at
random, to a transitional jobs program or to a program providing basic job search (JS) assistance but no
subsidized jobs. Both groups are being followed using state data on employment and recidivism. Random
assignment ensures that if significant differences emerge between the two groups, those differences can be
attributed with confidence to the different types of employment services each group received.
This is the first major report in the TJRD project. It describes how the demonstration was implemented
and assesses how the transitional jobs programs affected employment and recidivism during the first year
after people entered the project, a period when the recession caused unemployment rates to rise substantially in all four cities. Key findings include:


The TJRD project generally operated as intended. The TJ programs developed work slots
and placed a very high percentage of participants into transitional jobs. About 85 percent of the
men who were assigned to the TJ programs worked in a transitional job, reflecting a strong motivation to work. On average, participants worked in the TJs for about four months.



The TJ group was much more likely to work than the JS group early on, but the difference between groups faded as men left the transitional jobs; overall, the TJ group was no
more likely to work in an unsubsidized job than the JS group. The programs provided temporary jobs to many who would not otherwise have worked, but at the end of the first year, only
about one-third of the TJ group — about the same proportion as in the JS group — was employed in the formal labor market.



Overall, the TJ programs had no consistent impacts on recidivism during the first year of
follow-up. About one-third of each group was arrested and a similar number returned to prison.
Most of the prison admissions were for violations of parole rules, not new crimes. In one site, the
transitional jobs group was less likely to be reincarcerated for a parole violation.

These results point to the need to develop and test enhancements to the transitional jobs model and other
strategies to improve outcomes for former prisoners who reenter society. They also raise questions about
the assumed connection between employment and recidivism, since there were no decreases in arrests
even during the period when the TJ group was much more likely to be employed. This is not the final
word on the TJRD project; both groups will be followed up for another year, with two-year results
available in 2011.
iii

Contents
Overview
List of Tables, Figures, and Boxes
Preface
Acknowledgments
Executive Summary

iii
vii
xi

xiii
ES-1

Chapter
1

Introduction
Prisoner Reentry and Employment
Transitional Jobs and Former Prisoners
The Transitional Jobs Reentry Demonstration
Research Design and Study Intake
Timeframe and Data Sources
Characteristics of the Study Participants
Roadmap to the Report

Economic and Criminal Justice Contexts of the Transitional Jobs
Reentry Demonstration Sites
Economic Context
Criminal Justice Context
Baseline Characteristics and Criminal Histories of the Sample, by Site

Program Implementation
Background of the Transitional Jobs and Job Search Organizations
Study Recruitment
Job Search Program Services
Transitional Jobs Program Services
Summary

Program Participation

21
21
22
26
31
32
36
40
41
57
59
59
64
65

Transitional Jobs Group Participation
Job Search Group Participation
Conclusion

1
1
3
5
7
10
13
19

Impacts on Employment and Earnings
Impacts for the Full Sample
Impacts for Sites
Impacts by Study Enrollment Cohort
Impacts for Other Subgroups
Conclusion

67
69
73
83
87
88

Impacts on Recidivism
Impacts for the Full Sample
Impacts by Site
Impacts for Other Subgroups
The Connection Between Employment and Recidivism
Conclusion

Costs of the Programs in the Transitional Jobs Reentry Demonstration 111
Methodology of the Cost Analysis
Operating Costs
Net Cost of the Traditional Jobs Program
Conclusion

111
115
118
121

Qualitative Study of Transitional Jobs Group Members

123
123
126
137
144
149

The Field Work Plan
Core Findings from the First Round of Interviews, 2008
Core Findings from the Second Round of Interviews, 2009
Core Findings from the Third Round of Interviews, 2010
Conclusion

91
91
98
106
107
109

Reflections and Next Steps
An Update on the TJRD Transitional Jobs Programs
Reflections on the Transitional Jobs Reentry Demonstration
What Next?

151
151
152
153

Appendixes
A: Timeline for the Transitional Jobs Reentry Demonstration
B: Supplementary Exhibits for Chapter 2
C: Supplementary Exhibits for Chapter 5
D: Supplementary Exhibits for Chapter 6
E: Supplementary Exhibit for Chapter 7
F: Characteristics of the Qualitative Study Sample
G: Results of Subgroup Analyses

159
163
181
193
201
205
209

References

241

Earlier MDRC Publications on Transitional Jobs

245

List of Tables, Figures, and Boxes
Table
ES.1

Transitional Jobs and Job Search Organizations in the Demonstration Sites

ES.2

One-Year Impacts on Employment and Earnings: Full Sample

ES-11

ES.3

One-Year Impacts on Recidivism: Full Sample

ES-13

1.1

Transitional Jobs and Job Search Organizations in the Demonstration Sites

1.2

Characteristics of Sample Members at Baseline: Full Sample

1.3

Criminal History Characteristics of Sample Members at Baseline: Full Sample

2.1

Crime and Corrections Contextual Statistics: Comparison of TJRD State Averages
and National Averages

2.2

Characteristics of Sample Members at Baseline, by Site

3.1

Overview of the TJRD Study Sites: Transitional Jobs Providers

3.2

Overview of the TJRD Study Sites: Job Search Providers

3.3

Transitional Jobs Program Service Approaches in the Demonstration Sites

3.4

Characteristics of the Transitional Jobs and Job Placement Services in the
Demonstration Sites

Participation in Transitional Jobs and Other Activities for the Transitional Jobs Group,
by Site

4.2

Participation in Program Activities for the Job Search Group, by Site

5.1

One-Year Impacts on Employment and Earnings: Full Sample

5.2

One-Year Impacts on Employment and Earnings, by Site

5.3

One-Year Impacts on Employment and Earnings, by Exposure to Retention
Bonuses (St. Paul)

5.4

One-Year Impacts on Employment and Earnings, by Cohort

6.1

One-Year Impacts on Recidivism: Full Sample

4.1

vii

ES-4

6.2

One-Year Impacts on Recidivism, by Follow-up Time Periods: Full Sample

6.3

One-Year Impacts on Recidivism, by Site

6.4

One-Year Impacts on Recidivism, by Exposure to Retention Bonuses

104

7.1

Estimated Unit and Gross Costs for the Transitional Jobs Group (2008 Dollars)

116

7.2

Estimated Gross Costs for the Transitional Jobs Group, by Site (2008 Dollars)

119

7.3

Estimated Net Costs for the Transitional Jobs Group (2008 Dollars)

120

A.1

Timeline of Transitional Jobs Reentry Demonstration Study

161

B.1

Characteristics of Sample Members at Baseline (Chicago)

165

B.2

Characteristics of Sample Members at Baseline (Detroit)

167

B.3

Characteristics of Sample Members at Baseline (St. Paul)

169

B.4

Characteristics of Sample Members at Baseline (Milwaukee)

171

B.5

Criminal History Characteristics of Sample Members at Baseline (Chicago)

173

B.6

Criminal History Characteristics of Sample Members at Baseline (Detroit)

175

B.7

Criminal History Characteristics of Sample Members at Baseline (St. Paul)

177

B.8

Criminal History Characteristics of Sample Members at Baseline (Milwaukee)

179

C.1

Quarterly Impacts on Employment and Earnings (Chicago)

183

C.2

Quarterly Impacts on Employment and Earnings (Detroit)

185

C.3

Quarterly Impacts on Employment and Earnings (St. Paul)

187

C.4

Quarterly Impacts on Employment and Earnings (Milwaukee)

189

C.5

Characteristics of Sample Members at Baseline, by Exposure to Bonus

191

D.1

Quarterly Impacts on Combined Measures of Employment and Prison: Full Sample

195

D.2

One-Year Impacts on Recidivism (Chicago)

196

D.3

One-Year Impacts on Recidivism (Detroit)

197

D.4

One-Year Impacts on Recidivism (St. Paul)

198

viii

D.5

One-Year Impacts on Recidivism (Milwaukee)

199

D.6

Quarterly Impacts on Combined Measures of Employment and Arrest: Full Sample

200

E.1

Estimated Net Costs for the Transitional Jobs Group Using Alternative Costs
for Each Local Job Search Program, by Site (2008 Dollars)

203

F.1

Characteristics of Sample Members from the Qualitative Study, by Site

207

G.1

One-Year Impacts on Employment and Earnings, by Age: Full Sample

211

G.2

One-Year Impacts on Recidivism, by Age: Full Sample

214

G.3

One-Year Impacts on Employment and Earnings, by Cohort: Full Sample

216

G.4

One-Year Impacts on Recidivism, by Cohort: Full Sample

219

G.5

One-Year Impacts on Employment and Earnings, by Employment History:
Full Sample

221

G.6

One-Year Impacts on Recidivism, by Employment History: Full Sample

224

G.7

One-Year Impact on Employment and Earnings, by Criminal History: Full Sample

226

G.8

One-Year Impacts on Recidivism, by Criminal History: Full Sample

229

G.9

One-Year Impacts on Employment and Earnings, by Exposure to Retention Bonuses:
Full Sample

231

G.10

One-Year Impacts on Recidivism, by Exposure to Retention Bonuses: Full Sample

234

G.11

One-Year Impacts on Employment and Earnings, by Educational Attainment:
Full Sample

236

One-Year Impacts on Recidivism, by Educational Attainment: Full Sample

239

G.12
Figure
ES.1

Quarterly Impacts on Total and Unsubsidized Employment

1.1

Month of Random Assignment and Follow-Up Period for Two Hypothetical Sample
Members in the Impact Analysis

Average Monthly Unemployment Rates in the Study’s Metropolitan Statistical Areas:
Full Sample

2.1

ES-9

5.1

Quarterly Impacts on Employment: Full Sample

6.1

Quarterly Impacts on Employment and Arrests: Full Sample

108

7.1

Simplified Diagram of Major Cost Components in the Transitional Jobs and Job
Search Programs

114

Box
1.1

The Transitional Jobs Reentry Demonstration Sample Compared with a
National Sample of Prisoners

5.1

How to Read the Impact Tables in This Report

5.2

Participation in the St. Paul Transitional Jobs Program, by Bonus Exposure

6.1

Glossary of Recidivism Outcomes

9.1

The Transitional Jobs Reentry Demonstration and the Center for Employment
Opportunities Evaluation: Similarities and Differences

155

Preface
Transitional jobs provide temporary, paid work for individuals who have great difficulty getting and holding
employment, in order to prepare them to find jobs in the regular labor market. Among those individuals are
former prisoners, who, as they reenter society, must persuade employers to hire them despite their criminal
record and lack of recent work experience. But many of these individuals are very eager to work. The
straightforward theory behind transitional jobs is that people are best able to learn to work by working, and
program staff are best able to identify and address potential problems on the job by observing people working.
In addition, by providing an alternative to street life and criminal activity in the form of an immediate source
of legitimate income and the daily structure provided by a steady job during the critical period following release from prison, former prisoners might be less likely to commit crimes or violate the terms of parole.
In 2006, The Joyce Foundation initiated the Transitional Jobs Reentry Demonstration (TJRD) to test these
theories. Participants — former prisoners in four midwestern cities who had been released within the past 90
days — were assigned, at random, to a program that offered temporary jobs, support services, and permanent job placement assistance or to a program providing basic job search assistance but no transitional jobs.
Early results show mixed evidence: The transitional jobs programs dramatically increased employment and
earnings, as they provided work opportunities for many who would not otherwise have found jobs. But former prisoners who had access to transitional jobs were no more likely to find work in the regular labor market and were no less likely to return to prison than those who were offered basic job search assistance without transitional jobs.
On the one hand, neither the TJRD results to date nor the results of earlier evaluations indicate that transitional jobs, at least as currently designed and operated, help people get or hold permanent jobs. On the other
hand, there were very high rates of employment in the transitional jobs, which suggests that many former
prisoners need income support and are willing to work for it. Moreover, these employment results were obtained during the worst recession in 75 years.
New employment initiatives, like the $45 million that the U.S. Department of Labor has allocated for supporting and studying transitional jobs, might take steps to develop stronger models, with an extra focus on
improving long-term employability. Future programs, for instance, could be targeted more carefully to
people who are unlikely to find work in the regular labor market and to those who have already tried job
search services with no success. Or employment retention bonuses, which were offered in two sites in this
study with potentially promising results, could be tested further as a way to help difficult populations remain
employed. Programs must also focus more intently on improving the crucial transition to unsubsidized work.
This is the first major report in the TJRD project, and another year of data collection remains. As a starting
point, the study offers some direction about testing enhanced models of transitional jobs programs for a difficult population. Thus, these early results, while disappointing in some respects, can create a strong foundation for future research on strategies to improve outcomes for former prisoners and other vulnerable groups.
Gordon Berlin
President

Acknowledgments
The success of the Transitional Jobs Reentry Demonstration reflects the contributions
of many people, and the authors extend their appreciation to all those involved. In particular, the
project benefited tremendously from the ongoing support and commitment of Whitney Smith
and Jennifer Phillips of The Joyce Foundation. We thank our project officers, Peggy McGarry
and Bob Crane, formerly of the JEHT Foundation, and David E. Balducchi and Mary Vines
from the U.S. Department of Labor, for their input and guidance.
We are deeply grateful to the current and former program administrators and staff in
each of the four cities where the demonstration occurred, who have given generously of their
time and whose commitment and cooperation were unwavering throughout the project. They
include B. Diane Williams, Jodina Hicks, Kimberly Flennoy, and Judy Davis of the Safer
Foundation; Keith Bennett and John Rykert of Goodwill Industries of Greater Detroit; Tamela
Aikens, David Felton, and Erica Bailey of the Michigan Prisoner ReEntry Initiative; Julie
Kerksick, Tom Back, and Terron Edwards of the New Hope Project; Michael Wirth-Davis, Rob
Hope, Kelly Matter, and Sheila Olson of Goodwill-Easter Seals Minnesota; Claudia Wasserman, Chris Fotsch, Kate Ellefson, and Ann Pollard of the Amherst H. Wilder Foundation;
Wendel Hruska from Project RETURN; Michael Reaume and Bambi Hites of JVS in Detroit;
and Angela Reyes and Jessica Ascencio of the Detroit Hispanic Development Corporation.
The following current and former officials from the Department of Corrections in
each state in the demonstration were integral to the success of the evaluation and helped
guide our understanding of each state’s criminal justice system: Gladyse Taylor and her
predecessor, Deanne Benos, Illinois Department of Corrections; Patricia L. Caruso, Dennis
Schrantz, and Gary W. Stockman, Michigan Department of Corrections; Farris Bell,
Minnesota Department of Corrections; and Anthony Streveler and Mary Kay Kollat,
Wisconsin Department of Corrections.
Many staff from state agencies facilitated our access to the administrative data used in
this study, provided multiple updates of key data files, and patiently answered our questions.
Their names follow.
Illinois: Christine Devitt, Illinois Criminal Justice Information Authority; Steve Karr
and Jennifer Rozhan, Illinois Department of Corrections; Dan Flanagan, Barry Isaacson, and
Laura Chase, Illinois Department of Employment Security.
Michigan: Chad M. Canfield and Nick Romanov, Michigan State Police; Terry Lippincott, Michigan Department of Corrections; Larry McLaren, Michigan Department of Labor and
Economic Growth.

xiii

Minnesota: Katherine Engler, Minnesota Bureau of Criminal Apprehension; Grant
Duwe and Deb Kerschner, Minnesota Department of Corrections; Deb Serum, John Wiersma,
and Pete Jonas, Minnesota Department of Employment and Economic Development.
Wisconsin: Donald Stadler, Wisconsin Department of Corrections; Beatrice A. Dunning
and Gil Peterson, Wisconsin Department of Workforce Development.
From the Urban Institute, Shelli Rossman helped launch random assignment, conducted implementation research, and offered insightful reviews and advice throughout the
study. David Thacher from the University of Michigan conducted implementation research.
Vivian Byrd, Jennifer Cox, Alfred Defreece, Cory Honeyman, and Ariana LaBarrie of the
University of Michigan provided assistance with the qualitative research. In addition, Amy
Rynell and Melissa Young of the National Transitional Jobs Network were partners on the
project from the start and provided thoughtful suggestions on earlier drafts of this report.
We also thank the following people for their contributions to the project: Christy Visher,
Joe Antolin, David Disabato, Abby Coppock, Tom Hurley, Joanne Halfacre, Kathleen
Ludington, Stephanie Johnigan, and Mindy Tarlow.
At MDRC, we thank Charles Michalopoulos, who served as the impact adviser for the
project, offered thoughtful guidance on the analysis, and reviewed several drafts of this report.
David Butler and Margaret Bald reviewed and commented on drafts of the report. John Wallace
played key roles in the early stages of the project. Frieda Molina provided technical assistance
to the sites in the early phases of the project. Johanna Walter provided advice and reviews of the
cost analysis. Vanessa Martin was invaluable in launching random assignment and monitoring
sample recruitment. Gilda Azurdia was integral to the data management and analysis on the
project. Tojuana Riley and Emily Terwelp provided research assistance and coordinated the
production of the report. Julia Gomez provided research assistance for the analysis and the
initial stages of report production. Lauren Cates provided resource management. Alice Tufel
ably edited the report, and Stephanie Cowell and David Sobel prepared it for publication.
Finally, we are especially grateful to the men in the Transitional Jobs Reentry Demonstration, who have allowed us to gain knowledge from their experiences and spoke openly and
candidly to the researchers during the focus group sessions and the qualitative interviews.
The Authors

xiv

Executive Summary
More than two million people are incarcerated in prisons or jails in the United States,
and about 700,000 people are released from state and federal prisons each year. Men and
women who are released from incarceration often face daunting obstacles as they seek to
reenter their communities; many end up returning to prison.
The prisoner reentry issue has attracted growing attention in recent years, as states seek
ways to reduce recidivism as a means to control surging corrections costs and improve public
safety. Many experts believe that stable work is critical to a successful transition from prison to
the community, and most reentry initiatives include services to help former prisoners find
employment. Yet, little is known about what strategies are effective in helping former prisoners
find and hold jobs.
The Joyce Foundation’s Transitional Jobs Reentry Demonstration (TJRD) is designed
to help fill this gap in knowledge. Also funded by the JEHT Foundation and the U.S. Department of Labor,1 the TJRD project is testing employment programs for former prisoners in four
midwestern cities using a rigorous random assignment design,2 generally considered to be the
most reliable method for assessing the impact of social programs. The project focuses in
particular on transitional jobs programs that provide temporary, subsidized jobs; support
services; and job placement assistance. Transitional jobs are widely seen as a promising
employment model, both for former prisoners and for other disadvantaged groups.
This is the first major report in the TJRD project. It describes how the demonstration
was implemented, assesses how the transitional jobs programs affected employment and
recidivism during the first year after individuals entered the project, presents information on
program costs, and presents qualitative findings for a small group of program participants. The
report was prepared by MDRC, which is leading the evaluation, along with the Urban Institute
and the National Poverty Center at the University of Michigan.
As discussed further below:


1
2

The transitional jobs programs dramatically increased employment early in the follow-up period, as the programs provided work opportunities for
many who would not otherwise have found jobs. However, these gains were
driven entirely by the temporary, subsidized program jobs.

The JEHT Foundation ceased operating in January 2009.
The TJRD project began in 2006, with random assignment beginning in 2007. The study will end in 2011.

ES-1



Over the full one-year follow-up period, across all four sites, former
prisoners who had access to transitional jobs were no more likely to find
unsubsidized jobs, and no less likely to return to prison, than those who
were offered basic job search assistance without transitional jobs.

The follow-up period for this project coincided in part with the onset of the national recession, although the recession’s effects on the results are uncertain. These early results are not
the final word on the TJRD project; the research team will continue collecting data for another
year.

Background
After remaining relatively stable for decades, the per capita rate of incarceration in the
United States started rising dramatically in the late 1970s. Between 1975 and 2004, the number
of people incarcerated in state and federal prisons per 100,000 U.S. residents rose from 111 to
484.3 Corrections costs also surged during this period, and now approach $70 billion per year,
with most of that total borne by state and local governments.4
Prisoners returning to the community often have difficulty finding housing and reconnecting with their families and other social supports. Finding steady work can be particularly
daunting, since former prisoners often have low levels of education and skills and no recent
work experience, and because many employers are reluctant to hire people with criminal
records. Moreover, returning prisoners are heavily concentrated in a small number of struggling
urban neighborhoods that lack resources to assist in the reentry process. The most recent
national statistics show that two-thirds of released prisoners are arrested and about half return to
prison within three years.5
Many states have developed multifaceted prisoner reentry initiatives in recent years. At
the federal level, the Serious and Violent Offender Reentry Initiative, the Prisoner Reentry
Initiative, and, most recently, the Second Chance Act of 2008 have supported those efforts.
Many experts believe that stable employment is critical to a successful transition from prison to
the community, and most reentry initiatives include a strong focus on employment services.
And yet, the data on the relationship between crime and employment are mixed, and there is
little evidence about what kinds of program strategies are effective at increasing employment
for former prisoners.

Raphael and Stoll (2007).
Bureau of Justice Statistics (2008).
5
Langan and Levin (2002).
4

ES-2

The Joyce Foundation initiated the Transitional Jobs Reentry Demonstration in 2006 to
learn whether one promising employment strategy — transitional jobs — is effective at
increasing employment and reducing recidivism among former prisoners. Transitional jobs
programs provide temporary paid jobs, support services, and job placement help to hard-toemploy individuals. One of a number of different subsidized employment approaches that have
been implemented or tested in recent decades, the transitional jobs model is based on the
assumption that some people have difficulty finding and holding jobs because they do not
understand how to function in a work environment, and that people are best able to learn to
work by working. The model also assumes that program staff are best able to identify and
address workplace problems — tardiness, difficulty taking direction, and so on — by observing participants on the job, and that employers will be more likely to hire someone who has a
track record of successful employment.
Finally, in a reentry context, transitional jobs provide a source of legitimate employment during the critical period following release from prison, based on the assumption that
former prisoners who have steady jobs, income to meet their basic needs, and the daily structure
of work will be less likely to commit crimes or violate the terms of parole supervision.
Two major random assignment studies have tested transitional jobs programs in recent
years. An evaluation of the New York City-based Center for Employment Opportunities (CEO),
which targets former prisoners, found that the program generated a large but short-lived
increase in employment, driven entirely by the transitional jobs; there were no impacts on
unsubsidized employment. Nevertheless, CEO significantly decreased recidivism — particularly for participants who came to the program shortly after release — and these effects lasted for
at least three years. Another study, an evaluation of the Transitional Work Corporation, tested a
transitional jobs program for long-term welfare recipients in Philadelphia. Like CEO, that
program also generated short-lived employment gains that were mostly driven by transitional
jobs; it also led to reductions in the number of participants receiving welfare payments and the
actual dollar amounts of those payments.

The Transitional Jobs Reentry Demonstration
The TJRD project was designed from the start as a rigorous evaluation. Like the CEO
evaluation, it aims to learn whether transitional jobs programs are more effective than simpler,
less costly programs that provide basic job search and referral services but no subsidized
employment. To accomplish this goal, The Joyce Foundation used a competitive process to
select and fund employment programs for former prisoners in four cities within its midwestern
grantmaking area: Chicago, Illinois; Detroit, Michigan; Milwaukee, Wisconsin; and St. Paul,

ES-3

Minnesota.6 In Detroit, Milwaukee, and St. Paul, two different organizations were identified in
each city, one to run a transitional jobs program and the second to run a job search program; in
Chicago, the same organization provided both types of services. The transitional jobs programs
were selected based on their experience with the model and the target population, their ability
to raise public or private funds to support the program, their linkages with state or local
corrections agencies, and other factors. Table ES.1 lists the transitional jobs and job search
organizations in each city.

The Transitional Jobs Reentry Demonstration

Table ES.1
Transitional Jobs and Job Search Organizations in the Demonstration Sites
Site

Transitional Jobs Program

Job Search Assistance Program

Chicago

Safer Foundation
(through Pivotal Staffing Services)

Safer Foundation

Detroit

Goodwill Industries of
Greater Detroit

JVS and Detroit Hispanic Development
Corporation

Milwaukee

New Hope Project

Project RETURN

St. Paul

Goodwill/Easter Seals Minnesota

Amherst H. Wilder Foundation

SOURCE: MDRC field research.

The transitional jobs programs all shared some basic features, though they were structured somewhat differently. The Detroit and St. Paul programs were operated by Goodwill
Industries affiliates, and participants worked in jobs in existing Goodwill enterprises such as
retail stores or a light manufacturing plant. In Chicago, transitional jobs workers were employed
by a staffing agency established by the Safer Foundation. The staffing agency contracted with a
waste management firm that in turn had contracts with the City of Chicago to operate garbage
recycling plants; almost all program participants worked in those facilities. The New Hope
program in Milwaukee used a scattered site model: participants were placed in positions with
local nonprofit organizations or small businesses but remained employed by New Hope, which
6

Initially a fifth site was selected, but research there was discontinued in 2007.

ES-4

paid their wages. All of the transitional jobs programs offered an array of support services and
helped participants find permanent jobs.
The Milwaukee and St. Paul programs also offered bonus payments to participants who
found and held unsubsidized jobs. These payments were designed to give participants a stronger
incentive to stay employed — even in a low-paying job — and to stay in touch with program
staff. Some studies have found that similar kinds of earnings supplements have led to increases
in employment.7
The demonstration targeted men over 18 years of age who had been released from state
prison within the past 90 days.8 Starting in early 2007, men who met those criteria and volunteered for the study were assigned, at random, to one of two groups:


the transitional jobs group, whose members were referred to the TJRD
transitional jobs program in their city and were offered a transitional job,
support services, and job search and job placement help



the job search group, whose members were referred to the job search program in their city, where they could get help finding employment but were
not offered a transitional job

Just over 1,800 men entered the study before enrollment ended in September 2008.
The research team has collected several kinds of data for all members of the research
sample. Each sample member completed a brief form just before random assignment to provide
information on his demographic characteristics, prior work history, and educational attainment.
The transitional jobs and job search programs provided information on sample members’
participation in program activities, and state agencies in all four states provided administrative
records to measure sample members’ employment and criminal justice involvement. The
employment data show each sample member’s quarterly earnings in jobs covered by unemployment insurance (UI), while the criminal justice records show arrests, convictions, and state
prison stays, both before and after people entered the study.
Because men were assigned to one group or the other through a random process, any
significant differences in employment or recidivism that emerge between the groups during the
7

See, for example, Berlin (2000). Most of the positive results for earnings supplements have been for
low-income single mothers. One study that did not use random assignment found promising results for
“youth who have been or who are at risk of being under juvenile or criminal justice supervision” (Jenks,
MacGilliveray, and Needels, 2006).
8
While the number of female prisoners is growing, a large majority of prison inmates are male. The
project targeted men for research reasons. If the study had been open to both men and women, the number of
female sample members would likely have been too small to analyze separately.

ES-5

follow-up period can be attributed to the different types of employment services they received.
These differences are referred to as impacts or effects of the programs.
The project assessed program impacts for the full sample of about 1,800 men, for each
site, and for subsets of the sample defined by criminal history, age, and prior employment. It
also examined whether impacts varied for sample members who entered the project after the
recession led to a spike in unemployment in all four cities.
In addition to assessing program impacts, the research team studied the implementation
and costs of the transitional jobs and job search programs. Finally, 33 men in the transitional
jobs group participated in a series of in-depth interviews to shed light on their post-prison
experiences.

The Participants
Most of the sample members are African-American men in their thirties. Only about
one-fourth had graduated from high school, but about half had earned a General Educational
Development (GED) certificate. Most had worked in the past, but only about half had worked
six consecutive months for a single employer.
The sample members had extensive histories with the criminal justice system, with an
average of nine prior arrests and a total of six years in state prison. Almost all were under parole
supervision when they entered the study.
The characteristics of the sample members are generally similar from site to site, though
the St. Paul sample stands out in some respects because the programs in that site targeted much
of their recruitment to residents in halfway houses. Men living in those facilities after release
from prison were often classified as “high risk” and were under a particularly intensive form of
parole supervision. As discussed below, this situation has implications for the number of people
who were later returned to prison for violating parole conditions.

Program Implementation


The TJRD project generally operated as intended; the transitional jobs
programs were able to develop work slots and place a very high percentage of sample members into transitional jobs.

The transitional jobs programs worked hard to place men into subsidized jobs quickly
after they were randomly assigned; in three of the four sites, participants were usually at work
within a few days. Thus, despite the instability in participants’ lives, about 85 percent of the
men assigned to the transitional jobs group actually worked in a transitional job. On average,

ES-6

those who were placed in transitional jobs worked for 53 days over a period of about four
months. Data gathered from other transitional jobs programs in the participating cities suggest
that very few men in the job search group worked in transitional jobs.
With few exceptions, the transitional jobs were low-skill positions that were not designed to train participants in particular occupations. Rather, the jobs aimed to teach the “soft
skills” that many employers value — such as showing up to work on time, cooperating with coworkers, and taking direction from supervisors. In the in-depth interviews, some participants
expressed disappointment about the menial nature of the work on transitional job worksites.
The study found that the transitional jobs programs cost about $4,000 per participant, on
average. Nearly half of this amount went directly to participants in wages and other supports.


The transitional jobs programs initially focused on recruiting participants and operating the subsidized jobs; activities designed to move participants into regular jobs took longer to develop.

All of the sites had established linkages with local corrections agencies before the
project began. Nevertheless, during the initial months of the project, most of the sites spent
considerable time and energy working with their criminal justice partners to develop and refine
referral and enrollment procedures for the demonstration; in some sites, these procedures were
revised several times.
Similarly, while all of the programs already operated transitional jobs programs, most
found that changes were needed to accommodate the TJRD population. For example, in St. Paul
and Milwaukee, the programs needed to develop new worksites to accommodate sex offenders,
who are not permitted to work near vulnerable populations.
The sites were generally able to overcome these early challenges, but such issues diverted attention away from the task of establishing strong job development and job placement
activities for TJRD participants. The organizations running the transitional jobs programs
already had experience with these activities, but in some cases, their existing staff had not
worked extensively with former prisoners, a population that presents special challenges.
Although job placement services improved over time in most of the transitional jobs programs,
the research team concluded that these services were generally of average quality.
In three of the sites, the staff working with the job search group had as much, or more,
experience providing job placement services to former prisoners as the staff working in the
transitional jobs programs. The job search programs used a variety of approaches but, at a
minimum, they all helped participants develop a résumé and learn job-seeking skills. Thus, as
expected, the transitional jobs themselves were the key element of the “treatment difference”
between the groups.

ES-7

Program Impacts


The transitional jobs group was much more likely to work than the job
search group early on, as the programs provided temporary jobs to
many men who would not otherwise have worked. The difference between the two groups faded as men left the transitional jobs, however,
and the transitional jobs group was no more likely to work in an unsubsidized job than the job search group.

The top panel of Figure ES.1 shows the percentage of people employed in each quarter
of the one-year follow-up period.9 The dark line shows the employment rate for the transitional
jobs group and the lighter line shows the employment rate for the job search group. In this
figure, a sample member is considered to be employed if he worked either in a TJRD transitional job or in any other job covered by unemployment insurance in a particular quarter. The
asterisks at the bottom of the figure indicate quarters in which the difference between groups is
statistically significant, meaning that it probably did not occur by chance; that is, it was very
likely driven by the differences between the transitional jobs and job search programs.
In the first two quarters of the follow-up period, the transitional jobs group was much
more likely to be employed. For example, in Quarter 2, the employment rate was 75 percent for
the transitional jobs group and 42 percent for the job search group, a difference of 33 percentage
points, which is statistically significant. However, the difference between groups narrowed
quickly as people left the transitional jobs and, by the end of the year, the two groups were
almost equally likely to be employed.
The bottom panel of Figure ES.1 shows the percentage of each group working in an unsubsidized job in each quarter. Unsubsidized jobs are defined as jobs that are covered by
unemployment insurance but are not TJRD transitional jobs. This figure shows that, for the first
three quarters of the follow-up period, the job search group was significantly more likely to be
working in an unsubsidized job. This pattern suggests that some people in the transitional jobs
group could have found a regular job fairly quickly if they had not been given a transitional job.
It also means that the impact on overall employment shown in the top panel was driven entirely
by the transitional jobs. It was sometimes difficult for the programs to determine which potential participants “needed” a transitional job. It appears that some people may have volunteered
for the study in the hope of getting an immediate paid job but, once they were assigned to the
job search group, found a job on their own or with help from the job search programs.

A small percentage of sample members were employed in the quarter before they were randomly assigned. This could have occurred, for example, if an individual was released from prison in February, found a
job right away, but lost it in March, and was randomly assigned in April.

ES-8

The Transitional Jobs Reentry Demonstration

Figure ES.1
Quarterly Impacts on Total and Unsubsidized Employment

Percentage Employed

Total Employment
Transitional Jobs Group
Job Search Group

100
90
80
70
60
50
40
30
20
10
0
Pre-RA

1 ***
2 ***
3 ***
Quarter after Random Assignment (RA)

Unsubsidized Employment
Transitional Jobs Group
Job Search Group

100
Percentage Employed

90
80
70
60
50
40
30
20
10
0
Pre-RA

1 ***

2 ***
3 **
Quarter after Random Assignment (RA)

ES-9

Table ES.2 summarizes the impacts on employment and earnings. The transitional jobs
group was much more likely to work overall during Year 1, but was less likely to work in an
unsubsidized job. Similarly, total earnings were significantly higher for the transitional jobs
group, but earnings from unsubsidized jobs were higher for the job search group.


The employment results were similar across sites and across subgroups
of participants; there is preliminary evidence that the use of employment retention bonuses may have led to stronger results, and that the
transitional jobs programs may have been more effective in weaker labor markets.

The research team examined the employment results separately for each site. While the
site-specific results are less certain owing to the smaller sample sizes, the general pattern is
similar in all sites, with a large but short-lived increase in overall employment. The results were
also similar for subgroups of sample members defined by age, prior employment history, and
criminal history.
The pattern of impacts on unsubsidized employment differed somewhat across sites. In
St. Paul, the job search group had a much higher rate of unsubsidized employment than the
transitional jobs group. This may be because the transitional jobs program in that city placed
men into transitional positions very quickly. Also, the employment rate for the job search
group was unusually high in St. Paul, perhaps because the local economy was relatively
healthy for much of the follow-up period, and because there was a strong, experienced job
search program in place. In contrast, in Detroit, the transitional jobs group did not have a lower
rate of unsubsidized employment than the job search group, probably because relatively few
people in the job search group were able to find jobs in a labor market that was weak even
before the recession began.
Because unemployment rates rose dramatically in all four of the cities beginning in
mid-2008, the analysis also compared results for people who entered the study in 2007 and
early 2008 (the early cohort) with results for those who entered the study after March 2008 (the
late cohort). Since the study followed each person for one year after he entered the study, those
in the late cohort experienced the recession more directly during the study’s follow-up period.
The impacts on overall employment were similar for the two cohorts but, once again, the pattern
of impacts on unsubsidized earnings differed. In fact, in the late cohort, the transitional jobs
group had significantly higher earnings from unsubsidized employment than did the job search
group in the last half of the follow-up period. As with the site-by-site impacts, this pattern may
reflect the fact that the job search group had more difficulty finding employment as the economy weakened; it may also reflect improvements in the operation of the transitional jobs programs over time.

ES-10

The Transitional Jobs Reentry Demonstration

Table ES.2
One-Year Impacts on Employment and Earnings: Full Sample
Outcome

Transitional Job Search Difference
Jobs Group
Group
(Impact)

P-Value

Transitional employment in Year 1a (%)
Unsubsidized employment in Year 1 (%)
Total employment in Year 1 (%)

85.2
50.3
92.8

0.0
59.2
59.5

85.2 ***
-8.9 ***
33.2 ***

0.000
0.000
0.000

Number of quarters employed
Employed in all 4 quarters (%)

2.3
20.2

1.4
11.6

0.9 ***
8.6 ***

0.000
0.000

2,044
2,292
4,336

0
2,917
2,917

2,044 ***
-625 **
1,419 ***

0.000
0.013
0.000

893

881

Transitional job earnings in Year 1a ($)
Unsubsidized earnings in Year 1 ($)
Total earnings in Year 1b ($)
Sample size (total = 1,774)

SOURCES: MDRC calculations using payroll data from each site and unemployment insurance
wage records from each of the states in the demonstration (Illinois, Michigan, Minnesota, and
Wisconsin).
NOTES: Statistical significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * =
10 percent.
Results in this table are regression-adjusted for pre-random assignment characteristics.
aEarnings and employment from transitional jobs are based on payroll data from each of the
sites.
bTotal earnings may not be equal to the sum of transitional job earnings plus unsubsidized job
earnings due to rounding.

Another area where results may be more promising relates to the use of employment retention bonus payments. As noted earlier, both the Milwaukee and St. Paul transitional jobs
programs offered participants bonus payments for getting and holding unsubsidized jobs. The
payments could total up to about $1,500 over six to twelve months.
It is difficult to draw conclusions about the bonus offer in Milwaukee because the bonuses are only one of many factors that distinguished that site from the others. In St. Paul,
however, the bonus offer was implemented part-way through the study period, so it is possible
to compare results for sample members who enrolled early and were not eligible for the bonus
with results for those who enrolled later and were eligible for the payments. It appears that the
employment and earnings results are significantly better for those who entered the transitional
jobs program after the bonus offer was in place. It is not possible to attribute the stronger results
to the bonus offer alone, however, since other aspects of the transitional jobs program and the
local economy also changed over time. Moreover, there is no way to know whether the stronger

ES-11

results could have been obtained with bonuses alone, without transitional jobs. Nevertheless, the
results are promising enough to suggest that further testing would be worthwhile.


There were no consistent impacts on recidivism during the first year of
the follow-up period, though, in one site, the men in the transitional jobs
group were less likely to be reincarcerated for a parole violation.

Table ES.3 shows the impacts on several measures of recidivism: arrests, convictions,
incarceration in state prison, and a combined measure that includes all of these events. There is
only one statistically significant difference between the groups: the transitional jobs group spent
about seven fewer days in prison than the job search group, on average, during Year 1.
Further analysis (not shown) found that the reduction in prison days is concentrated in
the St. Paul site, where the transitional jobs group was significantly less likely than the job
search group to be incarcerated for a parole violation. As noted earlier, most of the sample
members in St. Paul were on an intensive form of parole supervision and the percentage of
sample members who returned to prison for a parole violation overall was much higher in that
site than in the other sites. It may be that parolees were somewhat less likely to violate the terms
of supervision while they were working in transitional jobs. Alternatively, it may be that parole
officers were less likely to send someone back to prison for a relatively minor violation if he
was working when the violation occurred. There were no statistically significant impacts on
recidivism in the other three sites.

Conclusions
The first-year employment results from the Transitional Jobs Reentry Demonstration
are generally consistent with earlier random assignment studies of transitional jobs programs.
They confirm that transitional jobs programs can be operated at a relatively large scale and can
create temporary work opportunities for hard-to-employ populations, resulting in large increases
in employment rates and earnings. However, there is little evidence that working in a transitional job, in and of itself, improves participants’ prospects in the regular labor market. Indeed,
participants who were interviewed for the qualitative study who were working a year or two
after entering the study typically reported that their experience in transitional jobs did not lead to
their unsubsidized job.
The fact that there were reductions in unsubsidized employment in some TJRD sites
suggests that careful targeting is critical; it is not efficient to use scarce resources to provide
subsidized jobs to people who can find unsubsidized jobs with much less intensive assistance.
This issue appears to be less salient when the unemployment rate is very high, since former
prisoners are less likely to be able to find unsubsidized jobs.

ES-12

The Transitional Jobs Reentry Demonstration

Table ES.3
One-Year Impacts on Recidivism: Full Sample
Transitional Job Search
Jobs Group
Group

Outcome

Difference
(Impact)

P-Value

Arrested (%)

38.8

35.3

3.5

0.119

Convicted of a crimea (%)
Convicted of a felony
Convicted of a misdemeanor

13.6
9.0
4.3

14.5
7.9
4.9

-1.0
1.1
-0.7

0.556
0.393
0.489

2.3
4.7
3.8
3.7

2.7
4.9
3.0
4.2

-0.4
-0.2
0.8
-0.6

0.556
0.864
0.356
0.547

Admitted to prison (%)
Admitted to prison for a new crime
Admitted to prison for a parole/probation violation
Admitted to prison for other reason

35.1
6.2
22.4
8.9

36.4
4.8
23.9
9.7

-1.3
1.4
-1.5
-0.8

0.537
0.201
0.407
0.512

Total days incarcerated in prison

40.2

47.5

-7.3 **

0.040

Arrested, convicted, or admitted to prison (%)

53.5

56.1

-2.6

0.250

Sample size (total = 1,808)

909

899

Conviction categoriesb (%)
Convicted of a violent crime
Convicted of a property crime
Convicted of a drug crime
Convicted of a public order crime

SOURCES: Calculations based on data from unemployment insurance records in each demonstration site,
Michigan State Police, Minnesota Bureau of Criminal Apprehension, State of Wisconsin Department of
Justice, Illinois Criminal Justice Information Authority, and the Department of Corrections in each state.
NOTES: Statistical significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10
percent.
Results in this table are regression-adjusted for pre-random assignment characteristics.
Subcategories may sum to more than the total due to multiple arrests, convictions, or prison admissions
per person during the follow-up period.
aEach conviction date is counted only as a single event. If there are multiple convictions on the same
date, only the most serious conviction is recorded in the analysis. Some convictions may have been
associated with an arrest that occurred prior to random assignment. These convictions are counted in the
analysis as occurring after random assignment. Total includes convictions for felony, misdemeanor, and
other crime classes.
bThe categorization of charges is based on definitions from Langan and Levin (2002).

It is interesting that the TJRD transitional jobs programs did not generally reduce recidivism in Year 1, since many people assume that employment and recidivism are causally
related. If this were true, some difference would have been expected between the groups in
arrests during the period when the transitional jobs group had an employment rate that was 40

ES-13

percentage points higher than the rate for the job search group. Of course, this study measured
UI-covered employment only, and it is possible that a more complete measure of employment
— including informal jobs — would show a smaller difference between groups. Moreover, the
recidivism results might have looked different if there had been a large increase in unsubsidized
employment rather than an increase only in temporary minimum-wage jobs. At a minimum,
however, the results suggest that the relationship between employment and recidivism is more
complex than many assume.
In considering future directions, it is important to note that transitional jobs programs
may have multiple goals. The immediate goal of these programs, like many other subsidized
employment programs, is to provide income support to struggling people in a way that is seen
as consistent with American values — that is, through work. The very low employment rates
for the job search group in this study suggest that such support is needed for many former
prisoners, and the high rates of participation among the transitional jobs group show that TJRD
sample members were highly motivated to work. When thinking back, some of the study
participants who were interviewed remembered their time in the transitional jobs program as a
rare period of relative stability in an otherwise chaotic reentry process.
However, transitional jobs programs also have a goal of increasing unsubsidized employment, and here the research record is much more sobering. The findings from this study to
date suggest that the next round of evaluations in this area should test different approaches —
for example, a “beefed up” transitional jobs program with more effective approaches to job
placement and post-placement services, models that mix work and skills training or literacy
programming, or different ways of using subsidies to build links to permanent jobs.

ES-14

Chapter 1

Introduction
This report presents results from the Transitional Jobs Reentry Demonstration (TJRD),
which is testing employment programs for former prisoners in four midwestern cities using a
rigorous random assignment research design.1 The TJRD project was developed by the Joyce
Foundation, with additional funding from the JEHT Foundation and the U.S. Department of
Labor.2 The project focuses in particular on transitional jobs programs that provide temporary
subsidized jobs, support services, and job placement assistance to former prisoners, in an effort
to increase participants’ employment rates and reduce the odds that they will return to prison.
This is the first major report in the study.3 It was prepared by MDRC, which is leading
the TJRD evaluation, along with the Urban Institute and the National Poverty Center at the
University of Michigan. The report describes the implementation and costs of the employment
programs, presents data on how they affected employment and recidivism during the first year
after people entered the study, and provides qualitative findings on a small number of the study
participants. A second year of follow-up is planned, with results available in 2011.
This chapter provides an overview of the TJRD project. It describes the program model,
research design, and study intake process, as well as the data sources used in the analysis and
the characteristics of the study participants.

Prisoner Reentry and Employment
The number of people incarcerated in the United States has more than quadrupled since
the 1970s.4 Today, more than 2 million people are incarcerated in federal and state prisons and

The random assignment approach involves a lottery-like process that places individuals into either a program group (which, in this study, receives a transitional job and job search services) or a control group (which,
in this study, receives basic job search services but not a transitional job). Random assignment ensures that the
demographic characteristics, educational and employment histories, and motivation levels of participants in
both the program and control groups are the same at the start of the study.
2
The JEHT Foundation ceased operating in January 2009.
3
A brief introduction to the project was published in 2009. See Bloom (2009).
4
On a per capita basis, the number of people in prison in the United States remained roughly constant —
about 110 per 100,000 residents — from the 1920s to the 1970s. By 2004, there were 484 prisoners per
100,000 residents (Raphael and Stoll, 2007).

local jails,5 and about 700,000 people are released from state prisons each year.6 Corrections
costs exceed $65 billion per year, with most of this total borne by state and local governments.7
People who are released from prison often face daunting obstacles as they move back to
their communities. They frequently have difficulties finding jobs and housing, and experience
problems reconnecting with family and other social supports. In addition, former prisoners are
concentrated in a relatively small number of urban neighborhoods that lack resources to assist in
the reentry process. Not surprisingly, many end up returning to prison. The most recent national
statistics show that two-thirds of former prisoners are rearrested and half are reincarcerated
within three years of release, most often for violations of parole conditions rather than for new
crime convictions.8 Given these statistics, reducing recidivism is critical both as a means to
reduce corrections costs and improve public safety, but also as a strategy for addressing the
interrelated problems of low-income families and communities.
The prisoner reentry issue has attracted increasing attention in recent years and a number of states have launched multifaceted prisoner reentry initiatives. The federal government
has provided special funding to support these efforts through the Serious and Violent Offender
Reentry Initiative, the Prisoner Reentry Initiative, and, most recently, the Second Chance Act
of 2008.
Many experts believe that stable employment is critical to a successful transition from
prison to the community, and most reentry initiatives include services to help participants find
jobs. In fact, it is very difficult to isolate the impact of employment on recidivism, and the
evidence on this issue is not very clear. Former prisoners who are employed may be less likely
to return to prison, but this does not mean that the relationship is causal. In other words, it may
be that people who avoid recidivism have characteristics that also make them more likely to
work. If that is the case, then it is not necessarily true that increasing employment among former
prisoners will lead to lower recidivism.
What is clear is that many former prisoners have great difficulty finding steady work.
Follow-up studies typically find that half or more of former prisoners are not employed at any
point during the year following release. A large proportion of former prisoners have low levels
of education and work experience or health problems that make them hard to employ, particularly in a changing labor market that offers fewer well-paying opportunities for workers without
a postsecondary education. While it is also difficult to isolate the impact of incarceration on
5

West (2010).
Sabol, West, and Cooper (2009).
7
Bureau of Justice Statistics (2008).
8
Langan and Levin (2002); Petersilia (2003).
6

labor market outcomes, several studies have found that earnings — and possibly employment as
well — are lower for individuals who have spent time in prison than for similar individuals who
have not.9 This is not surprising because individuals fall behind their peers in work experience
while incarcerated. Moreover, in many states, convicted felons are legally barred from specific
occupations. Finally, studies have shown that employers are quite reluctant to hire former
prisoners, particularly African-Americans, those with violent offenses, and those who were
recently released — and that employers are increasingly likely to conduct background checks
before making hiring decisions.10 As discussed further below, all of these barriers are likely to
be compounded when the labor market is weak.
There is very little reliable evidence about which types of employment services, if any,
are effective for former prisoners. Despite a long history of research in the criminal justice field,
including many experimental evaluations, there have been few rigorous studies of employmentfocused reentry models. For example, a meta-analysis of the effects of “community employment programs on recidivism among persons who have previously been arrested, convicted, or
incarcerated” found only eight such studies that used random assignment designs, and several of
those studies did not specifically target former prisoners. The authors note that “this systematic
review…is hampered by inadequate contemporary research.” It also concluded that the programs did not lead to reductions in recidivism.11
In fact, few programs of any kind have produced consistent reductions in recidivism in
rigorous evaluations.12 The most recent large-scale study, the SVORI evaluation, tested multifaceted reentry programs in 16 locations, mostly with a quasi-experimental research design
(random assignment was used in two sites), and found no overall impacts on recidivism.13 One
important exception to this pattern, the Center for Employment Opportunities (CEO) evaluation, is discussed below.

Transitional Jobs and Former Prisoners
Transitional jobs programs provide temporary paid jobs, support services, and job
placement help to hard-to-employ individuals. The transitional jobs model emerged in the

Western, Kling, and Weiman (2001).
Holzer, Raphael, and Stoll (2007); Pager (2007).
11
Visher, Winterfield, and Coggeshall (2005).
12
Some studies have indicated that drug treatment and cognitive behavioral therapy interventions have
reduced recidivism. See Wilson et al. (2006); Mitchell et al. (2007); and Landenberger and Lipsey (2005).
13
Lattimore and Visher (2009).
10

1990s in the welfare system, but its roots stretch back to a number of different subsidized
employment models that have been implemented or tested in the past.14
The transitional jobs model is based on the assumption that many people have difficulty
finding or holding jobs because they do not understand how to function in a work environment.
Further, the model hypothesizes that people are best able to learn to work by working, and that
placing participants into temporary jobs in a program environment will give program staff an
opportunity to identify and address workplace problems such as tardiness or difficulty taking
direction that would likely lead to dismissal in a regular job. Thus, the model is designed to
teach the vital “soft skills” that many employers say they value most. It is also assumed that
employers will be more likely to hire someone who has performed well in a job environment for
several weeks or months.
In recent years, transitional jobs programs have increasingly targeted former prisoners.
In the reentry context, these programs provide a source of legitimate income in the critical
period immediately after release. Thus, if there is indeed a causal relationship between employment and recidivism, transitional jobs programs may lead to immediate reductions in
recidivism, in addition to preparing participants for longer-term success in the labor market.
The first rigorous test of a transitional, subsidized employment model for former prisoners was the National Supported Work Demonstration, which ran from 1975 to 1980. The
Supported Work programs offered 12 to 18 months of highly structured paid work experience.
Participants worked in crews to promote peer group support, and the model emphasized
“graduated stress”; that is, expectations at the work site were supposed to increase over time
until they approximated the expectations in a regular, unsubsidized job.15 Almost all the programs helped participants find regular jobs, though the intensity and quality of this assistance
varied. Some of the Supported Work programs were “social enterprises” that sold products or
services to partly offset the costs of running the program.
Supported Work was tested for four hard-to-employ groups, including individuals who
had been incarcerated in the prior six months. The sample included about 2,300 former prisoners (94 percent male) across seven sites. The program initially generated large increases in
employment and earnings during the program period (the initial period, when many participants
were working in subsidized jobs), but there was little evidence of longer-term impacts on
employment or earnings. Overall, there were no impacts on recidivism except for some reduc14

See Bloom (2010).
Throughout this report, the terms “unsubsidized jobs” and “regular jobs” are used interchangeably to
represent employment in the regular labor market covered by unemployment insurance, as opposed to
“subsidized jobs,” for which wages are subsidized by an employment program.
15

tions among participants who were older than 26 years of age, perhaps because they had
reached a point in their lives when they were determined to avoid further incarceration and the
jobs program helped them further this goal.16
In 2004, MDRC initiated a random assignment evaluation of the New York City-based
CEO, one of the nation’s largest and mostly highly regarded employment programs for former
prisoners, as part of the U.S. Department of Health and Human Services’ Hard-to-Employ
project (also funded by the U.S. Department of Labor).17 CEO serves about 2,000 parolees each
year. Participants begin with a four-day pre-employment class and then are placed into a CEO
work crew. The crews, supervised by CEO staff, do maintenance and repair work under
contract to city and state agencies. Participants work four days per week and are paid the
minimum wage; they are paid daily at the work site. The fifth day is spent in the program office,
where participants meet with job coaches and job developers and participate in supplemental
activities, such as attending a fatherhood group meeting.
The evaluation, which includes almost 1,000 parolees, compared CEO’s core model
with a simpler, less costly model that provided job search assistance (but no subsidized jobs).
As in Supported Work, there were large but short-lived increases in employment rates driven
entirely by the subsidized jobs; there were no increases in unsubsidized employment. Interestingly, however, there were reductions in several measures of recidivism that persisted
throughout the three-year follow-up period, well after the employment gains had disappeared;
this pattern raises questions about the mechanism that produced the impacts and about the
connection between employment and recidivism. The impacts on recidivism were concentrated
among sample members who came to the program within 90 days of release from prison.
Since the CEO evaluation is similar in many ways to the TJRD study, it provides a key
point of reference for the TJRD findings. TJRD also offers an opportunity to examine whether
CEO’s effects can be replicated in other transitional jobs models and in other locations.

The Transitional Jobs Reentry Demonstration
The Transitional Jobs Reentry Demonstration was designed to determine whether transitional jobs are effective at increasing employment and reducing recidivism among former
prisoners. Like the CEO evaluation, it compares transitional jobs programs with less expensive
job search programs, which offered basic job search and placement assistance but did not
provide transitional jobs.
16
17

Uggen (2000).
Bloom, Redcross, Zweig, and Azurdia (2007); Redcross et al. (2009).

In mid-2006, the Joyce Foundation conducted a competition and selected employment
programs in four cities to participate in the project — Chicago, Detroit, Milwaukee, and St.
Paul.18 Each site received about $600,000 over three years, and the grantees were also expected
to leverage existing funds or raise funds from state or local agencies to support their programs.19
Each state’s Department of Corrections was an active partner in the project and provided
funding to support the employment programs.
In each of the four project sites, former prisoners were assigned at random to a transitional jobs program or to a job search program. Table 1.1 shows the organizations that operated
the transitional jobs and job search programs in each city. In Detroit, Milwaukee, and St. Paul,
separate organizations provided the two types of services, while in Chicago, the same organization provided both.
The TJRD transitional jobs programs were all structured somewhat differently, but
there were some basic similarities. All provided participants with temporary, minimum-wage
jobs that offered 30 to 40 hours of paid work each week. The transitional jobs that were provided to participants were not focused on building skills in any particular occupation, but all
aimed to identify and address behavior or performance issues that emerged at the worksites. All
provided a range of ancillary services and supports to participants, and all helped participants
look for unsubsidized jobs to follow the transitional jobs, often with the help of job developers
who approached employers to identify job openings for participants. The Milwaukee and St.
Paul transitional jobs programs also offered participants retention bonus payments for getting
and holding unsubsidized jobs. The payments could total up to $1,500 or so over six to nine
months.
The Detroit and St. Paul programs were operated by affiliates of Goodwill Industries,
and participants worked in jobs in existing Goodwill enterprises such as retail stores or a light
manufacturing plant. In Chicago, transitional jobs workers were employed by a staffing agency
established by the Safer Foundation. The staffing agency contracted with a major waste management firm that in turn had contracts with the City of Chicago to operate garbage recycling
plants; almost all program participants worked in those plants. The New Hope program in
Milwaukee used a scattered site model; that is, participants were placed in positions with
various local nonprofit organizations or small businesses, but they remained employed by New
Hope, which paid their wages.
18

A fifth site was selected for the study, but withdrew from the project early on because of competing organizational priorities. Results for that site are not included in the analysis.
19
Most of the funds went to the transitional jobs programs, but the job search programs shared the funding
in some sites.

The Transitional Jobs Reentry Demonstration

Table 1.1
Transitional Jobs and Job Search Organizations in the Demonstration Sites
Site

Transitional Jobs Program

Job Search Assistance Program

Chicago

Safer Foundation
(through Pivotal Staffing Services)

Safer Foundation

Detroit

Goodwill Industries of
Greater Detroit

JVS and Detroit Hispanic Development
Corporation

Milwaukee

New Hope Project

Project RETURN

St. Paul

Goodwill/Easter Seals Minnesota

Amherst H. Wilder Foundation

SOURCE: MDRC field research.

The job search programs also used a variety of approaches but, at a minimum, all of
them helped participants prepare a résumé, learn how to fill out job applications and interview
for jobs (including how to answer questions about their convictions), and identify job leads.
Some of the job search programs also provided referrals to housing and other assistance
programs.
The TJRD transitional jobs were designed based on the assumption that by increasing
employment levels (both subsidized and unsubsidized), the programs would lead to lower
recidivism. In addition to testing various transitional jobs models, the TJRD project also
provides an opportunity to learn about the operations and impacts of transitional jobs and job
search programs in a range of environments. There are important differences across the four
cities, for example, in labor market conditions, population characteristics, and criminal justice
practices, which are discussed in Chapter 2.

Research Design and Study Intake
The study includes four main components: an implementation analysis that examines how the transitional jobs and job search programs operated, an impact analysis that

assesses how the transitional jobs programs affected employment and recidivism, a cost
analysis, and a series of in-depth interviews with study participants to gain a better understanding of their experiences after leaving prison.20
The TJRD project targeted men who were age 18 or older and who had been released
from state prison within 90 days prior to enrollment in the study (a target that was based on
early lessons from the CEO study findings, which suggested that the program is more effective
for people who were recently released from prison).21 Men with all types of criminal histories
were accepted into the project, with no project-wide restrictions based on the number or type
of previous offenses (though there were some limitations in individual sites). Eligible participants had to be willing to work full time and could not have worked in a transitional job within
the past year.
The sites recruited men into the study from January 2007 through September 2008.
Slightly more than 1,800 men entered the study in all, with the site totals ranging from about
375 to 500 each.22 Former prisoners who were eligible and who agreed to be in the study were
assigned at random to one of two groups:


The transitional jobs (TJ) group. The 912 individuals who were randomly
assigned to this group were referred to the TJRD transitional jobs program in
their city and were offered a transitional job and other support services such
as pre-employment classes, job coaching, job search assistance, job placement, and post-placement services.



The job search (JS) group. The 901 individuals who were randomly assigned to this group were referred to the job search program in their city and
were offered basic job search and placement assistance but were not offered
a transitional job.

By tracking the two groups over time, the evaluation is able to assess whether the transitional jobs programs lead to employment and recidivism outcomes that differ from the
outcomes for the job search assistance programs. Any differences that emerge between the two
groups are considered “impacts” or “effects” of the transitional jobs programs because, owing to
the random assignment research design, the research groups were comparable on measured and
unmeasured characteristics when the study began.

See Appendix Table A.1, which provides a brief chronicle of the study timeline.
To read more about CEO’s findings, see Redcross et al. (2009).
22
Total sample size by site is as follows: Chicago, 374; Detroit, 426; Milwaukee, 507; St. Paul, 506.
21

The research team worked with staff in the demonstration sites to develop procedures
for enrolling individuals into the study. Each site hired or identified a staff member to serve as
the research coordinator; that individual’s salary was subsidized by the project. The research
coordinator served as the primary liaison to the research team and was responsible for overseeing the study enrollment and random assignment process.
Although the enrollment process differed somewhat from site to site, the general steps
were the same in all sites:


Recruitment. Men were recruited for the study while they were in prison or
shortly after they were released through information sessions, flyers, and other methods. Many men were referred to the project by their parole officer. In
each site, staff from the transitional jobs and job search programs forged
links with corrections and parole agencies to facilitate recruitment.



Informed consent. Men who were interested in the project attended an intake session at the job search program, the transitional jobs program, or some
“neutral” location. (Initially, the Detroit site enrolled people into the study
the day before they were released, so the sessions took place in prisons.) At
the session, the men received a brief, general description of the transitional
jobs and job search programs and watched a five-minute video describing the
project. The video stressed that the TJRD project was only open to people
who agreed to participate in the evaluation, that a random process would be
used to determine which type of employment service each person would receive, and that all data collected for the study would be kept private. Men
who wanted to participate in the project then signed a consent form.



Baseline data. Participants completed a brief questionnaire — called a Baseline Information Form (BIF) — to provide information about their demographic characteristics and prior employment and education history.



Random assignment. To conduct random assignment, the research coordinator logged onto a secure MDRC Web site and entered a few pieces of identifying information for each study participant. Once the data were submitted
for random assignment, a research group assignment (transitional jobs or job
search) was displayed immediately. Participants were notified of their research assignment and told where and when to report to their assigned program. In some sites, study participants were already in the office of one of
the programs when random assignment took place; in some others, they were
transported directly to their assigned program.

Timeframe and Data Sources
Data on each sample member were collected for one year after the date he was randomly assigned. Thus, the exact dates of the one-year follow-up period are different for each sample
member and include data for the quarter of random assignment plus an additional three quarters
of follow-up.
Figure 1.1 illustrates the random assignment and follow-up periods for the impact
analysis for two hypothetical sample members.23 As shown in the figure, Sample Member A
was randomly assigned in May 2007, which means that his follow-up period ended in March
2008 (Quarter 4 for Sample Member A). Sample Member B was randomly assigned in
September 2008, which means that his follow-up period ended in June 2009 (Quarter 4 for
Sample Member B).
The evaluation is using a variety of data sources:



Baseline data. As noted earlier, participants completed a short baseline questionnaire when they entered the study. These data provide a snapshot of the
characteristics of the study participants just before they were randomly assigned.



Program participation data. These data, collected from the management
information systems (MIS) of both the transitional jobs and job search programs, provide information on each individual’s participation in the program
for which he was eligible. The research team also obtained detailed payroll
records to measure employment and earnings in the transitional jobs.



Implementation research. The research team conducted two rounds of site
visits — one in summer 2007 and another in spring 2008 — to the transitional jobs and job search programs in each city. The site visits were devoted to collecting qualitative information on program operations and the
local criminal justice context. Financial expenditure reports, which were
used in the cost analysis, were also collected during these visits. Each site
visit lasted between two and three days and included interviews with program staff, criminal justice agency staff, observation of program activities,
and visits to transitional job worksites. Case file reviews and focus groups
were also conducted to get a better understanding of the participants and the
services they received.

The figure covers employment measures only, for which data are available quarterly.

A M

2007

A M

J
2008

M A M

Follow-up period, Sample Member B

J
2009

Sample Member B randomly assigned September
2008 and followed up through June 2009

NOTE: This figure covers employment measures only, for which data are available quarterly. Thus, while random assignment occurs in a particular month,
follow-up employment data are collected quarterly. The month in which each sample member is randomly assigned determines his first quarter of follow-up,
whether it is the first, second, or third month in that quarter.

Follow-up period, Sample Member A

Sample Member A randomly assigned May
2007 and followed up through March 2008

Month of Random Assignment and Follow-Up Period for Two Hypothetical Sample Members in the Impact Analysis

Figure 1.1

The Transitional Jobs Reentry Demonstration



In-depth interviews. A series of qualitative in-depth interviews were conducted with 33 men who were participating in the transitional jobs programs
in all four sites. Between 2008 and 2010, three waves of on-site interviews,
intermittent phone interviews, and visits to participants’ worksites and residential communities were undertaken. The objective of the interviews was to
acquire an in-depth understanding of the participants’ transition experience.24



Employment data. Data from the state unemployment insurance (UI) agency was collected in each of the four states in the study and are used to measure quarterly employment in UI-covered jobs for each sample member during the follow-up period. UI data do not cover all jobs; for example,
employment that is “off the books” and federal government jobs are not included.



State criminal justice administrative data. The analysis uses arrest and
conviction data from the State Police or a criminal history data depository in
each of the states. The Department of Corrections in each of the states provided prison admissions and incarceration data. Taken together, these data
provided information on a range of outcomes, including arrests, convictions,
parole violations, and incarceration in state prison for each member of the
study sample, both before and after study entry.

The data have some limitations that should be recognized. First, the TJRD project does
not include a client survey, so outcomes that were not available on administrative records data
files are not included in the analysis. Several examples of outcomes that could not be measured
without a survey include employment that is not covered by UI data, information about job
characteristics, housing status, income, substance abuse, child and family engagement, criminal activities that did not lead to an arrest or parole revocation, and participation in education
and employment activities outside the TJRD programs. Second, data were not collected from
local jail facilities; therefore, incarceration in local jails, which may be quite common, was also
not measured.25

Dr. Alford Young, from the University of Michigan, conducted the interviews. One research assistant
worked with him on each site visit.
25
Jails are most often run by local governments and are designed to hold individuals who are awaiting trial
or serving short sentences. Prisons are operated by state governments and the Federal Bureau of Prisons; they
are designed to hold individuals who are convicted of crimes and are usually serving longer-term sentences.

Characteristics of the Study Participants
Table 1.2 presents background characteristics of the research sample. These characteristics are based on information collected from the Baseline Information Form that all study
participants completed just before random assignment. As expected in a random assignment
design, there are very few significant differences in background characteristics between the two
research groups.
As the table shows, almost everyone in the sample (96 percent) reported being released
from state prison, and a similar percentage (97 percent) reported that they were under parole
supervision when they enrolled in the study. This is not surprising, since parole agencies were a
key source of referrals to the project.
The average age of study participants at baseline was 35 years, which is somewhat older than a national sample of former prisoners.26 Most of the TJRD sample (81 percent) is
black/African-American. Nationally, fewer than half of released prisoners are AfricanAmerican, but it is not surprising that the racial composition of the TJRD sample does not
mirror national statistics because the TJRD sites were in urban areas, where a concentrated
number of non-white former prisoners return home. (Box 1.1 summarizes some additional
similarities and differences between the TJRD sample and a nationwide sample of prisoners.)
About half of the study sample (52 percent) had at least one child under age 19, although only 13 percent lived with any of their children.27 Among those with children, 45
percent reported having a formal child support order.
The process of obtaining affordable housing upon release from prison is complicated
because most returning prisoners do not have jobs and are not eligible for many forms of public
assistance. In addition, public housing authorities can deny housing to individuals with a history
of drug use or criminal behavior. Thus, many sample members were living with friends or
relatives (48 percent) or in some type of transitional housing (30 percent); 5 percent reported
living in a shelter or were homeless.

Langan and Levin (2002). The national sample of former prisoners refers to the Bureau of Justice Statistics national study on the rearrest, reconviction, and reincarceration of former prisoners who were tracked for
three years after their release from prisons in 15 states in 1994. The former inmates represented two-thirds of
all prisoners released in the United States that year. The report includes prisoner demographic characteristics
(gender, race, ethnicity, and age), criminal record, types of offenses for which they were imprisoned, the effects
of length of stay in prison on likelihood of rearrest, and comparisons with a study of prisoners released in 1983.
27
There is a small but statistically significant difference between the two groups in the percentage of sample members living with their children. The impact calculations presented in Chapters 5 and 6 are adjusted to
account for this difference.

The Transitional Jobs Reentry Demonstration

Table 1.2
Characteristics of Sample Members at Baseline: Full Sample
Transitional Job Search
Jobs Group
Group

Characteristic

Full
Sample

Age (%)
18-24
25-34
35-44
45 or older

17.2
34.9
28.4
19.5

16.9
36.4
28.0
18.8

17.0
35.6
28.2
19.1

Average age (years)

34.8

34.5

34.6

Race/ethnicity (%)
White/non-Hispanic
Black/non-Hispanic
Hispanic
Other

9.8
82.3
4.0
3.8

10.6
79.0
5.4
4.9

10.2
80.7
4.7
4.3

Has any children under age 19 (%)

51.5

52.7

52.1

Lives with own children under age 19 (%)

14.0

11.7

12.9

Among those with children
Lives with own children under age 19 (%)
Average number of children
Has a child support ordera (%)

26.8
2.1
44.3

21.5
2.2
46.1

24.1
2.1
45.2

Education (%)
GED certificate
High school diploma
Associate's degree/2- or 4-year college/beyond
None of the above

48.6
21.3
5.4
24.7

43.8
24.5
6.1
25.7

46.2
22.9
5.8
25.1

Has occupational training certificate (%)

47.7

44.4

46.0

Marital status (%)
Never married
Married, living with spouse
Married, living away from spouse
Legally separated/divorced/widowed

75.5
3.3
4.7
16.5

78.2
2.9
4.0
14.9

76.8
3.1
4.4
15.7

Housing status (%)
Owns/rents house or apartment
Lives with friends or relatives
Has transitional housing
Lives in shelter/other

17.0
48.8
30.1
4.1

17.4
46.4
30.3
5.9

17.2
47.6
30.2
5.0

Ever employed (%)

86.0

86.9

86.5

Sig.

(continued)

Table 1.2 (continued)
Transitional Job Search
Jobs Group
Group

Characteristic

Full
Sample

Sig.
***

Ever employed 6 consecutive months for one employer (%)

49.5

55.7

52.6

Worked full time in the 6 months before incarceration (%)

42.7

46.5

44.6

Ever in drug or alcohol treatment (%)

56.4

54.8

55.6

Public assistanceb (%)
No government assistance
Temporary Assistance for Needy Families
Food stamps
Social Security
Medicaid
Housing assistance

53.1
9.7
33.4
1.5
11.2
5.3

54.8
8.2
32.3
1.5
12.2
6.7

53.9
9.0
32.9
1.5
11.7
6.0

Family assists with (%)
Finding a place to live
Finding a job
Dealing with a drug or alcohol problem
Financial support
None of the above

37.5
37.0
15.5
37.8
37.2

38.2
36.0
15.1
37.6
37.6

37.8
36.5
15.3
37.7
37.4

Facility type of most recent release (%)
Prison
Jail

96.4
3.6

95.5
4.5

96.0
4.0

Under community supervision (parole/probation) (%)

96.8

97.9

97.4

Sample size (total = 1,813)
SOURCE: MDRC calculations using the Baseline Information Form.

912

901

NOTES: In order to assess differences in characteristics across research groups, chi-square tests were used
for categorical variables, and t-tests were used for continuous variables. Statistical significance levels are
indicated as follows: *** = 1 percent; ** = 5 percent; * = 10 percent.
aThis measure is calculated only for those who have children and were ordered to pay child support.
Some participants were ordered to pay but did not report the amount; they are missing from this measure.
bFor Michigan, this variable is created only for those who were randomly assigned after January 8,
2008, when the site began random assignment after release.

Other issues affect former prisoners’ ability to reenter the community after incarceration, including the support they receive from their families and the extent to which they battle
substance abuse. In the TJRD sample, more than one-third of the participants reported that their
families helped or supported them in some way. More than half had received drug or alcohol
treatment at some point in their lives, prior to entering the study.

Box 1.1

The Transitional Jobs Reentry Demonstration Sample
Compared with a National Sample of Prisoners
The study sample in the Transitional Jobs Reentry Demonstration (TJRD) was similar to a
national sample of former prisoners (released in 1994) in some respects and different from
that national sample in other respects. The TJRD study was targeted to males, while the
national sample, though overwhelmingly male, included a small percentage of females.
The samples were similar in terms of criminal history: both the TJRD sample and the national sample had an average of nine prior arrests and four prior convictions.
Differences between the two samples show up in race, with most of the TJRD sample
black and the national sample evenly divided between black and white, and ethnicity, with
less than 5 percent of the TJRD sample Hispanic compared with 25 percent of the national
sample. The TJRD sample is somewhat older than the national sample, with nearly half of
the TJRD sample being 35 or older compared with about one third of the national sample.
Differences are also seen in the crime category of the most recent conviction; for example,
a greater proportion of the TJRD sample was convicted of a violent crime, and a greater
proportion of the national sample was convicted of a drug crime. Finally, length of the
most recent incarceration differed, with the average time served being 33 months for the
TJRD sample and 20 months for the national sample.
________________________________
SOURCE: Langan and Levin (2002) and MDRC calculations using the Baseline Information Form
and data from Michigan State Police, Minnesota Bureau of Criminal Apprehension, State of Wisconsin Department of Justice, Illinois Criminal Justice Information Authority, and the Department of
Corrections in each state.

Only about one in four participants had a high school diploma, but nearly half had a
General Educational Development (GED) certificate at baseline; it seems likely that some of the
men earned a GED certificate while incarcerated. Most had at least some employment history.
Approximately 87 percent of the sample reported that they had ever worked, but only
about half reported that they had worked for a single employer for six consecutive months at
any point in their life.28
28

There is a small but statistically significant difference between the two groups in the percentage of sample members who worked for six consecutive months at some point in their lifetime. The impact calculations
presented in Chapters 5 and 6 are adjusted to account for this difference.

Table 1.3 presents the criminal histories of study participants before random assignment.29 These data are from the criminal justice agencies in each of the states. Again, there are
few differences between the two research groups. As expected, all the study participants have a
history of arrest, conviction, and incarceration. The average lifetime total time spent in state
prison for study participants was about 71 months, or approximately six years, and the average
length of sample members’ most recent incarceration was 33 months.

The Transitional Jobs Reentry Demonstration

Table 1.3
Criminal History Characteristics of Sample Members at Baseline: Full Sample
Transitional
Jobs Group

Job Search
Group

Full
Sample

99.3

99.0

99.2

9.1

9.3

9.2

98.1

97.4

97.8

4.0
2.6
1.0

4.1
2.7
0.9

4.1
2.6
1.0

Ever convicted of a violent crimec (%)

57.1

58.6

57.8

Ever convicted of a drug crimec (%)

46.4

47.2

46.8

Ever convicted of sexual assault (%)

10.9

10.5

10.7

73.6

69.2

71.4

Characteristic

Sig.

Arrest history
Any prior arrests (%)
Average number of prior arrestsa
Conviction history
Any prior conviction (%)
b

Average number of prior convictions
Number of prior felony convictions
Number of prior misdemeanor convictions

State prison history
Lifetime number of months in state prison

(continued)

These data include incarceration in state prison but not local jails. All criminal history calculations include only events that occurred between January 1, 1980, and the date of random assignment.

Table 1.3 (continued)
Transitional
Jobs Group

Job Search
Group

Full
Sample

Crime category of most recent convictionb (%)
Violent crime
Drug crime
Property crime
Public order crime

33.4
24.4
22.5
19.7

36.1
25.9
21.0
17.0

34.8
25.1
21.7
18.3

Reason for most recent prison admission (%)
New crime
Parole/probation violation
Other

60.4
36.5
3.1

61.1
34.3
4.6

60.8
35.4
3.8

Length of most recent incarceration (months)

33.3

33.1

33.2

Days between latest state prison release
and random assignment

47.8

39.7

43.8

Sample size

898

906

1,804

Characteristic

Sig.

Most recent criminal justice events

SOURCES: MDRC calculations based on data from Michigan State Police, Minnesota Bureau of
Criminal Apprehension, State of Wisconsin Department of Justice, Illinois Criminal Justice Information
Authority, and the Department of Corrections in each state.
NOTES: All criminal history calculations include only events that occurred between January 1, 1980,
and the date of random assignment.
There are no significant differences between the research groups on any measures in this table.
Subcategories may sum to more than the total due to multiple arrests, convictions, or incarcerations
per person during the follow-up period.
aEach arrest date is counted only as a single event. If there are multiple crimes or charges on the same
date, only the most serious charge is recorded in the analysis.
bEach conviction date is counted only as a single event. If there are multiple convictions on the same
date, only the most serious conviction is recorded in the analysis. The total includes convictions for
felony, misdemeanor, and other crime classes.
cThe categorization of charges is based on definitions from Langan and Levin (2002).

Before random assignment, study participants had been arrested an average of nine
times — similar to a 1994 national cohort of former prisoners, as observed in Box 1.1.30 Before
random assignment, study participants had been convicted an average of four times, with about
three of those events being felony convictions and one a misdemeanor. Slightly over half of the
sample (58 percent) had been convicted of a violent crime at some point in their lifetime, and 47
percent of the sample had at some point in their lifetime been convicted of a drug crime.
30

Langan and Levin (2002).

Table 1.3 also shows that, on average, sample members enrolled in the study 44 days
after release; as noted earlier, the study targeted people who had been released fewer than 90
days earlier.

Roadmap to the Report
The remainder of the report is divided into the following chapters:
Chapter 2 presents the criminal justice context and the economic environment in each
of the study sites during the follow-up period. The chapter also describes the baseline characteristics and criminal histories of the sample by site, highlighting key differences.
Chapter 3 describes the implementation of the transitional jobs and job search programs, and Chapter 4 presents data on the rates of participation in TJRD program services for
both research groups.
Chapters 5 and 6 cover the one-year impacts of the transitional jobs programs on employment and earnings and on recidivism. Chapter 7 presents data on program costs. Chapter 8
describes findings from the in-depth interviews.
Chapter 9 reviews the status of the TJRD transitional jobs programs approximately one
year after the operational phase of the demonstration ended, and presents some of the lessons
the transitional jobs programs learned from TJRD. The chapter concludes by discussing the
implications of the first-year TJRD results for future policy and research.

Chapter 2

Economic and Criminal Justice Contexts of the
Transitional Jobs Reentry Demonstration Sites
The Transitional Jobs Reentry Demonstration (TJRD) focuses on a sample of men who
were released from state prisons and who returned to Chicago, Detroit, Milwaukee, and St. Paul
during portions of 2007 and 2008. This chapter describes the economic and criminal justice
contexts in the four sites during the demonstration period.

Economic Context
One of the key goals of the TJRD programs was to increase the employment and earnings of participants. This section provides information about economic conditions during the
timeframe of the project and the ways in which these conditions should be considered in the
context of the study.
Both the overall state of the economy and the variation in economic conditions across
the four cities are considerations for interpreting the findings of this study — although, in an
experimental framework, external factors, such as the economy, are the same for the transitional
jobs group and the job search group. Thus, while local economic conditions may affect the
levels of employment-related outcomes that are observed in the evaluation, it is unclear how or
whether economic conditions will influence the impacts of the TJRD programs. Prior studies on
the subject have found mixed results about whether the impacts of employment programs and
policies are affected by economic conditions. Some prior studies that looked at the interaction
between public policy and unemployment found that certain types of policies are somewhat
more effective at increasing employment when the economy is in a downturn, while the effects
of other types of policies are not sensitive to economic conditions.1
All of the TJRD programs enrolled participants between January 2007 and September
2008 and continued to serve study participants into 2009. Employment data for the current
report extend through June of 2009 for those who were randomly assigned during the last part
of the study enrollment period. During that timeframe, the national recession led to sharp
increases in unemployment in all four cities. In addition, the four study cities vary somewhat in
their local economic and employment conditions. For example, economic conditions in Detroit
have been depressed for some time, with very high levels of unemployment and job loss even
1

Herbst (2008); Greenberg et al. (2001); Michalopoulos (2005).

prior to the national downturn. St. Paul and Milwaukee, on the other hand, had enjoyed somewhat better local economic conditions prior to the national downturn, characterized by relatively
low levels of unemployment.
Figure 2.1 shows the unemployment rates in each of the metropolitan statistical areas
(MSAs) for the four cities during the study follow-up period.2 As the figure shows, unemployment rates were relatively steady throughout 2007 and the first half of 2008. But beginning in
mid-2008, all four cities experienced an increase in unemployment rates, followed by a very
steep jump in unemployment during the first half of 2009. The rates of unemployment differed
somewhat across MSAs, with the highest rates in Detroit and the lowest in St. Paul and Milwaukee. Nonetheless, unemployment in all of the cities was at its highest point in decades. By
early 2009, each of the cities experienced a doubling of unemployment and job loss rates (job
loss not shown).
While it is unclear how the local economic conditions will affect the impacts of the
TJRD programs on employment and recidivism, the rates of employment for the sample are
expected to be lower during the time period affected most by poor employment conditions. As
discussed in Chapter 1, obtaining employment is difficult for many former prisoners even when
economic conditions are good. It would not be surprising to find that the troubled economic
conditions in these cities made it unusually difficult for the staff in the transitional jobs programs to help participants make the transition to regular jobs. However, the same is likely to be
true for the job search programs, and it is possible that the services received by the transitional
jobs group are more valuable than they would be in better economic times, and program
impacts could be more positive.
The timeframe of the study offers an unusual opportunity to examine the impacts of transitional jobs programs in different labor market environments. In Chapters 5 and 6, impacts are
explored separately for those who were randomly assigned late in the enrollment period, with all
or most of their follow-up occurring during the worst part of the recession to date. The impacts of
the TJRD programs are also analyzed for each site separately (see Appendixes C and D).

Criminal Justice Context
In addition to examining employment impacts, this report assesses the impacts of the
transitional jobs programs on recidivism — that is, rearrest, reconviction, revocation of parole,
return to incarceration, and length of incarceration. As with economic conditions, the criminal

Bureau of Labor Statistics (n.d.).

The Transitional Jobs Reentry Demonstration

Figure 2.1
Average Monthly Unemployment Rates in the Study’s
Metropolitan Statistical Areas: Full Sample

Chicago

Detroit

Milwaukee

St. Paul

Average Unemployment Rate (%)

Jun-09

Apr-09

May-09

Mar-09

Jan-09

Feb-09

Dec-08

Oct-08

Nov-08

Sep-08

Jul-08

Aug-08

Jun-08

May-08

Mar-08

Jan-08

Feb-08

Dec-07

Oct-07

Nov-07

Sep-07

Jul-07

Aug-07

Jun-07

Apr-07

May-07

Mar-07

Jan-07

Feb-07

Apr-08

Late cohort
enrollment:
4/08 to 9/08

Early cohort enrollment:
1/07 to 3/08

Month of Follow-Up
SOURCE: Based on data from the U.S. Bureau of Labor Statistics showing the average unemployment rate in
each of the metropolitan statistical areas.

justice context in which the TJRD programs existed might influence both the levels of recidivism and the impacts.
In 2008, 735,454 people were released from state and federal prisons nationally.3 The
TJRD states accounted for 66,728 of those releases. Because Chicago, Detroit, Milwaukee, and
St. Paul (along with Minneapolis) are the major cities in the states of interest, a large share of
released prisoners would be expected to return to those cities. Indeed, in 2008, 56 percent of the
Illinois parole population lived in Cook County (Chicago).4 In the same year, 12 percent of the
Minnesota population that was subject to supervised release lived in Ramsey County (St. Paul).5
When Hennepin County (Minneapolis) is included in addition to Ramsey County, this proportion jumps to 33 percent.6 Similarly, from 2003 to 2007, Milwaukee parole agencies supervised
25 percent of the state parole population.7 Among those who were released to parole in Michigan in 2003, 34 percent returned to Wayne County, and the majority of those individuals (80
percent) returned to Detroit specifically.8
Most criminal justice laws and policies are set at the state level, shaping how people are
sentenced, the implementation of community supervision programs, and what state criminal
justice agencies record in their administrative data records. Table 2.1 provides some information
about the criminal justice context in each TJRD state. For each indicator of crime and corrections shown in the table, the position of each state relative to the national average is shown.
Illinois, for instance, had a crime rate of 3,469 per 100,000 population in 2007, which was 7
percent lower than the national average of 3,731. Additional information about state operations
that help provide a context for the TJRD data is presented below.
Illinois. The Illinois criminal justice administrative data indicate that felony convictions
nearly always lead to a prison admission. Also, prisoners in Illinois are able to earn “good time”
and can be released early onto parole, leading to relatively shorter stays in prison and relatively
longer stays on parole.
3

Sabol, West, and Cooper (2009).
Illinois Department of Corrections (2008).
5
Minnesota Department of Corrections (2009).
6
Although most people returning from prison in Minnesota have probation sentences, the focus here is on
the supervised release/intensive supervised release/parole population, given that few people on probation were
recruited into the TJRD sample. (Intensive supervised release is for high-risk offender and involves greater
contact with the supervising officer, as explained later in this chapter.) For the full St. Paul sample, 94 percent
reported being on parole and 1 percent reported being on community supervision (presumably supervised
release or parole). Only 1 percent reported being on probation and 3 percent reported not being on any type of
supervision.
7
Van Stelle and Goodrich (2009).
8
Solomon and Thomson (2004).
4

6.1
-9%

22
+70%

277
-13%

499
+10%

3,527
+27%

2.6
-61%

120
-62%

181
-60%

3,411
+25%

3,037
-7%

289
-38%

536
+13%
3,066
-6%

-11%

3,326

Minnesota

-3%

3,602

Michigan

8
+16%

395
+19%

397
-11%

2,491
-3%

2,838
-13%

291
-38%

-16%

3,129

Wisconsin

NOTES: Positive percentages indicate that the measure is higher than the national average; negative percentages indicate that the measure is lower
than the national average.
aThe corrections population is the number of people under corrections supervision per 100,000 people, and corrections supervision includes the
number of adults in prison, in jails, on probation, and on parole.

SOURCES: National Institute of Corrections Library (2010); for number of releases in 2007, Sabol, West, and Cooper (2009).

6.7

344
+7%

319

Parolees per 100,000 (2007)
Relative to national average

Taxpayers' cost (2008), percent of state general
fund spent on corrections
Relative to national average

371
-17%

447

2,609
+1%

2,936
-10%

533
+12%

-7%

3,469

Illinois

Number incarcerated per 100,000 (2007)
Relative to national average

2,572

3,264

Number of property crimes
Relative to national average

Corrections population per 100,000 (2004)a
Relative to national average

467

3,731

National Average

Number of violent crimes
Relative to national average

Relative to national average

Number of crimes per 100,000 (2007)

Variable

Crime and Corrections Contextual Statistics: Comparison of TJRD State Averages and National Averages

Table 2.1

The Transitional Jobs Reentry Demonstration

Michigan. As shown in Table 2.1, Michigan has a lower parole rate than do Illinois and
Wisconsin, and lower than the national average. This is likely because in the late 1990s the state
passed a “truth in sentencing” law that requires anyone sentenced to prison to serve the entire
specified minimum sentence. In other words, Michigan prisoners cannot earn their way out of
prison with good behavior; thus, longer prison stays and shorter stays on parole would be
expected.
Minnesota. For most prison sentences in Minnesota, a person spends two-thirds of his
sentence in prison and one-third in the community on supervised release. As a result of the way
the St. Paul site recruited people into the sample for TJRD (discussed below), about two-thirds
of the St. Paul sample members were on intensive supervised release, a rate much higher than
the countywide rate.9 Intensive supervised release is a supervision strategy for high-risk offenders that involves having contact with the supervising officer at least four times a week, including
unannounced home visits. As discussed further below, the high proportion of sample members
on intensive supervised release is likely related to the fact that the St. Paul staff targeted much of
their recruitment to halfway houses, facilities that are often used for high-risk offenders.
Wisconsin. Sentencing in Wisconsin changed as a result of laws that were implemented
in the late 1990s, which affect rates of incarceration.10 Prisoners who committed an offense
before June 1, 1984, can earn two types of “good time”: statutory good time and industrial good
time. Among prisoners who committed an offense between June 1, 1984, and December 31,
1999, an automatic mandatory release date became part of their sentence, meaning that they
serve two-thirds of the length of their sentence in prison. After the two-thirds date, prisoners can
be held only for “institution misconduct” (violations of prison rules). People who committed an
offense after December 31, 1999, are not eligible for sentence reduction, and they must serve
out their entire sentence incarcerated. Thus, depending on the point in time when the crime was
committed, length of prison stays in Wisconsin can be expected to vary.

Baseline Characteristics and Criminal Histories
of the Sample, by Site
Table 2.2 reports selected background characteristics and criminal history characteristics
of sample members at the time of random assignment, by site. The state context characteristics
that are presented in Table 2.1 help to explain some of the differences that are found across sites.

The same proportion of the transitional jobs and job search groups were on intensive supervised release.
Division of Community Corrections (2003).

The Transitional Jobs Reentry Demonstration

Table 2.2
Characteristics of Sample Members at Baseline, by Site
Full
Chicago Detroit Milwaukee St. Paul Sample

Characteristic
Average age (years)

35.1

36.6

33.2

34.2

34.6

Race/ethnicity (%)
White/non-Hispanic
Black/non-Hispanic
Hispanic
Other

3.2
89.7
6.5
0.5

6.6
88.6
2.8
1.9

6.9
88.6
2.4
2.0

20.2
65.5
4.8
9.5

10.2
80.7
4.7
4.3

Education (%)
GED certificate
High school diploma
Associate's degree/2- or 4-year college/beyond
None of the above

36.0
19.1
5.6
39.3

64.1
10.2
4.4
21.4

38.7
32.8
4.3
24.3

54.3
21.2
7.3
17.1

46.2
22.9
5.8
25.1

Housing status (%)
Owns/rents house or apartment
Lives with friends or relatives
Has transitional housing
Lives in shelter/other

19.0
58.7
15.2
7.1

17.2
68.4
11.0
3.3

19.3
55.6
23.5
1.6

13.1
18.7
63.3
4.8

17.2
47.6
30.2
5.0

Ever employed 6 consecutive months for one employer (%)

47.7

49.8

43.8

56.1

52.6

Worked full time in the 6 months before incarceration (%)

43.8

45.9

39.2

42.6

44.6

Average number of prior arrests

15.3

4.4

7.7

10.2

9.2

Ever convicted of a violent crimeb (%)

47.2

59.3

41.9

80.4

57.8

Ever convicted of a drug crimeb (%)

65.0

41.9

54.9

29.4

46.8

Ever convicted of sexual assault (%)

1.6

3.1

5.9

28.6

10.7

62.1

99.2

71.6

54.4

71.4

Crime category of most recent conviction (%)
Violent crime
Drug crime
Property crime
Public order crime

25.0
38.8
22.8
13.5

40.3
18.2
31.5
10.0

20.5
34.0
18.3
27.2

51.7
12.1
16.0
20.1

34.8
25.1
21.7
18.3

Length of most recent incarceration (months)

27.5

54.2

31.7

21.1

33.2

Sample size

374

426

507

506 1,813
(continued)

Lifetime number of months in state prison
c

Table 2.2 (continued)
SOURCES: MDRC calculations using the Baseline Information Form and data from Michigan State Police,
Minnesota Bureau of Criminal Apprehension, State of Wisconsin Department of Justice, Illinois Criminal
Justice Information Authority, and the Department of Corrections in each state.
NOTES: All criminal history calculations include only events that occurred between January 1, 1980, and the
date of random assignment.
Subcategories may sum to more than the total due to multiple arrests, convictions, or incarcerations per
person during the follow-up period.
aEach arrest date is counted only as a single event. If there are multiple crimes or charges on the same
date, only the most serious charge is recorded in the analysis.
bThe categorization of charges is based on definitions from Langan and Levin (2002).
cEach conviction date is counted only as a single event. If there are multiple convictions on the same
date, only the most serious conviction is recorded in the analysis. Total includes convictions for felony,
misdemeanor, and other crime classes.

As shown in Table 2.2, the average age of the sample members was similar across sites.
More sample members from St. Paul were white than in other sites and more people in St. Paul
were categorized as a race other than white, black, or Hispanic. Sample members in Chicago
reported the least amount of education, and more sample members in Detroit and St. Paul than
in Chicago and Milwaukee reported having either a General Educational Development (GED)
certificate or a high school diploma. Over half of the sample in St. Paul had been employed for
six consecutive months by the same employer at some point before going to prison, which is
slightly higher than in the other sites.
One key difference across sites is housing status. Almost two-thirds of the St. Paul
sample lived in transitional housing at the time of random assignment, which is much higher
than in the other sites. As discussed earlier, this finding likely reflects the focus on recruiting
residents of halfway houses, which are considered transitional housing.
Differences among sites in criminal history information reveal several points of interest.
First, the samples in Chicago and St. Paul had more prior arrests, on average, than those in
Detroit and Milwaukee. This finding may reflect in part the fact that both Illinois and Minnesota
report all arrests that individuals have in their history regardless of whether that arrest led to a
conviction — in other words, they do not “seal” and restrict access to information on arrests.11 In
addition, a much larger proportion of the St. Paul sample had ever been convicted of violent
crimes and sexual assaults, compared with the other sites. As noted earlier, the St. Paul recruitment strategy resulted in a high-risk sample that appears to represent a group of men who
committed more serious crimes than those in the samples from the other cities.
11

Office of the State Appelate Defender (2010); Minnesota Judicial Branch (n.d.).

At the other end of the spectrum, Detroit’s arrest rate is unusually low, likely because
the state criminal justice agency seals arrest information for arrests that lead to a verdict of “not
guilty.”12 This means that some data on arrests for which people are not convicted are not
released as part of individual criminal histories.
Finally, because Michigan has truth-in-sentencing, people from this state would be expected to spend more time in prison than in other states. This pattern is clear for both the
lifetime length of time spent in prison and for the length of the most recent incarceration, with
sample members from Detroit spending much more time in prison than those from the three
other cities.

This scenario holds except in the case of certain violent crimes, if the person has already been convicted of
a crime, or if the arrest record is otherwise ordered to be destroyed at the discretion of a judge despite the
particular crime code or a previous conviction. See www.legislature.mi.gov.

Chapter 3

Program Implementation
This chapter describes the organizations that operated the transitional jobs and job
search programs in the Transitional Jobs Reentry Demonstration (TJRD) and the services that
each provided. Information in this chapter comes largely from two site visits to each city — one
in summer 2007 and the other in spring 2008. The goals of the visits were to understand (1) how
each program was implementing its transitional jobs or job search services, (2) what types of
challenges the sites faced in operating the program, (3) how staff and participants viewed the
programs and services, and (4) what types of variation existed in the local environment, in
particular the social service and criminal justice contexts.
During each visit, researchers interviewed staff from the organizations that were operating the transitional jobs and job search programs as well as community corrections staff.
Researchers also observed program activities and visited transitional job work sites, during
which time they interviewed site supervisors. The research team also conducted case reviews
with staff from the transitional jobs programs to gain a better understanding of the issues faced
by participants and services they received.1 Separate focus groups with transitional jobs and job
search participants were conducted in summer 2007 in each city. Focus group participants were
recruited from lists of enrolled TJRD participants; group sizes ranged from 2 to 18 men. While
the views expressed by these men do not necessarily represent the experiences of all TJRD
participants, they provide insights into the participant perspective on the program.
After providing background information about each organization that was involved in
the demonstration, the chapter describes study recruitment procedures. The types of services
provided by the job search programs are then described, followed by details of the services
provided by the transitional jobs programs. Various challenges that the sites encountered in
implementing and running the transitional jobs portion of the demonstration are also described
in this section. The chapter ends with a summary of the major implementation findings.

In summer 2007, case reviews were conducted of active participants. In 2008, case reviews were focused
on individuals who had moved into unsubsidized employment after working in a transitional job. The research
team selected cases for review in advance of the site visit.

Background of the Transitional Jobs and Job Search
Organizations
As introduced in Chapter 1, The Joyce Foundation used a competitive process to select
and fund employment programs for former prisoners in four cities within its midwestern
grantmaking area: Chicago, Detroit, Milwaukee, and St. Paul. In Detroit, Milwaukee, and St.
Paul, two different organizations were identified, one to run a transitional jobs program and the
second to run a job search program. In Chicago, the same organization ran both the transitional
jobs and job search programs.
The four transitional jobs programs were operated by different types of organizations,
including a small community-based nonprofit organization (in Milwaukee), a large regional
nonprofit organization (in Chicago), and a local affiliate of a national nonprofit organization (in
Detroit and St. Paul). The four job search programs were also operated by different types of
organizations but were primarily community-based nonprofit organizations (in Detroit, Milwaukee, and St. Paul). Although the transitional jobs program organizations had different levels
of experience serving former prisoners, all had served this population in some capacity prior to
being selected for the demonstration. Furthermore, all transitional jobs providers had previous
experience providing transitional jobs to participants who met certain criteria (other than being
former prisoners). In contrast, the organizations that were operating job search programs had
more experience providing services to former prisoners. Tables 3.1 and 3.2 provide general
characteristics of the programs in each city; further details of each are provided below.
Chicago
The Safer Foundation provided services to both the transitional jobs and job search
groups in Chicago. Safer was founded in Chicago in 1972 to advocate on behalf of individuals
with a criminal record who were attempting to find jobs. The Safer Foundation continues to
provide services to individuals with a criminal record, with nine locations in Illinois and Iowa.
Safer’s goal is to “reduce recidivism by supporting, through a full spectrum of services, the
efforts of people with criminal records to become employed, law-abiding members of the
community.” All of the employment and support services provided by Safer are targeted to
former prisoners and are funded by a variety of organizations, including the Illinois Department
of Corrections (IDOC), the U.S. Department of Housing and Urban Development, and private
foundations.
Safer also operates Pivotal Staffing, LLC, a temporary staffing agency providing a full
range of hiring services for the employers that contract with them. Pivotal Staffing was started
in January 2005 specifically for a contract with Allied Waste, the company that held waste
management and recycling contracts with the City of Chicago. With the start of the contract,

The Transitional Jobs Reentry Demonstration

Table 3.1
Overview of the TJRD Study Sites: Transitional Jobs Providers
Characteristic

Chicago

Detroita

Milwaukee

Transitional jobs
provider

Safer Foundation

Goodwill Industries
of Greater Detroit

New Hope Project Goodwill/Easter
Seals Minnesota

Organizational type

Regional nonprofit
organization

Local affiliate of
national nonprofit
organization

Community-based
nonprofit
organization

Local affiliate of
national nonprofit
organization

Previous experience
Organization founded
serving former prisoner with purpose of
population
serving this population

Served this
population through
other programs;
population not a
specific focus until
2004

Served this
population through
other programs;
population not a
specific focus until
2007

Previous experience
providing transitional
jobs

Through industrial
site

Scattered site
model part of
original program

Through retail
outlet

Through contracted
company

St. Paul

SOURCE: MDRC field research.
NOTE: aThe Michigan Department of Corrections, Prisoner Reentry Initiative, chose the transitional jobs and job
search provider organizations and had overall leadership of the TJRD programs in Detroit.

temporary jobs with Allied Waste were available to Safer clients, who could gain work experience and a paycheck before moving into other jobs, an arrangement approximating a transitional jobs program.
Detroit
Compared with the other cities in the evaluation, Detroit was unique in that an additional organizational layer was involved in the project. The Michigan Prisoner Reentry Initiative
(MPRI), a statewide reentry effort, is operated by the Michigan Department of Corrections
(MDOC) and was the original applicant to participate in the demonstration. MDOC envisioned
the demonstration as part of the larger MPRI effort.

The Transitional Jobs Reentry Demonstration

Table 3.2
Overview of the TJRD Study Sites: Job Search Providers
Characteristic

Chicago

Detroita

Milwaukee

St. Paul

Job seach provider

Safer Foundation

First JVS, then
Detroit Hispanic
Development
Corporation

Project RETURN

Amherst H. Wilder
Foundation

Organizational type

Regional nonprofit
organization

Metropolitan-area
nonprofit (JVS)/
Community-based
nonprofit (Detroit
Hispanic)

Community-based
nonprofit
organization

Metropolitan-area
nonprofit

Served this population
through other
programs (JVS)/
contract to serve
former prisoners
(Detroit Hispanic)

Organization
Contract to serve
founded with
former prisoners
purpose of serving
this population

Previous experience
Organization founded
serving former prisoner with purpose of
serving this population
population

Goodwill Industries of Greater Detroit was selected to deliver services to the transitional jobs group. Goodwill was founded in Detroit in 1921, originally to serve only the seriously
disabled. Over time, Goodwill expanded its service population to include individuals who are
HIV-positive, public assistance recipients, dislocated workers (people who have been laid off
from their jobs), and former prisoners. Today, there are eight Goodwill locations in the metropolitan Detroit area. In 2004, the Ford Foundation provided funding to Goodwill to target male
and female former prisoners for services. Goodwill has an on-site industrial facility and contracts to perform light assembly and related work for a variety of clients, including the automotive industry. Participants in the other Goodwill programs (for those with disabilities,
dislocated workers, and others) work in the facility in transitional employment slots.
MPRI initially selected JVS (formerly Jewish Vocational Services), through a requestfor-proposals process, to offer services to the job search group. JVS is a long-time employment
and training organization, with more than 65 years of operating experience in Detroit and
southeastern Michigan. In the Detroit office, JVS provides workforce development services for

welfare recipients, food stamp recipients, homeless people, and those with disabilities who are
referred to the organization. JVS has a history of working with former prisoners in its various
programs, although not with any specific funding stream or program dedicated to that population. JVS’s contract with MPRI expired on December 31, 2007, and was not renewed because
JVS was also selected to operate a statewide transitional jobs initiative, and it would have been
too complicated for the organization to operate both programs simultaneously. Therefore, in
2008, Detroit Hispanic Development Corporation (Detroit Hispanic) assumed the operation of
job search services for TJRD sample members. Detroit Hispanic was founded in 1997 by its
current executive director as a youth gang violence prevention program in southwestern Detroit,
an area with a large Latino population. Detroit Hispanic also operated a welfare-to-work
program in its early years. Although the organization serves adults, programs for youth remain
its core focus. Before becoming involved in TJRD, Detroit Hispanic had been serving former
prisoners through MPRI.
Milwaukee
The New Hope Project ran the transitional jobs program in Wisconsin. New Hope was
created in 1991 as a demonstration project designed to help individuals leave poverty through
employment. To that end, New Hope has provided transitional jobs (previously called “community service jobs”) for individuals who are unable to secure unsubsidized employment. New
Hope’s programming has shifted several times over the years, both because of funding changes
and because of changes in the political climate. The original demonstration program, which was
evaluated by MDRC, ran from 1991 to 1999. Between 1999 and 2002, the organization primarily conducted policy analysis and advocacy work. In 2003, New Hope began offering services
again. From 2003 to 2004, the organization had a contract to provide services through the
Workforce Investment Act (WIA), and starting in 2004 it received funding from the Department
of Corrections via the State of Wisconsin to work primarily with noncustodial parents and those
with criminal records. In 2009, New Hope merged with the YWCA of Greater Milwaukee.
Project RETURN (Returning Ex-offenders To Urban Realities and Neighborhoods)
provided the job search services in Milwaukee. Project RETURN, which was incorporated in
1981, is a nonprofit organization that grew out of a “Social Concerns Committee” affiliated
with Milwaukee’s Cross Lutheran Church, which was charged with determining which communities were not well served in its area. The group identified four minimum-security prisons
located within a two-mile radius of the church and found few or no services available to the
formerly imprisoned. Prior to involvement in this demonstration, Project RETURN offered job
search services, a fatherhood program, alcohol and drug treatment services, and support groups
that focused on anger management, social skills training, and the like, all of which were directed
at the former prisoner population.

St. Paul
Goodwill/Easter Seals Minnesota (G/ESM) operated the transitional jobs program in St.
Paul. G/ESM is a large regional nonprofit organization that aims to help individuals, including
those with disabilities, who face employment barriers to achieve economic success. The
Minnesota chapter of Goodwill was founded in 1919 by a group of St. Paul civic and religious
leaders. After World War II, Goodwill increasingly emphasized efforts to help individuals with
disabilities overcome barriers to employment, and in 1984 its Minnesota chapter became
affiliated with the National Easter Seals Society, which had a long history serving those with
disabilities. Services to the disabled population remain an important focus of G/ESM, though
the organization works with other groups, including welfare recipients, dislocated workers, and
young fathers struggling with their responsibilities as parents. G/ESM has also been involved in
a small number of programs targeted at prisoners who are on probation and released prisoners in
Hennepin County. G/ESM is most closely associated with its 15 retail stores, which resell
donated goods throughout the state.
The Amherst H. Wilder Foundation provided services to the job search group in St.
Paul. The Wilder Foundation is one of the largest nonprofit organizations in the Twin Cities,
established in 1906 with a mandate to assist St. Paul’s needy, and today it provides a wide range
of services to assist the elderly, at-risk youth, the homeless population, and other groups. Wilder
has periodically worked with criminal justice clients; at the time of the demonstration, it
provided pre- and post-release employment services through a contract with the Minnesota
Department of Corrections.

Study Recruitment
The organizations in each city were charged with developing a strategy to recruit men
within three months of their release from prison, and those strategies varied by site. In Chicago,
study recruitment closely mirrored the Safer Foundation’s standard business practice, although
that process was subsequently augmented. In St. Paul, the job search program used an existing
referral pathway based on established ties with the corrections system and halfway houses,
although those efforts were also supplemented. Since recruitment into the demonstration meant
that only half of the men would be assigned to a transitional job (and the other half to the job
search program), existing recruitment efforts alone often did not yield the numbers needed to
conduct random assignment. Detroit and Milwaukee had to devise new recruitment practices. In
Detroit, recruitment was overseen by staff employed by the Michigan Prisoner Re-entry
Initiative, but in Milwaukee, staff in the transitional jobs and job search organizations both were
involved with recruitment.

In Chicago, the Safer Foundation initially relied on its established referral paths to recruit individuals into the demonstration. That is, referrals came from contacts that the organization had developed with the Illinois Department of Corrections, parole and probation
agencies, and the Illinois Department of Human Services. There were also a large number of
walk-ins because Safer is well known in the community. However, most clients who came via
those routes had been released from prison 6 to 12 months previously, making them ineligible
for the TJRD study. Safer then instituted other recruitment strategies to increase the number
of eligible referrals. Those strategies included meeting with inmates within 60 days prior to
their release from an IDOC or other facility to explain the study, and meeting with parole and
probation officers and staff from halfway houses to encourage them to refer men soon after
their release date.
Staff from New Hope and Project RETURN worked together closely to recruit participants in Milwaukee. Staff from both organizations conducted monthly presentations at correctional facilities throughout Wisconsin, particularly the Racine Correctional Facility and the
Milwaukee Secure Detention Facility. Staff also distributed flyers at local parole offices and
attended bimonthly meetings with community corrections personnel and their parolees in an
effort to “market” the demonstration. Similarly, staff from the job search organization in St.
Paul provided information about the study within correctional facilities at periodic “transition
fairs,” which are information sessions for soon-to-be-released inmates. They also met with
parole officers and staff from halfway houses and recruited internally from the existing employment program for former prisoners.
Unlike the other sites, enrollment into the study in Detroit was initially conducted before men were released from prison. From February through December 2007, MPRI staff and
representatives from the transitional jobs and job search organizations went into two facilities
(Mound and Ryan, both located in Detroit) monthly to conduct presentations about the demonstration and to recruit participants. Only men to be released on the second or fourth Tuesday of
each month were eligible, as this criterion represented the majority of the release population.
Former sex offenders were not eligible because, in Michigan, sex offenders are sent to specialized parole agents. In January 2008, in response to relatively low enrollment rates, the site
switched to a post-release process, and the study was opened to men released from all prisons in
Michigan who were returning to Wayne County and who had been out of prison for less than 90
days. An MPRI staff person continued to give monthly presentations within the prisons to
promote the demonstration, but potential participants were primarily referred by their parole
officers after they were released.
All the sites, regardless of their prior experience working with the criminal justice
system, had to forge relationships with correctional facilities and the local parole system to
ensure a consistent flow of potential client referrals. This relationship-building took time and

was often solidified only several months into the demonstration. Staff time was used to visit
correctional facilities to present information about the demonstration to soon-to-be-released
inmates, which required cooperation from the institutions’ leadership. Staff resources were
also spent educating parole supervisors and officers about the demonstration project, including eligibility criteria and the potential benefits of referring their clients to the demonstration.
During the second research visit to each site, many staff noted that they had just solidified
these relationships with local community corrections when it was time to start winding down
recruitment for random assignment.
Intake
Like recruitment generally, the intake process differed somewhat from site to site. In
Chicago, the Safer Foundation assessed all new clients and referred them to one of three
“pathways.” Those who were able to work but had relatively little work experience or education
were assigned to the “transitional employment pathway.” Random assignment took place
among clients within this pathway. After random assignment, individuals who were assigned to
the transitional jobs group were usually placed in a transitional job at Allied Waste’s recycling
plant within a day or two of random assignment. Individuals who were assigned to the job
search group could receive assistance in finding a job, but they were not eligible for placement
in a transitional job.
As noted earlier, in St. Paul, random assignment took place at the job search program
organization. Individuals who were assigned to the transitional jobs group were given a bus
token and Goodwill’s phone number, and were immediately sent to Goodwill (located about
four blocks away) for an appointment with a case manager that morning. They began work in a
transitional job the following day. Individuals who were assigned to the job search group
remained on-site to continue receiving services.
In Milwaukee, where random assignment occurred in the building where both organizations had their offices, participants who were assigned to the transitional jobs and job search
groups met with staff from the respective organizations immediately after random assignment to
learn about available services. Individuals who were assigned to the transitional jobs group were
given appointments to meet individually with their assigned case managers to begin choosing a
transitional job worksite.
Finally, in Detroit in 2007, when random assignment occurred within correctional facilities, participants were bused to the transitional jobs and job search program organizations upon
their release, where they were able to contact friends or family. When random assignment was
conducted post-release, individuals were taken by van from a neutral random assignment
location to the transitional jobs program, while those in the job search group made appointments
to meet with a case manager later that day or on another day.
38

Intake Challenges
Intake into the demonstration was not constant throughout the course of the project, and
sites had to devise new ways of attracting participants. Despite recruitment and other publicity
efforts, at times only a handful of participants attended an orientation session. Even in Detroit,
where staff were provided with information about release dates, it was not always possible to
predict the number of participants. In some cases, release dates changed, or some men changed
their minds about participating in the demonstration. After concerted efforts to increase enrollment, the sites had a large number of men attending orientations.
A different challenge that was of particular concern in Milwaukee was ensuring that
men who were randomly assigned were eligible for and could benefit from participating in the
demonstration. In previous program offerings, New Hope had asked participants to first search
for employment, and placed participants in a transitional job only if they were unable to secure
unsubsidized employment on their own. While this is a valid design for an employment assistance program, it was not the intended TJRD model, and staff had to reorient their philosophies
as well as develop new ways of talking about the demonstration. Although only half of those
who were randomly assigned would become eligible for the transitional jobs program, it was
vital that all men who were randomly assigned met the eligibility criteria. Before fine-tuning the
recruitment message, staff in Milwaukee randomly assigned some individuals who already had
jobs or who were not interested in being assigned to a transitional job. Employed individuals
should have been screened out of the intake process for the demonstration although they may
have qualified for services in the organization’s past.
Similarly, in St. Paul, staff reported that it was sometimes difficult to tell who truly
“needed” a transitional job. Potential clients heard about the demonstration at the job search
program site. Staff emphasized that the transitional jobs paid minimum wage and sometimes
urged men to look for a job before signing up for the demonstration if they thought they could
get a better job. The idea was to urge men to self-select in or out of the project depending on
their labor market prospects. In reality, however, there was very little risk for men in signing up
for the project since they would have a 50/50 chance of getting a paid job within 24 hours.
Some participants who could have done better may have reasoned that they could keep looking
for another job while working in the transitional job.
For participants, a major motivation for entering the study was the possibility of being
assigned to the transitional jobs group and getting a paid job relatively quickly, which often
made the assignment to the job search group less attractive. Staff in two of the cities noted that
some participants, when they found out they were assigned to the job search group, were
disappointed. As a staff person in St. Paul noted, “People say ‘you can go to the Wilder Job
Club and they’ll get you a job in this here program.’ So they come here with the anticipation
that they’re going to get hired. And then once they find out that it’s a random selection…and
39

they don’t get picked, they’re pissed off, and they just walk out.” Staff in Detroit also noted that
some participants voiced a similar disappointment when they were assigned to the job search
group. In Chicago, this issue was not raised. A number of the focus group participants noted
that Safer had a good reputation in the community, so that being assigned to the job search
group over the transitional jobs group was not a concern.

Job Search Program Services
As discussed in Chapter 1, the TJRD project was designed to test services that are built
around transitional jobs against a set of more limited job search services. Most of this chapter is
dedicated to the transitional jobs services; however, this section describes the services that were
available to those who were randomly assigned to the job search group.
All job search participants had access to computers, phones, and fax machines to assist
in finding and applying for jobs. Some participants were able to access other ancillary services
as well. For example, in Detroit, JVS provided access to clothing and packages of toiletries, and
in St. Paul participants received a clothing allowance.
Initially, all job search group members could participate in group activities offered by
the service providers. During this time they were instructed on appropriate workplace behaviors
and other soft skills, learned how to discuss their criminal records with employers, and received
information about formatting résumés and filling out applications. The group activities varied in
intensity. In Chicago these workshops ran for a full week and provided information on developing résumés, job search techniques, and appropriate workplace behavior. Offerings in St. Paul
were similar. In Milwaukee, participants could join the 16-session “Fatherhood Initiative,”
which focused on similar job-related issues, communication skills (including talking about
offense histories), and interpersonal skills, as well as on parenting, although not all participants
were fathers. The fatherhood group met for two hours a day, four days a week. In 2007, participants in Detroit could join a group job-readiness class that met three days a week; these classes
included non-TJRD participants as well. When Detroit switched job search service providers to
Detroit Hispanic, that organization did not offer group activities. Instead, case managers were
available to assist participants individually with tasks such as developing a résumé. In the other
cities, participants could meet one-on-one with case managers, but this practice was formalized
only in Milwaukee, where staff met twice weekly with participants until they found a job.
The degree to which job search group participants received assistance with finding employment (as opposed to help mastering job-seeking skills) also varied by city. In Chicago, staff
called “sector managers” worked with local employers to identify open positions and then
match those positions to participants (including those in the transitional jobs and job search
groups, and those enrolled through other Safer Foundation programs). In Milwaukee, a con40

tracted staff person served as a “job developer.” Like the sector managers, the job developer met
with area employers to learn about various job openings and then shared that information with
participants through their case managers. While this individual spent more time with participants in the transitional jobs group, he shared job leads with staff working in the Milwaukee job
search program. The initial Detroit job search provider also had a specialized job developer on
staff. No staff person in the St. Paul job search provider was designated as a “job developer,”
and this type of service was offered more informally. That is, staff talked with employers in the
community and occasionally hosted employers at their site so that participants could learn more
about what employers look for when they are hiring.
Focus group participants had mixed assessments of the services offered by the job
search providers. One man in Detroit said that for individuals who had limited work experience,
learning about appropriate workplace attire and how to fill out applications was probably useful.
On the other hand, he said, “If you’ve been in the employment field and you know how to fill
out an application, you know how to dress yourself properly, and you know how to groom
yourself, then their services [are] inadequate.” By contrast, in Milwaukee, focus group participants singled out the attention staff paid to them as an important part of the program. As one
said, “They [the staff are] hands-on with you. You know, a lot of people out here talk about
helping you get a job and this and that, [but] they sit down with you, you know; you send faxes
right there in the office.”

Transitional Jobs Program Services
At the heart of this demonstration was a test of the transitional jobs model as a way to
help returning former prisoners find unsubsidized employment. However, while “transitional
jobs” share a broad set of characteristics, including being temporary in nature and providing
more intensive support and supervision, the type of work offered, the setting of that work, and
the nature of the supports and supervision may differ across programs. Indeed, although all the
sites provided transitional jobs, each organization had slightly different staffing configurations,
provided different types of employment support and case management services, and offered
different experiences in the transitional job. Table 3.3 provides an overview of the structures of
the transitional jobs programs.
Staffing


Each transitional jobs program had a different staffing configuration,
but staff performed similar functions overall. The degree of specialization for any one staff member varied by transitional jobs organization.

Assessment, vouchers for transportation
or work boots; soft skills group
activities

Pre-employment support
services

SOURCE: MDRC field research.

Intake staff, case managers, job
developers, on-site transitional job (TJ)
supervisors

Existing contracts with Illinois
Department of Corrections; twicemonthly meetings with soon-to-be
released inmates; recruiting from
community activities; meetings with
community corrections personnel, Dept
of Human Services, community-based
organizations, and halfway houses

Study recruitment

Staffing

Chicago

Characteristic

Case managers, job
developers, on-site TJ
supervisors
ID recovery, child support
navigation, referral to
support services, vouchers
for transportation,
clothing, other work
necessities; voluntary
support group

Case managers, job
developer, TJ site liaison

ID recovery, small loans,
child support navigation,
vouchers for
transportation, referrals for
clothing, other work
necessities; soft skills
group activities as of fall
2007

ID recovery, child support
navigation, vouchers for
transportation, clothing,
other work necessities; $20
daily stipend for one week;
soft skills group activities

Marketing within
correctional facilities,
particularly to soon-to-bereleased inmates; flyers
provided to community
corrections personnel and
halfway houses

Monthly presentations at
correctional facilities in
state; marketing efforts
targeted to community
corrections personnel

Monthly presentations to
soon-to-be-released inmates
at Mound and Ryan
correctional facilities; after
January 2008, referrals from
parole officers

Case managers, job
developer, on-site TJ
supervisor

St. Paul

Milwaukee

Detroit

Transitional Jobs Program Service Approaches in the Demonstration Sites

Table 3.3

The Transitional Jobs Reentry Demonstration

Staff Experience with Former Prisoners
In all the transitional jobs programs except Chicago, former prisoners were not a focal
population in prior program offerings, though all the programs worked with that population to
some degree. Staff in these organizations, and new staff who were hired throughout the course
of the demonstration, had to learn about issues that are unique to the population of former
prisoners, such as complying with parole conditions. Both Milwaukee and Detroit had staff that
resembled the TJRD population demographically — that is, young African-American men from
the community, some of whom had a criminal record. Many case managers in Milwaukee,
however, were hired on a short-term basis through volunteer programs, were temporarily
relocated to Milwaukee from other parts of the country, and stayed in their position for approximately a year. In St. Paul, staff that resembled the TJRD population demographically were
hired later in the demonstration.
Staff Specialization
The extent to which staff were specialized in their jobs varied. All programs had staff
who performed case management functions, and they served as the primary contacts for the
participants. Each program also had at least one job developer, whose role was to make connections with local employers and help place participants in unsubsidized jobs. Staff in Milwaukee,
St. Paul, and Detroit tended to be less specialized, with the case managers assisting participants
with a variety of issues, including lessons in soft skills, legal aid referrals, or résumé development. Staff in Chicago were slightly more specialized, with separate intake staff, case managers
(called “retention specialists”), and job developers (called “sector managers”).
The organizations in Chicago, Detroit, and St. Paul ran their transitional jobs programs
on-site or at a single location, with additional staff supervising participants who were working
in a transitional job. Although these staff were not necessarily funded by the demonstration,
they interacted frequently with participants, and their role is described further in the discussion
of the transitional jobs that appears later in this chapter. In the second year of the demonstration,
Milwaukee, which ran a scattered site transitional jobs model, hired a staff person to develop
new transitional job positions and sites, and to serve as a liaison with the existing jobs sites.
As noted in Chapter 1, each demonstration site had a “research coordinator” position
that was subsidized by the evaluation. This staff person, often an existing employee from one of
the organizations involved in the demonstration, assumed responsibility for administrative
functions related to the evaluation. For example, the research coordinator conducted random
assignment and gathered all signed consent forms to send to MDRC. In Detroit, because the
Michigan Prisoner Re-entry Initiative was the coordinating body for the demonstration, additional staff hired by MPRI were involved with the program. Those staff included the research
coordinator, who conducted intake and random assignment, his supervisor, who was involved in
43

administrative issues related to the demonstration (for example, monitoring the flow of potential
participants from the prisons), and support staff.
Staff Turnover
As expected, the sites experienced some turnover in staff; in Milwaukee the turnover
was partially planned with the use of case managers who were hired on a short-term basis
through volunteer programs. In all sites, new staff had to be recruited and trained, and then
had to establish relationships with participants. Additionally, key positions sometimes went
unfilled for some length of time, requiring certain tasks to be reassigned to staff who might
not have had the relevant expertise. For example, in Detroit the job developer position was
not filled for many months. In Milwaukee, before a transitional jobs site coordinator was
hired, the job developer initially focused on developing transitional job sites, as opposed to
private sector job leads.
Employment Support Services and Case Management


The transitional jobs programs offered similar sets of services, which
were delivered in either group or individual settings. The degree of formalized contact with participants and types of support services offered
varied. However, all programs provided support for transportation,
housing, clothing or tools procurement, referrals to substance abuse or
mental health treatment, or referrals to training programs.

Program Orientation
Once they were randomly assigned, all transitional jobs participants attended an orientation to the transitional jobs program, during which time expectations about attendance and other
issues were explained. In Detroit, participants attended orientation on Tuesday and spent the rest
of the first week obtaining identification and getting acclimated to life outside of prison;
participants began their transitional jobs assignment the following Monday. For participants in
Chicago, placement into the transitional job usually occurred within a few days to a week,
depending upon the current staffing of the work site. In St. Paul, participants started working in
their transitional job the day after random assignment. In those sites, any type of employment
readiness and other group-based activities (explained in more detail below) were provided
around work schedules. After random assignment in Milwaukee, participants first met with their
case manager several times before beginning their transitional job placement; later in the
demonstration, participants also attended a group workshop series before starting to work.

Case Management
Each program provided some degree of case management services. All programs had
case managers who were the primary point of contact and coordination for participants. However, the degree to which that contact was formalized or routinized varied quite a bit, even within
a site, and sometimes differed depending upon a participant’s circumstances and his willingness
to comply with various activities and referrals. One service provided by all but the Chicago
program was identification recovery. Many former prisoners leave incarceration without any
valid identification, such as a social security card or state identification, which are required to
secure legal employment. One program noted that fully one-third of the participants could not
be placed in a transitional job until they secured identification. However, the degree of assistance that was provided varied. Staff in St. Paul, Detroit, and Milwaukee identified applying for
and obtaining social security cards and driver’s licenses as major challenges for participants but
provided some support navigating through the process for each. Staff in Chicago determined
that identification recovery (for example, obtaining a birth certificate, social security card, or a
driver’s license) was a major barrier for participants to be “work ready” because men were often
released without proper identification. However, Safer did not provide direct support to participants with that process, but referred them to an outside agency.
Pre-Employment Support Services
Other support services and materials that programs provided were child support navigation, vouchers for transportation, clothing, and tools or other materials needed for employment. Although staff noted that participants faced a host of health, mental health, housing, and
family challenges (including owing back child support), the programs themselves were not
able to offer comprehensive support services. As a staff person in St. Paul noted, for other
populations, such as the disabled, dedicated programming often exists. However, for former
prisoners, that is rarely the case. To the extent that program staff made referrals, substance
abuse programs were commonly used. In Detroit, staff referred participants with substance
abuse problems to treatment programs, and in Milwaukee participants were often mandated to
participate in Alcohol and Other Drug Abuse (AODA) programs as a condition of their
parole. In Chicago, participants with substance abuse issues were not allowed to begin their
transitional job until they completed or, later in the program, at least started that site’s “supportive service pathway.” Within this pathway, staff assisted clients in resolving issues that
could be barriers to successful employment, such as housing, substance abuse, or mental
health issues, through referrals to outside agencies. In St. Paul, staff noted that many participants were already in treatment programs, as mandated by their parole officers. The program
also started its own support group that participants could voluntarily attend to discuss issues
such as reconnecting with family members.

Housing challenges were common among many participants, and staff and the participants themselves noted a dearth of resources in this area. In St. Paul, many participants were
referred to the program from halfway houses and needed to quickly find another place to live. In
focus groups, some of the men reported that when they were released, family members with
whom they might have stayed had either moved away or were deceased. Another challenge
frequently mentioned by staff and participants was transportation. Most programs provided some
form of transportation support to participants, usually in the form of bus fare or transit passes.
Group Activities
In addition to case management services, all participants were offered the opportunity to
participate in group activities that focused largely on the development of soft skills such as
work-appropriate behavior, communicating about criminal history, and strategizing about how
to handle potentially difficult situations at work. In all cities except Milwaukee, group activities
were offered in combination with the transitional jobs. In Chicago, “retention groups” met
weekly at the recycling plant and twice a month at the Safer office. In Detroit, groups met each
afternoon after work and participants were paid for up to five hours a week for attending soft
skills trainings. St. Paul participants were not involved in employment readiness or skills
training until they had successfully demonstrated their seriousness about working, typically
after working at least 30 days in a transitional job. Participants could then join the 16-hour
employment readiness training as well as occupational skills training programs. Those in
training worked fewer hours in their transitional job but continued to receive full wages. In early
2008, Milwaukee implemented an eight-day group workshop for participants prior to starting
the transitional job. Participants met for two hours a day and covered topics such as goal setting
and interest identification, workplace expectations, and life skills.
Focus groups with program group participants revealed that participants had mixed assessments of the group activities and case management services, although comments tended to
be more positive than negative. In particular, participants singled out the efforts of individual
case managers, whom they perceived to be advocating on their behalf, whether to employers or
to other individuals. However, a theme that came up frequently during the focus group sessions
was that while the various services (and the transitional job) were very important, participants
had to be motivated in order to succeed.
The Transitional Job


The time from random assignment to placement in the transitional job
as well as the types of jobs varied across sites. The transitional jobs were
primarily designed to teach general employability skills, not specific job

skills. The case managers and the transitional jobs supervisors communicated regularly for the most part.
Timing of Transitional Job Placement
Participants in Chicago, Detroit, and St. Paul usually began their transitional job assignment the day after random assignment (St. Paul) or within a week thereafter (Chicago and
Detroit). In Milwaukee, placement typically occurred one to two weeks after entry into the
program. First, transitional job slots needed to be available; in the early months of the demonstration, the Milwaukee program did not have enough transitional job opportunities for the
number of participants who enrolled. Second, participants were required to interview for a
particular position. Finally, a number of the transitional jobs sites could not accept participants
with certain offense histories. For example, those who had been convicted of sex offenses could
not take janitorial positions in day care centers. This was also an issue in St. Paul, where
participants with violent or sexual offense histories could not work near individuals who were
considered mentally or physically vulnerable.
Transitional Jobs Models Used in the Demonstration
As shown in Table 3.4, the transitional jobs differed significantly across the programs.
There were three general models of transitional jobs: scattered site (Milwaukee), placement in
existing enterprises (Detroit and St. Paul), and contracted work (Chicago). In Milwaukee’s
scattered site model, participants were placed in positions with local nonprofit organizations or
small businesses but New Hope paid their wages. For those organizations or businesses, hosting
a transitional job slot meant getting an employee whose pay was fully subsidized. New Hope
(Milwaukee’s transitional jobs program) had used a version of this model in its prior programs.
Participants in Detroit and St. Paul worked in jobs in existing Goodwill enterprises. In Detroit
the jobs were in Goodwill’s light industrial facility, and in St. Paul participants worked in one of
Goodwill’s four retail-related facilities. In Chicago, as noted earlier, transitional jobs workers
were employed by a staffing agency that had been established by the Safer Foundation and was
contracted to work for Allied Waste, a waste management company that held a recycling
contract with the City of Chicago.
All sites paid participants the federal minimum wage or the state minimum wage,
whichever was higher. Participants were generally scheduled to work between 30 and 40
hours a week.2
2

The structure of the management information system data did not allow for calculations of the actual average hours worked per week.

Case management services
available; information about
job search techniques,
interview coaching

Daily soft skills group;
participants paid up to 5 hours a
week for attending; resume
development, soft skills
coaching, interview coaching;
identification recovery;
emergency food, clothing,
rent/utility assistance;
transportation support; job
search assistance; quarterly
men’s groups and bi-weekly
alumni luncheons

Weekly soft skills group
meeting at plant; twicemonthly soft skills group
meetings at Safer;a
participants eligible for gift
cards for retention at 30, 60,
90 days; participants
eligible for monthly bonuses
based on production;
interview coaching

Services available
during or after
transitional job
placement

Scheduling varied by
employer; 40-hour maximum
work week

5 days/week for 30 hours with
opportunity for overtime

Janitorial, general labor,
sorting/stocking,
warehousing, auto repair,
retail, maintenance, etc.

Typically within two weeks

5 days/week for 40 hours

Monday after random
assignment

Scattered site, nonprofits and
small businesses

Transitional job work
hours

Next day, if slots available

Timing of placement

Single site, existing enterprise

Milwaukee

Light manufacturing,
reclamation, piecework,
packaging for auto industry
contracts

Single site, contracted
business

Transitional job (TJ)
model

Detroit

Type of work performed Pulling recyclable materials
from conveyer belts of
garbage plant

Chicago

Characteristic

(continued)

Case management services
available; identification
recovery; computer lab access;
weekly men's group; job search
assistance; full wages paid for
training and job search; 16-hour
group employment readiness
class; skills training available
(e.g., forklift operator, auto
repair); transportation support

5 days/week for 32.5 hours; less
when searching for
unsubsidized employment

Collecting, processing, selling
merchandise at retail stores

Next day

Single site, existing enterprise
(4 retail divisions)

St. Paul

Characteristics of the Transitional Jobs and Job Placement Services in the Demonstration Sites

Table 3.4

The Transitional Jobs Reentry Demonstration

Weekly calls for 1 month by
case manager; job
developers followed up with
employers

Post-job placement
services

Not offered

Case management services
available; participants could
continue to attend soft skills
class

Job developer met with local
employers to learn about job
openings; particular focus on
small businesses

Detroit

Eligible for up to $1,500 over
12 months with proof of fulltime employment

Case management services
available

Job developer met with local
employers to learn about job
openings

Milwaukee

Offered after December 1,
2007; after completing 30-day
TJ, eligible for $175 monthly
and additional $350 after 6
months of work, for total
$1,400

Case management services
available; weekly job club
available

Job developers met with local
employers to learn about job
openings; particular focus on
small businesses

St. Paul

NOTE: aTopics addressed during the group meetings included workplace issues (tardiness, attendance), child support navigation, record expungement,
coping or work management, and job opportunities. Vouchers for work apparel and bus transportation were also provided as appropriate.
bThough Chicago did not offer retention bonuses to participants, production bonuses were available.

SOURCE: MDRC field research.

Not offeredb

Case managers provided job
leads during weekly group
meetings; job developers
contacted employers and
then selected clients to
interview for specific job
openings

Job development
approach

Retention bonus

Chicago

Characteristic

Table 3.4 (continued)

While working, participants had access to other services, including continued case
management. As noted above, participants in Chicago and Detroit could attend group meetings that were designed to assist with soft skills development. Also, in Chicago, participants
were eligible to receive gift cards for staying in a job for a specific amount of time, and cash
bonuses were also paid out when the plant in which the participant worked pulled out the
most recyclables.
Transitional Job Responsibilities
In general, the placements were designed to teach employability skills (for example,
the importance of punctuality, proper workplace etiquette), rather than to train participants in
specific occupations, which was consistent with participants’ impressions. For example, one
focus group participant from Milwaukee noted, “I think the disadvantages to a transitional job
is that all the transitional jobs is real menial type jobs. It’s not like you’re going to get welding, machine shop.”
The types of tasks that participants performed, though, varied tremendously. In Chicago, most participants worked in garbage plants pulling recyclable materials from conveyer belts,
which was physically strenuous and required constant focus. The work itself was not challenging, but the conditions of work, such as the odor of the garbage, the heat of the plants during
warm months, and the potential hazard presented by some discarded items, could be difficult to
endure. In Detroit, participants did light assembly and manufacturing work, piece work, and
packaging of various items, including nails and screws.3 In St. Paul, participants began work in
one of four Goodwill retail divisions: the salvage department, the retail shop floor, the ecommerce division, or the outlet store. Participants performed tasks that included sorting
donations, pricing merchandise, doing janitorial work, or handling odd jobs. Participants in
Milwaukee performed a variety of tasks, depending upon the transitional job site in which they
were placed. Maintenance, janitorial work, sorting or stocking merchandise, warehousing, auto
repair, retail, and general labor were among some of the tasks that participants performed; some
participants worked as mentors to youth but this type of work was the exception.

Goodwill’s plant shut down for two weeks each in July and December to re-tool for changes in production work. During those weeks, participants did not work or receive wages. In June 2008, three other transitional jobs opportunities were developed to compensate for the decline in plant business, which was related to
downturns in the auto industry. Approximately one dozen participants were placed with two construction firms
and a sanitation company.

Supervision Practices
Supervision practices varied across the transitional jobs programs. Participants in Milwaukee, because they were working in a variety of locations, were supervised primarily by
managers or other employees of the nonprofit organization or business in which they were
placed. However, New Hope case management staff were informed about problems and could
intervene when necessary. Participants submitted weekly time cards in order to get paid, but
these cards also had space for the job site manager to comment on the participants’ performance. Occasionally, case managers or the transitional jobs site liaison met with the site
supervisors to discuss specific participants and issues that might have arisen. A regular meeting
with the participant, case manager, and job site supervisor was also scheduled after 30 and 90
days at the transitional job. Participants could be terminated from their jobs, just like a regular
employee, for issues related to performance, absenteeism, or other problems.
In the other programs, communication between the case managers and the transitional
jobs supervisors was more regular and direct. In Chicago, participants were supervised by plant
managers who reported back to case managers about issues with specific participants. Participants could receive a written notice if they exhibited improper workplace behavior, such as
tardiness or absenteeism. After three notices, participants could be suspended or possibly
terminated. In both Detroit and St. Paul, on-site supervisors provided direct oversight of the
transitional jobs work, but they communicated directly with case managers about problems or
concerns. Additionally, case managers were easily able to come to the work site directly to
observe or intervene, if needed.
Participants’ Views
Participants had varying assessments of the transitional jobs experience. Many focus
group participants saw value in their transitional jobs, since the jobs provided a work history
following their incarceration, gave relatively quick access to legitimate wages, and provided
participants with feelings of general self-worth. One participant in Milwaukee noted how the
transitional job allowed him to build a work history and keep forward momentum: “I really
needed a transitional job … to build that work history while I look for another job. So I think in
that sense it helped to keep me on the right track...just moving forward, trying to get a permanent job.” A participant in Detroit said that the transitional job got him used to the work environment again: “It got me back into the idea of getting up, going to work … being on time,
get[ting] along with other workers and … getting back into the state of mind … Basically, it’s a
mindset.” The participants also noted that having an income, even though minimum wage, was
important to them and gave them a sense of worth. One participant in Chicago said, “I’ve been
able to save money, and kind of hold my head up now because I know I’ve earned my pay …
When I … meet my grandchildren, I can … do a little something and feel proud about it.”

On the other hand, a number of participants found certain aspects of their jobs to be rather negative. One Chicago focus group participant implied that his transitional job employer
only wanted a “body” to do the job: “When you working there, you know they don’t care
anything about you. And they work you very hard ... I don’t mind being a hard worker. I like
stepping up making a living. I don’t mind working hard, but there some days they take it to the
limit.” A participant in the St. Paul focus group noted similarly the hard work required by the
job: “It’s almost like sweatshop type work ... They make your work real hard for little pay.
Yeah, they know that you’re just grateful to have a job.”
Transitional Jobs Challenges
One challenge faced by case managers and participants in this demonstration was learning how to deal with the local corrections system beyond issues related to recruitment. Those
who are released to community supervision often have many mandated requirements, such as
participating in substance abuse programs, taking anger management courses, and meeting with
parole officers. Additionally, those on supervision sometimes have curfews and restrictions on
travel; individuals on a specialized or intensive supervision caseload may face even more
mandates that require multiple meetings or “check-ins” with parole officers within a single
week. As discussed in Chapter 2, in St. Paul, most of the participants referred from halfway
houses were on intensive supervised release (ISR), a program for high-risk parolees. ISR
parolees had more frequent meetings with parole officers and more mandates than the average
parolee. Staff in the transitional jobs sites indicated that participants faced a challenge in
balancing the conditions of supervision with their transitional job responsibilities while at the
same time looking for permanent positions. Programs that had strong relationships with community corrections departments were able to help participants negotiate with their parole
officers to modify check-in times or other conditions around their work schedules. While TJRD
program staff generally did not have an issue with responding to specific requests from parole
officers for information about individual participants, some staff noted that they did not want to
be viewed as an extension of the corrections system.
In Detroit, the relationship between program staff and community corrections staff was
facilitated by the arrangement set up by the MPRI. All study participants (in both the transitional jobs and job search programs) were assigned parole officers who were designated specifically
to work with TJRD participants. Initially, one parole officer each was assigned to the transitional jobs and job search groups, but more were added as enrollment in the study increased. When
recruitment into the study occurred before prisoners were released, the parole officers met with
participants right before release, as part of the orientation to the study. The parole officer(s)
assigned to the transitional jobs group met subsequently with participants on the first and third
Tuesdays of each month on-site, often at the transitional jobs program office. When the timing

of recruitment switched to the post-release period, participants were switched to one of the
specialized parole officers once they were randomly assigned.
A second challenge of transitional jobs involved balancing goals. The goal of the transitional jobs programs was to provide people with work opportunities in a somewhat more
forgiving environment than they would otherwise experience in order to ease them back into the
workforce. But, businesses also have particular goals to attain. The sites encountered a number
of issues related to these dual goals.
As noted in the discussion of recruitment, the number of men entering the transitional
jobs group varied from one week to the next. Although the size of the weekly (or biweekly) inflow of participants did not necessarily strain the resources of case management staff, the
unpredictability in the number of participants who needed to be placed in a transitional job
posed some problems in terms of filling transitional jobs slots. A larger than usual number of
participants one week sometimes meant that some men had to wait for a transitional jobs
position to open in the contracted or existing enterprise settings. A lighter than usual number of
participants could leave some positions unfilled, particularly as other men were completing their
transitional job assignment. Yet, as one transitional jobs supervisor noted, the work of the
organization needed to be done. For example, the Chicago program had a contractual obligation
with a business to meet certain production goals. The business was concerned that the constant
fluctuation of program participants in and out of positions and the need to train the new staff
moving into these positions would affect the production goals for the organization. In the end,
production was not compromised by the turnover, but this “mission versus business challenge,”
as some termed it, sometimes caused tension.
Additionally, all the sites offered services to individuals other than participants in the
TJRD project, and in some cases, clients from these other programs needed to be placed into
transitional jobs. In those instances, different internal programs were sometimes competing with
each other for the same transitional jobs slots.
In a couple of sites, the ebb and flow of available work at the transitional jobs locations
meant that demand for workers fluctuated. In Milwaukee, if the businesses and nonprofit
organizations that served as transitional jobs sites did not need additional labor, some participants had to wait to start their transitional jobs assignment until other participants vacated those
positions. In Detroit, as noted earlier, the plant at Goodwill closed operations for two weeks
every summer, coinciding with the automotive industry’s re-tooling for new production contracts. Men who were randomly assigned during this time period had to wait to begin working,
while those who had already been assigned to the transitional job did not work or have a
paycheck for two weeks.

Job Development and Permanent Job Placement


The emphasis on permanent job search and placement was not a strong
focus throughout the duration of the demonstration, nor did staff use
participants’ experience working in a transitional job as a “selling
point” to potential employers.

The Transition to a Permanent Job
Transitional jobs are designed to be temporary jobs. The transitional jobs in this study
were meant to last approximately 90 days, although sites could keep participants in the job
longer if enough slots were available. In all cities, the decision about when to start moving
someone out of the transitional job was fairly individualized. In Chicago and Milwaukee,
participants were sometimes allowed to stay in a position beyond 90 days. In Chicago, in fact,
participants could and did continue in the transitional job as regular employees, eligible to
purchase medical benefits and moving into regular jobs at the recycling plant without needing to
search for another job. In Detroit, the transitional position ended after 90 days, but participants
could continue to use program services. In St. Paul, participants were initially allowed to remain
in a transitional job beyond 90 days, as long as they did not exhibit behavior problems, but by
the second research visit, staff had instituted a time limit because they believed some participants were becoming too complacent in the transitional job and were not searching for unsubsidized employment.
Staff in all sites monitored individuals’ behavior in the transitional job as an indicator of
when someone was ready to focus on searching for permanent work. Although participants in
Milwaukee were encouraged from the first day to begin searching for jobs, staff said that in
practice most participants waited to look for work until near the end of their time in the transitional job. A common challenge expressed by focus group participants was the difficulty in
searching for permanent employment while also working in the transitional job. One participant
in Detroit summed up the challenges in this way:
We work Monday through Friday from eight to one. Anybody that’s been in the job
field, anybody that has been in society and seen somebody else work knows, that if
you looking for a job, you need to be there. ASAP. First thing Monday morning. The
problem we run into, we getting these lists [of jobs], and we don’t even get these lists
until one o’clock. By the time we get the list, half the people that’s taking applications, they finished accepting the applications. I know if I’m an employer, looking to
potentially employ someone, I’m going for the individuals that’s standing outside
that door when I pulled up. Or the individuals that came right behind me in that door.
I’m not looking for the laggers that’s going to come in after lunch.

During the job search process, the programs offered some services to participants, at a
minimum helping them prepare for interviews and, in St. Paul, offering a structured “employ54

ment readiness” workshop, which covered the same set of topics that other sites covered earlier
in the process. St. Paul also offered the opportunity for participants to enroll in one of their skills
training programs, such as forklift operation or auto repair. Participants could also enroll in
other vocational training such as construction work, for which they received certificates that
may have given them an edge with employers.
Finally, the Milwaukee and St. Paul programs provided incentives to participants who
worked in permanent jobs after working in a transitional job. These programs paid retention
bonuses at preset intervals to participants who met the qualifications. Participants in Milwaukee
were eligible to receive up to $1,500 over 12 months if they provided proof of full-time employment to their case manager. St. Paul instituted a similar incentive program with a total of
$1,400 offered after six months of permanent employment. In reflecting on the program, the
Milwaukee program director was surprised by how important that retention bonus was to
participants, no matter how much they were actually making from their permanent jobs.
The Role of Job Developers
Many social service and employment assistance organizations have job developers on
staff to assist disadvantaged populations — such as those who were formerly incarcerated or
those with disabilities — with finding employment, including the organizations providing
transitional jobs services. While the title of the staff person varied from site to site, all sites had
at least one staff person whose job it was to assist clients in finding permanent jobs that paid a
living wage (some were called “job developers” or “sector managers”). Although there is little
evidence-based research to affirm the most important components of job development, at least
one study team has observed the importance of job development as an individualized and
person-centered approach that is based on the skills and interests of the job seeker and how
those skills and interests match the needs and workplace culture of an employer.4
In the TJRD project, job developers did not necessarily find positions that matched participants’ interests and skills, nor did they necessarily pursue employment opportunities that
would place participants on a long-term employment path. Furthermore, few job development
staff used participants’ transitional job experience as a “selling point” to potential employers.
Some staff felt that the brief duration of the placement — three months or less — did not
provide enough work experience to be an enticement to employers who were looking for stable
employees. On the other hand, other program staff believed that discussing participants’ time in
transitional jobs with employers was extremely valuable. Staff reported that sharing attendance
records, reports of on-the-job behavior, and assessments of soft skills, such as motivation,
4

Wyckoff and Clymer (2005).

helped employers get a sense of what they might be able to expect from the employee. However, job developers had a difficult time finding unsubsidized jobs for former prisoners. Seemingly as a result, they placed participants into whatever permanent positions became available.
Chicago handled job placement somewhat differently from the other cities. Once a
job developer had a lead on a potential job, all case managers in the organization (not just
those who were working with the demonstration program) were alerted. Staff informed
participants about available positions, determined whether they met the qualifications, and
helped them prepare for interviews. The employer then interviewed the participants and made
hiring decisions.
Permanent Placement Challenges
Perhaps the biggest challenged faced by the programs was placing participants in permanent positions after the transitional jobs. First, programs found it challenging to find employers who were willing to hire former prisoners in permanent positions. Additionally, as already
noted, some participants had restrictions on the types of jobs they could seek given their offense
histories, and employers were reluctant to hire people with certain conviction histories, such as
crimes involving violence. Even if a participant did not have such restrictions, lack of reliable
transportation was sometimes an impediment to finding and keeping a job. In Milwaukee, many
of the available jobs were located in the suburbs, which were not accessible by public transportation. The St. Paul staff supervisor noted that over time program staff became more knowledgeable about the special circumstances that were involved in placing former prisoners in
permanent jobs. Staff had also developed more connections with employers who were willing
to hire this population, but not until later in the follow-up period.
As noted in Chapter 2, the economic slowdown and subsequent recession played a role
in the ability of participants to secure jobs. In Detroit and Milwaukee in particular, certain
sectors of the low-skilled labor market shrank. For example, the construction industry, which
for a number of years had provided relatively well-paying entry-level positions for people,
declined, and instead of hiring new workers this industry was laying workers off. Staff also
noted that former prisoners were part of the “last hired, first fired” group of workers. Thus, for
places with few available jobs, former prisoners were not at the front of the hiring queue.
Beyond structural concerns, staff reported that a number of personal issues made
placing former prisoners into permanent jobs a difficult task. First, some staff reported that
some participants seemed disinterested in and lacked motivation to find a permanent position.
Second, staff believed some clients stayed in transitional jobs as long as possible because they
were not ready to move on to positions outside of the program. And perhaps most important,
staff reported that the issues facing former prisoners were very different from and more
challenging than those of other populations typically served in transitional jobs programs.
56

Further, problems such as lack of housing, substance abuse problems, and untreated mental
health problems were ones that employment alone — whether in a transitional or permanent
position — could not address.

Summary
Aside from the offer of a transitional job, the services available to the transitional jobs
and job search groups were not very different from each other. All participants had access to
assistance with developing résumés, instruction in job search techniques, and learning how to
discuss criminal histories, and all sites provided participants with job leads.
Although the time that elapsed between random assignment and placement into a transitional job varied between Milwaukee and the other three demonstration cities, all participants
who were assigned to the transitional jobs group were given the opportunity to work within a
relatively short time following their release from prison. Given the difficulty that former
prisoners face in finding jobs, it is likely that transitional jobs participants entered the workforce
(that is, through a transitional job) sooner after release than participants in the job search group.
Members in the job search group, if they participated in the activities offered by these programs,
would have been involved in more group activities or unstructured job search before entering
the workforce.
Despite the variation across the transitional jobs programs, in general, each of the
sites implemented the transitional jobs programs fairly quickly and without major problems.
That said, the transitional jobs sites, even if they had previous experience running similar
programs, had to devise new recruitment strategies, and most had to forge new relationships
with parole officers and the larger criminal justice system, particularly so that they could
enroll newly released men. Perhaps because of that, the sites put a great deal of effort into
ensuring that the “front end” of the operations — recruitment and placement into the transitional job — was operating smoothly, while less emphasis was placed on the “back end” —
placement into an unsubsidized job. Additionally, the lack of a job developer in Detroit for
part of the demonstration and the time that the Milwaukee job developer spent finding
transitional job placements in the community as opposed to unsubsidized positions (again,
during the first part of the demonstration) may have also contributed to a lack of emphasis on
placements relative to “front end” activities.

Chapter 4

Program Participation
Chapter 3 describes the services provided by each of the organizations that participated
in the Transitional Jobs Reentry Demonstration (TJRD). This chapter presents data on the rates
of participation in those services, both for transitional jobs (TJ) group members and for job
search (JS) group members. The analysis uses data from the management information system
(MIS) of each of the programs included in the study.
Because survey data about participation in services were not collected for TJRD, it is
not possible to know the extent to which sample members in both the TJ and JS groups received
services from sources other than the TJRD programs. Therefore, the numbers reported in this
chapter almost certainly underestimate service receipt for both groups.

Transitional Jobs Group Participation
Table 4.1 shows TJ group members’ participation in the transitional jobs programs,
both for the full sample and by site. The first row in the table shows that participation was high,
with 93 percent of TJ group members engaging in at least one program activity. These participation rates varied somewhat across sites, ranging from 88 percent in Milwaukee to 99 percent
in St. Paul, but participation was high in all sites. As described below, this variation in participation rates may have been related to differences in the time it took sites to place participants into
transitional jobs.


The rate of participation in transitional jobs was high.

The second panel of the table shows that participation in the transitional jobs, the key
component of the program, was high, though the rate varied across sites. Overall, 85 percent of
TJ group members worked in a transitional job. This is high compared with the rate of 72
percent for TJ participation in the evaluation of the Center for Employment Opportunities
(CEO), a transitional jobs program for former prisoners in New York City (see chapters 1 and 9
for more details about the CEO evaluation).1
Transitional jobs participation was lowest in Milwaukee, where 70 percent of TJ group
members worked in a transitional job — comparable to the CEO participation rate of 72

Redcross et al. (2009).

The Transitional Jobs Reentry Demonstration

Table 4.1
Participation in Transitional Jobs and Other Activities for the
Transitional Jobs Group, by Site
Outcome

Chicago

Participated in any activity (%)

Detroit Milwaukee St. Paul

Full
Sample

92.6

91.3

87.5

99.2

92.7

87.3

87.7

69.9

97.2

85.2

Weeks worked in a TJ (%)
Never worked
Less than 1 week
1-4 weeks
5-8 weeks
9-12 weeks
13-24 weeks
More than 24 weeks

12.7
11.1
20.1
18.0
12.7
12.2
13.2

12.3
4.1
23.1
23.6
33.3
2.6
1.0

30.1
2.3
17.2
14.5
10.9
25.0
0.0

2.8
4.4
20.6
16.6
16.6
28.9
10.3

14.8
5.2
20.0
17.8
17.8
18.5
5.9

Average weeks worked a

10.3

6.7

7.0

12.0

9.0

Among those who worked in a TJ
Average weeks worked

11.7

7.7

10.0

12.3

10.6

83.6
55.6
66.7
NA

91.3
73.8
89.2
NA

79.7
66.0
40.6
NA

99.2
96.0
77.5
28.1

88.6
74.0
67.2
NA

Received a support service paymentb (%)

89.2

53.9

89.7

76.6

Placed in or reported an unsubsidized job (%)

25.6

38.3

32.8

Unsubsidized job characteristics
Average starting hourly wage ($)
Hours per week
Employer provided benefits (%)

NA
NA
NA

8.5
38.3
22.0

8.6
37.3
14.1

10.1
39.8
66.3

9.2
38.4
36.5

Received a retention bonusc (%)

45.9

76.3

54.4

Among those who received a retention bonus
Number of retention bonuses received
Total amount received in bonuses ($)

NA
NA

3.6
484

4.3
827

3.9
618

Transitional jobs
Worked in a transitional job (TJ) (%)
a

Other program activities
Participated in any activity other than TJ (%)
Had an individual meeting with a case manager
Participated in job readiness/job club activities
Participated in vocational training

Among those who were placed in or reported
an unsubsidized job

(continued)

Table 4.1 (continued)
SOURCES: MDRC calculations using payroll and MIS data provided by each of the programs.
NOTES: This table includes only participation activities that took place within one year of random assignment.
NA indicates that data were not available for that site.
aAverage weeks worked is created by dividing the total number of hours worked by the number of hours in a
work week (30-40) for each site. Weeks may not be consecutive.
bSupport service payments included cash payments and vouchers for transportation, education or vocational
training, clothing, food, medical bills, identification cards, rent, utility bills, and other expenses.
cCalculation includes only data from New Hope (Wisconsin) and Goodwill/Easter Seals (Minnesota), where
retention bonuses were offered. Calculation for Goodwill/Easter Seals does not include those who were randomly
assigned before December 1, 2007, when the program began offering retention bonuses.

percent, but lower than the rate in the three other TJRD sites. This lower rate may reflect dropoff from the program while participants were waiting to be placed in a transitional job, which
took a relatively long time in Milwaukee. As described in Chapter 3, the transitional jobs model
in Milwaukee was a scattered site model, in which TJ participants were placed in temporary
positions with nonprofit organizations and small businesses in the community. The process of
placing participants in those positions usually took about two weeks, compared with less than
one week in the other sites. Program staff had to first identify a transitional job that was suitable
for a participant, and once such a job was found, participants were required to interview for the
position. It is possible that difficulties making placements or participants’ dropping out of the
program while that process was occurring led to the lower TJ participation rate in Milwaukee.
In contrast, in St. Paul, placement in the transitional jobs occurred almost immediately,
on the day after random assignment. This is most likely why the transitional jobs participation
rate was so high in St. Paul, with 97 percent of TJ group members working for at least one day
in a transitional job. In Chicago and Detroit, placements were usually made within one week of
random assignment; in Detroit, participants attended three days of other program activities
before placement, and in Chicago, transitional jobs slots were not always immediately open but
were usually open within one week. These small delays in placement may explain why participation rates in Chicago and Detroit, at 87 percent and 88 percent, respectively, were between
the St. Paul and Milwaukee placement rates.
Table 4.1 shows that TJ group members typically spent one to three months in transitional jobs; 56 percent worked for between one and twelve weeks. About one-fourth stayed
longer in the transitional job, working for more than 12 weeks. Overall, including zeroes for
those who did not work, program group members averaged nine weeks (45 days) working in a
transitional job (based on a five-day work week). Among those who worked, the average was
10.6 weeks (53 days). These numbers are an estimate of TJ “dosage” rather than a measure of
consecutive weeks worked. They were calculated, using payroll data, by dividing the total
number of hours worked by the number of hours in a work week for each site. Therefore, these
61

weeks may not have been consecutive. For example, a sample member may have worked in a
transitional job for five weeks, left the program for two weeks, and returned to work for another
two weeks; that participation would be calculated as a “dosage” of seven weeks in the transitional job. Thus, these measures estimate the amount of time spent working in a transitional job
rather than the time elapsed between starting the program and exiting the program.2 Additional
analysis indicates that, among those who worked, the 10.6 weeks of work in transitional jobs
occurred over the course of about four months of engagement with the program.3
Time spent working in a transitional job was longest in the Chicago and St. Paul sites,
where participants who worked in a transitional job averaged about 10 to 12 weeks. The St.
Paul program site tended to allow participants to stay longer (or to return to the program if they
had left previously), while in Chicago, participants were sometimes encouraged to stay longer
in the transitional jobs.4 In those two sites, 13 percent of the participants in Chicago and 10
percent in St. Paul were in the transitional jobs for more than 24 weeks.


A large percentage of transitional jobs group members participated in
at least one other program activity, including case management, job
readiness activities, job search preparation, job development, and vocational training.

The second panel in Table 4.1 shows participation in activities other than the transitional job. Participation in those services was also high, with nearly 90 percent of TJ group members participating in at least one other activity. These activities included meetings with a case
manager, job readiness and job club activities — including job club and other group job
readiness classes, individual meetings about job search preparation and soft skills, and meetings
with a job developer — and vocational training. About three-fourths of TJ group members
participated in case management services, while about two-thirds participated in at least one
type of job readiness or job club activity.5 These rates indicate that there was some degree of
overlap, such that most participants were involved in more than one type of activity. Unfortu2

For more information about the timing of transitional job employment, relative to random assignment,
see Chapter 5, which presents quarterly measures of transitional job employment.
3
The four months of engagement with the program do not include any contact that participants had with
the programs after being placed in an unsubsidized job or dropping out of the transitional job. Thus, those four
months do not include retention services related to the bonuses in St. Paul and Milwaukee (see Chapter 3).
4
In the Chicago site, Safer considered some participants to be “placed” in a permanent job after they had
worked for 90 days in the transitional job, as they were allowed to stay with their job at Allied.
5
There was a good deal of variation across sites in the available data and in the structure of the program
components — for example, whether job development was done primarily in a group or individual setting. In
order to create a comparable measure across sites, it was necessary to combine the measures of job readiness,
job club, and job development activities into one category.

nately, data on the number of sessions or the amount of time participants spent in these other
activities was not available consistently across sites.
As with work in the transitional jobs, participation in other activities was highest in St.
Paul, at 99 percent overall. Nearly all TJ group members, 96 percent, were involved in case
management services, while 78 percent participated in at least one job readiness or job club
activity. In addition to case management and job readiness services, St. Paul offered forklift,
construction, and automotive skills training to certain participants. Twenty-eight percent of TJ
group members in St. Paul participated in at least one of these types of training.6
In addition to case management, job readiness, and job search services, all the TJ programs offered support service payments, which included cash payments or vouchers for
transportation, education or vocational training, clothing, food, medical bills, identification
cards, rent, utility bills, and other expenses. Among TJ group members in Detroit, St. Paul, and
Milwaukee (data were not available for Chicago), 77 percent received at least one support
service payment.


According to program records, about one-third of TJ group members
made the transition into an unsubsidized job.

According to program records, about 33 percent of TJ group members obtained an unsubsidized job. This measure includes direct placements by program staff and jobs that were
reported to staff by participants. Since these jobs include only those known to the program, they
are not a complete measure of the jobs obtained by TJ group members. However, this measure
provides some information about the proportion of TJ group members who were able to move
into unsubsidized positions with the help of, or while they were supported by, the program.
Although 93 percent of TJ group members participated in the programs and 85 percent worked
in a transitional job, only one-third obtained an unsubsidized job known to the program. Data
from unemployment insurance (UI) records, which provide a more complete picture because
they include jobs that were not reported to program staff, show similar results. Thirty-eight
percent of TJ group members obtained an unsubsidized job in the same quarter or in the quarter
after they worked in a transitional job (not shown in table).
The bottom panel of Table 4.1 shows the job characteristics that were reported by those
who were placed in unsubsidized jobs. The mean wage for these jobs was $9.16 an hour and the
jobs were mostly full time, averaging 38 hours per week. A little more than one-third of the jobs
6

The Milwaukee program referred participants to an outside organization that provided forklift training.
Among Milwaukee TJ group members, 13.7 percent reported to their case managers that they were participating in that training.

offered benefits of any kind, including paid sick days, paid vacation days, health insurance, or
other benefits.
The highest rates of placement in or reporting of unsubsidized jobs were in the two sites
that offered retention bonuses for remaining employed in an unsubsidized job: Milwaukee,
where 38 percent were placed, and St. Paul, where 33 percent were placed. Once they obtained
unsubsidized employment, participants could receive bonuses on a monthly basis if they provided proof of employment, for up to 12 months in Milwaukee and up to 6 months in St. Paul
(see Chapter 3 for more information). Among participants who obtained an unsubsidized job, 46
percent in Milwaukee and 76 percent in St. Paul received a bonus.7 Among those who received a
bonus, the average number of bonuses in both sites was four; in total, those in St. Paul received
about $830 on average while those in Milwaukee received about $485 on average.
It may be that the relatively high rate of unsubsidized job placement in those two sites
was related to the bonuses, either because participants had more incentive to obtain employment
or because they had more incentive to report employment that they obtained on their own.
However, it is not clear whether bonuses had any effect on placement numbers, as a number of
other factors, including the local job market or sample characteristics, could also have affected
job placement or reporting.

Job Search Group Participation
Table 4.2 shows JS group members’ participation in the job search programs for the full
sample and by site. As observed earlier, it is not possible to know the extent to which JS group
members received services from sources other than the TJRD job search programs because
survey data were not collected for TJRD. Therefore, the participation rates presented in this
section almost certainly underestimate the services that job search group members received. In
particular, it is possible that some JS group members may have accessed transitional jobs from
other programs; however, information gathered from other transitional jobs providers suggests
that very few JS group members worked in a transitional job.8


Nearly two-thirds of JS group members received at least one service
from the job search program in their city.

For St. Paul, the calculation of the percent receiving a bonus includes only those who were randomly
assigned after December 1, 2007, when the program there began offering the bonuses.
8
Several of the major providers of transitional jobs in the four sites provided the research team with information or data.

The Transitional Jobs Reentry Demonstration

Table 4.2
Participation in Program Activities for the Job Search Group, by Site
a
Chicago Detroit

Outcome
Participated in any activity (%)
Had an individual meeting with a case manager
Participated in job readiness/job club activities

Milwaukee

St.Paul

Full
Sample

59.5
36.2
45.4

53.9
44.0
47.4

72.0
NA
NA

66.4
NA
NA

63.8
39.2
46.2

Received a support service paymentb (%)

61.2

32.0

41.2

Sample size

185

193

250

253

881

SOURCES: MDRC calculations using management information system data provided by each of the programs.
NOTES: This table includes only participation in activities that took place within one year of random assignment.
NA indicates that data were not available for that site.
aIn Detroit, the job search service provider changed during the study. Participants who were randomly
assigned through December 31, 2007, were served by JVS; those who were randomly assigned after that date were
served by the Detroit Hispanic Development Corporation (DHDC). Aside from "Participated in any activity,"
calculations for Detroit do not include data for DHDC, from which information was not available. For all other
variables, calculations include only those who were randomly assigned before January 1, 2008.
bSupport service payments included cash payments and vouchers for transportation, education or vocational
training, clothing, food, medical bills, identification cards, rent, utility bills, and other expenses.

Table 4.2 shows that about 64 percent of JS group members participated in at least one
service at the job search program. The participation rates ranged from 54 percent in Detroit to
72 percent in Milwaukee. It is not clear why participation varied. In Chicago and Detroit, the
two sites that provided detailed participation information, about 40 percent of the total JS group
sample attended an individual case management meeting and about 46 percent participated in at
least one job readiness or job club activity, including job development. Among participants in
Detroit and St. Paul, 41 percent received a support service payment.

Conclusion
TJ group participation in the transitional jobs programs was high; 93 percent of TJ
group members participated in a program activity and 85 percent worked in a transitional job.
There was some variation across sites in the rates of participation, with participation generally
highest in St. Paul and lowest in Milwaukee, but in all sites, a large percentage of TJ group
members worked in a transitional job.
Programs reported that one-third of program participants obtained an unsubsidized job
that was known to the program. This suggests that although 85 percent of TJ group members
worked in a transitional job, only one-third made the transition directly or indirectly into an
65

unsubsidized job. However, these data may be incomplete, as participants may have obtained
jobs that were unknown to the program. Still, UI records show only a slightly higher percentage,
38 percent, making the transition to an unsubsidized job. Chapter 5 presents more comprehensive employment results using UI wage records.
Participation by JS group members in the job search programs was low relative to the
participation of TJ group members in the transitional jobs programs. Sixty-four percent of JS
group members participated in at least one service at the job search program, while 89 percent
of TJ group members participated in similar services at the transitional jobs program. However,
it is possible that JS group members also accessed services at other programs that are not
measured here. Nonetheless, available evidence suggests that very few job search group
members participated in a transitional job, which is the key treatment tested in this study; thus,
as expected, the transitional jobs themselves were the core feature of the treatment difference
across sites.

Chapter 5

Impacts on Employment and Earnings
The four transitional jobs programs in the Transitional Jobs Reentry Demonstration
(TJRD) aimed to increase the employment and earnings of former prisoners by placing transitional jobs (TJ) group members in immediate, wage-paying transitional jobs. It is believed that
transitional jobs can be beneficial because they provide participants with legitimate income
during the critical period just after release while also helping them gain basic skills and work
experience. Because the transitional jobs provided participants with an incentive to actively
engage in program services, program staff had ample opportunity to observe and address
workplace behaviors that may hinder future employment success.
The job search (JS) programs, which served the job search group, also sought to help
participants find and keep unsubsidized (that is, regular) employment. Unlike the transitional
jobs programs, however, these programs did not offer subsidized jobs to participants while they
looked for employment. Thus, they reflect what would have happened in the absence of the
transitional jobs programs and so provide a basis for comparison.
This chapter presents the one-year impacts of the study’s transitional jobs programs on
employment and earnings. Outcomes for TJ group members are compared with outcomes for JS
group members, and are described in terms of increases and reductions. (For example, a finding
that the TJ program did not increase unsubsidized employment means that the members of the
transitional jobs group were no more likely to find unsubsidized employment than were the job
search group members.) The difference between the two research groups on measures of
employment and earnings represents the impact of the transitional jobs programs on these
outcomes. (See Box 5.1 for a detailed explanation of how to read the impact tables in this
report.)
This analysis uses payroll data from each of the transitional jobs programs and unemployment insurance (UI) data from each of the four states in the demonstration. Notably, the
employment and earnings data in this chapter differ from the data that are reported about job
placement in Chapter 4, which were drawn from each program’s management information
system (MIS) and include only job placements made by or reported to program staff.
The ultimate goal of the TJRD programs was to increase the unsubsidized employment
and earnings of participants, so it is useful to examine employment and earnings results for
transitional jobs and unsubsidized jobs separately. The term “transitional job” in this study
refers to the transitional jobs that the TJRD programs provided to TJ group members. All other

Box 5.1

How to Read the Impact Tables in this Report
Most tables in this report use a similar format, illustrated below. Several employment and earnings outcomes are shown for the transitional jobs group and the job search group. For example,
about 72 (72.1) percent of the transitional jobs group and about 32 (31.6) percent of the job
search group were employed in Quarter 1.
The “Difference” column in the table shows the differences between the two research groups’
earnings and employment rates — that is, the transitional jobs program’s estimated impact on
employment. The estimated impact on employment in Quarter 1 can be calculated by subtracting
31.6 percent from 72.1 percent, yielding a 40.5 percentage point difference.
Differences marked with asterisks are “statistically significant,” meaning that it is quite unlikely
that the differences arose by chance; that is, they are likely attributable to the program. The
number of asterisks indicates whether the estimated impact is statistically significant at the 10
percent (one asterisk), 5 percent (two asterisks), or 1 percent (three asterisks) level — and the
lower the level, the less likely that the impact is a result of chance. For example, as shown in the
second row of data, the transitional jobs group model had a statistically significant impact of 32.9
percentage points on employment in Quarter 2. This impact is statistically significant at the 1
percent level — meaning that there is a 1 percent chance that the impact on employment in
Quarter 2 occurred by chance rather than as a result of the program. The p-value shows the exact
level of significance.

One-Year Impacts on Employment and Earnings
Transitional
Jobs Group

Job Search
Group

Difference
(Impact)

Total employment (%)
Quarter 1 (quarter of random
assignment)
Quarter 2

72.1
74.9

31.6
42.0

40.5
32.9

***
***

0.000
0.000

Total earnings ($)
Quarter 1 (quarter of random
assignment)
Quarter 2

789
1,467

307
844

482
624

***
***

0.000
0.000

Outcome

P-Value

SOURCES: MDRC calculations using payroll data from each site and unemployment insurance wage
records from each of the states in the Transitional Jobs Reentry Demonstration (Illinois, Michigan,
Minnesota, and Wisconsin).

UI-covered jobs that were not the TJRD transitional job are described as “unsubsidized” or
“regular” jobs in this report.1 A job was considered “unsubsidized” if the federal employer
identification number (FEIN), or tax ID, on the UI data file was different from the FEIN for the
transitional job employer in that site.2

Impacts for the Full Sample
Table 5.1 presents the impacts of the transitional jobs programs on employment and
earnings for each of the four quarters of the one-year follow-up period, for all sites combined.
Results are shown separately for transitional jobs, unsubsidized employment, and total employment (which combines transitional jobs and UI-covered unsubsidized jobs).
The first section of the table shows the proportion of the sample that worked in a transitional job for at least one day during each of the four follow-up quarters.3 As expected,
almost all (85 percent) of the TJ group members worked in a transitional job during the first
year, while none of the JS group members accessed a TJRD transitional job during the followup period.4 Chapter 4 reports that most participants worked in the transitional job for about 11
weeks, which most often occurred during the first two quarters after random assignment.5 By
Quarter 3, 26 percent of TJ group members worked in the transitional job; this decreased to 9
percent in Quarter 4.

Because some subsidized or transitional employment is reported in UI records, it is possible that a small
number of these “unsubsidized” jobs were transitional jobs provided by other organizations in the sites’
communities.
2
The Detroit and St. Paul sites reported that state laws did not require the reporting of transitional jobs
earnings to the state UI program and thus those earnings did not appear in UI data. The Chicago and Milwaukee sites’ transitional job earnings did appear in UI data; in those sites, the FEIN was used to distinguish which
jobs were the TJRD transitional job and which were not. All employment records that were not associated with
the employer ID of the transitional job were coded as “unsubsidized” in the analysis. Because two of the four
TJRD transitional jobs programs reported to UI and two did not, payroll data were used to identify transitional
employment for consistency across the four sites. The UI records in the two sites that reported to UI matched
very closely to payroll data, and so this analysis strategy does not affect the outcomes or impacts of the study.
3
Unlike the information about “weeks worked in a transitional job” that was reported in Chapter 4, this
table does not measure the dosage or the number of weeks worked in the transitional job.
4
While it is possible that some sample members accessed transitional jobs at other organizations in the
sites’ communities, it is likely to be a very small number. As noted in Chapter 4, several of the major providers
of transitional jobs in the four sites provided the research team with information or data that indicated that very
few JS group (and TJ group) members accessed transitional jobs from these other programs.
5
The information reflecting weeks worked in a transitional job that is presented in Chapter 4 may not be
consecutive and is not in “calendar” time. It is created by taking the total hours worked and dividing by the
number of work hours in a week. Thus, the timing of the “weeks worked” in Chapter 4 does not correspond
exactly to the quarterly data presented in this chapter.

The Transitional Jobs Reentry Demonstration

Table 5.1
One-Year Impacts on Employment and Earnings: Full Sample
Outcome

Transitional
Jobs Group

Job Search Difference
Group
(Impact)

P-Valuea

Employment
Transitional employmentb (%)
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

66.1
65.2
25.5
9.1
85.2

0.0
0.0
0.0
0.0
0.0

66.1
65.2
25.5
9.1
85.2

***
***
***
***
***

0.000
0.000
0.000
0.000
0.000

Unsubsidized employment (%)
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

16.6
25.9
30.0
28.8
50.3

31.7
41.8
34.3
31.4
59.2

-15.1
-15.9
-4.4
-2.5
-8.9

***
***
**
***

0.000
0.000
0.045
0.238
0.000

Total employment (%)
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

72.1
74.9
47.9
35.5
92.8

31.6
42.0
34.3
31.3
59.5

40.5
32.9
13.5
4.2
33.2

***
***
***
*
***

0.000
0.000
0.000
0.057
0.000

Number of quarters employed

2.3

1.4

0.9 ***

0.000

Employed in all 4 quarters (%)

20.2

11.6

8.6 ***

0.000

Transitional job earningsb
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

592
986
344
122
2,044

0
0
0
0
0

592
986
344
122
2,044

***
***
***
***
***

0.000
0.000
0.000
0.000
0.000

Unsubsidized earnings
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

197
481
757
857
2,292

307
841
902
866
2,917

-110
-360
-145
-9
-625

**
***
*

0.016
0.000
0.098
0.926
0.013

Earnings ($)

(continued)

Table 5.1 (continued)
Outcome

Transitional Job Search Difference
Jobs Group
Group
(Impact)

Total earningsc
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)
Sample size (total =1,774)

789
1,467
1,101
979
4,336

307
844
902
865
2,917

893

881

482
624
199
113
1,419

P-Valuea
***
***
**
***

0.000
0.000
0.026
0.260
0.000

SOURCES: MDRC calculations using payroll data from each site and unemployment insurance (UI) wage
records from each of the states in the demonstration (Illinois, Michigan, Minnesota, and Wisconsin).
NOTES: Statistical significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10 percent.
Results in this table are regression-adjusted for pre-random assignment characteristics.
aStandard errors are presented in this table for all impacts with a p-value of 0.000. Following are the
standard errors for all impacts with a p-value of 0.000 (presented in the order in which they appear on the
table). Employment: 1.542, 1.609, 1.468, 0.97, 1.177, 1.97, 2.19, 2.282, 2.11, 2.182, 2.263, 1.834, 0.06, and
1.708. Earnings: 21.737, 35.964, 26.895, 17.052, 69.527, 73.298, 49.66, 77.552, and 256.833.
bEarnings and employment from transitional jobs are based on payroll data from each of the sites.
cTotal earnings may not be equal to the sum of transitional job earnings plus unsubsidized job earnings due
to rounding.

The second section of Table 5.1 shows that TJ group members had lower rates of unsubsidized employment compared with JS group members in three of the four quarters of
follow-up. This reduction occurred because some TJ group members substituted the transitional
job for regular employment. If program staff believed that an individual could find a job on his
own, they tried to get that person to self-select out of enrolling in the program, but it is not
always easy to identify who really needs a transitional job among a large group of very disadvantaged men. The impact results suggest that some men volunteered for the study, and when
they were placed in the JS group, they went out and found a job on their own or with help from
the JS organizations. Alternatively, TJ group members were provided with a transitional job
immediately, and therefore some may have delayed seeking employment in the regular labor
market.
Substitution of transitional jobs for regular employment has been found in previous
evaluations of programs that provided a wage-paying transitional or community service job to
participants.6 The Center for Employment Opportunities (CEO) evaluation, which is most
similar to the present study, found some substitution of the transitional job for regular employment, but only very early in the study’s follow-up. The substitution in TJRD may have resulted
6

Redcross et al. (2009); Bos et al. (1999).

from the fact that participants had little time to seek employment on their own before being
provided with the transitional job. By design, study participants were recruited and enrolled in
the programs very shortly after being released from prison and in most cases started working in
the transitional job almost immediately upon enrollment. Nonetheless, the TJRD transitional
jobs programs did not increase the unsubsidized employment of TJ group members relative to
their JS counterparts at any point in the one-year follow-up period. Even as TJ group members
left the transitional jobs by Quarter 4, their rates of unsubsidized employment were not higher
than those of JS group members.


The TJRD transitional jobs programs generated a large increase in employment driven by the transitional jobs. This increase faded by the end
of the first year.

The third section of Table 5.1 shows that TJ group members had substantially higher
rates of total employment than JS group members. This large increase in employment was due
entirely to the transitional jobs provided to TJ group members. Nearly all TJ group members
worked at least one day during the year (92.8 percent) compared with about 60 percent of the JS
group, for a 33 percentage point increase in employment. During most quarters of the follow-up
period, only a little more than 30 percent of JS group members were employed; this represents
what employment levels would have looked like for TJ group members in the absence of the
transitional jobs programs. Low levels of employment are common for reentry populations;
similar quarterly employment rates were found among sample members in the CEO evaluation.7
The transitional jobs programs paid participants the state minimum wage for working in
the transitional job. During the follow-up period for the study, the minimum hourly wage
ranged from a low of $6.15 in St. Paul to a high of $8.00 in Chicago. The minimum wage
increased in all four cities during the follow-up period.8 The bottom panel of Table 5.1 shows
the impacts of the transitional jobs programs on earnings and is presented in the same format as
the employment results discussed above. The table shows that TJ group members earned about
$1,400 more than their JS group counterparts; this increase was driven by earnings from the
transitional jobs. The TJ group earned about $625 less, on average, from unsubsidized jobs than
the JS group. This reduction in unsubsidized earnings occurred because of the substitution of
transitional jobs for regular employment discussed earlier.

Redcross et al. (2009).
As a result of the Fair Minimum Wage Act of 2007, the federal government increased the federal minimum wage during the follow-up period. Where an employee is subject to both the state and federal minimum
wage laws, the employee is entitled to the higher of the two minimum wages. The minimum wage rate in
Chicago and Detroit is higher than the federal minimum; in Milwaukee it is the same; and in St. Paul it is lower
than the federal minimum.
8

Figure 5.1 presents quarterly employment rates over the follow-up period. This figure
provides an illustration of the results presented in the total employment section of Table 5.1 and
shows TJRD’s transitional employment combined with unsubsidized employment. The figure is
a useful illustration of how the patterns of employment changed over time. Differences that are
statistically significant are indicated by asterisks next to the quarter number at the bottom of the
graph. As reported above, the figure shows that the TJ group was much more likely than the JS
group to be employed throughout the early part of the follow-up period. The impacts on
employment faded over time and were much smaller in the last quarter of the follow-up period,
as almost all TJ group members had left the transitional jobs by that time.
One interesting point about the pattern of employment over the follow-up period is that
the rates of employment increased early on and declined somewhat in the second half of the
year for both research groups. While this pattern is expected for the TJ group (as they leave
their transitional jobs), it is not a typical pattern for the JS group. The biggest drop-off in
employment occurred between Quarter 2 and Quarter 3. This pattern suggests that some JS
group members were able to find jobs soon after release, but lost them quickly. As discussed in
Chapter 2, it is possible that declining economic conditions during the follow-up period
contributed to this pattern. A subgroup analysis (discussed below) examines the employment
outcomes separately for those who enrolled later and had most or all of their follow-up during
the worst part of the recession.

Impacts for Sites
While the four programs all operated a transitional jobs model, there were key differences in the implementation strategies across the four sites, as described in Chapter 3. Also,
Chapter 2 highlighted some differences in the local economic and criminal justice conditions in
each site. This section explores whether differences in the local context or differences in the
implementation of the program models translated into differences in the impacts of the transitional jobs programs on employment and earnings. Impact results shown by site should be
interpreted with caution because the sample sizes are smaller.


Despite differences in implementation strategies and local contexts, the
impact of the transitional jobs programs on employment and earnings
was similar across sites.

Table 5.2 presents the impacts of the transitional jobs programs on employment and
earnings for each of the sites separately. In general, the employment and earnings results are
similar across the sites. All four transitional jobs programs generated large and statistically
significant increases in total employment during the first year. In three of the four sites, the

The Transitional Jobs Reentry Demonstration

Figure 5.1
Quarterly Impacts on Employment: Full Sample

100
90

Transitional Jobs
Group
Job Search Group

Percentage Employed

70
60
50
40
30
20
10
0
Pre-RA

1 ***

2 ***

3 ***

Quarter after Random Assignment (RA)

The Transitional Jobs Reentry Demonstration

Table 5.2
One-Year Impacts on Employment and Earnings, by Site

Outcome

Transitional Job Search Difference
Jobs Group
Group
(Impact)

P-Valuea

Difference
Between
Site Impacts

Chicago
Transitional employment in Year 1b (%)
Unsubsidized employment in Year 1 (%)
Total employment in Year 1 (%)
Transitional job earnings in Year 1b ($)
Unsubsidized earnings in Year 1 ($)
Total earnings in Year 1c ($)

87.5
41.5
93.4

0.0
54.4
53.8

87.5 ***
-12.9 **
39.6 ***

0.000
0.016
0.000

†††
††
†††

2,471
1,829
4,300

0
2,369
2,383

2,471 ***
-540
1,918 ***

0.000
0.255
0.000

†††
††

88.0
38.5
92.4

0.0
33.1
34.7

88.0 ***
5.4
57.7 ***

0.000
0.284
0.000

†††
††
†††

1,461
1,700
3,161

0
1,252
1,245

1,461 ***
448
1,915 ***

0.000
0.389
0.000

†††
††

70.2
58.1
87.7

0.0
66.5
67.4

70.2 ***
-8.3 *
20.4 ***

0.000
0.055
0.000

†††
††
†††

1,831
2,906
4,737

0
3,004
2,986

1,831 ***
-98
1,751 ***

0.000
0.854
0.001

†††
††

97.1
59.1
98.7

0.0
74.5
74.0

97.1 ***
-15.5 ***
24.7 ***

0.000
0.000
0.000

†††
††
†††

2,410
2,582
4,992

0
4,394
4,386

2,410 ***
-1,813 ***
606

0.000
0.000
0.240

†††
††

Detroit
Transitional employment in Year 1b (%)
Unsubsidized employment in Year 1 (%)
Total employment in Year 1 (%)
Transitional job earnings in Year 1b ($)
Unsubsidized earnings in Year 1 ($)
Total earnings in Year 1c ($)
Milwaukee
Transitional employment in Year 1b (%)
Unsubsidized employment in Year 1 (%)
Total employment in Year 1 (%)
Transitional job earnings in Year 1b ($)
Unsubsidized earnings in Year 1 ($)
Total earnings in Year 1c ($)
St. Paul
Transitional employment in Year 1b (%)
Unsubsidized employment in Year 1 (%)
Total employment in Year 1 (%)
Transitional job earnings in Year 1b ($)
Unsubsidized earnings in Year 1 ($)
Total earnings in Year 1c ($)

(continued)

Table 5.2 (continued)
SOURCES: MDRC calculations using payroll data from each site and unemployment insurance wage records from
each of the states in the demonstration (Illinois, Michigan, Minnesota, and Wisconsin).
NOTES: Statistical significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10 percent.
Results in this table are regression-adjusted for pre-random assignment characteristics.
The H-statistic was used to test for statistically significant differences in impact estimates across sites.
Statistical significance levels are indicated as follows: ††† = 1 percent; †† = 5 percent; † = 10 percent.
aStandard errors are presented in this table for all impacts with a p-value of 0.000. Following are the standard
errors for all impacts with a p-value of 0.000 (presented in the order in which they appear in the table). Chicago
employment: 2.537 and 4.201. Chicago earnings: 232.153 and 500.667. Detroit employment: 2.494 and 4.169.
Detroit earnings: 92.587 and 531.609. Milwaukee employment: 2.982 and 3.703. Milwaukee earnings: 119.554.
St. Paul employment: 1.067, 4.206, and 2.912. St. Paul earnings: 113.327 and 507.815.
bEarnings and employment from transitional jobs are based on payroll data from each of the sites.
cTotal earnings may not be equal to the sum of transitional job earnings plus unsubsidized job earnings due to
rounding.

increase in employment was driven entirely by the transitional jobs: in Chicago, St. Paul, and
Milwaukee, the TJ group had lower rates of unsubsidized employment than the JS group during
the one-year follow-up period.
Three of the four sites (Chicago, Detroit, and Milwaukee) increased total earnings by
amounts ranging from $1,750 to $1,900. In Chicago and Milwaukee, the impact on earnings
was driven entirely by the transitional jobs. In Detroit, about one-fourth of the earnings impact
($448) was from unsubsidized employment (though the impact on unsubsidized earnings was
not statistically significant). The St. Paul transitional jobs program did not have an impact on
earnings during the first year. Appendix Tables C.1 through C.4 present quarterly employment
and earnings results for each of the sites.
Chapter 2 discusses some differences in local employment conditions that may have affected the ability of both TJ and JS group members to find and keep unsubsidized jobs. For
example, local unemployment rates were particularly high in Detroit and lower in St. Paul and
Milwaukee during the study’s follow-up period. The table shows that many JS group members
found employment fairly quickly in St. Paul, even without having access to a transitional job.
(About 53 percent of JS group members worked in Quarters 1 and 2, as shown in Appendix
Table C.3). The relatively high rates of employment for the JS group in St. Paul (compared with
the other three sites) may partially explain the larger substitution effects in that city (shown in
the fourth panel of Table 5.2). However, the economy may not be the only factor affecting
employment rates in St. Paul: the job search program that served JS group members (Amherst
Wilder Foundation) may have been particularly strong, or the large number of participants who
were under intensive supervised release in that site may have been under more pressure to seek

employment.9 Also, as discussed in Chapter 3, the transitional jobs program in St. Paul encouraged or required participants to work in transitional jobs for a certain amount of time before
recommending that they begin to focus on moving into regular employment. This approach led
to longer stays, on average, in the transitional job.
Detroit, the site with the lowest rates of employment among JS group members, was the
only site in which the JS group did not have higher rates of unsubsidized employment than the
TJ group during the first year. In fact, as the second panel of Table 5.2 shows, TJ group members in Detroit had slightly higher rates of unsubsidized employment in Year 1 than their JS
group counterparts (though the difference of 5 percentage points is not statistically significant).
In addition to a severely troubled economy, the Detroit program also had limited availability in
transitional jobs slots and thus had to move participants out of the transitional jobs quickly. The
economy, the shorter stays in the transitional jobs, or other factors could have contributed to
Detroit’s lower rate of substitution.
Similar to St. Paul, the Chicago program also put less emphasis on helping participants
make a quick transition into regular employment. In some cases, participants were encouraged
to stay longer in the transitional jobs. For example, Chicago’s transitional jobs program, Safer
Foundation, offered TJ group members financial and other incentives for reaching 30-, 60-, and
90-day milestones in the transitional job. Safer staff reported that once someone remained
employed in the transitional job for 90 days, that participant was considered “placed” and was
offered the option to purchase medical benefits.10 Even with the encouragement to stay on in the
transitional job, the program’s impact on employment and earnings was similar to the other sites
and faded by the end of the one-year follow-up period.11


The retention bonus in St. Paul showed promising effects.

As discussed in earlier chapters, the Milwaukee and St. Paul transitional jobs programs
implemented an employment retention bonus, or incentive, as part of the study design. It is
difficult to isolate the impact of the bonus in Milwaukee because all TJ group members were
9

Participants on supervised release may have been more motivated to find and keep employment because
they may have felt they could be found in violation of parole conditions and sent back to prison if they were not
working. This fear of violation is likely to be more prevalent among those under intensive supervision.
10
Safer staff also reported that participants were given raises in their hourly wage upon reaching 90 days
on the job. However, the payroll data for that site did not indicate a raise in wages for participants related to
staying employed in the transitional job for 90 or more days. All earnings from payroll and UI data sources are
recorded in the analysis.
11
Had this impact analysis counted the Chicago transitional job as “unsubsidized” after the participant
stayed for 90 days, the program would have generated increases in unsubsidized employment in Quarter 3, and
a small insignificant increase in Quarter 2. As shown in the total employment outcomes, impacts on having any
employment faded by the end of the follow-up period in Safer (see Appendix Table C.1).

eligible to receive the bonus and it is not possible to separate the effect of the bonus from the
effects of the other services provided by the TJ program in that site.
In St. Paul, however, the bonus program was put in place midway through the enrollment period, which allows for a comparison of the program’s impacts for sample members who
were eligible to receive the bonus with the impacts for sample members who were not offered
this incentive. The sample members who were not offered a bonus were randomly assigned
between January 2007 and November 2007, and those who were offered a bonus were randomly assigned between December 2007 and May 2008. Other contextual factors could also affect
any differences observed in the impacts for these subgroups. For example, implementation
research suggests that the programs improved their job development and placement services
over time. In addition, about 25 percent of the sample to whom a bonus was offered overlaps
with the late cohort that experienced much more of the economic downturn during their followup period than those who were not offered a bonus. Finally, caution should be taken when
interpreting these findings because the sample size is small, with only 172 people in the subgroup that was eligible to receive a bonus.
Table 5.3 presents the impacts of the St. Paul transitional jobs program on employment and earnings outcomes for the subgroups defined by bonus eligibility. The left side of
the table shows the outcomes and impacts for the subgroup that was not eligible to receive a
bonus, and the right side of the table shows them for the bonus-eligible subgroup. Statistical
significance levels for differences in impacts between subgroups are shown in the rightmost
column of the table.
The table shows that the impacts of the St. Paul transitional jobs program on employment and earnings appear more positive for the bonus-eligible sample. There are fewer, less
persistent reductions in unsubsidized employment. In fact, there are small positive impacts on
unsubsidized employment in later quarters for those who were eligible for the bonus. For
example, in Quarter 4, about 41 percent of TJ group members in the bonus subgroup worked in
an unsubsidized job compared with 32 percent of JS group members (although this difference of
9 percentage points is not statistically significant). Most striking is the difference in impacts on
being employed in all four quarters, which is much larger in the bonus subgroup. The daggers in
the rightmost column of the table show that the differences in impacts between the subgroups
are statistically significant.
The bottom panel of the table presents the program’s impacts on earnings for each of
the subgroups. For the bonus subgroup, the program generated an increase of $614 in earnings
in Quarter 4 compared with a reduction in earnings in that same quarter for the no-bonus
subgroup. This result suggests that TJ group members who were eligible for the bonus either
worked more or had higher wages than their JS group counterparts.

79
25.9

Employed in all 4 quarters (%)

88.4
83.1
52.6
43.1
97.5

Total employment (%)
Quarter 1 (random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)
2.7

15.9
24.9
31.4
34.8
57.5

Unsubsidized employment (%)
Quarter 1 (random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Number of quarters employed

88.7
76.5
34.5
12.1
96.2

Transitional employmentc (%)
Quarter 1 (random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Employment

Outcome

19.6

2.0

49.9
59.4
48.0
42.0
77.3

51.2
60.8
48.8
41.8
78.4

0.0
0.0
0.0
0.0
0.0

***

***
***
***

***
***
***
***
***

6.3

0.7 ***

38.4 ***
23.7 ***
4.6
1.1
20.2 ***

-35.3
-35.9
-17.4
-7.0
-20.9

88.7
76.5
34.5
12.1
96.2

0.179

0.000

0.000
0.000
0.406
0.840
0.000

0.000
0.000
0.001
0.203
0.000

0.000
0.000
0.000
0.000
0.000

40.3

2.9

97.6
75.9
62.9
49.0
99.3

15.2
29.6
41.7
41.4
60.0

98.9
65.2
34.5
13.0
98.9

14.4

1.6

50.1
46.2
35.9
31.3
69.3

49.9
46.0
36.2
31.9
69.1

0.0
0.0
0.0
0.0
0.0

***
***
***
***
***

***
***
***
**
***

25.9 ***

1.2 ***

47.5
29.8
27.0
17.7
30.0

-34.6 ***
-16.4 **
5.5
9.5
-9.1

98.9
65.2
34.5
13.0
98.9

0.000

0.000
0.000
0.001
0.022
0.000

0.000
0.036
0.500
0.221
0.251

0.000
0.000
0.000
0.000
0.000

(continued)

††

††
†

††
††
†

†††
††

Difference
Exposure to Retention Bonuses
Between
Not Exposed to Retention Bonuses
Exposed to Retention Bonuses
Transitional Job Search Difference
Transitional Job Search Difference
Subgroup
P-Valuea
Impactsb
P-Valuea
Jobs Group
Group (Impact)
Jobs Group
Group
(Impact)

St. Paul

One-Year Impacts on Employment and Earnings, by Exposure to Retention Bonuses

Table 5.3

The Transitional Jobs Reentry Demonstration

93
346
802
898
2,139
850
1,401
1,200
1,034
4,485

Unsubsidized earnings
Quarter 1 (random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Total earningsd
Quarter 1 (random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)
588
1,454
1,409
1,277
4,728

600
1,468
1,425
1,279
4,772

0
0
0
0
0
***
***
***
*
***

***
***
***
***
***

262 **
-54
-208
-243
-242

-507
-1122
-623
-381
-2634

758
1,055
398
136
2,347

0.027
0.790
0.370
0.291
0.704

0.000
0.000
0.007
0.097
0.000

0.000
0.000
0.000
0.000
0.000

962
1,636
1,615
1,580
5,794

35
563
1,150
1,442
3,189

928
1,074
465
138
2,605

527
1,278
1,132
966
3,903

550
1,252
1,128
984
3,913

0
0
0
0
0

***
***
***
***
***

435 ***
359
484
614 *
1892 **

-515 ***
-690 ***
22
458
-724

928
1,074
465
138
2,605

0.000
0.185
0.190
0.092
0.043

0.000
0.008
0.952
0.211
0.423

0.000
0.000
0.000
0.005
0.000

(continued)

††
†

†
†

††

NOTES: Statistical significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10 percent.
Results in this table are regression-adjusted for pre-random assignment characteristics.
aStandard errors are presented in this table for all impacts with a p-value of 0.000. Following are the standard errors for all impacts with a p-value of 0.000
(presented in the order in which they appear in the table beginning with the "Not exposed" group). Employment: 2.595, 3.439, 3.671, 2.649, 1.535, 1.173,
5.532, 5.328, 3.547,1.173, 4.913, 5.274, 5.117, 7.298, 4.769, 4.922, 3.547, 6.002,7.478, 5.517, 0.139, 0.207, and 7.032. Earnings: 46.876, 73.959, 54.024,
37.221, 138.116, 59.689, 124.674, 90.728, 210.534, 109.486, 193.515, 631.298, 112.169, and 121.953.
bThe H-statistic is used to assess whether the difference in impacts between the subgroups is statistically significant. Significance levels are indicated as
follows: ††† = 1 percent, †† = 5 percent, and † = 10 percent.
cEarnings and employment from transitional jobs are based on payroll data from each of the sites.
dTotal earnings may not be equal to the sum of transitional job earnings plus unsubsidized job earnings due to rounding.

Sample size (total = 506)
167
167
86
86
SOURCES: MDRC calculations using payroll data from St. Paul and unemployment insurance data from Minnesota.

758
1,055
398
136
2,347

Transitional job earningsc
Quarter 1 (random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Earnings ($)

Outcome

Difference
Exposure to Retention Bonuses
Between
Not Exposed to Retention Bonuses
Exposed to Retention Bonuses
Transitional Job Search Difference
Transitional Job Search Difference
Subgroup
P-Valuea
P-Valuea
Impactsb
Jobs Group
Group
(Impact)
Jobs Group
Group (Impact)

Table 5.3 (continued)

As discussed above, there are a number of reasons to be cautious about these results.
Most important, there could have been differences in the characteristics of the sample or in
program implementation later in the enrollment period, such as improvements in the job
placement function. As described in Chapter 3, over time program staff became more knowledgeable about the special circumstances involved in placing former prisoners. Site staff had
also developed more connections to employers who were willing to hire this population later in
the follow-up period.
Appendix Table C.5 presents selected baseline characteristics for the bonus and nobonus subgroups. While there are few differences in characteristics, a higher proportion of
people are under intensive supervised release in the bonus subgroup (not shown in table).12 Box
5.2 shows an interesting pattern of differences in participation and job placement rates, measured using data from the St. Paul program’s MIS. The data presented in Box 5.2 suggest that TJ
group members in the bonus subgroup participated in job readiness activities at much higher
rates than TJ group members in the no-bonus group. The bonus group also had much higher
placement rates in unsubsidized jobs. Both the higher rates of participation in job readiness
activities and the program’s higher placement rates for the bonus group could be the result of an
improvement in the job development function of the program or an effect of the employment
retention bonus. In any case, these differences in participation in job readiness activities and job
placement could have contributed to the larger impacts for the bonus group in St. Paul.
While there are reasons to be cautious, there are also reasons to be optimistic about
these results. The pattern of participation and impacts is consistent with what would be expected
for this type of incentive. The largest impacts appear later in the follow-up period, after many
left the transitional jobs for unsubsidized jobs. The pattern also shows more stability in employment, which is consistent with an earnings supplement for staying employed over time.
Finally, these results are consistent with prior research findings, which have shown promising
effects for earnings supplements.13 Based on these findings, further research and testing of
retention bonuses for this population is recommended.

There is also an impact on parole violations concentrated in the bonus subgroup, and this impact does
not appear to be related to the employment impacts (discussed in Chapter 6), suggesting that there may be other
differences about that particular sample that are unrelated to the retention bonus. It is also possible that the
recidivism reductions caused the employment impacts. Since TJ group members were less likely to be
incarcerated, they were more available for work.
13
Michalopoulos (2005); Martinson and Hendra (2006).

Box 5.2

Participation in the St. Paul Transitional Jobs Program, by Bonus Exposure
The table below shows participation rates for various program activities among transitional jobs
(TJ) group members in St. Paul who were eligible to receive a bonus for remaining employed
and those who were not. The bonus-eligible group participated in job readiness/job club activities
at much higher rates than the no-bonus group (95 percent versus 68 percent). In St. Paul, site staff
began to “market” the bonus to participants once they were “job ready” and had attended their
first job development session. It is possible that once participants knew about the bonus, they
became more motivated to find regular employment and started participating in job development
at higher rates. The table shows that the bonus group members were also more likely to be placed
in unsubsidized employment or to report it to a staff member. Because the data are from the
program’s management information system (MIS), the job placement outcome could be measuring a true difference in job placement rates or it could be measuring a difference in participants’
reporting of unsubsidized employment. The bonus subgroup may have been more likely to report
back to program staff about employment because of the incentive. It is unlikely that this difference is just measuring higher reporting among the bonus group, however, because state unemployment insurance records also show higher employment rates in Quarters 3 and 4 for the
bonus-eligible TJ group members compared with the no-bonus TJ group members (see Table
5.3). The bottom panel of the table below shows that the bonus group received an average of
three payments, totaling $631 over the one-year follow-up period.
Outcome
Transitional jobs
Worked in a transitional job (TJ) (%)
Average days workeda
Other program activities
Participated in any activity other than TJ (%)
Had an individual meeting with a case manager
Participated in job readiness/job club activities
Participated in vocational training
Placed in or reported an unsubsidized job (%)
Among those placed in or reporting an unsubsidized job
Received a retention bonus (%)
Among those who received a retention bonus
Number of retention bonuses received
Total amount received in bonuses ($)
Sample size (total = 253)

Bonus

No Bonus

98.8
64.2

96.4
57.6

100.0
95.3
95.3
22.1
44.2

98.8
96.4
68.3
31.1
26.9

76.3

0.0

3.3
630.9

0.0
0.0

167

***
***

SOURCES: MDRC calculations using payroll and MIS data from Goodwill/Easter Seals Minnesota.
NOTES: Data include only participation activities that occurred within a year of random assignment.
Statistical significance levels are indicated as: * = 10 percent; ** = 5 percent; *** = 1 percent.
a
Total hours worked divided by hours in a work day (6 to 8 hours) for each site.

Impacts by Study Enrollment Cohort
As discussed in Chapter 2, unemployment rates began to rise dramatically in 2008.
Economic conditions are expected to have implications for the levels of employment found in
the sample. But, because these conditions exist equally for both TJ and JS group members, it is
unclear how economic conditions will affect the impacts of the TJRD transitional jobs programs. Most studies, at least of welfare populations, have found modest effects of economic
conditions on the impacts of policies and programs that are aimed at improving employment
outcomes.14 However, the recession that took hold during the study follow-up period is unusual
and much more severe than any recession in decades. Unemployment rates have not been as
high since the early 1980s.15 Further, as Chapter 2 shows, all sites experienced a steep and
steady increase in unemployment rates during the same timeframe of the study follow-up
period. Since all the sites experienced a dramatic increase in unemployment rates at roughly the
same time, there is an unusual opportunity to assess the effects of the recession on the impacts
of the transitional jobs programs, while controlling (at least to some extent) for variations in
implementation of the model.
Table 5.4 presents the impacts of the transitional jobs programs on employment and
earnings for an “early” and “late” cohort of participants, defined by the timing of enrollment in
the study. The participants in the early cohort were randomly assigned between January 2007
and March 2008, and they experienced relatively good and stable economic conditions during
most of their follow-up.16 Alternatively, the late cohort includes individuals who were randomly
assigned between April and September 2008 and experienced the worst part of the economic
downturn for most or all of their one-year follow-up. (See Figure 1.1 in Chapter 1 for an
illustration of the follow-up periods for each of these cohorts.) Notably, the St. Paul site ended
enrollment in May of 2008; thus, only a small number of participants (54) from St. Paul are
present in the late cohort.17

Some studies have found that programs and policies are more effective when economic conditions are
favorable; see, for example, Herbst (2008); Bloom, Hill, and Riccio (2001); Greenberg, Michalopoulos, and
Robbins (2001).
15
In Detroit and St. Paul, the unemployment rates in June 2009 were within 1 percentage point of the
record-high unemployment rate in that state (recorded during the recession of the early 1980s). In Chicago and
Milwaukee, the unemployment rates were within 2 percentage points of the record highs (U.S. Bureau of
Labor Statistics, 2010).
16
Notably, those who were randomly assigned between January and March of 2008 could have had up to
two quarters of their follow-up during the economic downturn.
17
When the small number of St. Paul participants are given additional weight in the analysis, the impact
results are generally similar.

84
20.9

Employed in all 4 quarters (%)

72.2
74.7
50.5
38.5
92.7

Total employment (%)
Quarter 1 (random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)
2.4

16.7
26.8
30.5
30.6
52.3

Unsubsidized employment (%)
Quarter 1 (random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Number of quarters employed

66.2
64.9
28.3
10.9
84.6

Transitional employmentc (%)
Quarter 1 (random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Employment

Outcome

12.0

1.5

32.2
43.6
36.0
34.0
62.2

32.5
43.5
36.1
34.1
62.0

0.0
0.0
0.0
0.0
0.0

***
***
***
*
***

***

***
***
**

***
***
***
***
***

8.9 ***

0.9 ***

40.0
31.1
14.5
4.4
30.5

-15.8
-16.7
-5.6
-3.5
-9.6

66.2
64.9
28.3
10.9
84.6

Early
Transitional Job Search Difference
Jobs Group
Group (Impact)

0.000

0.000
0.000
0.000
0.081
0.000

0.000
0.000
0.025
0.159
0.000

0.000
0.000
0.000
0.000
0.000

P-Valuea

17.8

2.1

71.6
75.9
38.8
23.4
92.5

15.3
23.3
28.7
21.3
42.4

67.2
66.6
14.9
2.1
87.8

10.1

1.1

29.6
35.5
27.1
22.1
50.0

29.8
35.1
27.1
22.1
49.3

0.0
0.0
0.0
0.0
0.0

***
***
***
*
***

***

***
***
**

7.6 **

1.0 ***

42.1
40.4
11.7
1.3
42.5

-14.5 ***
-11.8 **
1.5
-0.8
-6.9

67.2
66.6
14.9
2.1
87.8

Late
Transitional Job Search Difference
Jobs Group
Group (Impact)

Cohort

One-Year Impacts on Employment and Earnings, by Cohort

Table 5.4

The Transitional Jobs Reentry Demonstration

0.041

0.000

0.000
0.000
0.016
0.774
0.000

0.001
0.014
0.743
0.854
0.185

0.000
0.000
0.000
0.067
0.000

P-Valuea

(continued)

†††

†

†††
†††

Difference
Between
Subgroup
Impactsb

Sample size (total = 1,774)

Total earnings
Quarter 1 (random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Unsubsidized earnings
Quarter 1 (random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Transitional job earningsc
Quarter 1 (random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Earnings ($)

Outcome

706

740
1,449
1,128
1,029
4,347

184
464
744
884
2,276

556
985
384
145
2,070

695

297
876
995
979
3,147

300
876
997
980
3,152

0
0
0
0
0

***

**
***
**

***
***
***
***
***

444 ***
573 ***
133
50
1200 ***

-115
-412
-252
-96
-876

556
985
384
145
2,070

Early
Transitional Job Search Difference
Jobs Group
Group (Impact)

0.000
0.000
0.206
0.682
0.000

0.016
0.000
0.015
0.424
0.003

0.000
0.000
0.000
0.000
0.000

P-Valuea

187

978
1,550
1,023
789
4,340

233
559
830
758
2,380

745
991
193
31
1,960

186

339
708
526
441
2,014

346
699
524
441
2,011

0
0
0
0
0

***

***
***
***

639
842
497
348
2,326

***
***
***
**
***

-113
-140
306 *
316 *
369

745
991
193
31
1,960

Late
Transitional Job Search Difference
Jobs Group
Group (Impact)

Cohort

Table 5.4 (continued)

0.000
0.000
0.003
0.038
0.000

0.380
0.363
0.065
0.058
0.458

0.000
0.000
0.000
0.117
0.000

P-Valuea

(continued)

†

†††
††
††

†††
†††

†††

Difference
Between
Subgroup
Impactsb

NOTES: Statistical significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10 percent.
Results in this table are regression-adjusted for pre-random assignment characteristics.
The early cohort includes sample members randomly assigned before April 1, 2008 (N = 1,401). The late cohort includes those randomly assigned on and
after that date (N = 373).
aStandard errors are presented in this table for all impacts with a p-value of 0.000. Following are the standard errors for all impacts with a p-value of 0.000
(presented in the order in which they appear on the table starting with the "Early" cohort). Employment: 1.729, 1.819, 1.721, 1.194, 1.350, 3.603, 3.639, 2.717,
2.428, 2.233, 2.508, 2.567, 2.375, 2.474, 2.593, 2.047, 4.853, 4.919, 4.191, 0.068, 0.134, and 1.956. Earnings: 23.478, 41.204, 31.991, 20.993, 81.701, 56.842,
78.116, 45.498, 131.375, 84.428, 52.874, 90.008, 300.786, 135.265, 155.029, and 498.726.
bThe H-statistic was used to test for statistically significant differences in impact estimates across subgroups. Statistical significance levels are indicated as
follows: ††† = 1 percent; †† = 5 percent; † = 10 percent.
cEarnings and employment from transitional jobs are based on payroll data from each of the sites.
dTotal earnings may not be equal to the sum of transitional job earnings plus unsubsidized job earnings due to rounding.

SOURCES: MDRC calculations using payroll data from each site and unemployment insurance data from each of the states in the demonstration (Illinois,
Michigan, Wisconsin, and Minnesota).

Table 5.4 (continued)

As expected, levels of employment in unsubsidized jobs are lower for both research
groups in the late cohort, and by as much as 12 percentage points in Quarter 4 for the JS group.
The impacts of the transitional jobs programs on employment outcomes were similar in both
cohorts and similar to the full sample results. The transitional jobs programs produced large
impacts on total employment, driven by the transitional jobs in both cohorts. The substitution
of transitional jobs for unsubsidized employment was less persistent, but still present, for the
late cohort.
Earnings impacts were also similar across the cohorts, for the first two quarters. In quarters 3 and 4, the transitional jobs programs generated a modest increase of about $300 in
earnings from unsubsidized jobs for the late cohort. There were no differences in the average
earnings of TJ group members across cohorts, but earnings for the JS group were much lower
among the participants in the late cohort. This pattern suggests that transitional jobs programs
may be more helpful in stabilizing employment when the counterfactual is likely to have lower
rates of employment, such as during an economic downturn. At the same time, the programs
improved their job development functions over time, which could have made the transitional
jobs programs more effective for the late cohort. A second year of follow-up, planned for this
study, will determine whether this pattern of effects continues, as all sample members will then
have employment outcomes measured during the recession.
Figure 5.1, discussed above, shows an unexpected pattern of declining employment
rates for the JS group after Quarter 2. While this pattern is expected for TJ group members as
they left the transitional jobs, it is unusual for the JS group. Table 5.4 shows that this pattern
was present, but less severe, in the early cohort. The economy was relatively stable during the
follow-up period for the early cohort, suggesting that it is only partially responsible for the
declining employment rates over time. Higher rates of incarceration are another possible
explanation for declining rates of employment. Over the course of the year, higher numbers of
sample members were incarcerated, and that may explain why fewer were able to work later
in the follow-up period. Appendix Table D.1 provides some evidence about whether the
declining employment rates are linked to higher rates of incarceration over time and shows
that even among those who are not in prison, employment rates decline over time. Thus,
higher incarceration rates also do not explain the decline in employment rates. Chapter 6
discusses recidivism effects.

Impacts for Other Subgroups
The research team examined results for other subgroups based on literature suggesting
that transitional jobs programs may have affected certain subgroups differently. For example,

prior research has shown that employment programs may be more successful for older men
coming out of prison.18 To address this question, subgroup impacts were measured for those
who were 26 or younger and those who were 27 or older. Impact results were also examined
separately for groups that may be more disadvantaged, such as those who had never worked
steadily in the past, those without a high school diploma or a General Educational Development
certificate, and those with a lengthier criminal history. The hypothesis for analyzing these
subgroups was that those who are more disadvantaged may benefit more from transitional
employment models.
Results for the subgroups are shown in Appendix G and are generally similar to results
for the full sample. However, there were larger impacts on earnings among those who had never
worked steadily, which appears to be the result of less substitution of unsubsidized employment
(Appendix Table G.5).

Conclusion
The employment results from this study, combined with findings from the Center for
Employment Opportunities (described in Chapter 1) and Transitional Work Corporation
evaluations of transitional jobs programs, provide strong evidence that transitional jobs are
effective at increasing employment for participants. Because of the transitional job, many of the
TJ program participants worked who otherwise would not have worked. However, the findings
show that these increases in employment are driven by the transitional job, and they provide no
evidence that the transitional jobs programs, as implemented, led to better unsubsidized employment outcomes.
The employment findings discussed above offer some suggestion that transitional jobs
may be more effective when levels of employment and earnings are likely to be particularly low
without the offer of a transitional job. For example, there was less substitution of the transitional
job for regular employment in Detroit and better earnings effects for the late cohort, both
situations characterized by extraordinarily low rates of employment for the JS group. Similarly,
the subgroup analysis suggests slightly more positive effects among those who had never
worked steadily. Because these impacts started to show up toward the end of the year, it will be
interesting to learn whether this pattern continues for the second year of follow-up. The CEO
evaluation also provided evidence that transitional jobs may be more effective for those who

Uggen (2000).

were recently released from prison and other high-risk groups.19 It may be worthwhile to target
scarce resources to populations that are most likely to benefit from transitional jobs, but identifying those who are most disadvantaged among a large group of disadvantaged men (and
women) can be difficult.
The findings for the retention bonus in the St. Paul site are promising, but are not necessarily connected to the transitional jobs. It is possible that had the job search program also
offered a retention bonus, similar effects would have been found for that program. Nonetheless,
retention bonuses should be tested further and as part of a comprehensive transitional jobs
program strategy.

Bloom, Redcross, Zweig, and Azurdia (2007); Redcross et al. (2009); Zweig, Yahner, and Redcross
(forthcoming).

Chapter 6

Impacts on Recidivism
When targeted to former prisoners, transitional jobs programs are designed not only to
improve employment outcomes but also, largely by increasing employment, to reduce criminal
activity. Employment is believed to be a key component in the successful transition back into
society, as work may provide stability, lead former prisoners to develop more positive roles and
social supports, and provide legitimate income that may discourage individuals from turning to
illegal sources of earnings.1 Programs that provide employment in the period immediately
following release, when recidivism rates are highest,2 are seen as especially promising.3
The results presented in Chapter 5 of this report show that the transitional jobs programs in the Transitional Jobs Reentry Demonstration (TJRD) produced a large increase in
employment, driven by the transitional jobs, especially in the first two quarters of the follow-up
period. Although the programs did not have a positive impact on unsubsidized employment, the
early impact on total employment could have led to decreases in recidivism.
This chapter presents the transitional jobs programs’ one-year impacts on key measures
of recidivism. The analysis uses administrative data on arrest and conviction, provided by the
Illinois Criminal Justice Information Authority, the Michigan State Police, the Minnesota
Bureau of Criminal Apprehension, and the State of Wisconsin Department of Justice, and
prison data provided by the Department of Corrections (DOC) in each of the states.

Impacts for the Full Sample
Table 6.1 shows one-year impacts of the transitional jobs programs on key measures of
recidivism, including arrest, conviction, and prison incarceration.4 The data include only

Solomon, Johnson, Travis, and McBride (2004).
Langan and Levin (2002).
3
Solomon, Johnson, Travis, and McBride (2004).
4
Due to differences in the available data, the first year of follow-up for criminal justice measures differs
slightly from the first year of follow-up for employment measures (presented in Chapter 5). Because unemployment insurance data are available only as quarterly measures, Year 1 employment measures include the
quarter of random assignment plus the next three quarters. For the criminal justice measures, Year 1 includes
the month of random assignment plus the next 11 months. Therefore, for individuals who were randomly
assigned in the second or third month of a quarter, the two periods do not exactly overlap. For example, for an
individual who was randomly assigned in March 2007, the Year 1 employment measures include Quarter 1 of
2007 (January to March) through Quarter 4 of 2007 (October to December), while the Year 1 recidivism
measures include March 2007 through February 2008. Measures that combine incarceration and employment,
such as those that appear in the bottom panel of Table 6.1, are adjusted so that incarceration measures line up
exactly with the quarterly employment measures in terms of timing.
2

The Transitional Jobs Reentry Demonstration

Table 6.1
One-Year Impacts on Recidivism: Full Sample
Transitional Job Search
Jobs Group
Group

Outcome
Arrested (%)

Difference
(Impact)

P-Value

38.8

35.3

3.5

0.119

13.6
9.0
4.3

14.5
7.9
4.9

-1.0
1.1
-0.7

0.556
0.393
0.489

2.3
4.7
3.8
3.7

2.7
4.9
3.0
4.2

-0.4
-0.2
0.8
-0.6

0.556
0.864
0.356
0.547

Admitted to prison (%)
Admitted to prison for a new crime
Admitted to prison for a parole/probation violation
Admitted to prison for other reason

35.1
6.2
22.4
8.9

36.4
4.8
23.9
9.7

-1.3
1.4
-1.5
-0.8

0.537
0.201
0.407
0.512

Total days incarcerated in prison

40.2

47.5

-7.3 **

0.040

Arrested, convicted, or admitted to prison (%)

53.5

56.1

-2.6

0.250

Incarcerated and employed
Incarcerated and not employed
Not incarcerated and employed
Not incarcerated and not employed

4.5
19.3
31.0
45.2

4.4
21.1
26.9
47.6

0.1
-1.8
4.1 *
-2.4

0.927
0.338
0.053
0.287

Sample size (total = 1,808)

909

899

Convicted of a crime (%)
Convicted of a felony
Convicted of a misdemeanor
Conviction categoriesb (%)
Convicted of a violent crime
Convicted of a property crime
Convicted of a drug crime
Convicted of a public order crime

Status in the last quarter of Year 1 (%)

SOURCES: Calculations based on data from unemployment insurance (UI) records in each site, Michigan
State Police, Minnesota Bureau of Criminal Apprehension, State of Wisconsin Department of Justice, Illinois
Criminal Justice Information Authority, and the Department of Corrections in each state.
NOTES: Statistical significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10
percent.
Results in this table are regression-adjusted for pre-random assignment characteristics.
Subcategories may sum to more than the total due to multiple arrests, convictions, or prison admissions
per person during the follow-up period.
aEach conviction date is counted only as a single event. If there are multiple convictions on the same
date, only the most serious conviction is recorded in the analysis. Some convictions may have been
associated with an arrest that occurred prior to random assignment. These convictions are counted in the
analysis as occurring after random assignment. Total includes convictions for felony, misdemeanor, and
other crime classes.
bThe categorization of charges is based on definitions from Langan and Levin (2002).
cMeasures whether a sample member was ever employed (in a transitional or UI-covered job) and
whether he was ever in prison at any point during Quarter 4. A person could be considered incarcerated and
employed during that quarter even if the employment and incarceration were not simultaneous.

“unsealed” events and only those events that occurred in the state in which the individual was
randomly assigned.5 Box 6.1 describes the key recidivism outcomes in detail.


The transitional jobs programs did not affect most key measures of recidivism in the first year of follow-up.

During the year following random assignment, there were no significant impacts on the
percentage of the sample that was arrested, convicted, or admitted to prison. The first row of
Table 6.1 shows that 39 percent of the transitional jobs (TJ) group and 35 percent of the job
search (JS) group was arrested, a difference that is not statistically significant.6 Generally,
average national recidivism rates are somewhat higher than the rates found among the TJRD
sample. According to the latest available statistics, measured among prisoners who were
released in 1994, 44.1 percent were arrested, 21.5 percent were convicted of a crime, and 10.4
percent were admitted to prison for a new crime within one year of release.7 There are several
possible reasons for the differences between these rates and those of the TJRD sample, including differences in the time periods and sample characteristics. As shown in Box 1.1 in Chapter
1, the two samples differ in some demographic and criminal history characteristics, including
age, race, ethnicity, and length of last incarceration. In addition, the two samples may differ on
unmeasured characteristics, such as level of motivation.
Table 6.1 also shows no significant impacts on the percent convicted of a crime, either
overall or by crime class or category. About 14 percent of sample members in both groups were
convicted of a crime during the follow-up period, with between 8 and 9 percent convicted of a
felony and about 5 percent convicted of a misdemeanor. For both groups, the convictions were
spread fairly evenly across crime categories, which include violent, property, drug, and public
order crimes. Less than 5 percent of study participants in either group were convicted of a crime
in each of the four categories.

Records of arrests and convictions are sometimes sealed (that is, removed from public criminal history
records) or expunged (that is, removed entirely from criminal history records) according to state law. For
example, states may seal or expunge juvenile records or records of arrests that did not result in conviction.
6
Compared with the arrest rate among sample members in the Center for Employment Opportunities
(CEO) evaluation, which tested a transitional jobs program in New York City, the arrest rate for the TJRD
sample is high. Among CEO’s early release subgroup (which provides the best comparison with the TJRD
sample because it included those who were randomly assigned within three months of release from prison), 24
percent of the control group was arrested. See Redcross et al. (2009). The difference between the CEO and
TJRD arrest rates may be due to variation in state rules about the sealing of data. While New York State
automatically seals some arrest records when they do not result in a conviction, some of the states included in
the TJRD evaluation do not. Therefore, the TJRD sample may have more arrests included in the unsealed data
provided by the states.
7
Langan and Levin (2002).

Box 6.1

Glossary of Recidivism Outcomes
Admissions to prison. Admissions to state prison for any reason.
Admissions to prison for a new crime. Admissions to state prison with a new sentence
following a conviction for a new crime. New crime admissions do not include prison
admissions based only on violations of parole conditions.
Admissions to prison for a technical parole violation. Admissions to prison after a
parolee has violated a condition of his parole from a previous incarceration. As a condition of release to parole, parolees must meet certain conditions, which may include reporting to a parole officer, abstaining from drugs and alcohol, participating in substance abuse
treatment, attending anger management classes, or a number of other conditions. Depending on its severity, a violation of these rules may lead to the revocation of parole, resulting
in a return to prison. Technical rule violations are not usually preceded by an arrest or
conviction.
Admissions to prison for other reason. Admissions to prison for a reason other than a
new crime or parole violation. These reasons include returns after an escape and, in Wisconsin, admissions to prison for a Temporary Parole and Probation Hold. These holds,
which typically last only a few days, but may be as long as 90 days, are not technically
parole revocations; they may be used as a warning to a parolee who has broken a technical rule, to investigate a possible violation, or as a precursor to a revocation. Temporary
holds, which can also occur in jail, occurred frequently in state prison in Milwaukee because the nearby Milwaukee Detention Facility was used for this purpose.a
Arrests. Unsealed arrests. Depending on state rules for the sealing of arrest records, the
data may include arrests that did not lead to a conviction.
Convictions. Dispositions of guilty, whether by trial or plea. Some convictions may have
been related to an arrest that occurred prior to random assignment.
Felony or misdemeanor convictions. Convictions with felony or misdemeanor charges.
For each conviction date, only the charge with the highest class, in order of felony,
misdemeanor, and other, is included.
Violent, property, drug, or public order convictions. Convictions with charges in the
given crime category.b Crimes were categorized as follows:
 Violent crime: Homicide, manslaughter, kidnapping, sexual assault, robbery, assault,

extortion, and other crimes against the person.
 Property crime: Arson, burglary, larceny, stolen vehicles, fraudulant activities, stolen

property, damage to property, smuggling, and other property offenses.
(continued)

Box 6.1 (continued)
 Drug crime: Drug trafficking, drug possession, and other drug offenses.
 Public order crime: Weapons offenses, traffic offenses, nonviolent sex offenses,

obscenity, family offenses, commercialized sex offenses, obstructing the police or
the judiciary, bribery, disturbing the public peace, invasion of privacy, and other
public order offenses.
______________________________
NOTES: aState of Wisconsin Department of Corrections (2003). Holds that resulted in a parole
revocation were counted only as admissions for technical violations.
b
Crimes were categorized based on the 1994 Bureau of Justice Statistics Special Report
classifications (see Langan and Levin, 2002).

The transitional jobs programs did not reduce the percentage admitted to prison in Year
1. Thirty-five percent of the TJ group and 36 percent of the JS group, which does not represent a
statistically significant difference, were admitted to prison during that time. Most of the prison
admissions were for technical parole violations, which occur when a parolee violates a condition of his release, such as abstaining from drugs, reporting to a parole officer, or being home by
a specified curfew. Twenty-two percent of the TJ group and 24 percent of the JS group were
returned to prison after a parole violation, a difference that is not statistically significant —
though, as discussed below, there was a significant impact on this measure in the first half of
Year 1. Lower percentages, about 6 percent in the TJ group and about 5 percent in the JS group,
were admitted to prison on a new sentence after being convicted of a new crime. Finally, just
less than 10 percent of the sample was admitted for another reason; this category includes
mostly admissions in Milwaukee for temporary parole and probation holds, which do not occur
in state prison in the other three study sites. Temporary parole and probation holds, which
typically last only a few days but may be as long as 90 days, are not technically parole revocations and may be used to warn a parolee who has broken a technical rule, to investigate a
possible violation, or as a precursor to a parole revocation. (See Box 6.1 for more information
on reasons for incarceration.)
Despite the lack of significant impacts on the percentage admitted to prison, the transitional jobs programs reduced the number of days spent in prison in Year 1 by about seven days.
On average, in the year following random assignment, the TJ group spent 40.2 days in prison,
compared with 47.5 days among the JS group,8 a statistically significant difference. This impact
appears to be driven by a reduction in parole revocations in the first six months of the follow-up

Measures of the number of days in prison include zeroes for those who were not admitted to prison.

period (shown in Table 6.2 and discussed below). An analysis of impacts by site, discussed
below, also shows that this impact was driven by the results in only one site, St. Paul.
The last line before the bottom panel of Table 6.1 shows the percentage of sample
members who were arrested, convicted, or admitted to prison in the year following random
assignment. The transitional jobs programs did not have a significant impact on this recidivism
composite measure. More than half of the sample members in both groups — 54 percent of the
TJ group and 56 percent of the JS group (not significantly different) — experienced at least one
criminal justice event during the follow-up period.
The bottom panel of Table 6.1 shows impacts on combined measures of employment
status and prison status in the last quarter of the year. These outcomes measure whether a
sample member was ever employed — in a transitional job or a job covered by unemployment
insurance (UI) — and whether he was ever in prison at any point during that quarter. A person
could be considered incarcerated and employed during the same quarter even if the employment
and incarceration were not simultaneous.9
The combined measures show that by the end of the year, the two groups were experiencing similar prison and employment outcomes. The only impact on these combined measures is an increase in the percentage employed but not incarcerated (31 percent of the TJ group
and 27 percent of the JS group). This impact appears to be driven not by incarceration but by
the 4 percentage point impact on total employment in that quarter, which was driven by a small
number of people who were still working in transitional jobs (see Table 5.1 in Chapter 5). Over
45 percent of the sample in both groups was neither employed nor incarcerated in Quarter 4.
Because no survey data were collected for TJRD, it is not possible to know what these individuals may have been doing to support themselves during this time period. See Appendix Table
D.1 for impacts on combined measures for all quarters.


There was a reduction in parole revocations and time in prison in the
first half of the follow-up period, which was driven by impacts in one
site, St. Paul.

If being employed reduces the likelihood of crime, it might be expected that recidivism
impacts would occur at the same time as, or soon after, employment impacts, such that when

Individuals are not likely to have been simultaneously in prison and employed in UI-covered jobs. Although individuals may have jobs in prison, these are not UI-covered and therefore do not appear in the
employment data used in this report. Because the UI data are available only by quarter and do not provide
specific dates of employment, it is not possible to test whether employers reported earnings for time periods
that overlapped with incarceration spells.

The Transitional Jobs Reentry Demonstration

Table 6.2
One-Year Impacts on Recidivism, by Follow-up Time Periods: Full Sample
Outcome

Transitional Job Search
Jobs Group
Group

Difference
(Impact)

P-Value

Months 1 to 6
Arrested (%)

20.9

18.9

1.9

0.302

4.2

5.4

-1.2

0.250

Admitted to prison (%)
Admitted to prison for a new crime
Admitted to prison for a technical parole violation
Admitted to prison for other reason

20.6
1.7
13.3
6.0

22.7
1.6
16.1
5.4

-2.1
0.2
-2.8 *
0.5

0.249
0.787
0.071
0.600

Total days incarcerated in prison

11.3

16.8

-5.5 ***

0.000

Arrested, convicted, or admitted to prison (%)

33.5

35.9

-2.3

0.281

Arrested (%)

23.9

21.0

3.0

0.125

Convicted of a crimea (%)

10.0

9.6

0.3

0.810

Admitted to prison (%)
Admitted to prison for a new crime
Admitted to prison for a technical parole violation
Admitted to prison for other reason

18.4
4.4
10.4
3.9

18.1
3.2
10.0
5.2

0.3
1.2
0.4
-1.3

0.881
0.182
0.762
0.175

Total days incarcerated in prison

28.9

30.7

-1.9

0.469

Arrested, convicted, or admitted to prison (%)

36.3

36.0

0.3

0.901

Sample size (total = 1,808)

909

899

Convicted of a crimea (%)

Months 7 to 12

SOURCES: Calculations based on data from Michigan State Police, Minnesota Bureau of Criminal
Apprehension, State of Wisconsin Department of Justice, Illinois Criminal Justice Information Authority,
and the Department of Corrections in each state.
NOTES: Statistical significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10
percent.
Results in this table are regression-adjusted for pre-random assignment characteristics.
Subcategories may sum to more than the total due to multiple arrests, convictions, or incarcerations per
person during the follow-up period.
aSome convictions may have been associated with an arrest that occurred prior to random assignment.
These convictions are counted in the analysis as occurring after random assignment.

employment is increased, recidivism is reduced. If this were true, reductions in recidivism
would be most likely to occur early in the follow-up period, since the largest impacts on
employment occurred in the first two quarters. This hypothesis is considered in Table 6.2,

which shows impacts separately for the first six months of the follow-up period and the second
six months of the follow-up period.
As with outcomes for the year as a whole, there were few statistically significant impacts during either of these time periods. However, the table shows a small reduction in admissions to prison for a parole violation in the first six months, when 13 percent of TJ group
members and 16 percent of JS group members had their parole revoked. This impact probably
produced the reduction of 5.5 days in prison during this time period. The timing of this impact
suggests that it is directly related to the transitional jobs employment. The impacts by site,
discussed in the following section, indicate that this reduction in parole violations was driven by
the results in the St. Paul site and did not occur in the other three sites.

Impacts by Site
Table 6.3 shows key recidivism impacts by site (see Appendix D for full recidivism impact tables for each site). Each panel of the table shows impacts for one of the four sites, and the
column on the far right shows whether the impacts are significantly different across sites. As
discussed in Chapter 2, the criminal justice context and a number of the sample characteristics
differed substantially across sites. The results in Table 6.3 suggest that those differences led to
variation across sites in the rates of recidivism.


The rates of recidivism varied across sites.

The arrest rates were especially high in Chicago, where 44 percent of the job search
group was arrested during the one-year follow-up period. The number of prior arrests at baseline,
shown in Table 2.2 in Chapter 2, was also highest for the Chicago sample. These relatively high
rates of arrest may reflect the available data, as Illinois does not automatically seal the records of
arrests that do not result in a conviction.10 While the other three states also do not automatically
seal all arrest records, the process that is necessary to have an arrest record sealed may be less
onerous in those states.11 The hypothesis that rules for sealing affect the arrest rates seems
plausible, since the conviction rates, which would generally not be affected by rules for sealing
10

In Illinois, criminal records include all arrests, even in cases in which charges were dropped or the person was found not guilty, unless the individual successfully petitions to have his record removed. These
petitions require the individual to obtain his own criminal records and detailed information about the case, pay
a fee, and possibly appear in court. See Office of the State Appellate Defender (2010).
11
In Michigan, arrest records may not be automatically sealed once a person has had a prior conviction. See
Michigan Legislative Council (n.d.). In Minnesota, records of arrests that did not lead to conviction may not be
sealed, and may require a petition, fee, and a court hearing. See Minnesota Judicial Branch (n.d.). In Wisconsin, sealing is not automatic, but a petition to have them removed requires only a one-page form. See Crime
Information Bureau (2009).

The Transitional Jobs Reentry Demonstration

Table 6.3
One-Year Impacts on Recidivism, by Site
Transitional Job Search Difference
Jobs Group
Group (Impact)

Outcome

Difference
Between
b
P-Value Site Impacts

Chicago
Arrested (%)

45.8

44.2

1.6

0.761

Convicted of a crimea (%)

17.6

14.9

2.8

0.475

Admitted to prison (%)
Admitted to prison for a new crime
Admitted to prison for a parole/probation violatio
Admitted to prison for other reason

22.9
7.9
15.5
0.0

20.4
8.7
12.3
0.0

2.6
-0.8
3.2
0.0

0.545
0.776
0.380
0.000

Total days incarcerated in prison

22.3

24.7

-2.4

0.709

Arrested, convicted, or admitted to prison (%)

50.6

48.1

2.5

0.627

39.4

33.5

6.0

0.206

Convicted of a crime (%)

14.1

18.9

-4.8

0.180

Admitted to prison (%)
Admitted to prison for a new crime
Admitted to prison for a parole/probation violatio
Admitted to prison for other reason

15.4
7.2
7.8
0.4

12.4
5.0
7.3
0.1

3.0
2.1
0.5
0.4

0.377
0.378
0.842
0.446

Total days incarcerated in prison

18.4

16.4

2.0

0.701

Arrested, convicted, or admitted to prison (%)

43.0

40.8

2.2

0.647

33.7

30.7

3.1

0.455

Convicted of a crime (%)

10.2

11.9

-1.7

0.554

Admitted to prison (%)
Admitted to prison for a new crime
Admitted to prison for a parole/probation violatio
Admitted to prison for other reason

50.9
3.2
22.6
31.9

52.1
3.1
21.2
34.0

-1.3
0.1
1.3
-2.1

0.780
0.975
0.718
0.623

Total days incarcerated in prison

49.6

57.0

-7.3

0.344

Arrested, convicted, or admitted to prison (%)

59.6

62.2

-2.6

0.559

††
†

Detroit
Arrested (%)
a

††
†

Milwaukee
Arrested (%)
a

††
†

(continued)

Table 6.3 (continued)

Outcome

Transitional Job Search Difference
Jobs Group
Group (Impact)

Difference
Between
b
P-Value Site Impacts

St. Paul
Arrested (%)

37.7

35.3

2.4

0.586

Convicted of a crimea (%)

14.0

12.6

1.4

0.640

Admitted to prison (%)
Admitted to prison for a new crime
Admitted to prison for a parole/probation violatio
Admitted to prison for other reason

44.4
6.4
38.9
0.0

53.2
3.9
50.3
0.0

-8.7 **
2.5
-11.4 ***
0.0

Total days incarcerated in prison

60.4

83.3

-22.9 *** 0.005

Arrested, convicted, or admitted to prison (%)

58.8

68.7

-9.9 **

0.045
0.200
0.008
0.000

††
†

0.019

SOURCES: Calculations based on data from unemployment insurance records in each site, Michigan State
Police, Minnesota Bureau of Criminal Apprehension, State of Wisconsin Department of Justice, Illinois
Criminal Justice Information Authority, and the Department of Corrections in each state.
NOTES: Statistical significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10
percent.
Results in this table are regression-adjusted for pre-random assignment characteristics.
Subcategories may sum to more than the total due to multiple arrests, convictions, or incarcerations per
person during the follow-up period.
aSome convictions may have been associated with an arrest that occurred prior to random assignment.
These convictions are counted in the analysis as occurring after random assignment.
bThe H-statistic was used to test for statistically significant differences in impact estimates across sites.
Statistical significance levels are indicated as follows: ††† = 1 percent; †† = 5 percent; † = 10 percent.

data, are fairly similar across sites, ranging from 12 percent to 19 percent. There may be some
other reason for the large number of arrests in Chicago; for example, the sample in Chicago may
be especially prone to arrest, or policing in Chicago may be especially stringent.
There was also a good deal of variation across sites in prison admissions, especially in
admissions resulting from a technical violation of parole. Returns to prison were rarest in
Detroit, where only 7.3 percent of the job search group was returned to prison after a parole
revocation. Because of Michigan’s “truth in sentencing” law, described in Chapter 2, prisoners
in that state tend to serve out a relatively high percentage of their sentence before they are
eligible for parole. As a result, they may have had relatively little time left on parole after
release, and therefore less time for their parole to be revoked.12
12

Time left on parole was not measured at baseline.

100

In St. Paul, the proportion of participants who returned to prison after a parole revocation
was especially high.13 There, because of a recruitment strategy that focused on halfway houses
(see Chapter 3), a high proportion of the study sample — 65.6 percent — was on intensive
supervised release (ISR), a parole status usually assigned to parolees with a history of sex offense,
violence, or other serious crimes.14 Parolees in this status are more closely monitored than those
on standard supervised release, have stricter parole conditions, and face more stringent penalties
when those conditions are broken.15 The high rate of parole revocations in St. Paul is likely due to
the high rate of individuals who are under intensive supervised release in the sample.
Returns to prison in Milwaukee were also high, though most were neither new sentences nor parole revocations but temporary parole and probation holds (which, as noted above,
may be used to warn a parolee who has broken a technical rule, to investigate a possible
violation, or as a precursor to a parole revocation).16 Temporary holds, which can also occur in
jail, occurred frequently in state prison in Milwaukee because a nearby state prison facility was
used for this purpose.17 Since only state prison data were collected for this project, and these
types of holds do not take place in state prisons in the other three states, the measured incarceration rates are higher for Milwaukee than for the other sites.


Despite variation in the rates of recidivism across sites, there was little
variation in impacts, as few impacts are statistically significant. However, there was a reduction in parole revocations and time in prison in
St. Paul.

Table 6.3 shows that, as was the case for the full sample, there were no statistically significant impacts in any of the sites on the percentage arrested or the percentage convicted of a
crime in the year following random assignment. There were some significant differences on
conviction classes and categories in some of the sites (shown in Appendix tables D.2 through
13

Because of Minnesota’s determinate sentencing laws, most individuals sentenced to prison in that state
must spend two-thirds of their sentence in prison and one-third on supervised release. Thus, the length of
supervision and the timing of release from prison are determined by a judge at sentencing and not by a parole
board. See Minnesota Sentencing Guidelines Commission (2009). Therefore, most individuals under community supervision in Minnesota are under “supervised release,” which refers to release from an indeterminate
sentence. However, for simplicity, this report refers to all individuals on post-release supervision as “parolees.”
14
At year’s end 2007, 24.9 percent of all prison releases under community supervision in Ramsey and Hennepin counties, which encompass St. Paul and Minneapolis, were on intensive supervised release. See Minnesota
Department of Corrections (2008a).
15
Minnesota Department of Corrections (2008b), (2009).
16
The “other” category in this analysis includes only those cases in which the person was held in state
prison and which did not lead to a parole revocation. Holds that resulted in a parole revocation were counted as
admissions for technical violations. See Division of Community Corrections (2003).
17
According to a communication from staff of the Wisconsin Department of Corrections, the Milwaukee
Secure Detention Facility is used for temporary parole and probation holds.

101

D.5), but there is no discernible pattern to these differences. Given the size of the site samples
and the number of outcomes measured, it is likely that these differences occurred by chance and
do not reflect true impacts of the transitional jobs programs.
Transitional jobs program impacts on prison admissions and time spent in prison do differ by site, as there are statistically significant reductions in parole revocations and time spent in
prison in St. Paul, but no significant impacts on these measures in the other three sites. This
finding suggests that the full sample impacts on parole revocation and time in prison are driven
primarily by the impacts in St. Paul. In that site, there was a decrease of 11 percentage points in
prison admissions for parole violations, and TJ group members spent 23 fewer days in prison,
on average, than their JS group counterparts.18 These impacts are significantly different from the
impacts in the other three sites.
It is not entirely clear why the St. Paul program reduced parole violations, while the
other TJ programs did not. However, further analysis suggests that these impacts are driven by
impacts among the parolees on intensive supervised release.19 As discussed above, individuals in
that status are subject to relatively close monitoring and have more stringent parole conditions
than typical parolees. As a result, they may be at relatively high risk of violating parole. It may
be that transitional jobs are more effective at improving the behavior of these high-risk individuals or that they are effective at improving the specific behaviors that are monitored more
closely in this population. Given the high level of monitoring, any behaviors that may lead to
parole violation are more likely to be noticed by parole officers, and therefore any reduction in
these behaviors is more likely to affect violations.
Alternatively, it may be that the transitional jobs programs did not improve behavior but
that, when parole violations were committed, the Department of Corrections (DOC) took
employment, including employment in a transitional job, into account when deciding how to
respond to the violation. When parole officers discover low-level parole violations, such as
breaking curfew or failing a drug test for the first time, DOC has some discretion in deciding the
level of sanction to apply to the parolee. According to Minnesota DOC guidelines, DOC may
revoke parole and send the individual back to prison or it may impose some other sanction, like
requiring the parolee to participate in drug treatment or modifying the parolee’s curfew. In
making these decisions, employment can be considered a mitigating factor, such that the DOC
18

Measures of the number of days in prison include zeroes for those who were not admitted to prison.
Additional analysis was conducted to determine whether the impact in Minnesota might be driven by
sample characteristics. That analysis suggested that the difference between the impacts in St. Paul and the
impacts in the other sites was driven largely by those who were living in transitional housing. Since 81 percent
of ISR individuals live in transitional housing, it is likely that this impact is related to the large percentage of
ISR individuals in the St. Paul sample. Other sample characteristics, including age, prior violent conviction,
and race, did not appear to drive the difference.
19

102

may be more likely to revoke parole if a person is not employed and more likely to impose a
lesser penalty if he is employed.20 Given this situation, the large impacts on employment early
in the follow-up may have resulted in TJ group members being less likely to have their parole
revoked when violations occurred. This process may have been especially common in St. Paul
if intensive monitoring meant that ISR individuals were more likely than typical parolees to be
caught violating low-level parole conditions.
Impacts of the Retention Bonus in St. Paul
Because the four sites in TJRD varied on multiple dimensions, including economic,
criminal justice, and programmatic factors, it is difficult to determine whether particular
program components may be effective. However, a change in the program in the St. Paul site
allows for a test of the efficacy of retention bonuses. Partway through the follow-up period, the
St. Paul site began to offer cash incentives to program participants who obtained unsubsidized
employment and were able to document their continued employment on a monthly basis for up
to six months.21
As described in detail in Chapter 5, using the St. Paul sample, a subgroup analysis was
conducted to compare impacts among those sample members who were not eligible to receive a
retention bonus because they were randomly assigned before the bonuses were offered, with
impacts among those in the sample who were eligible to receive such a bonus because they
were randomly assigned when the bonuses were offered. Table 6.4 shows impacts on key
measures of recidivism by bonus group. There are no differences in arrest or conviction impacts. However, the results suggest that the impacts on parole violations in St. Paul were
concentrated in the bonus-eligible sample. In that subgroup, there was a reduction of 25 percen-

The Minnesota Department of Corrections (2009) provided the following explanation: “A supervising
agent has the responsibility to supervise and monitor the offender’s compliance with his/her conditions of
release. Agents can and do frequently administer warnings and cautions to offenders when low-level violations
of release conditions occur, such as missing curfew or failing a drug test for the first time. They, also, have the
discretion, in consultation with a Field Supervisor, to formally initiate an immediate local sanction for lowlevel violation behavior that falls short of requiring a new or modified release condition to restructure supervision in the community or an arrest warrant following a determination of imminent risk to public safety. If
violation behavior rises to the level of release condition restructure or warrant issuance, an agent must contact
the Hearings and Release Unit [HRU] of the DOC. At this point, HRU will, along with agent input and DOC’s
formal Revocation Guidelines, determine an appropriate sanction to be applied either through a restructure to
remain in the community or a revocation hearing to consider a return to prison. When considering sanctions at
any level, employment, as a constructive community activity, is considered a mitigating factor.”
21
The Milwaukee site also offered retention bonuses throughout the follow-up period. However, because
the bonus was offered to all TJ program group members in Milwaukee, it is not possible to separate the effect
of the bonus from the effects of other program services and local context.

103

104

59.1
60.8

Total days incarcerated in prison

Arrested, convicted, or admitted to prison (%)

165

Sample size (total = 502)

8.7
20.3
34.1
36.9

Incarcerated and employed
Incarcerated and not employed
Not incarcerated and employed
Not incarcerated and not employed

Status in the last quarter of Year 1e(%)

46.8
6.0
41.6
0.0

4.3
4.4
4.1
4.1

15.8
9.8
6.1

39.0

Admitted to prison (%)
Admitted to prison for a new crime
Admitted to prison for a technical parole violation
Admitted to prison for other reason

Conviction categoriesd (%)
Convicted of a violent crime
Convicted of a property crime
Convicted of a drug crime
Convicted of a public order crime

Convicted of a crime (%)
Convicted of a felony
Convicted of a misdemeanor

Arrestedb (%)

Outcome

165

7.6
27.0
34.4
31.0

65.8

80.0

49.0
3.1
47.5
0.0

2.9
2.3
1.3
4.9

11.3
5.8
4.7

35.1

1.1
-6.7
-0.3
5.9

-5.0

-20.9 **

-2.1
3.0
-5.9
0.0

1.4
2.1
2.8
-0.8

4.6
4.0
1.4

3.9

0.722
0.153
0.953
0.276

0.334

0.037

0.694
0.184
0.264
0.000

0.495
0.284
0.133
0.735

0.240
0.185
0.602

0.473

1.3
23.5
47.7
27.5

53.9

65.0

39.5
8.1
32.3
0.0

5.0
2.8
0.0
3.5

11.3
7.7
3.6

35.9

8.0
25.3
23.2
43.4

75.2

87.6

61.6
4.7
57.2
0.0

5.4
5.3
0.0
4.6

14.2
6.2
9.1

35.1

-6.8 **
-1.8
24.5 ***
-15.9 **

-21.2 ***

-22.6

-22.1 ***
3.4
-24.9 ***
0.0

-0.4
-2.5
0.0
-1.1

-2.9
1.5
-5.5

0.8

0.044
0.795
0.002
0.032

0.006

0.134

0.005
0.395
0.001
0.000

0.910
0.429
0.000
0.744

0.600
0.721
0.188

0.922

(continued)

†††
††

†

††

Difference
Exposure to Retention Bonuses
Between
Not Exposed to Retention Bonuses
Exposed to Retention Bonuses
Transitional Job Search Difference
Transitional Job Search Difference
Subgroup
Jobs Group
Group (Impact) P-Value
Jobs Group
Group (Impact)
P-Value Impactsa

St. Paul

One-Year Impacts on Recidivism, by Exposure to Retention Bonuses

Table 6.4

The Transitional Jobs Reentry Demonstration

105

NOTES: Statistical significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10 percent.
Subcategories may sum to more than the total due to multiple arrests, convictions, or incarcerations per person during the follow-up period.
aThe H-statistic is used to assess whether the difference in impacts between the subgroups is statistically significant. Significance levels are indicated as
follows: ††† = 1 percent, †† = 5 percent, and † = 10 percent.
bEach arrest date is counted only as a single event. If there are multiple crimes or charges on the same date, only the most serious charge is recorded in the
analysis.
cEach conviction date is counted only as a single event. If there are multiple convictions on the same date, only the most serious conviction is recorded in the
analysis. Some convictions may have been associated with an arrest that occurred prior to random assignment. These convictions are counted in the analysis as
occurring after random assignment. Total includes convictions for felony, misdemeanor, and other crime classes.
dThe categorization of charges is based on definitions from Langan and Levin (2002).
eIncarceration and employment status in Quarter 4 after random assignment.

SOURCES: Calculations based on unemployment insurance data from Minnesota Bureau of Criminal Apprehension and the Minnesota Department of Corrections.

Table 6.4 (continued)

tage points in the percentage returned to prison for a parole violation, compared with no
significant impacts among the participants in the no-bonus group. The impact on parole violations among those in the bonus group appears to have driven significant impacts on prison
admission and on the composite recidivism measure, with TJ group members being 21 percentage points less likely to be arrested, convicted, or admitted to prison than their JS group
counterparts.
It is not clear why these subgroup differences exist, given that the impacts on parole violations occurred early in the follow-up period, before most participants could have worked
long enough in a permanent job to receive a bonus. Given the timing of the impacts, it is not
likely that the subgroup differences are attributable to the retention bonuses themselves. It is
possible that they result from higher rates of transitional jobs participation among TJ group
members in the bonus group. Table 5.3, in Chapter 5, shows that 99 percent of bonus-eligible
TJ group members worked in a transitional job in the first quarter of the follow-up period
compared with 89 percent of no-bonus TJ group members. It is possible that because, as
discussed above, employment can be a mitigating factor when parole revocations are considered, the high rate of transitional jobs employment among the bonus group led to larger
impacts on parole violations. Alternatively, the subgroup differences may reflect a difference in
the proportion of ISR parolees in the two groups; in the bonus group, 72 percent were on ISR
compared with 62 percent in the no-bonus group.

Impacts for Other Subgroups
In addition to impacts across sites and the subgroup analysis of retention bonus eligibility in St. Paul, other subgroup analyses were conducted to determine whether the impacts of the
transitional jobs programs differed by age, random assignment cohort, employment history, or
criminal justice history. The results of those analyses are shown in Appendix G.
Past research suggests that the impacts of employment programs on recidivism vary by
age, with reductions in crime among those age 27 years and older but little impact on crime
among those under age 27 years.22 A subgroup analysis was run to test whether the impacts of
the TJ programs differed by age subgroup. The results of this analysis do not show a pattern of
systematic differences in recidivism impacts by age subgroup. (See Appendix Table G.2.)
In order to determine whether the impacts of the TJ programs differed depending on the
strength of the economy, a subgroup analysis was conducted that compared impacts by random
assignment cohort. As shown in Figure 2.1 in Chapter 2, there was a steep rise in the unem22

Uggen (2000); Uggen and Staff (2001).

106

ployment rates in all four cities starting in November or December of 2008.23 The cohort
subgroup analysis is designed to compare program impacts among those who experienced a
relatively good economy before the change in unemployment rates with impacts among those
whose follow-up period fell largely within the months in which unemployment rates were
increasingly high. However, there could have been other changes over time that could produce
differences between the two cohorts. The late cohort is defined as those who were randomly
assigned on or after April 1, 2008, and the early cohort is defined as those who were randomly
assigned before April 1, 2008.24 The results, presented in Appendix Table G.4, show that there
is no consistent pattern of differences in recidivism impacts by cohort.
Finally, subgroup analyses were also conducted to compare the impacts of the TJ programs by employment history (those who had worked for at least six months with the same
employer in the past compared with those who had not), by criminal justice history (those with
more than three past convictions compared with those with three or fewer past convictions), and
by educational attainment (those who had a high school diploma or a General Educational
Development certificate compared with those who did not). The results are shown in Appendix
Tables G.6, G.8, and G.12. In all cases, there was not a consistent pattern of statistically significant differences in subgroup impacts.

The Connection Between Employment and Recidivism
As already observed, the theory behind the transitional jobs model for former prisoners
is that transitional jobs will both increase unsubsidized employment and, via increases in
transitional and unsubsidized employment, reduce recidivism. This logic assumes that employment is causally and inversely related to crime, meaning that all other things being equal, a
former prisoner who becomes employed is less likely to commit a new crime than he would be
if he were not employed. If this were true, positive impacts on employment would likely be
associated with negative impacts on, or decreases in, recidivism. Since the transitional jobs
programs led to large increases in employment, driven by transitional jobs, TJRD provides a test
of whether employment in the form of transitional jobs — that is, temporary, minimum wage,

The National Bureau of Economic Research (NBER) dated the beginning of the recession as December
2007, but unemployment rates, presented in Figure 2.1, suggest that this decline did not strongly affect
employment until late 2008. See www.nber.org/cycles/dec2008.html for NBER recession dates.
24
Study enrollment occurred at different rates and during somewhat different time periods in the four sites.
The site that ended study enrollment earliest was St. Paul, which randomly assigned participants through May
2008. The latest site was Chicago, which enrolled participants through September 2008. Because of the
different enrollment periods, some sites are under- or overrepresented in each cohort. However, an analysis that
weighted each site equally within cohorts did not show substantially different results.

107

low-skilled work — leads to reduced recidivism. However, the recidivism impacts do not fit the
expected pattern.
Figure 6.1 shows the quarterly impacts of the transitional jobs programs on total employment and on arrests. The figure includes arrests because they are likely to be the cleanest
measure, among those available, of criminal activity in real time.25 If these results were consistent with the hypothesis that employment of any kind is causally related to crime, the
The Transitional Jobs Reentry Demonstration

Figure 6.1
Quarterly Impacts on Employment and Arrests: Full Sample

40.5***

Quarter after Random Assignment

-1.1

32.9***

2.6

13.5***

2.5
Impact on employment

4.2*
4

Impact on arrests

-0.1
-5

Impact (percentage)

SOURCES: MDRC calculations using payroll data from each site and unemployment insurance, wage
records from each of the states (Illinois, Michigan, Minnesota, and Wisconsin), and criminal justice
data from Michigan State Police, Minnesota Bureau of Criminal Apprehension, State of Wisconsin
Department of Justice, and Illinois Criminal Justice Information Authority.
NOTES: Significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10 percent.
Results in this table are regression-adjusted for pre-random assignment characteristics.
25

Unlike convictions and prison admissions, arrests usually occur at the same time as, or soon after, the
criminal activity takes place. Further, unlike parole officers, who may take employment status into account
when deciding whether to revoke parole, police officers are less likely to be influenced by or aware of
employment status when making arrests. Therefore, arrests may be a better measure of criminal activity than
parole violations.
108

figure would show an inverse relationship between employment and arrest impacts. When there
were positive impacts on employment, there would be negative impacts on arrests, and these
impacts would be largest in the early quarters when the employment impacts were largest.
However, Figure 6.1 does not show this pattern. While there were large positive impacts on
employment, driven by transitional jobs, there were no significant impacts on arrests at any
point in the follow-up period. The two sets of impacts do not appear to be related to each other.
This pattern of employment and arrest impacts suggests that employment, at least in the
form of transitional jobs, may not be causally related to recidivism. Given that there were no
increases in unsubsidized employment, these results do not provide evidence of whether higher
quality, permanent employment would produce recidivism impacts. However, these results
suggest that simply having employment of any kind may not affect a former prisoner’s propensity to recidivate. It will be important to discover whether a different pattern of results emerges
during the second year of follow-up.

Conclusion
The TJ programs had little impact on key measures of recidivism in the year following
random assignment. For the full sample, there were no statistically significant impacts on arrests
or convictions. There was a small reduction in days spent in prison, which was driven by
impacts early in the follow-up period in the St. Paul site. There was not a strong pattern of
subgroup differences in recidivism impacts by age, cohort, employment history, criminal
history, or education.
Despite variation across sites in the rates of recidivism, there was little variation across
sites in the recidivism impact results. As for the full sample, there were few significant impacts.
The exception is in St. Paul, where there was a reduction in parole revocations. This impact
appears to drive the small set of full sample impacts described above. The St. Paul results may
be related to sample characteristics in that site, where a high percentage of sample members
were on intensive supervised release, which is a more stringent form of parole.
While TJRD cannot test whether permanent, higher quality, or unsubsidized employment reduces recidivism, a comparison of the pattern of employment impacts with the pattern of
arrest impacts suggests that minimum wage, temporary employment may not be causally related
to recidivism.

109

Chapter 7

Cost of the Programs in the Transitional Jobs Reentry
Demonstration
The previous chapters in this volume reported on the implementation and impacts of the
Transitional Jobs Reentry Demonstration (TJRD) programs. This chapter presents estimates of
the cost of providing TJRD services to research sample members. The cost of services provided
to the typical transitional job (TJ) group member is compared with the cost of services provided
to the typical job search (JS) group member.
The chapter begins with a brief description of the methodology and data sources used in
the cost analysis. The costs of providing transitional jobs and related services to the TJ group
are then presented, followed by a comparison of those costs with the costs of the TJRD job
search programs to provide services to the JS group.

Methodology of the Cost Analysis
The cost analysis measures the per person expenditures for providing transitional jobs
or job search services to the research sample plus operating costs during the one-year follow-up
period.1 Participants in the transitional jobs group received services from the programs for about
four months each, on average, in Chicago, Detroit, and Milwaukee, and for an average of five
months each in St. Paul. The cost analysis considers the value of all resources required for
service provision, including staff salaries and fringe benefits, office space, supplies, and
overhead costs. The general approach is to place dollar values on all resources that are used,
whenever possible, either by directly measuring them or by estimating them. The costs are
estimated and presented as an average cost per sample member.
This cost analysis compares the cost for the TJRD transitional jobs programs to provide
services to the TJ group members with the cost of services provided to JS group members by
the TJRD job search programs. The analysis uses the cost of services provided by the St. Paul
job search program as a proxy for the job search programs in all four sites. The cost of the St.
Paul job search program, operated by the Wilder Foundation, represents the median cost of
three of the TJRD job search programs and is considered representative of the cost to run the

Because costs are estimated for the first year following study enrollment, they cover services received by
individuals during that time, but not any follow-up services received during subsequent years.

111

TJRD job search programs.2 To show how the cost comparisons are affected by the decision to
use St. Paul as the job search program proxy, results are presented using alternative costs, by
site, for two other job search programs in Appendix Table E.1.3
Data Sources
The data for the cost analysis includes in-depth interviews with program staff, detailed
financial reports from the TJRD transitional jobs and job search programs, and participation
records from each program’s management information system (MIS). Financial expenditure
reports from each transitional jobs and job search program cover fiscal year 2008; the exact
months and years varied depending on the program.4 Study participants enrolled in the TJRD
programs between January 2007 and September 2008. Therefore, fiscal year 2008 represents a
“steady state” period of operation for the programs, as experienced by TJRD sample members.
Expenditure data were combined with data from each program’s MIS on the number of individuals served in various activities during the same time period, to produce cost estimates for each
program component. This analysis presents costs per participant, assuming that each participant
receives the same average amount of services. In this study, service receipt was measured for
one year after enrollment in the programs.5

The cost of services for the Chicago job search program, which was run by the Safer Foundation, was
similar to the cost in St. Paul. The cost of Project RETURN, which served JS group members in Milwaukee,
was much lower than the costs for the Chicago and St. Paul job search programs and was unusually low in
comparison with the cost of other programs offering similar services (based on prior studies). For example,
Aos et al. (2001) found that the average cost per participant for programs offering similar services was $772 (in
2001 dollars), which is roughly equivalent to the cost of the TJRD programs in 2008 dollars.
3
The job search programs in Chicago, St. Paul, and Milwaukee provided financial expenditure data. Data
from the job search provider in Detroit were not collected.
4
That is, fiscal year expenses covered March 1, 2007, through February 29, 2008, for the programs in Detroit and St. Paul, and from July 1, 2007, through June 30, 2008, for the Milwaukee and Chicago programs.
5
To the extent that a participant receives services for longer or shorter time periods, this analysis could
over- or underestimate the actual cost of providing services to each participant. A sensitivity analysis was
conducted to account for the actual number of months each individual worked in a transitional job. The cost
that was derived using this approach was very similar to the costs presented in this chapter. The average costs,
accounting for varying lengths of participation, were within $100 of the costs presented in this chapter. The
exact difference in costs that were found in the sensitivity analysis varied by site. The costs that were derived in
the sensitivity analysis for Chicago and Detroit were within a few dollars of the costs presented in this chapter.
The cost for Milwaukee was within $200, and the cost for St. Paul had the biggest difference, with $500 less
per TJ group member in the sensitivity analysis. Thus, while it is possible that the average costs presented in
this chapter may slightly overestimate the actual cost per TJ group member, the general similarity of costs
found in the sensitivity analysis suggests that the costs presented here are likely to be an accurate estimate.

112

Cost Components
Figure 7.1 illustrates the main cost components for the transitional jobs and job search
groups. It shows that the expenditures for the TJRD transitional jobs program services (Box A)
make up the gross cost for each TJ group member (Box B). The gross cost includes (1) expenses related to the transitional jobs, such as wages, fringe benefits, and field/site supervision
expenses, and (2) expenses for other program activities, such as intake and orientation, case
management, job readiness and job search services, and participant supports.6 Box C shows the
TJRD job search program services — including intake and orientation, employment readiness
and job search services, and in some cases, participant supports — which reflect the gross cost
that accrued to JS group members (Box D).7 As described above, all gross costs include staff
salaries and fringe benefits, overhead, and administration.
The net cost of the TJRD transitional jobs programs — that is, the cost per TJ group
member over and above the cost per JS group member — is represented in Box E of Figure
7.1. The net cost is obtained by subtracting the gross cost per JS group member from the gross
cost per TJ group member.
Limitations of the Analysis
This analysis is not a cost-benefit analysis and therefore does not account for any benefits of the transitional jobs programs that might offset their costs. In other words, during the first
year, the transitional jobs programs increased TJ group member’s employment and earnings,
but the analysis does not account for any benefits from that increase, such as income tax
benefits that accrue to government budgets or higher Earned Income Tax Credits (EITC) for
participants, nor does it account for expected benefits stemming from the transitional jobs
themselves. For example, TJ group members performed job functions while working in the
transitional jobs, and presumably those job functions have some value. This analysis does not
account for any offsetting benefits of the labor output from those jobs.
The wages, fringe benefits, supervision, and administration of the transitional jobs are
presented as part of operating costs, without accounting for the funding of those positions. In
Chicago, for example, transitional job wages were paid by the transitional job worksite provider
in that site (Allied Waste) and were not subsidized by the TJRD program (run by Safer). This
6

Examples of the participant supports provided by the transitional jobs programs include transportation
costs (for example, bus passes), photo identification fees, housing rent/deposits, emergency food cards, driving
fines (including fines for reinstating a license that has been suspended), work supplies, auto insurance,
dental/medical care payments, and child support. Vocational training was offered only by the St. Paul transitional jobs program.
7
For example, the job search program in St. Paul (Wilder Foundation) provided participant supports for
transportation costs, photo identification fees, work supplies, and other employment needs.

113

The Transitional Jobs Reentry Demonstration

Figure 7.1

Simplified Diagram of Major Cost Components in the
Transitional Jobs and Job Search Programs

Transitional Jobs Program

Job Search Program

C
Transitional Job Services:

Job Search Services:
 Intake and orientation
 Employment readiness and job
search
 Participant supports

Transitional Job Expenses
 Transitional job wages
 Transitional job fringe benefits
 Field/site supervision
 Other transitional job expenses
Other Expenses
 Intake and orientation
 Case management
 Job readiness/job club services
 Participant supports
 Counseling services
 Vocational training
 Job retention bonuses

D
Sum =
Total cost per Transitional Job
group member

Sum =
Total cost per Job Search
group member

E
Difference =
Net cost per Transitional Job group
member (E = B minus D)

114

analysis includes transitional job wages as a program cost, without regard to whether the wages
were subsidized by the programs or by another source.
This cost analysis also does not account for the cost of possibly displacing other workers as more TJ group members worked in the transitional jobs. This analysis does not measure
nonmonetary costs or benefits of the transitional jobs programs, such as a possible increase in
program participants’ self-esteem and/or readiness for change.
Finally, this cost analysis includes only the costs associated with services that are provided by the TJRD transitional jobs and job search programs. It is likely that TJ and JS group
members received services from other programs in the community during the follow-up period.
The extent to which TJ or JS group members participated in other services outside of TJRD is
not known. It is also not known whether there were differences between the research groups in
participation rates in other services.8

Operating Costs
The operating costs for transitional jobs programs cover expenditures for the average
transitional jobs group member and have been estimated for all of the main components of the
program. The average operating cost per transitional jobs group member for a specified activity
was calculated by first estimating a unit cost for each activity (that is, the average operating cost
per participant in that activity). To obtain the gross cost per TJ group member, the unit cost per
participant was multiplied by the participation rate in that activity. Participation rates in various
activities, also shown in Chapter 4, varied across the sites.


The cost of providing transitional jobs program services, for the average
length of time in the program of four months, was about $4,300 per TJ
group member. Nearly half of those costs consisted of direct payments to
participants in the form of transitional job wages, participant supports,
and job retention bonuses.

Table 7.1 presents the estimated unit and gross costs of operating the TJRD transitional
jobs programs, with each cost representing the average across all four transitional jobs programs. The average cost of services was $4,309 per TJ group member, which covers the cost of
transitional jobs services over the average length of time that participants received services in

Client surveys are sometimes helpful in providing information about services received by research sample members; however, as noted in earlier chapters, this project did not include a client survey.

115

The Transitional Jobs Reentry Demonstration

Table 7.1
Estimated Unit and Gross Costs for the Transitional Jobs Group (2008 Dollars)

Cost Componenta

Average Cost
per Participant in
This Service
Component

Average Rate of
Gross Cost per
Participation in
Average
This Service Transitional Jobs
Component Group Member

Transitional job expenses
Transitional job wages
Transitional job fringeb
Field/site supervision
Other transitional job expensesc

$2,041
$202
$394
$37

85.2%
85.2%
85.2%
85.2%

$2,279
$1,739
$172
$336
$32

Other expenses
Intake and orientation
Case management
Job readiness/job club activitiesd
Participant supportse
Counseling servicesf
Vocational training
Job retention bonuses

$228
$835
$817
$130
$272
$1,120
$887

92.7%
74.0%
67.2%
76.6%
59.5%
20.8%
17.8%

$2,031
$211
$618
$549
$100
$162
$233
$158

Total costs

$4,309

SOURCES: MDRC calculations using financial expenditure reports for 2008 and in-depth interviews with
program staff, combined with data on participation rates from each program's management information
system.
NOTES: aAll sites provided data for total transitional jobs (TJ) and other expenses, but not all sites
provided information allowing these totals to be broken into individual cost components. Accordingly, the
individual cost components that are presented represent the average across sites that provided such data.
For all sites, administrative oversight and overhead costs are included under other expenses. Recruitment
costs attributed to the research study are not included in costs.
bFor example, workers' compensation insurance.
cFor Chicago, includes only bonuses provided to TJ workers for retaining TJ employment, and work
equipment (for example, safety vests).
dIncludes job readiness/job club and job development activities.
e Can include transportation (for example, bus cards), housing rent/deposit, emergency food cards,
driving fines (including fines to reinstate a suspended driver's license), work supplies, auto insurance,
photo ID fees, dental/medical care payments, and child support.
fAvailable in Chicago only, this cost and participation rate was provided by program staff and includes
counseling services for anger management and substance abuse.

the TJ program, approximately four months.9 This cost figure is in line with those reported
nationally and those found in other random assignment evaluations of two other transitional
9

The average time spent in a transitional job was 11 weeks, but those weeks may not have been consecutive. On average, the total length of time that participants spent in the TJRD program was approximately four
months.

116

jobs programs, the Center for Employment Opportunities (CEO) and the Transitional Work
Corporation (TWC).10 The CEO program in New York, which serves former offenders, showed
an average cost of $4,200 per TJ group member,11 while TWC in Philadelphia, which serves
welfare recipients, had an average cost of $3,500 per TJ group member.12
This analysis showed that the largest share of expenses for transitional jobs group
members were for costs related to the provision of the transitional jobs. The average cost for
wages was $1,739 per TJ group member. (Average cost includes zeroes for TJ group members who did not work in a transitional job.) An estimated $336 was spent on supervision and
an estimated $172 was spent on transitional jobs fringe benefits such as workers’ compensation insurance.
With regard to the gross costs for other transitional jobs program activities, the largest
share of expenses was for case management ($618) and job readiness ($549). Job readiness
costs also include those for job development and retention services because such costs could not
be disentangled from one another. (For some sites, the same program staff were responsible for
providing both types of services, which often overlapped.)
In addition, an estimated $211 was spent on the initial recruitment, intake, and orientation into the program for each transitional jobs group member; this number excludes costs
that are attributable to research-related activities. Participant support payments worth $100,
on average, were provided to transitional jobs group members for things like transportation
costs, photo identification fees, housing rent/deposits, and emergency food cards. Also, in two
of the four transitional jobs programs (Milwaukee and St. Paul), participants were eligible for
job retention bonuses, which cost an estimated $158 per transitional jobs group member.
(That number includes zeroes for TJ group members who did not receive a bonus.)13 As
described in previous chapters, the retention bonuses were special incentives for sustaining
unsubsidized employment that were offered in two of the TJRD transitional jobs programs
(Milwaukee and St. Paul).
Additionally, data from St. Paul show that an estimated $233 per transitional jobs group
member was spent on vocational training. Some participants in Chicago also received additional

Nationally, the cost of transitional jobs programs ranges from approximately $2,000 to $14,000 per person, with wages making up 13 percent to 59 percent of this total, depending on the program. See Kirby et al.
(2002) for a study of six transitional jobs programs targeting long-term, hard-to-employ welfare recipients; see,
also, National Transitional Jobs Network (2009).
11
Redcross et al. (2009).
12
Bloom et al. (2009).
13
Chapter 4 shows that among those who received a bonus, each person received an average of $618 in
payments during the first year.

117

counseling support services, such as substance abuse treatment and anger management counseling, at an estimated cost of $162 per TJ group member.
Costs by Site
Table 7.2 shows the estimated gross costs per transitional jobs group member, separately for each of the four transitional jobs programs. In terms of total operating expenses, three of
the four programs were within a range of a few hundred dollars of the average cost of $4,309
per TJ group member. The average cost per participant of Milwaukee’s transitional jobs
program, at $3,551, was quite a bit lower than the average for the four demonstration programs.
The Chicago and St. Paul TJ programs had the highest average costs per participant, at $4,756
and $4,889, respectively.
The cost differences among the sites were driven mainly by differences in the implementation of the transitional jobs model or by the provision of additional services in certain
sites. Specifically, both Chicago’s and St. Paul’s TJ program participants had longer stays or
worked more hours, on average, in the transitional jobs, and thus the costs for TJ wages and
other related expenses were higher in those two sites.14 As stated above, St. Paul also offered
extra services such as vocational training and job retention bonuses. Chicago offered extra
counseling services to participants, such as anger management counseling and substance
abuse treatment. Milwaukee’s costs are lower because fewer program group members worked
in a transitional job and because the site operated a scattered site TJ model; thus, the daily
supervision of participants on the worksites was provided by staff at the worksites, which is
less costly for the transitional jobs program.
Notably, in Detroit, the costs of other program services (such as job readiness and
case management) could not be disentangled given the structure of the financial data provided
by the site; yet, the overall share of costs for other program services represents a portion that
is in line with those provided by two of the other sites — that is, approximately half of total
operating costs.

Net Cost of the Transitional Jobs Programs
The net cost of the transitional jobs programs is the difference between the cost of the
TJRD transitional jobs program services and the cost of basic employment readiness and job
search services, as provided by the TJRD job search programs.

In addition, differences in the minimum wage across sites may also account for site differences in costs.
The minimum wage was highest in Chicago at $8.00 per hour and lowest in St. Paul at $6.15 per hour.

118

The Transitional Jobs Reentry Demonstration

Table 7.2
Estimated Gross Costs for the Transitional Jobs Groups, by Site (2008 Dollars)
Cost Componenta

Chicago

Detroit

Milwaukee

St. Paul

Transitional job expenses
Transitional job wages
Transitional job fringeb
Field/site supervisionc
Other transitional job expensesd

$3,230
$2,265
$317
$614
$34

$1,801
$1,334
$125
$342
NA

$1,906
$1,496
$217
$193
NA

$2,177
$1,824
$72
$281
NA

Other expensese
Intake and orientation
Case management
Job readiness/job club activitiesf
Participant supportsg
Counseling servicesh
Vocational training
Job retention bonuses

$1,526
$121
NA
$1,064
$119
$222
NA
NA

$2,239
NA
NA
NA
$166
NA
NA
NA

$1,645
NA
$912
$623
$45
NA
NA
$65

$2,712
$457
$781
$571
$216
NA
$319
$368

Total costs

$4,756

$4,041

$3,551

$4,889

SOURCES: MDRC calculations using financial expenditure reports for 2008 and in-depth interviews with
program staff, combined with data on participation rates from each program's management information
system.
NOTES: NA indicates that data were not available for that site.
aAdministrative/oversight and overhead costs are included under other expenses. Recruitment costs
attributed to the MDRC research study are not included in this table.
bFor example, workers' compensation insurance.
cFor Detroit, exact field/site supervision costs were not provided; thus, the same proportion of
programming costs that field/supervision costs represented in St. Paul were applied to Detroit.
dFor Chicago, includes only bonuses provided to TJ workers for retaining TJ employment, and work
equipment (for example, safety vests).
eFor Detroit, other expenses could not be separated into different components, except for participant
supports.
fFor Chicago, this category also includes case management time that could not be separated from job
readiness/job club activities.
gCan include transportation (for example, bus cards), housing rent/deposit, emergency food cards,
driving fines (including fines to reinstate a suspended driver's license), work supplies, auto insurance, photo
ID fees, dental/medical care payments, and child support.
hAvailable in Chicago only, this cost and participation rate was provided by program staff, and
includes counseling services for anger management and substance abuse.

119



The average cost of the TJRD transitional jobs programs per program
group member was $3,340 more than the cost of services provided by
the TJRD job search programs.

The first column of Table 7.3 shows the costs of transitional jobs and other programming services for the average TJ group member, as discussed in the previous section. The
second column of Table 7.3 presents estimates of expenditures by the TJRD job search
programs to provide employment readiness and job search services to the JS group. The cost
for the services provided by the TJRD job search program averaged about $969 per JS group
member compared with the TJ programs’ average cost of $4,309 per program group member.
Thus, the net cost of services for the transitional jobs group member was $3,340 over and
above the average cost of services for the typical job search group member (represented by
Box E in Figure 7.1).
The Transitional Jobs Reentry Demonstration

Table 7.3
Estimated Net Costs for the Transitional Jobs Group (2008 Dollars)
Gross Cost per
Average
Transitional Jobs
Group Member

Gross Cost per
Job Search
Group Memberb

Net Cost per
Average
Transitional Job
Group Member

Transitional job expenses

$2,279

Other expenses (job readiness, case management, etc.)

$2,031

$969

$1,061

Total costs

$4,309

$969

$3,340

Cost Componenta

SOURCES: MDRC calculations using financial expenditure reports for 2008 and in-depth interviews with
program staff, combined with data on participation rates from each program's management information system.
NOTES: aCosts include only services provided by the transitional jobs (TJ) and job search (JS) programs in
TJRD.
bBased on costs obtained from the St. Paul JS program, which represents those of all JS programs in this
analysis. See Appendix Table E.1 for costs of other TJRD job search programs.

As discussed above, the job search program in St. Paul was used to represent the average cost of services that were provided to all job search group members in this cost analysis.
Costs were also estimated for two other TJRD job search programs, and net costs, using these
alternative costs for the local job search program, are presented in Appendix Table E.1. Notably, the alternative net costs are similar to the average net cost presented here: $3,571 in
Chicago, $3,920 in St. Paul, and $3,358 in Milwaukee.

120

Conclusion
This analysis has shown that the net cost of providing transitional jobs and other employment-related services was $3,340 more per TJ group member than the average cost to
provide basic job search services to the JS group. Of this total, more than half of the costs were
reflected in direct payments to transitional jobs group members in the form of wages, supportive
payments, and job retention bonuses. The next largest share of costs was for case management
and job readiness services. Costs for sites were generally similar, with the biggest differences
being driven by differences in the implementation strategies and in the cost for “extra” services
that were provided in two of the sites.
The focus of this analysis has been on the monetary costs associated with the TJRD
programs. This analysis does not attempt to estimate the monetary benefits resulting from
participation in the transitional jobs programs, and consequently, any exploration of their cost
effectiveness is beyond the scope of this analysis. However, because the transitional jobs
programs generated few impacts on unsubsidized employment, earnings, and recidivism during
the first year, the amount of any corresponding monetary benefits that might be associated with
these programs is questionable. The programs likely added some value to the communities in
which they were launched — for example, via the “ripple effect” of having an additional source
of monetary funds flowing through otherwise disadvantaged communities.

121

Chapter 8

Qualitative Study of Transitional Jobs Group Members
In addition to the quantitative analyses that were described in the preceding chapters, a
qualitative study was undertaken as part of the Transitional Jobs Reentry Demonstration
(TJRD) project. The qualitative study was conducted to acquire an in-depth understanding of
the experiences of a selection of men who were placed in transitional jobs. Rather than evaluating the program, the qualitative study was designed to document and illustrate the experiences,
reactions, opinions, and beliefs of these men as they reunited with their friends and families,
encountered transitional employment, and then attempted to integrate more formally into
society generally and the world of work in particular. The initial sample included 33 men drawn
from all four focal cities: Chicago, Detroit, Milwaukee, and St. Paul. (See Appendix F for
demographic portraits of the sample.) The transitional jobs agency staff selected the cases to be
included (so they may reflect men who represent the best-case scenarios for each agency), who
were then contacted by the researcher for the qualitative study. Those who agreed to be interviewed made up the sample. The names given in this report for the men who were interviewed
are fictional.

The Field Work Plan
The data for the qualitative study were collected from three rounds of in-person interviews that were conducted from 2008 to 2010, as well as intermittent telephone conversations
and visits to the residential communities of 14 of the men. The men received gift cards for use at
Target department stores; they received a $25 card for completing the first formal interview and
a $20 card for each of the next two formal interviews. During the interviews, specific attention
was given to the men’s views on the world of work and work opportunity in their respective
cities and municipal regions, and their hopes, dreams, and desires in relation to the workplace,
the residential community, and (if relevant) the family. A particular point of investigation
concerned how these men thought about and managed their public identities as previously
incarcerated men and how the transitional job informed their efforts in that direction.
The first round of on-site interviews took place during January through March 2008.
(Accordingly, all references to the men’s ages are based on the year 2008.) They were conducted at the transitional jobs program agencies. All of the first-round interviews took place
with men who were actively engaged in their transitional jobs; the researcher for the qualitative
study interviewed 17 of the men, and a research assistant who was trained in qualitative research interviewing techniques conducted 16 of the interviews.

123

While the men did not always provide all the requested information, the interview questions covered each participant’s history of family relations, educational attainment, schooling
experiences, and employment prior to incarceration. Each participant was then asked about his
history of convictions and periods of incarceration, and was asked to provide a general account
of his incarceration experience, in particular employment-related training or schooling opportunities that he may have pursued while incarcerated. The next portion of the interview involved
documenting each participant’s assessment of the local world of work and work opportunity to
ascertain the extent to which each respondent maintained ideas about the kind of job opportunities that existed in his municipal region and the extent to which those perceived opportunities
meshed with his desires and interests concerning longer-term employment. Each participant was
then asked to discuss expected family obligations or personal interests that he wished to pursue
now that he was no longer incarcerated. The interview concluded with each participant’s
assessment of the social support that he believed he would need to secure a good job and (if
relevant) to contribute to his family’s and his own well-being. The interviews also addressed the
participants’ difficulties in dealing with the societal and technological changes that had taken
place during their incarceration and for which they were unprepared.
In the year that followed the first-round interviews, the researcher for the qualitative
study had follow-up telephone conversations with 20 of the participants.1 These conversations
provided brief reports on the men’s experiences with their transitional jobs as well as any issues
that pertained to their transition from incarceration. Included in these conversations were
updates on the men’s experiences with any additional or subsequent work, family interactions,
potential changes in their outlook toward work and family, and potential changes in their
perception of support needs.
The second round of interviews took place in the late winter of 2009. With the exception of the Detroit-based sample, these interviews were also held at the agency that was charged
with assigning the transitional jobs; the Detroit participants were interviewed at public establishments in or near their areas of residence. This second round of interviews generated more
information about work and family experiences since the first interview, as well as each man’s
assessment of the TJRD program from the perspective of having completed the transitional job
some months earlier. The second round of interviews also allowed for more thorough documentation and investigation of any changes that the men may have had in their outlook toward work
and family, and in their perceptions of social support needs since the first round of interviews.
The third round of interviews took place in late winter of 2010. The Chicago and Milwaukee interviews were held at the transitional jobs program agencies. The Detroit interviews
1

Telephone interviews were conducted once with 3 of the men, twice with 12 of the men, and three or
more times with 5 of them.

124

took place in public establishments in the men’s residential communities. None of the St. Paul
participants could be contacted for a third round of interviews in time for their inclusion in this
report. However, caseworkers provided telephone updates on the status of some of these men.
As was the case with the second-round interviews, the third-round interviews generated
documentation of each participant’s account of more recent work and family experiences,
assessments of the TJRD program since completing the transitional job, changes in outlook
toward work and family, and changes in perception of support needs.
The Data Analysis Design
With the assistance of a research team of University of Michigan students, the interview
data were analyzed by the application of a coding scheme that categorized the commentary
according to the topics described above — views on the local world of work, expectations and
desires for future employment, expectations and desires for future quality of life, experiences
with the transitional job, and interpretations of the social significance of having been incarcerated. The small sample allowed for manual coding of the data, which also allowed for a more
intimate connection with the cases than is often possible through computer-based qualitative
data analysis programs.
Comparative analyses of the data by subgroups were done according to the four sites
where the demonstration took place, project city of residence, patterns of incarceration (single
terms of 10 years or more, serial terms totaling 10 years or more, single terms of 9 years or less,
serial terms totaling less than 9 years), and age group (39 years and younger; 40 years and
older). The data that were collected during the second and third rounds of interviews also
compared men who were employed after their transitional job experience with those who were
not employed. The paucity of racial and ethnic differences in the sample prevented meaningful
comparative analyses along those lines.
Limitations and Caveats of the Qualitative Study
The relatively small sample for the qualitative study (33 men from more than 1,800 in
the full program evaluation) prevents formally generalized findings. Accordingly, the qualitative study provides in-depth accounts by men who experienced the transitional job phase of the
demonstration project, with detailed self-portraits of their transition from prison to working in a
temporary, subsidized job to their life situation two years after their exposure to such jobs.
Unless otherwise stipulated, each finding reflects a simple majority sentiment of the men in a
particular subcategory (for example, men from the two large cities or men who experienced
incarceration for a period of less than 10 years). Minority perspectives are given when they
offer particularly illustrative or insightful information.

125

As the qualitative study involved four sites, less information was gathered about the
men’s experiences at work following the transitional job and in their places of residence than
would have been collected had the study occurred in one site. Furthermore, as the sample
diminished over the three years of field work, the qualitative study draws upon fewer men (13
of the original 33 cases) for its year-to-year follow-up.
Finally, as in the full sample, the majority of the men in the qualitative study were
African-American. Both the qualitative researcher and the research assistant who conducted a
portion of the first-round interviews are also African-American, and both have extensive
research exposure and life experiences in communities that are similar to those where the men
resided. These similarities facilitated interaction and conversation with most of the men in the
qualitative study. In the course of the field work, the interviewers had extensive informal
conversations with the study participants (often involving sports, local and national politics,
and the tribulations of family life and raising children) that supplemented and enriched the
formal exchanges. Those conversations are referred to where appropriate to add depth and
substance to this chapter.

Core Findings from the First Round of Interviews, 2008
During the winter of 2008, formal, in-person interviews took place with six men from
Chicago, eleven men from Detroit, eight men from Milwaukee, and eight men from St. Paul. In
addition, a number of interviews with the men took place via telephone during the spring.
Return to Society
The common sentiment of the men who were interviewed was that, as previously incarcerated men, they were at an extreme disadvantage compared with most job seekers. Gary, who
was 35 years of age and had served six years for sexual assault in St. Paul, sounded like many
of the men who were interviewed when explaining his outlook:
I think it’s more of a challenge for people coming out of prison with their history,
criminal history; it’s a little bit more difficult because, you know, we’re having
people who have clean records who’s been at jobs for years and now they’re losing
their jobs and out there trying to get a job . . . They would probably be more likely
to get a job before somebody who, you know [has been in prison].

Another common sentiment was that reentry into society meant having to acquire all of
the resources and accoutrements necessary for job readiness as quickly as possible, because
incarceration had precluded easy access to such things as identification, appropriate clothing for
work, transportation, a place to live, and so forth. Brock, a 39 year-old Detroit man who had
served 20 years for killing a man while engaged in a narcotics transaction, put it as follows:
126

It’s been kind of hard to get my state ID and stuff like that [because] I was incarcerated for a long time, so the transition’s been pretty, you know, hectic . . . I
mean, if you go by what’s on the news every day, jobs are scarce, you know what
I’m saying? People are striking all over the place. People getting foreclosed on . . .
It’s barely people making ends meet, trying to keep their heads above water.

In assessing his situation, Brock went on to say:
It was a lot easier being locked up than it is being out here . . . Being locked up,
even though it’s drastic on the mental [level], you get three meals a day, you can
sleep in a warm bed, warm room, you got your TV, you don’t have to . . . really
do nothing but get up, go to school, maybe get a job, you know what I’m saying?
And you’re confined to that one little area. But you come to the streets, man, you
got to hustle to get a state ID, and what I mean by hustle, you got to drive here, try
to get this paperwork, do this, try to make sure all your paperwork’s straight, and I
still ain’t got a state ID. Trying to get your driver’s license, trying to get a place to
stay, trying to get some medical insurance, trying to get benefits coming in.

The men with shorter sentences (fewer than 10 years) were less anxious about acquiring the necessary materials to reestablish themselves as citizens and potential employees; for
the most part, they already possessed many of those items. Hence, Brock’s comments reflect
one end of the spectrum of reactions to reentry. However, even those who served considerably shorter sentences spoke of the hurried pace that is necessary for getting prepared for life
after incarceration.
In a rather telling commentary, Walter, a 23-year-old man who was incarcerated for
narcotics distribution and who returned to Milwaukee after his release, said the following about
his incarceration experience: “It humbled me, because I know that you can get everything took
away from you and they don’t really care about that. So, I know that there is another side to life
besides the streets and being out. It can always get took away from you, so it’s very humbling.”
Walter’s observation points to the stress and difficulty that formerly incarcerated individuals
face when entering new social and emotional situations upon their release from prison.
Barry introduced another dimension of the humility that many men experience after
serving time:
What I’ve seen is that a lot of guys feel like there’s something owed to them
when they come home . . . Some of them just feel like everything’s supposed to
fall in place just like that, you know, job, house, wife, and it doesn’t happen that
way. There’s nothing for me to be ashamed about, so I don’t try to hide it at all.

These remarks help elucidate the mindsets of the men upon their initial reentry into society. Essentially, prior to incarceration, most of them saw themselves as highly efficacious
men. For most of them, the past was a time when they were extremely capable of acquiring

127

what they wanted to get out of life, at least in terms of material resources. However, incarceration put them into positions of dependency and subordination, and upon being released they
began to wrestle with how to assume their independence again. Yet, that often meant that they
had to behave in ways that were largely foreign to them or inconsistent with their past experience. Hence, what was at stake for them was not simply finding secure positions in the world
of work, but attaining security and comfort in their private lives. The need to balance those two
goals, and the difficulty of doing so, was apparent in everything they said about their experiences with transitional jobs in particular and their return to society more generally.
Social Support During the Reentry Process
Virtually all the men who had served either a single short-term sentence (less than 10
years) or had experienced cyclical patterns of incarceration and reentry for a period of less than
10 years reported having more elaborate and more reliable social networks of support (family,
friends) upon their release from prison than did the men who had served longer terms. They
viewed this support base as equal in value to their transitional job in allowing them to begin to
adjust to society. In contrast, those who experienced long-term incarceration expressed significant concern that they had few people to whom they could turn for assistance while making the
transition back to society. These men were more emphatic about having to find immediate work
upon completing their transitional jobs, as they were not secure about whom they could rely
upon for various kinds of support (particularly financial) once the transitional job ended.
Theodore, from Milwaukee, provides one example. At 43, Theodore had served multiple terms totaling 13 years for narcotics distribution. He mentioned his ex-wife as his only sure
form of social support upon his release. As he explained:
If you try to do it by yourself with a background like mine, it’s depressing. It’s
not good, and you’ve got to take a lot of nos. This is anybody. Even without a
criminal record, you’ve got to take a lot of nos. But, with a criminal background
like mine, trying to do it yourself, it’s depressing.

In fact, there was a lot of discussion from these men about the importance of social
networks in adjusting to their new lives. The men with more extensive social networks were
considerably more comfortable about asking for support, and each believed that the various
people who made up his base of support would help him in his efforts to find employment and
to secure other resources that were essential for his independence and stability. Those who
believed that they had less support were much more preoccupied with securing employment. As
far as they were concerned, they could rely only on themselves. Hence, they insisted that their
survival depended upon securing consistent and rewarding employment. Self-sufficiency, then,
was considerably more important for men who believed that they had fewer social contacts and
resources to rely upon.

128

Coming Back to the Neighborhood
The men from the two larger cities (Chicago and Detroit) were very focused on the
need to remove or buffer themselves from the social encounters and contexts that proliferated in
their communities and that often led to incarceration. A significant concern for these men was
how to avoid interactions with the potentially problematic people and circumstances that surface
in municipalities with more extensive crime and poverty problems.
Martin, a 21-year-old man who returned to his Detroit neighborhood after serving a
four-year term, said:
People are being robbed and all kinds of stuff, and I wouldn’t want to get robbed,
you know? That’s why I think . . . [returning to the old neighborhood will] affect
me. The guy that got killed in front of my building — they robbed him and they
killed him. And I think being over here and if someone tried to rob me, me being
an ex-offender and all that and me with my personality, I’m not just going to sit
around and let somebody just rob and kill me.

In comparison, the men from Milwaukee and St. Paul, the two relatively smaller
cities, were less concerned about potentially negative community effects because they had
either relocated to areas that (at least as they saw it) were not characterized by crime and
delinquency, or they believed that they were capable of avoiding problematic circumstances
in their original neighborhoods because they considered those neighborhoods to be much
less saturated with crime.
Thus, the men in Chicago and Detroit were much more emphatic about securing jobs
that could allow them both to relocate and to earn sufficient material support. In contrast, the
men from the smaller cities rarely mentioned neighborhood relocation as an immediate goal.
Confronting Societal and Technological Changes
The length of time that men served in prison also distinguished the way they described
their reentry experiences. One topic of intense preoccupation for them was the extent to which
they viewed technology as overrunning the work sphere. They often said that the workplace,
and society more generally, “moves faster today than it did before.” Men who had served 10 or
more years described particular challenges related to the technological transformations that had
taken place during the time they had been incarcerated, and they were the most anxious about
being able to reenter a society and workplace that were vastly more computerized than when
they were first incarcerated.
Vasquez, a 37-year-old man from Chicago who served 17 years for his involvement in
a gang-related murder, explained:

129

Man, the technology has changed dramatically; it’s like, that stuff is crazy.
You know, trying to come back and get used to the cell phones and the computer. When I first got out, I tried to go get an application, they sent me to a
computer, you know, there’s no more paper. And I’m like, “What am I supposed to do with this?”

All the men who served lengthy terms were especially anxious about how they would
learn to function in a vastly more technological society. In fact, Brock reported that in the first
weeks of his release, he struggled to ride the bus in Detroit because he was not familiar with
electronic fare cards, which had replaced the paper tickets that were in use prior to his incarceration. Brock said that as a grown man he felt embarrassed about asking people at bus stops to
explain to him how to use devices that seemed to be so common for everyone else. His brief
account reflects the degree of anxiety that the men felt upon their release, much of which
pushed them to reconsider how prepared they were to reenter society and the world of work.
In another example, Mel, a 56-year-old man from St. Paul who served 17 years for homicide, said, “So, I figured I had a pretty good handle on computers except for the fact that I
knew nothing about the Internet, nothing about e-mail. It caught me pretty much off guard.”
The feeling of being behind the times that was shared by those who served longer terms
permeated their vision of the contemporary world of work in their respective municipalities.
Here, however, many of the findings were tied to the fact that almost all of these men had life
experiences that — whether by choice, given their involvement in illicit activities, or by
circumstance, given the kinds of deprived neighborhoods in which they resided, or by a combination of both — had led to their exclusion from mainstream employment. Many of these men
spoke of their prison terms as having “frozen them in time.” Upon release, they said, the pace of
life seemed to pick up rapidly, and this meant that they had to find ways of “catching up.” In
particular, many of them were unsure of what they needed to know about technology to secure
satisfactory employment. Accordingly, for the majority of these men, much of their concern
about employment was an insecurity about their ability to function effectively in both society
and in the modern world of work.
Envisioning the Modern World of Work
When looking at the way in which these men conceptualized the world of work that
they aspired to enter upon completing their prison terms, one common finding stood out among
all others —that as previously incarcerated men, their prospects for work would not be as good
as those of non-offenders.
As Martin, who served a four-year term, explained, “I feel like it ain’t possible [to find a
job], because I got a lot of family members that never been to prison with a high school diplo-

130

ma, and they can’t even find a job, so I know with me [it will be even harder], and I got a
felony.”
Aside from recognizing the paucity of jobs, the men offered very little in the way of explaining what was available or even possible in various municipal labor market sectors. That is,
they had extraordinarily little to say about precisely what kinds of jobs they believed to be
available in their respective cities and whether the availability of such jobs could survive the
economic downturn. Essentially, the men expressed a strong desire for employment in order to
help them achieve stability in their lives, but they demonstrated very little understanding of the
world of work to which they aspired — for example, they had very little sense of what kinds of
jobs may be more or less plentiful in the skilled blue collar or minimally skilled service sectors.
Similarly, they demonstrated little or no knowledge of strategies for securing work beyond
relying on case mangers for information about jobs and, as one man said, “just walking the
streets” to try to find employment.
Stephon, a 32-year-old man from Milwaukee who had served a total of six years for
multiple terms, said:
To be honest, you know, when you’re getting out of jail, you have to try to readjust
back into the world. Once you get locked up, it’s like you’re dead, because you’re
gone, you don’t hear nothing . . . When you get back out there, it’ll be hard for you
to adapt because the only thing you know is one thing, and that’s fast money. You
don’t know how to sit at a job and work at a job eight hours and collect a paycheck
every Friday. You don’t know how to do that, because the one thing you’re used to
is getting that money every day, fast money coming every day . . . Temptations is
out there, man.

Like a number of the men who spoke in similar terms, Stephon was trying to make the
case that not only were men like him at a strong disadvantage in having to adjust to the contemporary world of work, but that they were constantly comparing their present-day disadvantage
with the memory of their former ability to garner material resources.
When first interviewed, each of the men who participated in the qualitative study was
actively involved in a transitional job. Hence, whatever problems they identified in their vision
of the modern world of work, they still had had some slight engagement with it. Unpacking that
vision more fully, then, necessitates some consideration of how they thought about that job
experience in relation to the larger world of work, which is described in the next section.
Engaging with the Transitional Job as Preparation for the World of Work
When asked to explain what they found interesting or even worthwhile about the transitional jobs program prior to entering it, the men said they believed that their participation would
allow them to begin building work histories — which, because work history is an important
131

stepping stone to better and permanent employment, they saw as the essential step toward
establishing economic and emotional stability and their principal need upon being released from
prison. In a comment that applies to virtually all the men in this study, Sam, a 34-year-old man
from St. Paul who served multiple terms totaling eight years, said, “Yeah, he [his parole officer]
said they give you a chance to build yourself back up and also your work history. I’m like, man,
that’s pretty good. To work yourself up, that takes time.”
A few others said that they hoped to acquire some formal work-related skills that they
never before had had a chance to cultivate. Mel reflected:
[You need] skills. You need to have a certain kind of personality. You need to
be able to present yourself as employable . . . You need to show them that, no
matter what your past is, that the future is not my past, and that I have goals, I
have skills, I have what they need. I need to impress upon them that I can be
the employee that they need.
And Chris, a 38-year-old man from Milwaukee who had served 20 years over multiple
terms, observed:
Transitional jobs do help you, especially if you’re getting out of prison because it
motivates you to work. It motivates you to work, and then, through them four
months that you’re working, you realize, yeah, I get a pay check. Now I get a
pay stub. I ain’t got to go sell drugs or rob nobody. I work for a living now.

Overall, these men were pleased to be earning an income but were thoroughly unsure
about how that work experience would prepare them to get the kind of skilled blue collar jobs
they hoped to find (which often were higher-skilled, factory-based jobs that involve working
with machinery or small-scale construction). These men were not thrilled by their work, but
they explained that they viewed their jobs as holding places until they could obtain more
rewarding employment; they consistently said that their work gave them few skills that would
be relevant for better employment.
Perceptions by Age
The men’s perceptions of how and whether their transitional job experience would prepare them for the world of work differed by age group. The men in the study who were over the
age of 50 were considerably more anxious about finding unsubsidized employment and beginning the process of constructing personal work histories — with or without a transitional job —
because they were acutely aware that they probably had fewer years to do so compared with
younger men. Many of them referred to the challenge of beginning to build a work history past
a point when others in their age group were thinking about retiring.

132

An example came from 53-year-old Joseph, who was from Chicago: “Because of my
background, and I don’t have no high school diploma . . . You know, they want…somebody out
there, you know, young, not old, not no 53-year-old man. The age make a lot of difference.”
These men explained in great detail and with strong conviction how they were concerned that their bodies could not hold out for many more years. Hence, age was a central
indicator of anxiety about the prospect of securing full-time employment, regardless of the
transitional job.
Quality of the Work Assignment
When turning to the actual work experience, there was a clear dichotomy of views
about the transitional job experience. The critical distinction was tied to the different qualities
of the work assignments. The men in Detroit and St. Paul (who all whom worked for the
Goodwill Industries agency in their respective cities) worked either in retail (St. Paul) or in
light-scale factory work settings (Detroit and St. Paul). Each of them expressed slightly more
comfort about their transitional jobs than did the men in Chicago. The Chicago-based men,
who all worked in transitional jobs at the recycling plant, were the least enthusiastic about
their experiences.
As Terrence, a 36-year-old Chicago-based man who had served a series of short sentences for nonviolent crimes, explained: “I ain’t got no insurance . . . I’m asthmatic, I got sinus
problems. The job that I’m working now, it bothers my sinus, it bothers my asthma. They don’t
care. They work me to death.”
The more middle-of-the-road perspective was offered by Travis, a 62-year-old sex offender in Milwaukee who worked in a retail capacity. He explained, “Work felt great because I
was able to do something and . . . even though I’m on a monitor, the freedom of it was there.”
For Travis, then, this freedom outweighed any negative aspects of the work experience itself.
Thus, while the men who were working in retail or light-scale factory sectors voiced
concerns that their jobs did not provide them with work-related skills that could help them to
gain better employment, they certainly did not find their jobs as difficult or as disappointing as
those in the recycling plant. Unlike Travis, however, almost all of the other Milwaukee-based
men worked in a more quasi-professional capacity, as peer counselors to adolescent offenders,
and they appeared to be considerably more satisfied with their work experiences. In fact, they
consistently talked about how their work as counselors both provided them with a professional
identity and allowed them to employ a set of skills (involving building social connections with
troubled adolescents) to perform a job that they believed that were uniquely qualified to do.
Indeed, the Milwaukee men who worked in this capacity all said that they hoped to continue
doing that kind of work in a longer-term capacity following the transitional job.
133

One of these men, Hakim, who had served a 20-year term for narcotics distribution, said:
Actually, it didn’t really feel like work at all, because, I don’t know, it just
seemed like something I was always destined to do . . . Community outreach
programs and just working with other community-based organizations in the
neighborhood, I mean, that’s just what I’m about . . . We were doing a gang
summons — pre-summons — trying to put together community workshops for
everybody to come out so we could stop the violence.

Basil, a 47-year-old man who had served a total of nine years for multiple offenses involving narcotics distribution, sounded more like the men who were incarcerated for longer
periods:
I really don’t care too much of what I do at this point. I just want to work. It’s
simple for me. Some people think that that’s probably taking things lightly, but
it’s not. It’s just that in my situation, I’ll take a laboring job. It pays my bills. It
keeps me out of trouble.

Indeed, most of the men who served lengthier sentences said that they simply wanted to
work; their preoccupation was almost solely on getting a job rather than investing much thought
in the kinds of jobs they preferred.
Short-Term and Long-Term Goals
The men were also asked about their short- and long-term goals concerning employment and overall quality of life. Not surprisingly, each of the men aspired to jobs that included
benefits packages, at least some degree of upward mobility, and some degree of freedom and
independence in the workplace. In terms of overall quality of life, the long-term goal of the
majority of the men was to become a homeowner, and for many of those in Chicago and
Detroit, the desire was to own a home in a residential neighborhood or suburb that was less
“congested” than their current neighborhoods in the inner city — that is, they wanted to live in a
neighborhood that had less temptation for pursuing or being exposed to criminal activity.
Each of the men was quite aware of the fact that he was far from being able to achieve
his ultimate goal of home ownership. Accordingly, each spoke quite specifically about shortterm objectives concerning work and residence. As for the short term, which meant the period
following the completion of the transitional job, the men sought secure blue collar employment
opportunities. Those who served shorter sentences were more elaborate about the kinds of jobs
or careers they sought. Sam, for instance, said that he wanted to acquire a forklift operating
license and place himself in a sector where he could make use of that skill. And Gary, who had
served a six-year term, said that he would be comfortable starting a job at $10.00 per hour, but
that he would prefer to earn $13.00 or $14.00, and that he wanted to work as an installer of
copper cable or some similar material.
134

Overall, and given the quality of the job market at the time, these men were quite clear
that lower-tier jobs were most likely in their immediate future. Gary essentially spoke for them
all when he said:
There are not a lot of good jobs. The type of jobs that exist for people like me are
warehouse, some construction, a lot of manual labor, you know, no type of —
just manual labor type of jobs I guess, you know. Slaughterhouse, you know —
jobs of that nature. Nothing . . . in the kind of . . . corporate world where you can
kind of break in, you know, show people that you are worthy of having a job of
that nature.

For most of these men, then, there was no strong connection between the transitional
job and their ultimate desires in the world of work. Transition for them meant stepping into a
job with better wages, prospects for upward mobility, and more intrinsic rewards from the work
experience. As they saw it, however, navigating that transition involved dealing with their very
public identities as previously incarcerated men.
Managing the Public Identity Associated with Previous Incarceration
At the very least, individuals who are released from prison and are seeking work are required to inform potential employers of their incarceration history, which can make it difficult
to get hired. Basil observed:
Mostly it’s the felony conviction that hinders employment for me . . . Felon,
just the word, period, frightens people because they don’t really know. Some
people don’t give you a chance because of the word “felon.”
Martin, a 21-year-old participant from Detroit who had served a four-year term, explained:
I’ll go in the store and ask them for an application. I’ll ask them if they’re hiring
and they say, “Yeah.” Then I ask them straight up, right there, “Do y’all hire
people with felonies?” They be like, “No, I don’t know. I don’t think you’re
going to get the job.”

As might be expected, these men regarded their former incarceration as the primary obstacle standing in their way of getting a job after the transitional job ended. All the men had
strong hope that having a transitional job would minimize the social stigma of having to
approach the labor market as previously incarcerated men. Yet almost all of them explained that
they were beginning to realize how powerful that status is in the job search process. Vasquez,
from Chicago, commented on the durability of the negative public image maintained by
previously incarcerated men:
I did 17 years in prison for murder. So, when you fill out the job applications,
people look at that and [I] guess they’re scared. What they don’t know, though,
is I’m a valuable worker. The nine months I was working (at the transitional job
135

in recycling, and then for a few months afterwards at the same location), I never
missed a day. I was there, never late, all the time, but they don’t see this. It’s hard
to land a job interview, because people look at the job application and see that
murder and that scares them off. They don’t know I’m one of the best workers
they’ll ever have, but I can’t get to the job interview, so I think that’s an obstacle.

He went on to say:
What people don’t know is that the average person that goes to jail for a murder
is not necessarily a killer. It’s not like we go around killing people, we’re serial
killers. We just made an accident. You know, when you coming up in the lifestyle of the gang life, you know, you run into accidents where some person may
get killed, but it doesn’t mean that our intentions was to kill them, you know, it
just so happens he got killed. When you young, you don’t know what you doing
. . .We’re not killers. We go to jail and we realize we’re not killers. We go to jail,
we do the time and we get smarter, but we’re not killers, man.

At the opposite end of the incarceration continuum was Terrence, who had been incarcerated for nonviolent crimes. As he explained, “I ain’t hurt nobody or kill nobody, so it
shouldn’t be held against me for life for something that I was doing when I was young.”
Many of these men believed they could best counter the negative images associated
with previously incarcerated men by developing positive identities as constructive and committed employees in the world of work. Accordingly, for these men the value of work was both
what it delivered in terms of material support and the role that it played in their effort to construct a more sanitized public identity. In particular, the men who had been convicted of sexual
assault and homicide/murder were the most emphatic about the significance of securing
employment as a way to recreate themselves as worthy and positive public citizens. What was
particularly interesting was the extent to which the men were convinced that others would judge
their character based on their good habits and high levels of performance at work rather than on
their criminal history. By way of example, Gary said:
People can make a mistake in their life, but it doesn’t mean that their life is over.
Being an ex-offender, I mean, it’s whatever emphasis you put on it. If you put a
negative emphasis on it that you can’t achieve anything, then that’s what it is.
That’s how I look at it.

When asked if he believes that others look at him in a particular way because of his incarceration, he said:
I’m sure they do initially until they really get to know me, get to meet me, really,
and give me an opportunity to express myself and, you know, just talk to them.
Basically people judge me, first of all, on the ex-offender thing — oh, he’s an
ex-con or whatever. Then, you know, once they get to sit down and have a dis-

136

cussion with me and give me a chance to really express myself, you know, I
hopefully change their minds.

Finally, Sam was the only man in the qualitative study who believed that he could put a
positive spin on having been incarcerated, especially when coupled with his participation in the
transitional jobs project. He said:
I look at it [being in prison] as an experience. It was an experience for me because maybe what I did was wrong and I admit that, but maybe it had to happen
for me to get to this point, you know? If it wouldn’t never have happened, God
knows where I’d be at now. So, right now, being an ex-offender is a great thing.
I don’t mind ever telling anybody that I’m an ex-offender, and I smile and I feel
good about it. It’s a great thing because a lot of positive has came out of the
whole ex-offender thing. I mean, you still [find] doors being closed, but a lot
more has opened, you know, because people see your dedication, your hard
work, effort to not go back to where you came from. So, the whole ex-offender
thing, I’m okay with it.

While working in the transitional jobs and immediately after the jobs ended, these men
were strongly focused on establishing various kinds of security for themselves — financial
security in particular, but also emotional security and stability, and the social validation of
themselves as decent people. Furthermore, when they talked about their transitional job experience and their future possibilities in the world of work, they did not focus solely on wages, but
also on the other things they believed they needed (such as family and friends, better skills, a
better public persona, a home, a better neighborhood) in order to attain these varied forms of
security and stability.

Core Findings from the Second Round of Interviews, 2009
During 2009, formal interviews were conducted in person with 13 of the original 33 research participants — as many as could be contacted directly in all four sites. These interviews
were supplemented by periodic phone conversations with 20 of the men. The pattern and results
of the site visits were as follows:
Detroit. Three of the original 11 men in Detroit were interviewed for the second round,
and updates were obtained from relatives via the case managers of the remaining eight
participants. Two of the three men who were interviewed in this sample were working
in full-time jobs at the time of the second interview.
Chicago. Three of the original six men were interviewed for the second round in Chicago, and updates were obtained on the other three cases. Only one of the six men from
this site who were originally interviewed in Round 1 was working full-time by midsummer 2009, according to a later telephone interview; he was self-employed, doing
137

small-scale landscaping and outdoor work. Another died in a house fire in March of
2009. Two were terminated from full-time jobs for inconsistent attendance. The two
others who were interviewed during this second round were working part time in their
original transitional jobs at the recycling plant.
Milwaukee. Of the eight men who were interviewed in Round 1, five were interviewed
in Round 2; updates were obtained on two; and one dropped out of the program shortly
after the first round of interviews was conducted. Of the five men who were interviewed, two were working in full-time jobs by mid-summer 2009, according to telephone interviews that were conducted later; two were not working (one had been laid
off from a retail trade job in the spring of 2009); and one began receiving disability
benefits in 2009. Two men were reincarcerated; one, who was interviewed on the telephone, was reincarcerated for domestic abuse and robbery, and no information was
available for the other reincarceration case.
St. Paul. Second-round interviews were completed with two men of the original eight;
updates were obtained for five men; and no information was available for one man. The
two interviewed men and one man who was not interviewed were unemployed. One
man had been reincarcerated for violating the terms of his parole for a sex offense conviction. Three were working at full-time jobs.
One Year Later: The Return to Society
One year after the first round of interviews, the men who left prison and returned to
what they argued were fairly extensive networks of support had begun to rely more heavily on
those networks for finding full-time employment, since their transitional jobs experience had
not provided the means to do that. They expressed gratitude that those networks were available
to them, and reported that they often felt as though they could count on nothing else, given the
lack of job opportunities that were available to them. As Chris explained:
Everybody’s still supportive of me. If we have a problem with each other, we
always be able to sit down and talk about it. Yeah, they was real supportive
when I was out of a job. Everybody always looked out and did what they could,
and if they had it, they always asked me if I needed anything, but I was the kind
of person, I would end up saying, no, I don’t need it, when I may have did. I just
let my pride get in the way.

A different perspective, from someone who had less of a social support system in place,
was offered by Sam:
Sometimes I’ll just be a little too independent and trying to do it myself, so I kind
of lack on that [knowing how to rely on others], you know? It’s been pretty
much me and, you know, in a way, I like it. I enjoy it because it’s something new
138

to me because the things I used to do and the people I used to be involved with
and things like that — you know, it’s [being more independent] something new
to me. It shows me experiences, it shows me responsibility, and it kind of helps
me, you know, stand up as a man and accept whatever and just don’t give up because a situation is bad, you know?

That being said, one year after the first formal interviews had been conducted, the men
essentially reported little to no significant change in their patterns of social relations. Moreover,
each of the men who was interviewed in the second round reported little to no tension, pressure,
or compulsion to return to the social relations and activities that had led to their incarceration.
Almost to a person they stated that they kept an appropriate distance from distracting or threatening situations, and had by now created public identities in their neighborhoods as men who,
as one participant put it, “no longer lived that kind of life.” Some reported that they had casual
interactions with old friends and associates who continued to engage in illicit or delinquent
activities, but those interactions, the men reported, were very inconsistent and intermittent. For
instance, Walter said, “It’s a lot of people that have offered me things and tried to get me back
into the game, but I’m ‘no.’ They took too much from me the first time.”
Envisioning the Modern World of Work
In the second year of the study, the interviews generated comments about envisioning
the modern world of work that were in most respects very similar to those made during the first
round of interviews. The men continued to have a strong sense of their disadvantage in the labor
market as formerly incarcerated men. In addition, they were keenly aware of the larger economic crisis that affected employment prospects in their respective cities and in the country more
generally, and they knew that work opportunities were severely restricted because of the state of
the economy. Furthermore, and more problematically, the men had no sense of how to navigate
this crisis in order to find consistent employment. As most of them had remained on the margins
of the work world during the preceding year, they had little to draw upon to enrich their understanding of possibilities in local labor market sectors. Many of the men simply said, “It’s just
bad out here. I don’t know what to do about it, but it’s just bad.”
Engaging with Work One Year After Leaving the Transitional Job
The second-round interviews revealed that frustration and anxiety were beginning to
emerge for the men who were not working. Many of them commented that they had begun to
question whether their earlier investments in the transitional jobs program provided them with
any meaningful advantages, given their lack of success finding work.
Garfield, a 39-year-old man who had served 14 years for aggravated sexual assault
while high on drugs, described his experiences trying to find work after completing his transi-

139

tional job in the retail store of St. Paul’s Goodwill center. He found a job in retail that was so
distant from his residence that it involved hours of round-trip transportation to get to and from
work. As he explained, “It didn’t work out too good. Just transportation-wise, I wasn’t driving.
There was a bus. I could have been responsible and got on the bus every day, but I just said it
was too much of a hassle.”
Transportation problems factored into a number of cases when the men either struggled
with the jobs that they found after the transitional job experience or could not find work at all.
As Benjamin, a 32-year-old man from Chicago who had served multiple terms totaling 11
years, said:
The reason why I lost that job was because it was out in . . . Cicero [a western
suburb of Chicago] and I was staying on 51st and State [on the south side of
Chicago] at the time . . . I had to be to work two hours early on the bus, and it
was cold out and my car had broke down at the time. And one day it broke down
and I didn’t get to work and I lost the job.

While none of the men reported having to resort to criminal activity, about a third of
them said that living for a year without work caused them to think more often about earlier
periods in their lives when their illicit activities and involvements allowed them to be more
economically secure. As Jerry, a 44-year-old man who had served a total of 19 years for various
sentences in Detroit, put it, “If I feel that I can’t get a job, I can always go to dope because that’s
always a job, you know what I’m saying?”
And Tony, a 32-year-old man who had served 10 years for narcotics distribution in Detroit, said:
I got a lot of friends who are still in the life now . . . or just selling little bags here
and there to make their little extras or whatever, which most of the people who I
know who are working right now, pretty much that’s what they do for their little
extras, because it’s hard out here, man. I used to hear people say that all the time,
like, man, it’s hard . . . Now [as a result of being unemployed], I see what they’re
talking about, because they was trying to do the right thing. It’s hard to do the
right thing, man. It’s easier to do wrong.

For about half of the men, the security of the transitional job resulted in an emotional
letdown when full-time employment did not follow. As Theodore explained:
That kind of surprised me in a negative way . . . not being able to get employment shortly thereafter this one [transitional job], because I thought I was on a
good roll, you know? So, that’s kind of surprising . . . they [employers] just
haven’t called me and I’m circulating résumés and applications and I don’t
know their reasons, you know? I’m kind of surprised that nobody would even
throw me a bone by now . . . I don’t know if it was because I got hooked up
with New Hope [doing offender counseling in Milwaukee], which made my
140

transition kind of not so bad, but now that I’m not with them, I do see the
workforce here in Milwaukee being, you know, very few and rare.

An even more despondent case was Nick, a 29-year-year old man from St. Paul who
had served a total of seven years for various convictions. In assessing his months of failed job
searching since completing his transitional job, he said:
Then it got to the point where I was getting discouraged, nobody is trying to hire
me, so I gave myself an excuse not to keep pounding and pounding. So I just
gave up and went to the old me again.

While he did not admit to returning to crime, he explained that the “old me” was the
kind of person who thought more about how to hustle to make ends meet rather than seek
formal employment. He said, “Because a criminal job, it don’t take much. All you need is the
right people or a little bit of money and whatnot. Doing it legal is hard, because you can’t have
no failings.”
At the completion of the formal second-round interviews, the men were asked to reflect
on the transitional jobs one year after completing them. The reactions were mixed. The few men
who were employed since completing those jobs readily admitted that the transitional jobs had
served them well in helping them make the transition from prison. Gary commented:
I think the program is good program. I think it did what it could do for me and I’m
appreciative of it. It gave me a work history, gave me a support group that I could
go to twice a week, I think it was, to be able to just talk some of the struggles. I
don’t really see at this point — I can’t really think of anything that they could have
really done more. I mean, that was really pretty much the basic stuff of what a person needs. A person needs a support group, a person needs a job when they come
out, you know. I really don’t know anything else more they could have did.

Similarly, Chris said:
Without them . . . I would have been lost, because I was contacted when I was in
prison, and by me being in there a long period of time, I wouldn’t have known
how to come out here and adjust to the job situation on the streets because I got a
letter from them and I went and hooked up with them finally. They helped me out
through a lot of stuff. Yeah, it was a big help. Without them, I wouldn’t have got
no first job.

Others who had been unemployed since completing the transitional job were not clear
that they had benefited at all from the program, as represented here by Terrence:
[The transitional job agency] get what they want out of us. That’s how I feel.
Like now, I’m here [to be interviewed]. I don’t have to be here. I was asked to
come here, but I don’t never get a call to see do I need a job, but they call me for
this, you know what I’m saying? So, I feel like why would you call me now for

141

this and you won’t even call me when I’m out here and I’m not working to see
how I’m doing? I show up at all y’all million programs. I don’t show up anymore, because y’all just send me on blank missions. If y’all going to send me on
blank missions, I can go on a blank mission by myself.

When asked to explain how he felt about his inability to find secure employment following his transitional job experience, Theodore said:
I felt like I needed more. One of my biggest things is, you know, I’m 44 and I
been locked away. I want to get somewhere where I can finally have somebody
offer me some benefits, you know, because it’s not so much the money that I
make hourly.

In a similar vein, Hakim, also from Milwaukee, said:
It’s only a matter of time, a matter of days, weeks, months, or years, and we’ll be
back into the same loop again, which is called the revolving door of the penitentiary. They let us come out here. They want you to do better — no, you want to
do better because you have to for your benefit and your fam’s benefit. But, if I
got jobs closing the door in my face every time based on what I did way back in
the days, what am I supposed to do? I got to eat, my fam got to eat, I got to pay
bills, I got to live. So, what you want me to do? You want me to do the right
thing, but you’re not giving me a chance.

Hakim was not criticizing the transitional job experiences so much as he was despondent about the lack of help in Milwaukee for previously incarcerated men who were trying to
find work. As both Theodore and Hakim had worked as peer counselors for their transitional
jobs assignment, and both had enjoyed their work, it is not surprising that they were especially
disappointed with their lack of employment nearly one year later.
Also not surprisingly, Sam — one of the few men who found work following a period
of unemployment after the transitional job had ended — incorporated the perspectives of both
the employed and the unemployed in his assessment of the transitional job one year later:
I really struggled [after completing the transitional job and finding no employment], I mean big time, but I didn’t give up. I had to make a way, and it had to
be the right way and it wasn’t going to be the streets or whatever, it was going to
be, I’m going to be a stand-up guy, responsible, and I’m going to make this
through, you know.

One year after completing the transitional job, then, most of the men who were interviewed expressed their frustration at remaining unemployed and questioned the value of their
transitional job experience. In fact, the better their experience with the transitional job, the more
frustrated they seemed to be if they had not found employment upon completing it. A major

142

part of the intensified frustration was rooted in what they saw as an enduring effect of their
identities as previously incarcerated men.
Managing the Public Identity Associated with Previous Incarceration:
One Year Later
Unlike the few men who were employed, the unemployed men who were interviewed
emphasized how much their status as previously incarcerated men had continued to be a
powerfully enduring barrier to reentering society in general and the world of work in particular.
Alternatively, the employed men observed that their transitional jobs experience had enabled
others to view them in a different, more positive light. Hence, while these men certainly
maintained strong identities as previously incarcerated men in discussing their sense of self —
as men who had survived or “bounced back from the bottom” — they believed that their status
as such was rapidly diminishing as a crucial part of their public identities. As Chris said,
“People don’t see me as just an ex-offender anymore; I am their coworker, or the guy that’s
good at driving or greeting the customers.” When asked whether his understanding of what it
means to be a previously incarcerated man had changed in the past year, Gary said:
I think it’s changed in the fact that . . . I’ve changed some people’s minds, especially the people I work with, about offenders, you know, ex-offenders. You
know, no matter what their crime may be, I think I’ve given them a different outlook on it . . . I have a sexual case in my history, and, you know, I believe that
they looked at me like, oh, you know, this dude, he’s just a sex offender, blah,
blah, blah. Even though they did give me a chance, I think they still had a negative look at me, and I think once . . . I really was able to sit down and articulate a
lot of different things of how I used to look at life versus what I learned in prison,
what I was willing to go through to be able to change my life and come out here
and be a positive person, I think it changed their view about me dramatically.

Those who were not working, however, remained preoccupied with the extent to which
their identities as previously incarcerated men may have been keeping them out of work,
especially in challenging economic times. As Theodore explained:
I remember, long before I got in the system, watching movies and hearing people
say, you paid your debt to society once you do your time. It angers me because
society’s not forgiving like that, like the myth that they put out there. You know,
there will be people that’s going to discriminate against me because of my
record, but me, I keep just believing in myself that there’s going to be those that
won’t and who will give me play. But, other than that, you know, I just get kind
of — I get angry sometimes, and when I see applications and they pose this
question, you know, like as if it’s nothing, oh, we just need to know, we just
want you to be totally honest. I fill it out and I got all this information, great information about me, but question number nine is that question [“Have you ever
been convicted of a felony?”], and I watch your eyes as you looking at my

143

résumé, you looking at my name, and then you got right to nine. Okay, so, I
mean, that angers me a little bit, but what can I do?

In a similar fashion, Vasquez said:
Man, it’s hard to escape that identity. Once you get that on your background, it’s
just hard. But, I’m maintaining, you know. I’m not letting it get the best of me.
I’m still focused and trying to make it better . . . Man, it’s like, man, I hope things
will be better with this identity, but as of now, it just doesn’t seem, you know —
people won’t hire you. They hire people with nonviolence, you know, if you had
a drug case or something that wasn’t violent. So, it’s hard to get over that.

The situation was bleak for most of the men one year after completing the transitional
job. Although they were aware of the economic downturn in the country, they were also
frustrated by their personal situations and highly critical of the transitional jobs project —
attitudes that were exacerbated by their feelings of being haunted by the public response to their
history of incarceration. The small subgroup of men who were working held contrasting
perspectives. For them, the transitional jobs did what they were designed to do, and they were
overcoming the social stigma associated with their prior incarceration.

Core Findings from the Third Round of Interviews, 2010
During 2010, formal in-person interviews were conducted in Chicago, Detroit, and
Milwaukee with 13 of the original 33 research participants. It was very difficult to obtain
updated information on the men in St. Paul; however, case managers from the St. Paul Goodwill
agency did report on the employment status of two of the men who were initially interviewed.
The results of the site visits and agency update in 2010 were as follows:
Detroit. Complete updates were secured for four of the Detroit men, as well as anecdotal reports on the seven others from both case managers and other study participants.
Eight of the men remained unemployed. One had begun to receive dialysis treatments
and was unable to work. Two had fallen out of contact with the qualitative study researcher as well as with the staff of the transitional jobs agency (Goodwill of Detroit).
Chicago. Three of the six men from this site who were originally interviewed were
working in full-time jobs by the spring of 2010, according to telephone interviews.
(Recall that one had died in a house fire in March of 2009.) One of the two men
who were terminated from full-time jobs was self-employed, doing groundskeeping for homeowners, while the other had been unemployed since losing his job.
Milwaukee. Three of the men could not be contacted. Two were incarcerated —
one since the time of the second-round interview, and one who had been on disabili-

144

ty the year before. One man had continued to work in his full-time job, and the final
two were unemployed.
St. Paul. As noted above, none of the St. Paul participants could be contacted for a
third round of interviews in time for their inclusion in this report. According to the
caseworkers’ telephone updates, one of the men was employed full time and one
had been reincarcerated.
Social Support and the Return to Society
The third round of interviews revealed a dramatic transformation in the men’s reported
use of friends and family as support providers. Ten of the men who were interviewed during
this third round and who were unemployed explained that they had been increasingly frustrated
in the past year with their inability to secure employment. As they described it, their relatives
were beginning to tire of serving as their means of support and had begun to express frustration
over the inability of the men to assume roles as providers (although, importantly, none of the
men reported that relatives or significant others encouraged them to resume illicit activities).
While none of these men described their situations as crises, they no longer maintained that the
households were as unconditionally supportive and secure as they had been in the recent past.
Steve, a 28-year-old man who had served a total of 12 years for multiple convictions,
said that his partner, who stayed with him throughout his turbulent adult life, had always stood
by his side. He said, “If I ever fell out or ever went to jail or anything, all I had to do was pick
up the phone and I was out. I had lawyers. I had money. I had bail and all that, so it wasn’t no
problem with that.” When discussing his relationship more recently, he simply said, “Now,
she’s like, it’s time for you to step up.” As he was not working (and had not worked since
completing his transitional job in Detroit), he was deeply concerned about what he could offer,
not only for himself, but for his partner.
Theodore was in a similar situation. He said:
I’m battling with that constantly, and then you got the peanut gallery, which
could be my ex-wife, my friend-lady now, you know. They want to get attitudes,
you know . . . hit me with, you know, “You’re just this and that,” and it all plays
a part on my psyche because I’m already battling with, okay, you know I know
how to make money [illicitly], but I don’t want to do that. So, you know, it’s a
bad spot. You don’t know which way to fall. I’m trying to stay positive and say,
no, I don’t need that, because I know how it ends.

In talking further about the transformation in expectations that his relatives had of
him, and how being incarcerated for so long had hindered his capacity to meet those expectations, he said:

145

I’ve always been distant, but it’s kind of how I kind of feel that they expect me,
now that I’m kind of around, to just take on this role in the family. I don’t know.
My friend’s girl tells me all the time, you’re probably not the family type person.
I don’t know. I love my kids. I love my family. I don’t know, man. I walked on
the dark side for so long. Now I’m out of that, but I still got some of those conditions. I’m conditioned to it. It’s probably not healthy.

Essentially, while these men struggled constantly with the social ramifications of being
previously incarcerated, one critical point of change for them was in their intimate social
relations. Here many of the men began to experience significant and problematic changes. They
were now expected to function in the ways that men were more traditionally expected to
function in the household. Yet, most of the men were not employed, nor did they have much
recent experience in environments that prepared them to tend to the emotional and social needs
of others.
Reflections on the Transitional Job: Two Years Later
More than two years after completing their transitional jobs, only a few of the men continued to say that their experience with those jobs had failed to connect them to longer-term
employment. Instead, during the third round of interviews, they said that those jobs had created
a degree of stability for them in what was otherwise a very turbulent and anxiety-ridden period
of their lives. Looking back to that period after two years allowed them to articulate more fully
how much hope, yet also how much insecurity, they had had about the reentry process. Moreover, as many of these men had also gone without work since completing the transitional job,
unemployment was no longer a sudden or immediate realization for them. Instead, it had
become a long-term reality. Consequently, when they were asked two years later to reflect upon
what it meant to have had a transitional job, their replies were uniformly positive and they now
regarded the transitional job as a bright moment in a recent history of malaise. As Chris stated:
[The transitional job] was good, because it gave me my first job and prepared me
for long-term work. It helped me out a lot because I was thinking I wasn’t ever
going to be able to work or get a job, so by going through that, it helped me out a
lot. I was able to work two jobs through them . . . I thank them for it, because
without them, I might have ended up back in jail.

Some men spoke about important emotional gains that they realized they had gotten
from the experience, even in the absence of securing full-time employment. Travis reflected
most of the men’s sentiments during the third round:
[The transitional job] gives you self-pride, and when you get that feeling, you
know that you can do it and that you’re welcomed there by the people and you’re
accepted. You feel great. That’s something that I believe inmates . . . when they
come out, it’s something they have to learn — feeling.

146

And Hakim said:
I really believe in the program. Since doing one of those transitional jobs, I even
moved on to other things before my [re]incarceration. Working with Vessels of
God, doing work with Resurrection Lutheran Church, Feed My Sheep. As a
community resource coordinator, I was involved in many tasks, many community-based things, and I wouldn’t change the experience, man, for nothing in this
world because I was able to try to put together gang summits to try to stop the
violence, and I rubbed elbows with so many important folks in the community. It
just felt good to be around some of the positive-ness.

Even those men who had spoken more critically about the transitional jobs during the
second round of interviews responded one year later more positively, as Theodore expressed:
Even though I’m not affiliated or plugged necessarily with these guys [at New
Hope] . . . I’ve got these guys as references and we talk and call and this and
that . . . It’s great because I still think I’m very confident about myself, but it
just gave me great confidence to get out the joint to be making that little money
every week or whatever. It made me feel like, yeah, I’m changing . . . I just
wish it was longer [laughs]. I have nothing but great things to say about that
reentry thing, man.

In fact, almost all the men who went through a third round of interviews stated that they
wished that the transitional jobs could have lasted longer. When asked about how long they had
hoped the jobs could have lasted, they said, not surprisingly, that the jobs should somehow have
lasted until they found full-time employment. Ultimately, many of the men found that their only
experience with employment over the previous two years had been the transitional job. Thus,
the transitional job opportunity was the only period of their post-incarceration lives (or the most
recent period following incarceration for those who had served multiple terms) when they were
able to provide some material support for themselves, avoid the idleness and tension that
accompanies unemployment, and begin to view themselves as being capable of successfully
reengaging with society.
Managing the Public Identity Associated with Previous Incarceration:
Two Years Later
For those men who remained unemployed two years after the transitional job had
ended, a common and lingering concern was their seemingly inescapable public identity as
formerly incarcerated men. As Terrence put it, “I did my crime so long ago and I served my
time, but I still have the ‘X’ on my back.”
Theodore concurred:
The biggest crock that they can give you in the world is that a man can be convicted in these courts, and then serve his time and he’s paid his debt. That’s the
147

biggest crock in the world, because it’s so double jeopardy . . . Though I made
those decisions to do those crimes and got myself in those situations, I’m still
paying another debt each time they say no or each time they turn me away . . . It
kind of fulfills the dream that people make us out to be and that’s why they turn
us away, for the reason that, oh, he’s bad luck, or whatever, so we don’t touch that
. . . I just had a birthday the twenty-third of March, so I’m five years from 50. I’m
getting scared, like, how long can I keep struggling? I’m getting up there.

Those men who were unemployed, but who had been issued relatively short sentences
(of less than a few years) and had served a single short term, experienced considerable frustration with the idea that the rest of their lives could be marked by something that they had done
for a short period (and, in some cases, on a single day) in their past. The men were extremely
assertive about how much they felt they could offer to employers, and to society more generally, compared with the impression that was conveyed by their past incarceration. In the end,
these men reflected a level of despondency that had not surfaced as obviously during the earlier
interviews.
In contrast to the unemployed men, the few men who had found and maintained work
over the preceding year or two replied that their conduct and performance at work had been
central to their efforts to reframe themselves as people who were more than simply previously
incarcerated men. Each man had stories about how his associates, co-workers, bosses, and other
significant contacts had begun to view him differently, thus allowing him to escape from the
public identity that continued to “imprison” the other men. Ronald, who was self-employed as a
residential groundskeeper — and who was by far the most financially successful of the men in
the qualitative study — observed:
Well, being an ex-offender means that, I mean, you do have a second chance. It
all depends on how you conduct yourself, you know? So, if you put forth the effort, being an ex-offender don’t mean that you are a bad person. But, you got to
put forth the effort and — how can you put it? Keeping yourself stable, challenges, just overcome them and just go out there, you know.

And Chris, who spent most of his post-incarceration years working as a delivery man
and stock person for a grocery store, said:
[Being a former prisoner] means that I’m not doing the things that I used to do.
I’m bettering myself. A life of crime wasn’t for me. It’s not for me now, but back
then when I was offending, I didn’t care about it. I just wanted to commit crimes.
Now, I know because I’m better out here than in there [prison], and everybody
needs me out here. I’m useless in there . . . My family, when things go wrong or
support or something, whenever they need something, they need me here, or
even just to talk to them, or even just family gatherings, they’d rather see me
there than coming up to see me [in prison] . . . [People] respect me for what I’m
doing and how I was able to make that change.

148

While most of the men did not have good news to report during the third round of interviews, each seemed to have come to terms with his personal circumstances such that he
looked back at the transitional job as a valuable experience.

Conclusion
We all make decisions, good and bad, and you have to reap what you sow . . .
being an ex-felon doesn’t mean that the person was actually a waste.
— Hakim

The core finding in the qualitative study is that the transitional job served as an opportunity for these men to achieve significant stability and focus in what was otherwise an intense
and turbulent period in their lives. Initially, the transitional job allowed them to feel efficacious
about connecting to the world of work and constructing personal work histories. The existence
of a weak economy throughout the three-year period following their release from incarceration
meant that the transition to consistent and materially rewarding work did not become a reality.
Hence, within a year after completing their transitional jobs, many of the men began to question
their capacity to secure employment for themselves and the value of the transitional jobs
program more generally. Yet, looking back on their experiences approximately two years
removed from the transitional job, the men reevaluated that period of their lives such that they
emphasized and ultimately found great value in experiencing the structure and stability that
came with having transitional jobs.
It is not difficult to speculate about why there was such a shift in thinking about the value of the transitional jobs from a period shortly after those jobs ended to two years later. In the
period following the transitional job experience, the men were exposed to other individuals who
had not been incarcerated and who had had little or no work during the prior three years. Hence,
they experienced their local economies not simply in terms of how they could best fit into them
in order to remedy the effects of being previously incarcerated, but also in terms of the extensive degree to which their neighbors and friends were suffering the consequences of a weak
economy.
That being said, prolonged unemployment had some critical lingering effects for most
of these men, and they had no capacity to resolve or minimize those effects. For instance, they
believed that the identity associated with their incarceration would endure irrespective of how
much they stressed that they had paid their debt to society through both incarceration and proper
behavior while on parole. As many of the men said, they had made mistakes at specific points in
the past, but they felt that they now had to pay for the consequences of those mistakes for the
rest of their lives. Coupled with this feeling was the sentiment that, as men, they were expected
to contribute materially to their families. While many said that in the year following their

149

release, relatives and dependents did not ask much of them, but rather tried to serve as sources
of support, the level of that support began to diminish as their families began to assert that the
men had to begin making some form of return on those investments. Perhaps somewhat
ironically, then, men who believed that they were well entrenched in supportive networks
began, within a year of their release, to express concern and anxiety about their inability to
reciprocate.
For those few men who were gainfully employed at the time of the third round of interviews — three years after their release from prison and two or more years after their transitional
jobs had ended — the remarkable finding was that they believed that their public identity was
no longer rooted in their history as incarcerated individuals. All of these men spoke about the
impressions they had made at work with coworkers, supervisors, or customers and clients,
which convinced other people that they were more than once-upon-a-time-prisoners.

150

Chapter 9

Reflections and Next Steps
This chapter describes where the transitional jobs programs in the Transitional Jobs
Reentry Demonstration (TJRD) stood approximately one year after the operational phase of
the demonstration ended, and discusses some of the lessons that program managers learned
from TJRD. It ends by discussing the implications of the first-year TJRD results for future
policy and research.

An Update on the TJRD Transitional Jobs Programs
Enrollment into the TJRD project ended between May 2008 and September 2008 in all
sites. Thus, the operational phase of the project was largely over by spring 2009. The research
team conducted telephone interviews with managers in all four of the transitional jobs programs
in spring 2010 to learn about their current programming and hear their reflections on TJRD.
Although none of the organizations was still operating its TJRD program, in Chicago, Detroit,
and St. Paul they were still running some kind of transitional jobs program for former prisoners;
in Milwaukee, the New Hope project had undergone significant organizational changes and was
not operating any transitional jobs programs at the time of the spring 2010 telephone interviews.
The Safer Foundation’s contract to staff garbage recycling plants for Allied Waste
ended when the City of Chicago changed its approach to recycling in fall 2008. Safer worked
with the City to develop a new work crew model, and more than 200 Safer clients were
transferred from Allied to the new program. Since that time, Safer and the City have further
developed this model into a larger transitional jobs program using funds from the American
Recovery and Reinvestment Act. As of March 2010, Safer was managing more than 200
transitional jobs work slots in crews working on reclamation projects, neighborhood clean-up,
and other tasks. Relative to the TJRD approach, the new program model has a more explicit
focus on building skills: participants work 20 to 30 hours per week and engage in 15 hours of
paid soft skills training. There is also a stronger emphasis on helping participants move into
unsubsidized jobs.
Goodwill/Easter Seals in St. Paul received funding from the U.S. Department of Education to operate an employment program for former prisoners with disabilities, mostly mental
health conditions. Unlike in TJRD, participants begin with an assessment to determine whether
they should start with a transitional job or move directly to job placement activities. The
organization also has state and federal funding to operate transitional jobs programs for a
broader group of former prisoners, for individuals with disabilities, and for welfare recipients.

151

Goodwill Industries in Detroit was still receiving funding from the Michigan Prisoner
Reentry Initiative to operate a program that offers transitional jobs using a scattered site model.
The program pays wages for 25 hours of work per week and encourages the host employer to
pay for additional hours and to hire the participant after the subsidy ends.
Across the three organizations that were still running a TJ program, the changes in program structure and target populations reflect both operational lessons from TJRD and the
ongoing struggle to find stable sources of funding for transitional jobs programs. In several
cases, the organizations changed the model or target group either to qualify for a particular
funding stream or to reduce per-client costs to reflect the amount of funding available.

Reflections on the Transitional Jobs Reentry Demonstration
Administrators voiced generally positive views about their experiences in TJRD and reflected on some lessons they took from the demonstration.
First, some sites discussed the challenge of determining who “needs” a transitional job.
The TJRD project was designed to test the impact of transitional jobs, so sites were generally
expected to offer such a position to everyone who was randomly assigned to the transitional
jobs group. Given this reality, the research team encouraged sites to conduct some screening
before random assignment, to try to weed out people who could find a job on their own relatively easily. Site staff explained to potential participants that the transitional jobs were temporary,
minimum wage positions, in the hope that people who thought they could do better in the
regular labor market would self-select out of the project; in some cases, potential participants
were encouraged to look for a job for a couple of weeks before enrolling in TJRD.
The challenge, of course, is that the potential participants were jobless, recently released
prisoners with few sources of legitimate income at their disposal. When presented with the
opportunity to enroll in a project that offered a 50/50 chance of an immediate paid job (albeit at
minimum wage), people may have jumped at the chance even if they thought they might find an
unsubsidized job with some additional effort. They might have assumed that they could look for
a better job while working in the transitional job. Those who were assigned to the job search
group then began looking for a job on their own, or with the assistance of the job search
program, and some of them succeeded, which explains why the transitional jobs group had a
lower rate of unsubsidized employment in three of the sites. The reduction in unsubsidized
employment was particularly large in St. Paul, where the local economy was relatively strong,
the job search program had substantial experience, and the transitional jobs program put people
to work within 24 hours of random assignment. Perhaps not surprisingly, when they were
interviewed in 2010, staff from the St. Paul transitional jobs program reported that they saw

152

advantages to their current model, which started with an assessment rather than an immediate
transitional job.
Second, program managers discussed the pros and cons of various transitional jobs
models. For example, staff from the Safer Foundation expressed great satisfaction with their
current crew-based transitional jobs program, which was put into place after the demonstration
ended. The model that operated during the demonstration, in which transitional jobs group
members worked in a recycling plant, was attractive to Safer because it generated revenue from
the waste management company that employed the transitional workers (via Safer’s staffing
agency). However, program staff reported that the arrangement was less than ideal in other
ways because a private company is less likely to support a model that results in constant,
planned turnover or allows workers to spend 15 hours a week in soft skills training. Thus, in
some respects, the model that operated during the demonstration was not truly “transitional” —
workers were employed full time, leaving little time for training or job search, and had financial
incentives to work longer at the recycling plant. The demonstration model also gave Safer very
little control over the day-to-day worksite experience while, under the current model, participants are organized into small crews with a Safer supervisor, who provides guidance and
encourages peer support.
The Detroit Goodwill agency had moved to a scattered site model, in which transitional
workers are employed in private companies. During the demonstration, virtually all participants
worked in Goodwill’s office, in light manufacturing or assembly work performed under
contract to outside companies. As in Chicago, the demonstration model earned revenue, but the
scattered site model allows participants to work in regular worksites, where they have a chance
of rolling over into a permanent position, and may also provide a more diverse set of work
opportunities.
Third, some sites that had not worked extensively with former prisoners before the
demonstration noted that they had learned a great deal about the nature and magnitude of the
challenges facing this population. Although the participants’ criminal record was the most
visible issue, many had few family or social supports, and no stable place to live. Moreover,
staff struggled to help younger participants understand the consequences of the lifestyle choices
they were making, notably selling drugs. These issues are discussed in the summary of the indepth interviews in Chapter 8. The transitional jobs programs also reported that they developed
much closer linkages with criminal justice agencies through their collaborative work on TJRD.

What Next?
Over the years, subsidized employment programs have been designed to achieve different goals. Some have aimed primarily to provide work-based income support to jobless individ-

153

uals; others have sought to use the subsidized work experience to prepare hard-to-employ
participants for permanent jobs; and still others have sought to emphasize both goals. Transitional jobs programs have been described primarily as a training strategy to help hard-to-employ
individuals enter and succeed in the labor market.
Although transitional jobs programs first emerged in the 1990s in the welfare system,
several factors explain the rise of transitional jobs in the prisoner reentry field in recent years.
First, many employers are quite reluctant to hire people with criminal records, but it is assumed
that they might be more willing to consider someone who has demonstrated his reliability for
several months in a transitional job; in other words, working in a transitional job is thought to
improve a participant’s chance of getting hired. Second, since transitional jobs provide an
opportunity to identify and address issues that will likely cause trouble in a regular job, the
model may improve participants’ ability to hold a job. Third, a transitional job provides a source
of legitimate income for a group of men with very high jobless rates who are unlikely to qualify
for other forms of income support (other than the Supplemental Nutrition Assistance Program,
formerly the Food Stamp program). And, finally, given the widely accepted view that people
who are employed are less likely to commit crimes, even temporary jobs are seen as a strategy
to lower recidivism, especially during the critical period just following release from prison.
The results of the TJRD project to date provide mixed evidence on those assumptions.
On the one hand, the very high rates of participation in transitional jobs suggest that many
former prisoners need income support and are willing to work for it. The transitional jobs
operated on a relatively large scale and provided real work opportunities to participants. In
retrospect, participants in the qualitative study recalled their time in the transitional jobs programs as a rare period of stability in an otherwise chaotic reentry process.
On the other hand, neither the TJRD results to date nor the results of earlier evaluations
provide much support for the assumption that transitional jobs, at least as currently designed and
operated, help people get or hold permanent jobs. Indeed, the qualitative study participants who
found unsubsidized jobs generally reported that they had found those jobs on their own or
through family connections, not through the transitional jobs programs.
The story with respect to recidivism is more complex. If there is indeed a strong causal
connection between employment and recidivism, fewer arrests would have been expected
among the transitional jobs group during the period when that group’s employment rate was 30
to 40 percentage points higher than the rate for the job search group. As described in Chapter 1,
the Center for Employment Opportunities evaluation, which used a very similar design, did find
reductions in recidivism. (See Box 9.1 for a comparison of the two studies.) Although there may
be a broader connection between earnings from “real” jobs and crime, it does not appear to be

154

Box 9.1

The Transitional Jobs Reentry Demonstration and
the Center for Employment Opportunities Evaluation:
Similarities and Differences
The pattern of employment results is quite similar in the Transitional Jobs Reentry Demonstration (TJRD) and the Center for Employment Opportunities (CEO) evaluation, but
the impacts on recidivism differ. In TJRD, there are no consistent impacts on measures of
recidivism, while in CEO there were reductions in recidivism for sample members who
came to the program within 90 days of release. It is not possible to pinpoint the reasons
for the different results, but they may be explained by some of the differences between the
two programs and study designs.
Similarities


Both studies compare a transitional jobs program with a job search assistance program.



The demographic characteristics of study participants are generally similar in the two
studies.



Both studies targeted parolees who had served time in state prison.

Differences


CEO’s transitional jobs program uses a work crew model; none of the TJRD programs used this model during the demonstration.



CEO had more extensive experience with reentry than most of the TJRD programs.



The job search programs in TJRD generally provided more extensive services than
the job search program in the CEO evaluation.

the case that policymakers can ensure lower recidivism simply by offering temporary, minimum-wage jobs to former prisoners. That said, it will be important to watch the recidivism
trends during Year 2 of this study.
The emerging results from transitional jobs evaluations — along with the results from
earlier studies described in Chapter 1 — suggest the need to clarify the goals of subsidized
employment programs. The transitional jobs programs that have been tested have not led to
long-term increases in unsubsidized employment, but that is not the only possible goal for a
subsidized employment program. During the Great Depression, the Works Progress Administration and other subsidized work programs were mainly intended to provide income support in

155

a way that also improved public facilities and infrastructure; many have argued that the same
needs exist today.
In addition, even when the labor market is healthier, certain groups experience great
difficulty finding and holding jobs. The employment rates for the job search group in the
TJRD study were typically below 40 percent even before labor market conditions began to
deteriorate. The question is whether society is willing to provide even work-based income
support to these groups, particularly without strong evidence that subsidized jobs will improve long-term employability or reduce recidivism and crime. In the past, public support for
this type of assistance has been strongest for parents who are raising children; single adults
like those targeted in TJRD are largely left out of the safety net. Of course, many former
prisoners are parents and, while relatively few live with their children, many more are
expected to pay child support.

Future Research
The inability of current transitional jobs models to generate higher rates of unsubsidized
employment and consistent reductions in recidivism underscores the importance of testing new
subsidized employment approaches that might do a better job of improving long-term employability. The promise of various possible approaches depends on figuring out why the
transitional jobs programs that have been tested so far have not increased long-term employment.
One hypothesis is that the failure to achieve long-term employment gains is the result of
weaknesses at the “back end” of the transitional jobs program — that is, the period when
participants are making the transition to unsubsidized employment and follow-up. Thus,
enhanced versions of the transitional jobs model might be tested that include stronger job
development, job placement, and post-placement services. The results among study participants
who were offered the employment retention bonuses in St. Paul suggest that financial incentives
might be part of a post-placement model.
A second hypothesis is that the programs fall short because they do not teach specific
occupational skills that will allow participants to qualify for higher-paying jobs. This hypothesis
suggests the need for models that combine subsidized work with skill-building. Such approaches might range from simple models in which participants split their time between traditional
work sites and classroom training to apprenticeship-like models in which hands-on training is
embedded into the work experience.
A third hypothesis is that transitional jobs do not last long enough to make a real difference in long-term outcomes. Although this third hypothesis is plausible, it appears that relatively few TJRD sample members left their transitional jobs because they had reached a “time
limit” and were let go. Most of the transitional jobs group members eventually found an
156

unsubsidized job, quit, or were terminated for cause before that point. At a minimum, it may be
worth testing models in which subsidized jobs are seen as fallback positions that are available
both before people get unsubsidized jobs and when they lose those jobs.
A fourth hypothesis is that it is inherently difficult for people to make the transition
from an artificial, subsidized work experience into a regular job, and that employers do not
necessarily value experience in a “program” job when they are making hiring decisions. This
hypothesis may argue for strategies that temporarily subsidize employment in private sector
jobs where participants can eventually “roll over” into a permanent, unsubsidized position. It is
not clear, however, how many employers would be willing to hire former prisoners, even with
subsidies.
In addition to hypotheses about how to improve employment impacts in transitional
jobs programs for former prisoners, the early TJRD results also suggest some directions for
future prisoner reentry research. For example, while the provision of subsidized jobs does not
necessarily ensure lower recidivism, such jobs are able to engage former prisoners and might be
a good platform for other interventions that address recidivism more directly — for example,
behavioral strategies that aim to improve former prisoners’ decision-making skills. Similarly,
the results from the St. Paul site suggest that subsidized jobs could be a component of efforts to
reduce technical parole violations, which account for a large share of reincarceration.
Stepping back, it seems clear that the early TJRD results, while disappointing in some
respects, can create a strong foundation for future research on strategies to improve outcomes
for former prisoners and other vulnerable groups.

157

Appendix A

Timeline for the Transitional Jobs Reentry
Demonstration

The Transitional Jobs Reentry Demonstration
Appendix Table A.1
Timeline of Transitional Jobs Reentry Demonstration Study
Date

Activity

2006
January - July

Site selection.

August October

Initial site visits: Focused on understanding the two service streams and beginning to plan
procedures for identifying and recruiting study subjects and conducting random
assignment.

October

Kick-off meeting, Wisconsin: Site staff, state Department of Corrections representatives,
the research team, and funders met for the official project launch.

2007
January

Random assignment begins.

February March

Early assessment site visits: The research team visited each site to assess and discuss any
issues the sites were having with research procedures and their sample take-up.

March

Cross-site meeting: Staff from each site met in Chicago to review the project’s progress.
The all-site meeting focused on many of the issues identified during the early assessment
visits.

July - September

First implementation site visit: Devoted to collecting information on program operations
and the local criminal justice context.

2008
January

Peer-to-peer visit: Site staff and members of the evaluation team visited the
Goodwill/Easter Seals site in St. Paul.

May - June

Second implementation site visit: Designed to update the findings from the first
implementation site visit, focusing on areas where the programs changed over time. The
team also collected cost information from each site.

September

Random assignment ends.

January - March

In-depth interviews, round 1.

2009
January - March

In-depth interviews, round 2.

2010
January - March

In-depth interviews, round 3.

October

One-year follow-up report.
161

Appendix B

Supplementary Exhibits for Chapter 2

Appendix Table B.1
Characteristics of Sample Members at Baseline
Chicago
Transitional Job Search
Full
Jobs Group
Group Sample

Characteristic
Age (%)
18-24
25-34
35-44
45 or older

16.4
34.4
29.6
19.6

17.3
33.5
28.6
20.5

16.8
34.0
29.1
20.1

Average age (years)

35.2

35.0

35.1

Race/ethnicity (%)
White/non-Hispanic
Black/non-Hispanic
Hispanic
Other

3.2
89.7
6.5
0.5

3.9
87.8
8.3
0.0

3.6
88.8
7.4
0.3

Has any children under age 19 (%)

54.0

45.9

50.0

Among those with children
Lives with own children under age 19 (%)
Average number of children
Has a child support order (%)

28.4
2.0
27.5

25.9
2.2
24.7

27.3
2.1
26.2

Among those with child support order a
Average monthly support order ($)

849

644

769

Amount of child support owed a (%)
None
Less than $5,000
$5,000 - $14,999
$15,000 or more
Don’t know

4.0
44.0
16.0
4.0
32.0

16.7
38.9
11.1
11.1
22.2

9.3
41.9
14.0
7.0
27.9

Education (%)
GED certificate
High school diploma
Associate's degree/2- or 4-year college/beyond
None of the above

36.0
19.1
5.6
39.3

30.9
22.8
8.0
38.3

33.5
20.9
6.8
38.8

Has occupational training certificate (%)

41.8

31.6

36.7

Marital status (%)
Single
Married, living with spouse
Married, living away from spouse
Legally separated/divorced/widowed

80.0
2.2
3.3
14.4

76.4
2.8
5.1
15.7

78.2
2.5
4.2
15.1

Sig.

(continued)

165

Appendix Table B.1 (continued)
Transitional Job Search
Jobs Group
Group

Characteristic

Total

Housing status (%)
Owns/rents house or apartment
Lives with friends or relatives
Has transitional housing
Lives in shelter/other

19.0
58.7
15.2
7.1

21.0
58.6
10.5
9.9

20.0
58.6
12.9
8.5

Ever employed (%)

84.2

84.1

Ever employed 6 consecutive months for one employer (%)

47.7

53.4

50.6

Worked full time in the 6 months before incarceration (%)

43.8

45.7

44.7

Medical condition restricts ability to work (%)

1.7

4.0

2.8

Ever in drug or alcohol treatment program (%)

39.9

39.8

Public assistance (%)
No government assistance
Temporary Assistance for Needy Families
Food stamps
Social Security
Medicaid
Housing assistance

35.6
1.1
63.2
1.1
1.7
1.1

44.5
0.6
53.2
0.0
0.6
1.7

40.1
0.9
58.2
0.6
1.2
1.4

Family assists with (%)
Finding a place to live
Finding a job
Dealing with a drug or alcohol problem
Financial support
None of the above

31.8
35.2
6.8
37.5
40.9

37.1
37.1
8.2
41.2
37.6

34.4
36.1
7.5
39.3
39.3

Facility type of most recent release (%)
Prison
Jail

99.4
0.6

98.9
1.1

99.2
0.8

Sample size

189

185

374

Sig.

*
*

166

The Transitional Jobs Reentry Demonstration

Appendix Table B.2
Characteristics of Sample Members at Baseline
Detroit
Transitional Job Search
Jobs Group
Group

Characteristic

Full
Sample

Age (%)
18-24 years
25-34 years
35-44 years
45 or older

12.5
29.5
37.1
21.0

12.1
35.7
27.2
25.0

12.3
32.6
32.1
23.0

Average age (years)

36.4

36.6

36.5

Race/ethnicity (%)
White/non-Hispanic
Black/non-Hispanic
Hispanic
Other

6.3
89.1
2.7
1.8

9.0
84.6
0.9
5.4

7.7
86.9
1.8
3.6

Has any children under age 19 (%)

49.1

50.9

50.0

Among those with children
Lives with own children under age 19 (%)
Average number of children
Has a child support order (%)

39.1
1.9
46.4

28.1
1.8
40.4

33.5
1.9
43.3

453

305

Amount of child support owed a (%)
None
Less than $5,000
$5,000 - $14,999
$15,000 or more
Don’t know

4.8
4.8
7.1
45.2
38.1

8.1
16.2
16.2
27.0
32.4

6.3
10.1
11.4
36.7
35.4

Education (%)
GED certificate
High school diploma
Associate's degree/2- or 4-year college/beyond
None of the above

63.0
11.1
4.2
21.8

60.8
10.8
5.2
23.1

61.9
11.0
4.7
22.4

Has occupational training certificate (%)

52.3

58.5

55.4

Marital status (%)
Single
Married, living with spouse
Married, living away from spouse
Legally separated/divorced/widowed

77.0
6.3
2.7
14.0

74.4
5.5
2.3
17.8

75.7
5.9
2.5
15.9

Among those with child support order
Average monthly support order ($)

Sig.

[* ]

(continued)

167

Appendix Table B.2 (continued)
Transitional Job Search
Jobs Group
Group

Characteristic

Full
Sample

Housing status (%)
Owns/rents house or apartment
Lives with friends or relatives
Has transitional housing
Lives in shelter/other

17.8
67.6
11.0
3.7

15.7
65.5
16.6
2.2

16.7
66.5
13.8
2.9

Ever employed (%)

84.5

87.4

85.9

Ever employed 6 consecutive months for one employer (%)

48.8

59.5

54.1

Worked full time in the 6 months before incarceration (%)

44.5

52.1

48.3

Medical condition restricts ability to work (%)

13.6

9.5

11.6

Ever in drug or alcohol treatment program (%)

52.7

46.8

49.8

Public assistanceb (%)
No government assistance
Temporary Assistance for Needy Families
Food stamps
Social Security
Medicaid
Housing assistance

45.8
2.1
52.1
2.1
8.3
3.1

51.0
3.1
41.7
3.1
12.5
6.3

48.4
2.6
46.9
2.6
10.4
4.7

Family assists with (%)
Finding a place to live
Finding a job
Dealing with a drug or alcohol problem
Financial support
None of the above

54.7
43.9
30.4
43.5
24.8

48.6
40.0
23.2
36.4
31.8

51.6
41.9
26.7
39.9
28.3

Facility type of most recent release (%)
Prison
Jail

99.1
0.9

100.0

99.5
0.5

Sample size

224

448

SOURCE: MDRC calculations using the Baseline Information Form.
NOTES: In order to assess differences in characteristics across research groups, chi-square tests were
used for categorical variables, and t-tests were used for continuous variables. Statistical significance
levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10 percent.
When significance levels are shown in square brackets, 20 percent or more of the categories for the
variable in question have cell sizes with fewer than 5 individuals.
aCalculated only for those who have children and are ordered to pay child support. Some
participants were ordered to pay but did not report the amount; they are missing from this measure.
bCreated only for those who were randomly assigned after January 8, 2008, when the site began
random assignment after release.

168

Sig.

The Transitional Jobs Reentry Demonstration

Appendix Table B.3
Characteristics of Sample Members at Baseline
St. Paul
Transitional Job Search
Jobs Group
Group

Characteristic

Full
Sample

Age (%)
18-24 years
25-34 years
35-44 years
45 years or older

20.2
34.0
26.5
19.4

16.2
38.7
28.1
17.0

18.2
36.4
27.3
18.2

Average age (years)

34.2

34.1

34.2

Race/ethnicity (%)
White/non-Hispanic
Black/non-Hispanic
Hispanic
Other

20.2
65.5
4.8
9.5

21.7
61.7
7.1
9.5

21.0
63.6
5.9
9.5

Has any children under age 19 (%)

52.6

55.7

54.2

Among those with children
Lives with own children under age 19 (%)
Average number of children
Has a child support order (%)

11.3
2.2
41.4

8.5
2.2
46.8

9.9
2.2
44.2

272

288

280

Amount of child support owed a (%)
None
Less than $5,000
$5,000 - $14,999
$15,000 or more
Don’t know

5.7
28.3
26.4
26.4
13.2

9.4
15.6
31.3
28.1
15.6

7.7
21.4
29.1
27.4
14.5

Education (%)
GED certificate
High school diploma
Associate's degree/2- or 4-year college/beyond
None of the above

54.3
21.2
7.3
17.1

52.1
21.1
5.4
21.5

53.2
21.1
6.4
19.3

Has occupational training certificate (%)

44.4

37.1

40.8

Marital status (%)
Single
Married, living with spouse
Married, living away from spouse
Legally separated/divorced/widowed

70.2
1.2
8.7
19.8

77.1
0.4
6.5
15.9

73.6
0.8
7.6
17.9

Among those with child support order
Average monthly support order ($)

Sig.

(continued)

169

Appendix Table B.3 (continued)
Transitional Job Search
Jobs Group
Group

Characteristic

Total

Housing status (%)
Owns/rents house or apartment
Lives with friends or relatives
Has transitional housing
Lives in shelter/other

13.1
18.7
63.3
4.8

11.5
17.1
64.7
6.7

12.3
17.9
64.0
5.8

Ever employed (%)

89.9

88.8

89.3

Ever employed 6 consecutive months for one employer (%)

56.1

53.1

54.6

Worked full time in the 6 months before incarceration (%)

42.6

42.9

42.7

Medical condition restricts ability to work (%)

4.5

6.8

5.6

Ever in drug or alcohol treatment program (%)

60.5

64.5

62.5

Public assistance (%)
No government assistance
Temporary Assistance for Needy Families
Food stamps
Social Security
Medicaid
Housing assistance

52.2
27.7
12.9
2.4
27.7
11.2

52.4
22.6
15.7
2.4
27.8
14.1

52.3
25.2
14.3
2.4
27.8
12.7

Family assists with (%)
Finding a place to live
Finding a job
Dealing with a drug or alcohol problem
Financial support
None of the above

25.6
21.5
10.2
26.8
55.3

27.4
21.8
12.9
29.4
55.6

26.5
21.7
11.5
28.1
55.5

Facility type of most recent release (%)
Prison
Jail

96.3
3.7

95.1
4.9

95.7
4.3

Sample size

253

506

170

Sig.

The Transitional Jobs Reentry Demonstration

Appendix Table B.4
Characteristics of Sample Members at Baseline
Milwaukee
Transitional Job Search
Jobs Group
Group

Characteristic

Full
Sample

Age (%)
18-24 years
25-34 years
35-44 years
45 or older

19.5
39.5
22.7
18.4

22.3
36.3
27.5
13.9

20.9
37.9
25.0
16.2

Average age (years)

33.6

32.8

33.2

Race/ethnicity (%)
White/non-Hispanic
Black/non-Hispanic
Hispanic
Other

6.9
88.6
2.4
2.0

5.8
85.8
5.4
2.9

6.4
87.2
3.9
2.5

Has any children under age 19 (%)

50.4

55.8

53.1

Among those with children
Lives with own children under age 19 (%)
Average number of children
Has a child support order (%)

16.4
2.3
28.9

15.5
2.4
34.3

16.0
2.3
31.6

Among those with child support order a
Average monthly support order ($)

343

226

274

Amount of child support owed a (%)
None
Less than $5,000
$5,000 - $14,999
$15,000 or more
Don’t know

0.0
8.5
25.4
29.6
36.6

3.8
17.7
19.0
34.2
25.3

2.0
13.3
22.0
32.0
30.7

Education (%)
GED certificate
High school diploma
Associate's degree/2- or 4-year college/beyond
None of the above

38.7
32.8
4.3
24.3

28.7
40.9
6.5
23.9

33.8
36.8
5.4
24.1

Has occupational training certificate (%)

49.6

Marital status (%)
Single
Married, living with spouse
Married, living away from spouse
Legally separated/divorced/widowed

75.9
3.7
3.3
17.0

84.0
3.0
2.5
10.5

79.9
3.3
2.9
13.8

Sig.

(continued)

171

Appendix Table B.4 (continued)
Transitional Job Search
Jobs Group
Group

Characteristic

Total

Housing status (%)
Owns/rents house or apartment
Lives with friends or relatives
Has transitional housing
Lives in shelter/other

19.3
55.6
23.5
1.6

22.0
51.9
20.7
5.4

20.7
53.7
22.1
3.5

Ever employed (%)

83.6

87.4

85.5

Ever employed 6 consecutive months for one employer (%)

43.8

56.7

50.2

Worked full time in the 6 months before incarceration (%)

39.2

44.9

42.0

Medical condition restricts ability to work (%)

11.6

5.6

8.6

Ever in drug or alcohol treatment program (%)

66.8

62.4

64.7

Public assistance (%)
No government assistance
Temporary Assistance for Needy Families
Food stamps
Social Security
Medicaid
Housing assistance

69.0
0.0
26.8
0.4
2.1
2.9

67.1
0.0
30.3
0.9
3.9
2.2

68.1
0.0
28.5
0.6
3.0
2.6

Family assists with (%)
Finding a place to live
Finding a job
Dealing with a drug or alcohol problem
Financial support
None of the above

39.1
48.1
13.6
44.0
26.7

40.9
46.4
15.7
43.4
24.3

40.0
47.3
14.6
43.7
25.5

Facility type of most recent release (%)
Prison
Jail

92.2
7.8

89.5
10.5

90.9
9.1

Sample size

256

251

507

Sig.

***

172

The Transitional Jobs Reentry Demonstration

Appendix Table B.5
Criminal History Characteristics of Sample Members at Baseline
Chicago
Transitional Job Search
Jobs Group
Group

Characteristic

Full
Sample

Arrest history
Any prior arrests (%)

99.5

Average number of arrestsa
Number of prior felony arrests
Number of prior misdemeanor arrests

15.2
4.7
5.4

15.4
4.8
5.8

15.3
4.7
5.6

97.3

95.1

96.2

4.5
3.2
0.6

4.3
3.3
0.5

4.4
3.2
0.6

Ever convicted of a violent crimec (%)

48.1

46.2

47.2

Ever convicted of a drug-related crimec (%)

64.9

65.2

65.0

2.2

1.1

1.6

Number of months in state prison

63.6

60.5

62.1

Days between latest state prison release
and random assignment

70.8

55.3

63.1

Length of most recent incarceration (months)

27.3

27.7

27.5

Crime category (%)
Violent crime
Drug crime
Property crime
Public order crime

25.0
37.8
22.2
15.0

25.0
39.8
23.3
11.9

25.0
38.8
22.8
13.5

Admission reason (%)
New crime
Parole violation
Other

83.9
15.1
1.1

80.8
18.7
0.5

82.3
16.8
0.8

Sample size

186

182

Conviction history
Any prior convictionb (%)
Average number of prior convictionsb
Number of prior felony convictions
Number of prior misdemeanor convictions

Ever convicted of sexual assualt (%)
State prison history

Most recent criminal justice event

173

368
(continued)

Appendix Table B.5 (continued)
SOURCES: Calculations based on data from Illinois Criminal Justice Information Authority
and the Department of Corrections in Illinois.
NOTES: All criminal history calculations include only events that occurred between January
1, 1980, and the date of random assignment.
In order to assess differences in characteristics across research groups, chi-square tests
were used for categorical variables, and t-tests were used for continuous variables. There are
no statistically significant differences between the research groups on any measures in this
table.
Subcategories may sum to more than the total due to multiple arrests, convictions, or
incarcerations per person during the follow-up period.
aEach arrest date is counted only as a single event. If there are multiple crimes or charges
on the same date, only the most serious charge is recorded in the analysis.
bEach conviction date is counted only as a single event. If there are multiple convictions
on the same date, only the most serious conviction is recorded in the analysis. Total includes
convictions for felony, misdemeanor, and other crime classes.
cThe categorization of charges is based on definitions from Langan and Levin (2002).

174

The Transitional Jobs Reentry Demonstration

Appendix Table B.6
Criminal History Characteristics of Sample Members at Baseline
Detroit
Transitional Job Search
Jobs Group
Group

Characteristic

Full
Sample

Sig.

Arrest history
Any prior arrests (%)

98.1

96.2

97.2

4.3
3.1
0.4

4.5
2.9
0.6

4.4
3.0
0.5

100.0

98.6

99.3 *

3.7
2.7
0.5

3.9
2.8
0.5

3.8
2.7
0.5

55.9

62.7

59.3

43.7

40.1

41.9

1.9

4.2

3.1

Number of months in state prison

99.5

99.0

99.2

Days between latest state prison release
and random assignment

21.7

21.1

21.4

Length of most recent incarceration (months)

51.0

57.3

54.2

Crime category (%)
Violent crime
Drug crime
Property crime
Public order crime

35.5
21.8
31.3
11.4

45.0
14.7
31.8
8.5

40.3
18.2
31.5
10.0

Admission reason (%)
New crime
Parole violation
Other

62.4
37.1
0.5

66.8
32.7
0.5

64.6
34.9
0.5

Sample size

213

212

425

Average number of arrestsa
Number of prior felony arrests
Number of prior misdemeanor arrests
Conviction history
Any prior convictionb (%)
b

Average number of prior convictions
Number of prior felony convictions
Number of prior misdemeanor convictions
Ever convicted of a violent crimec (%)
c

Ever convicted of a drug-related crime (%)
Ever convicted of sexual assualt (%)
State prison history

Most recent criminal justice event

(continued)

175

Appendix Table B.6 (continued)
SOURCES: Calculations based on data from Michigan State Police and the Department of
Corrections in Michigan.
NOTES: All criminal history calculations include only events that occured between January 1,
1980, and the date of random assignment.
In order to assess differences in characteristics across research groups, chi-square tests were
used for categorical variables, and t-tests were used for continuous variables. Statistical
significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10 percent.
Subcategories may sum to more than the total due to multiple arrests, convictions, or
incarcerations per person during the follow-up period.
aEach arrest date is counted only as a single event. If there are multiple crimes or charges on
the same date, only the most serious charge is recorded in the analysis.
bEach conviction date is counted only as a single event. If there are multiple convictions on
the same date, only the most serious conviction is recorded in the analysis. Some convictions
may have been associated with an arrest that occurred prior to random assignment. These
convictions are counted in the analysis as occurring after random assignment. Total includes
convictions for felony, misdemeanor, and other crime classes.
cThe classification of charges is based on definitions from Langan and Levin (2002).

176

The Transitional Jobs Reentry Demonstration

Appendix Table B.7
Criminal History Characteristics of Sample Members at Baseline
St. Paul
Transitional Job Search
Jobs Group
Group

Characteristic

Full
Sample

Sig.

Arrest history
Any prior arrests (%)

100.0

10.0
NA
NA

10.4
NA
NA

10.2
NA
NA

97.2

4.3
2.9
1.4

4.5
2.9
1.5

4.4
2.9
1.4

Ever convicted of a violent crimec (%)

79.4

81.3

80.4

Ever convicted of a drug-related crimec (%)

24.5

34.3

29.4

Ever convicted of sexual assault (%)

34.0

34.7

34.3

Number of months in state prison

56.5

52.3

54.4

Days between latest state prison release
and random assignment

23.1

19.9

21.5

Length of most recent incarceration (months)

21.7

20.6

21.1

Crime category (%)
Violent crime
Drug crime
Property crime
Public order crime

52.0
8.6
18.0
21.3

51.4
15.6
14.0
18.9

51.7
12.1
16.0
20.1

Admission reason (%)
New crime
Parole violation
Other

66.4
33.6
NA

66.1
33.9
NA

66.3
33.7
NA

Sample size

250

500

Average number of arrestsa
Number of prior felony arrests
Number of prior misdemeanor arrests
Conviction history
Any prior convictionb (%)
Average number of prior convictionsb
Number of prior felony convictions
Number of prior misdemeanor convictions

State prison history

Most recent criminal justice event

(continued)

177

Appendix Table B.7 (continued)
SOURCES: Calculations based on data from Minnesota Bureau of Criminal Apprehension and
the Department of Corrections in Minnesota.
NOTES: All criminal history calculations include only events that occured between January 1,
1980, and the date of random assignment.
NA indicates data were not available for that site.
In order to assess differences in characteristics across research groups, chi-square tests were
used for categorical variables, and t-tests were used for continuous variables. Significance levels
are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10 percent.
Subcategories may sum to more than the total due to multiple arrests, convictions, or
incarcerations per person during the follow-up period.
aEach arrest date is counted only as a single event. If there are multiple crimes or charges on
the same date, only the most serious charge is recorded in the analysis.
bEach conviction date is counted only as a single event. If there are multiple convictions on the
same date, only the most serious conviction is recorded in the analysis. Some convictions may
have been associated with an arrest that occurred prior to random assignment.These convictions
are counted in the analysis as occurring after random assignment. Total includes convictions for
felony, misdemeanor, and other crime classes.
cThe classification of charges is based on definitions from Langan and Levin (2002).

178

The Transitional Jobs Reentry Demonstration

Table B.8
Criminal History Characteristics of Sample Members at Baseline
Milwaukee
Transitional Job Search
Jobs Group
Group

Characteristic

Full
Sample

Sig.

Arrest history
Any prior arrests (%)

99.6

100.0

99.8

7.7
4.2
2.4

7.8
4.1
2.6

7.7
4.1
2.5

98.0

98.4

98.2

3.8
1.9
1.2

3.5
1.8
1.1

3.7
1.8
1.1

42.4

41.4

41.9

56.9

53.0

54.9

7.1

4.8

5.9

Number of months in state prison

76.0

67.0

71.6

Days between latest state prison release
and random assignment

77.1

63.7

70.4

Length of most recent incarceration (months)

34.3

29.2

31.7

Crime category (%)
Violent crime
Drug crime
Property crime
Public order crime

19.6
32.4
19.6
28.4

21.5
35.6
17.0
25.9

20.5
34.0
18.3
27.2

Admission reason (%)
New crime
Parole violation
Other

35.9
54.3
9.8

37.1
47.4
15.5

36.5
50.9
12.6

Sample size

256

251

507

Ever convicted of a drug-related crime (%)
Ever convicted of sexual assault (%)
State prison history

Most recent criminal justice event

(continued)

179

Table B.8 (continued)
SOURCES: Calculations based on data from State of Wisconsin Department of Justice and the
Department of Corrections in Wisconsin.
NOTES: All criminal history calculations include only events that occured between January 1,
1980, and the date of random assignment.
In order to assess differences in characteristics across research groups, chi-square tests were
used for categorical variables, and t-tests were used for continuous variables. Statistical
significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10 percent.
Subcategories may sum to more than the total due to multiple arrests, convictions, or
incarcerations per person during the follow-up period.
aEach arrest date is counted only as a single event. If there are multiple crimes or charges on
the same date, only the most serious charge is recorded in the analysis.
bEach conviction date is counted only as a single event. If there are multiple convictions on
the same date, only the most serious conviction is recorded in the analysis. Some convictions
may have been associated with an arrest that occurred prior to random assignment.These
convictions are counted in the analysis as occurring after random assignment.Total includes
convictions for felony, misdemeanor, and other crime classes.
cThe classification of charges is based on definitions from Langan and Levin (2002).

180

Appendix C

Supplementary Exhibits for Chapter 5

The Transitional Jobs Reentry Demonstration

Appendix Table C.1
Quarterly Impacts on Employment and Earnings
Chicago
Outcome

Transitional
Jobs Group

Job Search Difference
Group
(Impact)

P-Valuea

Employment (%)
Transitional employmentb
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

65.4
60.2
22.1
10.6
87.5

0.0
0.0
0.0
0.0
0.0

Unsubsidized employment
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

14.9
18.2
24.9
22.6
41.5

24.8
37.1
30.7
27.7
54.4

Total employment
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

71.8
68.6
44.1
31.3
93.4

24.5
36.9
30.7
28.0
53.8

Number of quarters employed

2.2

Employed in all 4 quarters (%)

65.4
60.2
22.1
10.6
87.5

***
***
***
***
***

0.000
0.000
0.000
0.000
0.000

-10.0 **
-18.9 ***
-5.8
-5.2
-12.9 **

0.019
0.000
0.228
0.264
0.016

47.3
31.7
13.4
3.4
39.6

***
***
***
***

0.000
0.000
0.009
0.491
0.000

1.2

1.0 ***

0.000

16.6

8.9

7.7 **

0.033

Transitional job earningsb
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

657
1,072
493
250
2,471

0
0
0
0
0

Unsubsidized earnings
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

116
307
655
751
1,829

175
635
764
795
2,369

Earnings ($)

183

657
1,072
493
250
2,471

***
***
***
***
***

0.000
0.000
0.000
0.000
0.000

-60
-328 ***
-109
-44
-540

0.290
0.007
0.553
0.834
0.255
(continued)

Appendix Table C.1 (continued)
Outcome

Transitional
Jobs Group

Total earningsc
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)
Sample size (total = 374)

Job Search Difference
Group
(Impact)

773
1,379
1,148
1,001
4,300

174
651
768
790
2,383

189

185

598
728
380
211
1,918

P-Valuea
***
***
*
***

0.000
0.000
0.053
0.323
0.000

SOURCES: MDRC calculations using payroll data from Safer and unemployment insurance data from the
State of Illinois.
NOTES: Statistical significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10
percent.
Results in this table are regression-adjusted for pre-random assignment characteristics.
aStandard errors are presented in this report for all impacts with a p-value of 0.000. Following are the
standard errors for all impacts with a p-value of 0.000 (presented in the order in which they appear in the
table). Employment: 3.665, 3.721, 3.230, 2.381, 2.537, 4.694, 4.751, 5.046, 4.201, and 0.135. Earnings:
58.859, 97.242, 87.000, 61.965, 232.153, 77.344, 142.773, and 500.667.
bEarnings and employment from transitional jobs are based on payroll data from Chicago.
cTotal earnings may not be equal to the sum of transitional job earnings plus unsubsidized job earnings
due to rounding.

184

The Transitional Jobs Reentry Demonstration

Appendix Table C.2
Quarterly Impacts on Employment and Earnings
Detroit
Outcome

Transitional
Jobs Group

Job Search Difference
Group
(Impact)

P-Valuea

Employment (%)
Transitional employmentb
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

66.9
73.0
12.8
3.2
88.0

0.0
0.0
0.0
0.0
0.0

66.9
73.0
12.8
3.2
88.0

Unsubsidized employment
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

11.4
20.8
20.4
17.8
38.5

7.2
23.0
18.8
19.3
33.1

4.2
-2.1
1.5
-1.4
5.4

Total employment
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

70.3
77.0
31.6
20.5
92.4

6.2
23.8
19.3
19.1
34.7

64.1
53.2
12.3
1.4
57.7

Number of quarters employed

2.0

Employed in all 4 quarters (%)

***
***
***
**
***

0.000
0.000
0.000
0.011
0.000
0.175
0.624
0.715
0.728
0.284

***
***
***
***

0.000
0.000
0.007
0.743
0.000

0.7

1.3 ***

0.000

10.3

3.6

6.6 **

0.016

Transitional job earningsb
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

517
830
81
32
1,461

0
0
0
0
0

517
830
81
32
1,461

Unsubsidized earnings
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

191
523
486
500
1,700

96
336
372
447
1,252

95
187
113
54
448

Earnings ($)

185

***
***
***
*
***

0.000
0.000
0.007
0.073
0.000
0.394
0.232
0.400
0.822
0.389
(continued)

Appendix Table C.2 (continued)
Outcome

Transitional
Jobs Group

Total earningsc
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)
Sample size (total = 388)

Job Search Difference
Group
(Impact)

708
1,354
567
533
3,161

88
341
373
444
1,245

195

193

620 ***
1,013 ***
194
89
1,915 ***

P-Valuea
0.000
0.000
0.157
0.711
0.000

SOURCES: MDRC calculations using payroll data from Goodwill and unemployment insurance data from
the state of Michigan.
NOTES: Statistcial significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10
percent.
Results in this table are regression-adjusted for pre-random assignment characteristics.
aStandard errors are presented in this report for all impacts with a p-value of 0.000. Following are the
standard errors for all impacts with a p-value of 0.000 (presented in the order in which they appear in the
table). Employment: 3.588, 3.364, 2.564, 2.494, 4.035, 4.470, 4.169, and 0.117. Earnings: 41.327, 59.996,
92.587, 122.558, 161.706, and 531.609.
bEarnings and employment from transitional jobs are based on payroll data from Detroit.
cTotal earnings may not be equal to the sum of transitional job earnings plus unsubsidized job earnings
due to rounding.

186

The Transitional Jobs Reentry Demonstration

Appendix Table C.3
Quarterly Impacts on Employment and Earnings
St. Paul
Outcome

Transitional
Jobs Group

Job Search Difference
Group
(Impact)

P-Valuea

Employment (%)
Transitional employmentb
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

92.1
72.5
34.2
12.1
97.1

0.0
0.0
0.0
0.0
0.0

92.1
72.5
34.2
12.1
97.1

***
***
***
***
***

0.000
0.000
0.000
0.000
0.000

Unsubsidized employment
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

15.8
27.8
35.6
36.4
59.1

50.6
54.4
43.8
39.1
74.5

-34.9
-26.6
-8.2
-2.8
-15.5

***
***
*

0.000
0.000
0.065
0.522
0.000

Total employment
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

92.1
81.4
56.4
44.4
98.7

49.4
54.2
43.6
39.0
74.0

42.6
27.1
12.8
5.4
24.7

***
***
***

Number of quarters employed

2.7

Employed in all 4 quarters (%)

***

0.000
0.000
0.004
0.216
0.000

1.9

0.9 ***

0.000

30.0

18.6

11.4 ***

0.003

Transitional job earningsb
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

807
1,057
414
131
2,410

0
0
0
0
0

807
1,057
414
131
2,410

***
***
***
***
***

0.000
0.000
0.000
0.000
0.000

Unsubsidized earnings
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

93
463
944
1,081
2,582

562
1,351
1,300
1,181
4,394

-469
-888
-357
-99
-1,813

***
***
*

0.000
0.000
0.066
0.599
0.000
(continued)

Earnings ($)

187

***

Appendix Table C.3 (continued)
Outcome

Transitional
Jobs Group

Total earningsc
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)
Sample size (total = 506)

Job Search Difference
Group
(Impact)

901
1,520
1,358
1,213
4,992

555
1,355
1,297
1,179
4,386

253

P-Valuea

346 ***
165
61
34
606

0.000
0.298
0.754
0.857
0.240

SOURCES: MDRC calculations using payroll data from Goodwill/Easter Seals and unemployment
insurance data from the state of Minnesota.
NOTES: Statistical significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10
percent.
Results in this table are regression-adjusted for pre-random assignment characterisitics.
aStandard errors are presented in this report for all impacts with a p-value of 0.000. Following are the
standard errors for all impacts with a p-value of 0.000 (presented in the order in which they appear in the
table). Employment: 1.730, 2.851, 2.978, 2.089, 1.067, 3.944, 4.298, 4.206, 3.623, 4.006, 2.912, and
0.113. Earnings: 37.259, 63.500, 46.362, 29.985, 113.327, 80.580, 152.886, 507.815, and 88.140.
bEarnings and employment from transitional jobs are based on payroll data from St. Paul.
cTotal earnings may not be equal to the sum of transitional job earnings plus unsubsidized job
earnings due to rounding.

188

The Transitional Jobs Reentry Demonstration

Appendix Table C.4
Quarterly Impacts on Employment and Earnings
Milwaukee
Outcome

Transitional
Jobs Group

Job Search Difference
Group
(Impact)

P-Valuea

Employment (%)
Transitional employmentb
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

41.6
56.2
28.9
9.5
70.2

0.0
0.0
0.0
0.0
0.0

Unsubsidized employment
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

22.5
34.8
36.6
35.4
58.1

36.9
46.0
38.2
34.6
66.5

Total employment
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

55.2
73.1
55.4
42.0
87.7

37.1
45.9
38.4
34.6
67.4

Number of quarters employed

2.3

Employed in all 4 quarters (%)

41.6
56.2
28.9
9.5
70.2

***
***
***
***
***

0.000
0.000
0.000
0.000
0.000

-14.4 ***
-11.2 **
-1.6
0.8
-8.3 *

0.001
0.012
0.717
0.859
0.055

18.1
27.2
17.0
7.4
20.4

***
***
***
*
***

0.000
0.000
0.000
0.090
0.000

1.6

0.7 ***

0.000

21.6

11.9

9.7 ***

0.004

Transitional job earningsb
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

406
972
361
91
1,831

0
0
0
0
0

406
972
361
91
1,831

Unsubsidized earnings
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

355
623
899
1,029
2,906

318
839
966
881
3,004

36
-216
-67
148
-98

Earnings ($)

189

***
***
***
***
***

0.000
0.000
0.000
0.000
0.000
0.735
0.151
0.715
0.449
0.854
(continued)

Appendix Table C.4 (continued)
Outcome

Transitional
Jobs Group

Total earningsc
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)
Sample size (total = 506)

Job Search Difference
Group
(Impact)

761
1,596
1,260
1,120
4,737

315
828
965
879
2,986

256

250

447 ***
767 ***
296
241
1,751 ***

P-Valuea
0.000
0.000
0.110
0.218
0.001

SOURCES: MDRC calculations using payroll data from New Hope and unemployment insurance data from
the state of Wisconsin.
NOTES: Statistical significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10
percent.
Results in this table are regression-adjusted for pre-random assignment chracteristics.
aStandard errors are presented in this report for all impacts with a p-value of 0.000. Following are the
standard errors for all impacts with a p-value of 0.000 (presented in the order in which they appear in the
table). Employment: 3.172, 3.244, 2.989, 1.912, 2.982, 4.453, 4.277, 4.482, 3.703, and 0.121. Earnings:
40.556, 73.874, 49.792, 23.430, 119.554, 157.420, and 184.390.
bEarnings and employment from transitional jobs are based on payroll data from Milwaukee.
cTotal earnings may not be equal to the sum of transitional job earnings plus unsubsidized job earnings
due to rounding.

190

The Transitional Jobs Reentry Demonstration

Appendix Table C.5
Characteristics of Sample Members at Baseline, by Exposure to Bonus
Characteristic

Exposed to
Bonus

Not Exposed
to Bonus

Full
Sample

Age (%)
18-24 years
25-34 years
35-44 years
45 or older

14.5
39.5
26.2
19.8

20.1
34.7
27.8
17.4

18.2
36.4
27.3
18.2

Average age (years)

34.7

33.9

34.2

Race/ethnicity (%)
White/non-Hispanic
Black/non-Hispanic
Hispanic
Other

22.7
58.7
11.6
7.0

20.1
66.1
8.4
5.4

21.0
63.6
9.5
5.9

Has any children under age 19 (%)

57.0

52.7

54.2

4.7

5.7

5.3

Among those with children
Lives with own children under age 19 (%)
Average number of children
Has a child support order (%)

4.7
2.4
29.7

5.7
2.1
21.0

5.3
2.2
23.9

Among those with child support order a
Average monthly support order ($)

282

278

280

Amount of child support owed a (%)
None
Less than $5,000
$5,000 - $14,999
$15,000 or more
Don’t know

12.2
18.4
30.6
26.5
12.2

4.4
23.5
27.9
27.9
16.2

7.7
21.4
29.1
27.4
14.5

Education (%)
GED certificate
High school diploma
Associate's degree/2-or 4-year college/beyond
None of the above

53.9
21.0
5.4
19.8

52.8
21.3
6.9
19.1

53.2
21.1
6.4
19.3

Has occupational training certificate (%)

43.5

39.4

40.8

Marital status (%)
Never married
Married, living with spouse
Married, living away from spouse
Legally separated/divorced/widowed

66.1
0.6
11.1
22.2

77.6
0.9
5.8
15.6

73.6
0.8
7.6
17.9

Lives with own children under age 19 (%)

(continued)

191

Appendix Table C.5 (continued)
Exposed to
Bonus

Not Exposed
to Bonus

Full
Sample

Housing status (%)
Owns/rents house or apartment
Lives with friends or relatives
Has transitional housing
Lives in shelter/other

9.9
15.1
70.9
4.1

13.6
19.3
60.4
6.6

12.3
17.9
64.0
5.8

Ever employed (%)

91.2

88.3

89.3

Ever employed 6 consecutive months for one employer (%)

58.1

52.7

54.6

Worked full time in the 6 months before incarceration (%)

46.2

40.9

42.7

Medical condition restricts ability to work (%)

7.0

4.9

5.6

Ever in drug or alcohol treatment program (%)

64.9

61.3

62.5

Public assistanceb (%)
No government assistance
Temporary Assistance for Needy Families
Food stamps
Social Security
Medicaid
Housing assistance

51.5
30.4
13.5
1.2
24.0
15.8

52.8
22.4
14.7
3.1
29.8
11.0

52.3
25.2
14.3
2.4
27.8
12.7

Family assists with (%)
Finding a place to live
Finding a job
Dealing with a drug or alcohol problem
Financial support
None of the above

27.1
20.6
12.4
25.3
57.6

26.2
22.2
11.1
29.6
54.3

26.5
21.7
11.5
28.1
55.5

Facility type of most recent release (%)
Prison
Jail

95.9
4.1

95.7
4.3

Under community supervision (parole/probation) (%)

95.9

97.5

97.0

Sample size (total = 506)
SOURCE: MDRC calculations using the Baseline Information Form.

172

334

Characteristic

NOTES: In order to assess differences in characteristics across research groups, chi-square tests were
used for categorical variables and t-test were used for continuous variables. There are no statistically
significant differences between research groups on any measure in this table.
aCalculated only for those who have children and are ordered to pay child support. Some participants
were ordered to pay but did not report the amount; they are missing from this measure.
bFor Detroit this variable is created only for those who were randomly assigned after January 8, 2008,
when the site began random assignment after release.

192

Appendix D

Supplementary Exhibits for Chapter 6

The Transitional Jobs Reentry Demonstration

Appendix Table D.1
Quarterly Impacts on Combined Measures of Employment and Prison: Full Sample
Transitional
Jobs Group

Outcome (%)

Job Search Difference
Group
(Impact)

P-Value

Employed and in prison
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

4.9
8.4
6.7
4.5
31.1

2.6
5.6
4.7
4.4
20.5

2.3
2.9
2.0
0.1
10.6

**
**
*
***

0.010
0.017
0.069
0.927
0.000

Employed and not in prison
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

67.2
66.5
41.1
31.0
61.7

29.0
36.4
29.6
26.9
39.0

38.3
30.2
11.5
4.1
22.6

***
***
***
*
***

0.000
0.000
0.000
0.053
0.000

Not employed and in prison
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

1.1
7.8
15.4
19.3
2.9

5.9
13.6
19.3
21.1
14.7

-4.8
-5.8
-3.8
-1.8
-11.8

***
***
**

0.000
0.000
0.028
0.338
0.000

Not employed and not in prison
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

26.8
17.2
36.8
45.2
4.3

62.5
44.5
46.4
47.6
25.8

-35.8
-27.3
-9.6
-2.4
-21.5

***
***
***

Sample size (total = 1,769)

890

879

***

0.000
0.000
0.000
0.287
0.000

SOURCES: Calculations based on data from Michigan State Police, Minnesota Bureau of Crime Apprehension,
State of Wisconsin Department of Justice, Illinois Crime Information Bureau, and the Department of Corrections
in each state.
NOTES: Statistical significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10 percent.
Results in this table are regression-adjusted for pre-random assignment characteristics.
Measures in this table are based on fiscal quarters and therefore may not sum exactly to the percent for Year
1 outcomes shown in Table 6.1 (which is based on monthly data).
aThese outcomes measure whether a sample member was ever employed (in a transitional or UI-covered job)
and whether he was ever in prison at any point during that quarter. A person could be considered incarcerated and
employed during the same quarter even if the employment and incarceration were not simultaneous.

195

The Transitional Jobs Reentry Demonstration

Appendix Table D.2
One-Year Impacts on Recidivism
Chicago
Transitional Job Search
Jobs Group
Group

Outcome

Difference
(Impact)

P-Value

Arrested (%)

45.8

44.2

1.6

0.761

Convicted of a crimea (%)
Convicted of a felony
Convicted of a misdemeanor

17.6
11.8
8.0

14.9
9.9
3.4

2.8
1.8
4.5 *

0.475
0.573
0.057

1.4
5.4
6.5
6.9

1.3
7.0
4.4
2.8

0.1
-1.6
2.1
4.1 *

0.938
0.517
0.383
0.077

Admitted to prison (%)
Admitted to prison for a new crime
Admitted to prison for a parole/probation violation
Admitted to prison for other reason

22.9
7.9
15.5
0.0

20.4
8.7
12.3
0.0

2.6
-0.8
3.2
0.0

0.545
0.776
0.380
0.000

Total days incarcerated in prison

22.3

24.7

-2.4

0.709

50.6

48.1

2.5

0.627

Incarcerated and employed
Incarcerated and not employed
Not incarcerated and employed
Not incarcerated and not employed

0.9
14.7
30.4
54.0

0.7
17.5
27.3
54.6

0.2
-2.8
3.2
-0.6

0.822
0.467
0.517
0.913

Sample size (total = 374)

189

185

Conviction categoriesb (%)
Convicted of a violent crime
Convicted of a property crime
Convicted of a drug crime
Convicted of a public order crime

Arrested, convicted, or admitted to prison (%)
c

Status in the last quarter of Year 1 (%)

SOURCES: Calculations based on data from Illinois UI, Illinois Criminal Justice Information Authority and
the Department of Corrections in Illinois.
NOTES: Statistical significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10
percent.
Results in this table are regression-adjusted for pre-random assignment characteristics.
Subcategories may sum to more than the total due to multiple arrests, convictions, or incarcerations per
person during the follow-up period.
aEach conviction date is counted only as a single event. If there are multiple convictions on the same
date, only the most serious conviction is recorded in the analysis. Some convictions may have been
associated with an arrest that occurred prior to random assignment.These convictions are counted in the
analysis as occurring after random assignment.Total includes convictions for felony, misdemeanor, and
other crime classes.
bThe classification of charges is based on definitions from Langan and Levin (2002).
cIncarceration and employment status in Quarter 4 after random assignment.

196

The Transitional Jobs Reentry Demonstration

Appendix Table D.3
One-Year Impacts on Recidivism
Detroit
Transitional Job Search Difference
Jobs Group
Group
(Impact)

Outcome
Arrested (%)

P-Value

39.4

33.5

6.0

0.206

14.1
8.8
3.3

18.9
11.5
3.3

-4.8
-2.7
0.1

0.180
0.371
0.963

2.1
7.4
1.9
2.6

4.0
6.7
4.7
3.5

-2.0
0.8
-2.9
-1.0

0.259
0.759
0.118
0.571

Admitted to prison (%)
Admitted to prison for a new crime
Admitted to prison for a parole/probation violation
Admitted to prison for other reason

15.4
7.2
7.8
0.4

12.4
5.0
7.3
0.1

3.0
2.1
0.5
0.4

0.377
0.378
0.842
0.446

Total days incarcerated in prison

18.4

16.4

2.0

0.701

43.0

40.8

2.2

0.647

Incarcerated and employed
Incarcerated and not employed
Not incarcerated and employed
Not incarcerated and not employed

0.6
12.8
20.0
66.6

0.5
8.8
18.7
72.0

0.1
4.0
1.3
-5.4

0.868
0.226
0.763
0.272

Sample size (total = 425)

213

212

Convicted of a crime (%)
Convicted of a felony
Convicted of a misdemeanor
b

Conviction categories (%)
Convicted of a violent crime
Convicted of a property crime
Convicted of a drug crime
Convicted of a public order crime

Arrested, convicted, or admitted to prison (%)
c

Status in the last quarter of Year 1 (%)

SOURCES: Calculations based on data from the Michigan State Police and the Department of Corrections in
Michigan.
NOTES: Statistical significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10
percent.
Results in this table are regression-adjusted for pre-random assignment characteristics.
Subcategories may sum to more than the total due to multiple arrests, convictions, or incarcerations per
person during the follow-up period.
aEach arrest date is counted only as a single event. If there are multiple crimes or charges on the same
date, only the most serious charge is recorded in the analysis.
bEach conviction date is counted only as a single event. If there are multiple convictions on the same
date, only the most serious conviction is recorded in the analysis. Some convictions may have been
associated with an arrest that occurred prior to random assignment. These convictions are counted in the
analysis as occurring after random assignment. Total includes convictions for felony, misdemeanor, and
other crime classes.
cIncarceration and employment status in Quarter 4 after random assignment.

197

The Transitional Jobs Reentry Demonstration

Appendix Table D.4
One-Year Impacts on Recidivism
St. Paul
Transitional Job Search Difference
Jobs Group
Group (Impact)

Outcome

P-Value

Arrested (%)

37.7

35.3

2.4

0.586

Convicted of a crimea (%)
Convicted of a felony
Convicted of a misdemeanor

14.0
9.0
5.1

12.6
6.1
6.4

1.4
2.9
-1.4

0.640
0.223
0.524

4.2
4.0
2.8
3.7

4.1
3.2
0.8
5.0

0.1
0.8
2.0 *
-1.3

0.957
0.632
0.087
0.486

Admitted to prison (%)
Admitted to prison for a new crime
Admitted to prison for a parole/probation violation
Admitted to prison for other reason

44.4
6.4
38.9
0.0

53.2
3.9
50.3
0.0

-8.7 **
2.5
-11.4 ***
0.0

0.045
0.200
0.008
0.000

Total days incarcerated in prison

60.4

83.3

-22.9 ***

0.005

58.8

68.7

-9.9 **

0.019

Incarcerated and employed
Incarcerated and not employed
Not incarcerated and employed
Not incarcerated and not employed

6.1
21.2
38.1
34.6

7.8
26.6
31.2
34.3

-1.8
-5.4
6.9 *
0.2

0.450
0.157
0.098
0.955

Sample size (total = 502)

251

Conviction categoriesb (%)
Convicted of a violent crime
Convicted of a property crime
Convicted of a drug crime
Convicted of a public order crime

Arrested, convicted, or admitted to prison (%)
c

Status in the last quarter of Year 1 (%)

SOURCES: Calculations based on data from St. Paul unemployment insurance records, Minnesota Bureau
of Criminal Apprehension, and the Department of Corrections in Minnesota.
NOTES: Statistical significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10
percent.
Results in this table are regression-adjusted for pre-random assignment characteristics.
Subcategories may sum to more than the total due to multiple arrests, convictions, or incarcerations per
person during the follow-up period.
aEach conviction date is counted only as a single event. If there are multiple convictions on the same
date, only the most serious conviction is recorded in the analysis. Some convictions may have been
associated with an arrest that occurred prior to random assignment. These convictions are counted in the
analysis as occurring after random assignment. Total includes convictions for felony, misdemeanor, and
other crime classes.
bThe classification of charges is based on definitions from Langan and Levin (2002).
cIncarceration and employment status in Quarter 4 after random assignment.

198

The Transitional Jobs Reentry Demonstration

Appendix Table D.5
One-Year Impacts on Recidivism
Milwaukee
Transitional Job Search
Jobs Group
Group

Outcome
Arrested (%)

33.7

30.7

10.2
7.5
1.9

11.9
5.1
5.6

1.1
2.8
4.3
2.7

Admitted to prison (%)
Admitted to prison for a new crime
Admitted to prison for a parole/probation violation
Admitted to prison for other reason
Total days incarcerated in prison

Convicted of a crime (%)
Convicted of a felony
Convicted of a misdemeanor

Difference
(Impact)

P-Value

3.1

0.455

-1.7
2.4
-3.6 **

0.554
0.276
0.036

1.3
3.5
3.2
4.4

-0.2
-0.7
1.2
-1.7

0.867
0.647
0.499
0.329

50.9
3.2
22.6
31.9

52.1
3.1
21.2
34.0

-1.3
0.1
1.3
-2.1

0.780
0.975
0.718
0.623

49.6

57.0

-7.3

0.344

59.6

62.2

-2.6

0.559

Incarcerated and employed
Incarcerated and not employed
Not incarcerated and employed
Not incarcerated and not employed

8.6
25.1
33.4
32.9

6.4
28.7
28.2
36.7

2.2
-3.6
5.1
-3.8

0.360
0.373
0.215
0.378

Sample size (total = 507)

256

251

Conviction categoriesb (%)
Convicted of a violent crime
Convicted of a property crime
Convicted of a drug crime
Convicted of a public order crime

Arrested, convicted, or admitted to prison (%)
c

Status in the last quarter of Year 1 (%)

SOURCES: Calculations based on data from Milwaukee unemployment insurance records, State of
Wisconsin Department of Justice, and the Department of Corrections in Wisconsin.
NOTES: Statistical significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10
percent. The significance level indicates the probability that one would incorrectly conclude that a difference
exists between research groups for the corresponding variable.
Results in this table are regression-adjusted for pre-random assignment characteristics.
Subcategories may sum to more than the total due to multiple arrests, convictions, or incarcerations per
person during the follow-up period.
aEach conviction date is counted only as a single event. If there are multiple convictions on the same
date, only the most serious conviction is recorded in the analysis. Some convictions may have been
associated with an arrest that occurred prior to random assignment. These convictions are counted in the
analysis as occurring after random assignment. Total includes convictions for felony, misdemeanor, and
other crime classes.
bThe classification of charges is based on definitions from Langan and Levin (2002).
cIncarceration and employment status in Quarter 4 after random assignment.

199

The Transitional Jobs Reentry Demonstration

Appendix Table D.6
Quarterly Impacts on Combined Measures of Employment and Arrest: Full Sample
Transitional
Jobs Group

Job Search
Group

Difference
(Impact)

Employed and arrested
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

3.5
9.4
6.3
2.7
36.5

1.3
3.4
2.5
2.0
18.8

2.2
6.1
3.7
0.7
17.7

***

0.003
0.000
0.000
0.341
0.000

Employed and not arrested
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

68.7
65.6
41.7
32.9
56.5

30.5
38.6
31.7
29.2
40.8

38.3
27.0
10.0
3.7
15.7

***
***
***
*
***

0.000
0.000
0.000
0.087
0.000

Not employed and arrested
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

0.7
4.9
7.1
9.8
2.6

4.0
7.9
8.3
10.4
16.5

-3.3 ***
-3.0 ***
-1.1
-0.7
-14.0 ***

0.000
0.009
0.371
0.642
0.000

Not employed and not arrested
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

27.1
20.1
44.9
54.6
4.4

64.2
50.1
57.5
58.4
23.9

-37.2
-30.0
-12.6
-3.7
-19.5

0.000
0.000
0.000
0.113
0.000

Sample size (total = 1,765)

887

878

Outcome (%)

P-Value
***
***
***

***
***
***
***

SOURCES: Calculations based on data from Michigan State Police, Minnesota Bureau of Criminal Apprehension,
State of Wisconsin Department of Justice, Illinois Crime Information Bureau, and the Department of Corrections
in each state.
NOTES: Statistical significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10 percent.
The significance level indicates the probability that one would incorrectly conclude that a difference exists
between research groups for the corresponding variable.
Results in this table are regression-adjusted for pre-random assignment characteristics.
Measures in this table are based on fiscal quarters and therefore may not sum exactly to the percent for Year 1
outcomes shown in Table 6.1 (which is based on monthly data).

200

Appendix E

Supplementary Exhibits for Chapter 7

The Transitional Jobs Reentry Demonstration

Appendix Table E.1
Estimated Net Costs for the Transitional Jobs Group Using Alternative Costs for
Each Local Job Search Program, by Site (2008 Dollars)

Cost Componenta

Gross Cost per
Transitional Jobs
Group Member

Gross Cost per
Net Cost per
Job Search Transitional Jobs
Group Member
Group Member

Chicago
Transitional job expenses

$3,230

Other expenses (job readiness, case management, etc.)

$1,526

$1,185

$341

Total operating costs

$4,756

$1,185

$3,571

Transitional job expenses

$1,801

Other expenses (job readiness, case management, etc.)

$2,239

Total operating costs

$4,041

Transitional job expenses

$1,906

Other expenses (job readiness, case management, etc.)

$1,645

$194b

$1,451

Total operating costs

$3,551

$194

$3,357

Transitional job expenses

$2,177

Other expenses (job readiness, case management, etc.)

$2,712

$969

$1,743

Total operating costs

$4,889

$969

$3,920

Detroit

Milwaukee

St. Paul

203

Appendix F

Characteristics of the Qualitative Study Sample

The Transitional Jobs Reentry Demonstration

Appendix Table F.1
Characteristics of Sample Members from the Qualitative Study, by Site
Characteristic (number)

Chicago Detroit Milwaukee

St. Paul

Total

Age
19-24
25-30
31-39
40-49
50 or older

0
1
3
1
1

2
1
5
1
2

1
1
3
2
1

0
3
3
1
1

3
6
14
5
5

Race/ethnicity (%)
White/non-Hispanic
Black/non-Hispanic
Hispanic

0
4
2

0
11
0

1
6
1

2
6
0

3
27
3

Education
8th grade
Some high school
High school diploma/GED certificate
Some post secondary
Bachelor's degree

0
4
2
0
0

0
3
5
3
0

0
2
6
0
0

1
1
4
1
1

1
10
17
4
1

Work history
None (since age 16)
Laborer, janitor, or custodian
Fast food
Retail clerk/cashier
Automobile inspector
Construction/skilled repair
Barber

1
1
3
0
0
1
0

3
3
2
2
1
0
0

3
2
1
1
0
0
1

1
4
2
0
0
1
0

8
10
8
3
1
2
1

Employeda
Round 2 interview
Round 3 interview

4
3

2
0

3
2

4
1

13
6

Incarceration history
One term (1-9 years)
One term (10 years or more)
Mutiple terms (1-9 years)
Mutiple terms (10 years or more)

1
1
2
2

2
4
0
5

3
1
1
3

3
2
3
0

9
8
6
10

Sample size

SOURCES: Based on information collected from the qualitative study.
NOTES: aEmployment status is given as reported at the time of the Round 2 (2009) and Round 3 (2010)
interviews. Categories do not sum up to the total sample size because some sample members were
incarcerated or deceased, or their employment status was unknown.

207

Appendix G

Results of Subgroup Analyses

211
14.2

Employed in all 4 quarters (%)

71.8
69.3
41.8
31.9
93.6

Total employment (%)
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)
2.1

16.9
23.3
26.9
26.8
50.9

Unsubsidized employment (%)
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Number of quarters employed

66.5
60.2
20.6
7.0
84.3

Transitional employment c (%)
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Employment

Outcome

9.2

1.3

31.1
36.2
31.0
28.3
56.9

31.0
36.4
30.7
28.4
57.0

0.1
-0.4
0.2
-0.1
-0.4

***
***
***
***
***

***

***
***
**

5.0

0.9 ***

40.8
33.1
10.8
3.6
36.7

-14.1 ***
-13.1 ***
-3.8
-1.6
-6.2

66.4
60.5
20.4
7.1
84.7

Age 26 or Younger
Transitional Job Search Difference
Jobs Group
Group (Impact)

0.108

0.000

0.000
0.000
0.018
0.416
0.000

0.001
0.003
0.372
0.715
0.196

0.000
0.000
0.000
0.000
0.000

P-Valuea

22.2

2.4

72.2
77.1
50.0
37.0
92.9

16.7
27.1
31.2
29.8
50.4

66.0
67.0
27.1
9.8
85.7

12.5

1.4

31.7
43.8
35.4
32.1
60.1

31.8
43.6
35.5
32.2
59.7

-0.3
0.1
-0.2
-0.2
0.0

***
***
***
*
***

***

***
***
*

***
***
***
***
***

9.8 ***

0.9 ***

40.6
33.2
14.5
4.9
32.8

-15.1
-16.6
-4.3
-2.3
-9.3

66.4
66.9
27.3
9.9
85.7

Age 27 or Older
Transitional Job Search Difference
Jobs Group
Group (Impact)

Age at Baseline

One-Year Impacts on Employment and Earnings, by Age: Full Sample

Appendix Table G.1

The Transitional Jobs Reentry Demonstration

0.000

0.000
0.000
0.000
0.057
0.000

0.000
0.000
0.091
0.350
0.000

0.000
0.000
0.000
0.000
0.000

(continued)

†
††

Difference
Between
Subgroup
P-Valuea Impactsb

212
849
1,179
886
872
3,785

Total earningsd
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)
216

273
386
646
764
2,069

Unsubsidized earnings
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Sample size (total =1,774)

576
793
240
107
1,716

Transitional job earningsc
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Earnings ($)

Outcome

244

258
574
567
544
1,943

259
574
563
545
1,941

-1
0
4
-1
2

591
604
319
328
1,842

14
-189
83
219
128

576
793
236
109
1,714

Age 26 or Younger
Transitional Job Search Difference
Jobs Group
Group (Impact)

***
***
**
**
***

***
***
***
***
***

0.000
0.000
0.037
0.043
0.000

0.902
0.155
0.576
0.170
0.791

0.000
0.000
0.000
0.001
0.000

P-Valuea

677

777
1,578
1,179
1,028
4,562

179
526
802
900
2,407

598
1,052
377
128
2,155

637

318
927
1,019
972
3,236

319
928
1,022
975
3,244

0
-1
-3
-3
-8

***

***
***
**

***
***
***
***
***

458 ***
651 ***
161
56
1,326 ***

-140
-402
-220
-75
-837

598
1,054
381
131
2,163

Age 27 or Older
Transitional Job Search Difference
Jobs Group
Group (Impact)

Age at Baseline

Appendix Table G.1 (continued)

0.000
0.000
0.141
0.655
0.000

0.004
0.000
0.041
0.548
0.005

0.000
0.000
0.000
0.000
0.000

(continued)

†

†††

†††
††

Difference
Between
Subgroup
P-Valuea Impactsb

213

NOTES: Statistical significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10 percent.
Results in this table are regression-adjusted for pre-random assignment characteristics.
aStandard errors are presented in this table for all impacts with a p-value of 0.000. Following are the standard errors for all impacts with a p-value of 0.000
(presented in the order in which they appear in the table, beginning with the "Age 26 or Younger" group). Employment: 2.986, 3.272, 2.731, 1.741, 2.399,
1.829, 1.864, 1.759, 1.177, 1.370, 2.298, 2.573, 2.649, 4.23, 4.468, 3.832, 2.476, 2.517, 2.632, 2.122, .117, .070, and 2.060. Earnings: 41.85, 66.21, 45.15,
125.10, 25.886, 42.961, 32.992, 20.370, 83.789, 88.355, 124.033, 141.491, 492.158, 52.662, 92.892, and 305.269.
bThe H-statistic is used to assess whether the difference in impacts between the subgroups is statistically significant. Significance levels are indicated as
follows: ††† = 1 percent; †† = 5 percent; and † = 10 percent.
cEarnings and employment from transitional jobs are based on payroll data from each of the sites.
dTotal earnings may not be equal to the sum of transitional job earnings plus unsubsidized job earnings due to rounding.

SOURCES: MDRC calculations using payroll data from each site and unemployment insurance data from each of the states in the demonstration (Michigan,
Minnesota, Wisconsin, and Illinois).

Appendix Table G.1 (continued)

214

52.3
66.7

Total days incarcerated in prison

Arrested, convicted, or admitted to prison (%)

6.0
22.3
25.9
45.8
223

Incarcerated and employed
Incarcerated and not employed
Not incarcerated and employed
Not incarcerated and not employed

Sample size (total = 1,808)

Status in the last quarter of Year 1f (%)

46.6
3.8
30.2
14.6

2.0
3.2
6.2
4.1

15.1
10.5
3.7

47.4

Admitted to prisone (%)
Admitted to prison for a new crime
Admitted to prison for technical parole violation
Admitted to prison for other reason

Conviction categoriesd (%)
Convicted of a violent crime
Convicted of a property crime
Convicted of a drug crime
Convicted of a public order crime

Convicted of a crime (%)
Convicted of a felony
Convicted of a misdemeanor

Arrested (%)

Outcome

249

4.5
27.8
23.8
43.9

65.1

64.9

45.0
5.1
30.8
11.8

3.5
3.6
4.1
6.0

16.3
8.3
5.9

42.5

1.5
-5.5
2.1
1.9

1.6

-12.6

1.7
-1.3
-0.6
2.9

-1.5
-0.4
2.2
-1.9

-1.2
2.2
-2.2

4.9

Age 26 or Younger
Transitional Job Search Difference
Jobs Group
Group (Impact)

0.485
0.196
0.614
0.686

0.721

0.118

0.715
0.515
0.881
0.301

0.357
0.821
0.302
0.382

0.727
0.428
0.300

0.300

P-Value

686

4.0
18.0
33.0
45.0

48.8

35.7

31.0
6.9
19.8
6.8

2.3
5.2
3.2
3.4

13.0
8.6
4.3

35.7

650

4.2
19.0
27.8
49.0

53.2

41.4

33.4
4.7
21.4
9.2

2.5
5.5
2.4
3.7

13.9
7.7
4.7

32.8

0.099

0.147

0.321
0.093
0.469
0.069

0.858
0.783
0.389
0.759

0.615
0.558
0.750

0.241

(continued)

†

Difference
Between
Subgroup
P-Value Impactsa

-0.2
0.861
-1.0
0.645
5.2 ** 0.039
-4.0
0.126

-4.4 *

-5.7

-2.4
2.1 *
-1.5
-2.4 *

-0.2
-0.3
0.8
-0.3

-0.9
0.9
-0.4

3.0

Age 27 or Older
Transitional Job Search Difference
Jobs Group
Group (Impact)

Age at Baseline

One-Year Impacts on Recidivism, by Age: Full Sample

Appendix Table G.2

The Transitional Jobs Reentry Demonstration

215

NOTES: Statistical significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10 percent.
Subcategories may sum to more than the total due to multiple arrests, convictions, or incarcerations per person during the follow-up period.
aThe H-statistic is used to assess whether the difference in impacts between the subgroups is statistically significant. Significance levels are indicated as
follows: ††† = 1 percent, †† = 5 percent, and † = 10 percent.
bEach arrest date is counted only as a single event. If there are multiple crimes or charges on the same date, only the most serious charge is recorded in the
analysis.
cEach conviction date is counted only as a single event. If there are multiple convictions on the same date, only the most serious conviction is recorded in the
analysis. Some convictions may have been associated with an arrest that occurred prior to random assignment. These convictions are counted in the analysis as
occurring after random assignment. Total includes convictions for felony, misdemeanor, and other crime classes.
dThe categorization of charges is based on definitions from Langan and Levin (2002).
eIncludes all reasons for prison admission such as sentences for new crimes, technical parole violations, and other reasons.
fIncarceration and employment status in Quarter 4 after random assignment.

Appendix Table G.2 (continued)

216
20.9

Employed in all 4 quarters (%)

72.2
74.7
50.5
38.5
92.7

Total employment (%)
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)
2.4

16.7
26.8
30.5
30.6
52.3

Unsubsidized employment (%)
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Number of quarters employed

66.2
64.9
28.3
10.9
84.6

Transitional employmentc (%)
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Employment

Outcome

12.0

1.5

32.2
43.6
36.0
34.0
62.2

32.5
43.5
36.1
34.1
62.0

0.0
0.0
0.0
0.0
0.0

***
***
***
*
***

***

***
***
**

***
***
***
***
***

8.9 ***

0.9 ***

40.0
31.1
14.5
4.4
30.5

-15.8
-16.7
-5.6
-3.5
-9.6

66.2
64.9
28.3
10.9
84.6

Early
Transitional Job Search Difference
Jobs Group
Group (Impact)

0.000

0.000
0.000
0.000
0.081
0.000

0.000
0.000
0.025
0.159
0.000

0.000
0.000
0.000
0.000
0.000

P-Valuea

17.8

2.1

71.6
75.9
38.8
23.4
92.5

15.3
23.3
28.7
21.3
42.4

67.2
66.6
14.9
2.1
87.8

10.1

1.1

29.6
35.5
27.1
22.1
50.0

29.8
35.1
27.1
22.1
49.3

-0.4
0.2
0.0
0.0
0.5

***
***
***
*
***

***

***
***
**

7.6 **

1.0 ***

42.1
40.4
11.7
1.3
42.5

-14.5 ***
-11.8 **
1.5
-0.8
-6.9

67.6
66.4
14.9
2.1
87.3

Late
Transitional Job Search Difference
Jobs Group
Group (Impact)

Cohort

One-Year Impacts on Employment and Earnings, by Cohort: Full Sample

Appendix Table G.3

The Transitional Jobs Reentry Demonstration

††

†

†††
†††

(continued)

0.041

0.000

0.000
0.000
0.016
0.774
0.000

0.001
0.014
0.743
0.854
0.185

0.000
0.000
0.000
0.067
0.000

Difference
Between
Subgroup
P-Valuea Impactsb

217
740
1,449
1,128
1,029
4,347

Total earningsd
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)
706

184
464
744
884
2,276

Unsubsidized earnings
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Sample size (total = 1,774)

556
985
384
145
2,070

Transitional job earningsc
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Earnings ($)

Outcome

695

297
876
995
979
3,147

300
876
997
980
3,152

0
0
0
0
0

***

**
***
**

***
***
***
***
***

444 ***
573 ***
133
50
1,200 ***

-115
-412
-252
-96
-876

556
985
384
145
2,070

Early
Transitional Job Search Difference
Jobs Group
Group (Impact)

0.000
0.000
0.206
0.682
0.000

0.016
0.000
0.015
0.424
0.003

0.000
0.000
0.000
0.000
0.000

P-Valuea

187

978
1,550
1,023
789
4,340

233
559
830
758
2,380

745
991
193
31
1,960

186

339
708
526
441
2,014

346
699
524
441
2,011

-7
9
2
0
3

***

***
***
***

639
842
497
348
2,326

***
***
***
**
***

-113
-140
306 *
316 *
369

752
982
191
32
1,957

Late
Transitional Job Search Difference
Jobs Group
Group (Impact)

Cohort

Appendix Table G.3 (continued)

†

†††
††
††

†††
†††

†††

(continued)

0.000
0.000
0.003
0.038
0.000

0.380
0.363
0.065
0.058
0.458

0.000
0.000
0.000
0.117
0.000

Difference
Between
Subgroup
P-Valuea Impactsb

218

NOTES: Statistical significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10 percent.
The early cohort includes sample members randomly assigned before April 1, 2008 (N = 1,401). The late cohort includes those randomly assigned on and
after that date (N = 373).
aStandard errors are presented in this table for all impacts with a p-value of 0.000. Following are the standard errors for all impacts with a p-value of 0.000
(presented in the order in which they appear in the table starting with the "Early" cohort). Employment: 1.729, 1.819, 1.721, 1.194, 1.350, 3.603, 3.639, 2.717,
2.428, 2.233, 2.508, 2.567, 2.375, 2.474, 2.593, 2.047, 4.853, 4.919, 4.191, 0.068, 1.956, and 0.134. Earnings: 23.478, 41.204, 31.991, 20.993, 81.701, 56.842,
78.116, 45.498, 131.375, 84.428, 52.874, 90.008, 300.786, 135.265, 155.029, 498.168.
bThe H-statistic is used to assess whether the difference in impacts between the subgroups is statistically significant. Significance levels are indicated as
follows: ††† = 1 percent, †† = 5 percent, and † = 10 percent.
cEarnings and employment from transitional jobs are based on payroll data from each of the sites.
dTotal earnings may not be equal to the sum of transitional job earnings plus unsubsidized job earnings due to rounding.

SOURCES: MDRC calculations using payroll data from each site and unemployment insurance data from each of the states in the demonstration (Michigan,
Minnesota, Wisconsin, and Illinois).

Appendix Table G.3 (continued)

219

42.7
55.7

Total days incarcerated in prison

Arrested, convicted, or admitted to prison (%)

4.8
19.8
33.7
41.7
708

Incarcerated and employed
Incarcerated and not employed
Not incarcerated and employed
Not incarcerated and not employed

Sample size (total = 1,808)

Status in the last quarter of Year 1f (%)

36.8
6.2
24.5
8.3

2.8
5.1
3.8
3.6

14.3
9.9
3.8

40.4

Admitted to prisone (%)
Admitted to prison for a new crime
Admitted to prison for technical parole violation
Admitted to prison for other reason

Conviction typed (%)
Convicted of a violent crime
Convicted of a property crime
Convicted of a drug crime
Convicted of a public order crime

Convicted of a crime (%)
Convicted of a felony
Convicted of a misdemeanor

Arrestedb (%)

Outcome

699

4.7
21.1
29.3
44.9

56.7

49.4

37.3
5.1
25.2
9.0

2.8
4.3
2.9
4.4

14.1
7.7
5.1

35.0

0.1
-1.3
4.4 *
-3.2

-1.0

-6.7

-0.5
1.1
-0.7
-0.6

0.0
0.9
0.9
-0.8

0.2
2.2
-1.3

5.5 **

Early
Transitional Job Search Difference
Jobs Group
Group (Impact)

0.904
0.533
0.078
0.220

0.696

0.106

0.826
0.371
0.745
0.627

0.966
0.431
0.352
0.460

0.914
0.140
0.242

0.032

P-Value

201

2.8
18.4
20.8
57.9

46.6

31.5

29.4
5.8
16.2
10.4

0.4
3.7
3.7
3.8

10.4
5.5
6.0

33.3

200

3.7
20.3
18.4
57.6

53.7

40.5

33.0
4.2
18.2
12.5

2.6
6.8
3.9
3.7

16.2
9.1
4.5

35.9

-0.9
-1.8
2.4
0.3

-7.1

-8.9

-3.6
1.6
-2.1
-2.1

-2.3 *
-3.1
-0.2
0.0

-5.8 *
-3.6
1.4

-2.7

Late
Transitional Job Search Difference
Jobs Group
Group (Impact)

Cohort

One-Year Impacts on Recidivism, by Cohort: Full Sample

Appendix Table G.4

The Transitional Jobs Reentry Demonstration

†

(continued)

0.636
0.659
0.570
0.954

0.145

0.230

0.408
0.466
0.584
0.453

0.085
0.187
0.919
0.984

0.099
0.183
0.545

0.584

Difference
Between
Subgroup
P-Value Impactsa

220

NOTES: Statistical significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10 percent.
Results in this table are regression-adjusted for pre-random assignment characteristics.
Subcategories may sum to more than the total due to multiple arrests, convictions, or incarcerations per person during the follow-up period.
The early cohort includes sample members randomly assigned before April 1, 2008 (N = 1,407). The late cohort includes those randomly assigned on and
after that date (N = 401).
aThe H-statistic is used to assess whether the difference in impacts between the subgroups is statistically significant. Significance levels are indicated as
follows: ††† = 1 percent, †† = 5 percent, and † = 10 percent.
bEach arrest date is counted only as a single event. If there are multiple crimes or charges on the same date, only the most serious charge is recorded in the
analysis.
cEach conviction date is counted only as a single event. If there are multiple convictions on the same date, only the most serious conviction is recorded in
the analysis. Some convictions may have been associated with an arrest that occurred prior to random assignment. These convictions are counted in the analysis
as occurring after random assignment.Total includes convictions for felony, misdemeanor, and other crime classes.
dThe categorization of charges is based on definitions from Langan and Levin (2002).
eIncludes all reasons for prison admission such as sentences for new crimes, technical parole violations, and other reasons.
fIncarceration and employment status in Quarter 4 after random assignment.

SOURCES: Calculations based on data from unemployment insurance in each site, Michigan State Police, Minnesota Bureau of Criminal Apprehension, State
of Wisconsin Department of Justice, Illinois Criminal Justice Information Authority, and the Department of Corrections in each state.

Appendix Table G.4 (continued)

221
19.8

Employed in all 4 quarters (%)

72.6
72.9
45.5
31.9
92.8

Total employment (%)
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)
2.2

13.4
23.4
28.1
25.9
46.6

Unsubsidized employment (%)
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Number of quarters employed

67.3
64.6
23.9
7.7
85.7

Transitional employment c (%)
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Employment

Outcome

8.7

1.2

28.9
39.3
29.8
25.6
56.8

30.0
39.1
29.8
25.7
56.8

0.0
0.0
0.0
0.0
0.0

***
***
***
***
***

***
***
***
*
***

11.1 ***

1.0 ***

43.7
33.6
15.7
6.3
35.9

-16.6 ***
-15.7 ***
-1.7
0.2
-10.2 ***

67.3
64.6
23.9
7.7
85.7

0.000

0.000
0.000
0.000
0.050
0.000

0.000
0.000
0.607
0.960
0.004

0.000
0.000
0.000
0.000
0.000

20.1

2.4

72.6
76.5
49.7
38.0
92.5

19.8
27.5
30.9
30.8
53.3

66.9
65.7
27.5
10.2
85.2

14.7

1.5

34.5
45.1
38.3
36.6
61.9

34.2
44.6
38.1
36.5
61.6

0.0
0.0
0.0
0.0
0.0

***

***
***
***

***
***
**
*
**

***
***
***
***
***

5.5 **

0.8 ***

38.2
31.3
11.4
1.5
30.6

-14.4
-17.2
-7.2
-5.7
-8.3

66.9
65.7
27.5
10.2
85.2

0.032

0.000

0.000
0.000
0.001
0.647
0.000

(continued)

0.000
0.000
0.025
0.072
0.011

0.000
0.000
0.000
0.000
0.000

Difference
Employment History
Between
Worked for Same Employer for 6 Months
Never Worked for Same Employer for 6 Months
Transitional Job Search Difference
Transitional Job Search Difference
Subgroup
P-Valuea Jobs Group
P-Valuea Impactsb
Jobs Group
Group (Impact)
Group (Impact)

One-Year Impacts on Employment and Earnings, by Employment History: Full Sample

Appendix Table G.5

The Transitional Jobs Reentry Demonstration

222
761
1,378
1,022
829
3,990

Total earningsd
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)
420

164
416
704
717
2,002

Unsubsidized earnings
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Sample size (total =1,658)

598
962
318
111
1,989

Transitional job earningsc
Quarter 1 (quarter of random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Earnings ($)

Outcome

363

264
620
649
592
2,125

269
618
654
592
2,134

0
0
0
0
0

***
***
***
***
***

498
758
373
237
1,866

***
***
***
*
***

-105
-202 **
50
125
-132

603
960
323
111
1,998

0.000
0.000
0.002
0.065
0.000

0.131
0.035
0.662
0.324
0.694

0.000
0.000
0.000
0.000
0.000

412

820
1,525
1,145
1,081
4,570

223
523
777
954
2,477

596
1,002
368
127
2,093

463

369
1,068
1,165
1,180
3,783

361
1,061
1,162
1,181
3,766

0
0
0
0
0

***

**
***
***

***
***
***
***
***

450 ***
457 ***
-20
-99
788 *

-138
-538
-386
-227
-1289

588
995
365
128
2,077

††

†
††

††
††
†
††

(continued)

0.000
0.000
0.889
0.556
0.055

0.043
0.000
0.008
0.176
0.002

0.000
0.000
0.000
0.000
0.000

Difference
Employment History
Between
Never Worked for Same Employer for 6 Months
Worked for Same Employer for 6 Months
Transitional Job Search Difference
Transitional Job Search Difference
Subgroup
P-Valuea Jobs Group
P-Valuea Impactsb
Jobs Group
Group (Impact)
Group (Impact)

Appendix Table G.5 (continued)

223

NOTES: Statistical significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10 percent.
Results in this table are regression-adjusted for pre-random assignment characteristics.
aStandard errors are presented in this table for all impacts with a p-value of 0.000. Following are the standard errors for all impacts with a p-value of 0.000
(presented in the order in which they appear in the table, beginning with the "Never Worked for Same Employer for 6 months'" group). Employment: 2.427,
2.538, 2.271, 1.427, 1.846, 2.141, 2.230, 2.098, 1.432, 1.635, 2.854, 3.258, 2.978, 3.199, 3.205, 3.354, 3.424, 2.759, 3.043, 3.101, 2.616, 0.09, 0.087, and 2.523.
Earnings: 33.867, 56.733, 42.714, 27.67, 113.912, 30.581, 49.373, 37.063, 22.789, 92.487, 118.948, 77.303, 105.868, 349.274, 72.370, and 122.187.
bThe H-statistic is used to assess whether the difference in impacts between the subgroups is statistically significant. Significance levels are indicated as
follows: ††† = 1 percent, †† = 5 percent, and † = 10 percent.
cEarnings and employment from transitional jobs are based on payroll data from each of the sites.
dTotal earnings may not be equal to the sum of transitional job earnings plus unsubsidized job earnings due to rounding.

SOURCES: MDRC calculations using payroll data from each site and unemployment insurance data from each of the states in the demonstration (Michigan,
Minnesota, Wisconsin, and Illinois).

Appendix Table G.5 (continued)

224

44.1
59.5

Total days incarcerated in prison

Arrested, convicted, or admitted to prison

4.6
21.4
27.2
46.8
428

Incarcerated and employed
Incarcerated and not employed
Not incarcerated and employed
Not incarcerated and not employed

Sample size (total = 1,691)

Status in the last quarter of Year 1f (%)

25.3
10.3

24.7
11.4

373

2.7
25.0
22.7
49.6

57.1

51.5

39.3
5.9

3.0
4.4
3.4
3.5

14.0
7.9
4.4

36.5

39.4
6.5

2.5
4.1
5.7
2.8

14.5
10.7
3.0

43.9

Admitted to prisone (%)
Admitted to prison for a new crime
Admitted to prison for a technical
parole violation
Admitted to prison for other reason

Conviction categoriesd (%)
Convicted of a violent crime
Convicted of a property crime
Convicted of a drug crime
Convicted of a public order crime

Convicted of a crime (%)
Convicted of a felony
Convicted of a misdemeanor

Arrestedb (%)

Outcome

1.9
-3.6
4.5
-2.8

2.5

-7.4

-0.6
1.1

0.1
0.6

-0.5
-0.3
2.3
-0.7

0.5
2.9
-1.5

7.4 **

0.173
0.242
0.141
0.410

0.459

0.172

0.833
0.569

0.965
0.743

0.699
0.838
0.126
0.568

0.834
0.172
0.277

0.029

420

4.6
18.5
33.6
43.3

48.3

36.8

20.6
6.6

31.8
6.2

2.0
5.4
2.4
4.5

13.1
8.1
5.3

34.8

470

5.8
18.2
30.8
45.2

54.6

45.4

23.3
8.8

33.9
4.0

2.2
5.2
2.8
5.1

14.8
8.1
5.7

34.4

0.305
0.166

0.467
0.134

0.851
0.841
0.734
0.696

0.470
0.980
0.816

0.893

-1.2
0.3
2.8
-1.9

0.437
0.917
0.375
0.557

-6.2 * 0.060

-8.6 * 0.093

-2.7
-2.3

-2.2
2.2

-0.2
0.3
-0.4
-0.6

-1.7
0.0
-0.4

0.4

(continued)

†

Difference
Employment History
Between
Never Worked for Same Employer for 6 Months
Worked for Same Employer for 6 Months
Transitional Job Search Difference
Transitional Job Search Difference
Subgroup
Jobs Group
Group (Impact)
P-Value Jobs Group
Group (Impact) P-Value Impactsa

One-Year Impacts on Recidivism, by Employment History: Full Sample

Appendix Table G.6

The Transitional Jobs Reentry Demonstration

225

NOTES: Statistical significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10 percent.
Results in this table are regression-adjusted for pre-random assignment characteristics.
Subcategories may sum to more than the total due to multiple arrests, convictions, or incarcerations per person during the follow-up period.
aThe H-statistic is used to assess whether the difference in impacts between the subgroups is statistically significant. Significance levels are
indicated as follows: ††† = 1 percent, †† = 5 percent, and † = 10 percent.
bEach arrest date is counted only as a single event. If there are multiple crimes or charges on the same date, only the most serious charge is recorded
in the analysis.
cEach conviction date is counted only as a single event. If there are multiple convictions on the same date, only the most serious conviction is
recorded in the analysis. Some convictions may have been associated with an arrest that occurred prior to random assignment.These convictions are
counted in the analysis as occurring after random assignment. Total includes convictions for felony, misdemeanor, and other crime classes.
dThe categorization of charges is based on definitions from Langan and Levin (2002).
eIncludes all reasons for prison admission such as sentences for new crimes, technical parole violations, and other reasons.
fIncarceration and employment status in Quarter 4 after random assignment.

Appendix Table G.6 (continued)

226
22.1

Employed in all 4 quarters (%)

71.9
77.0
53.1
38.6
93.7

Total employment (%)
Quarter 1 (random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)
2.4

16.5
26.3
33.9
32.2
53.6

Unsubsidized employment (%)
Quarter 1 (random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Number of quarters employed

66.7
65.8
28.5
9.3
85.9

Transitional employmentc (%)
Quarter 1 (random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Employment

Outcome

12.7

1.5

31.0
46.0
37.8
34.9
62.8

31.0
45.7
37.6
34.8
62.6

0.0
0.0
0.0
0.0
0.0

***
***
***
***
***

***

***
***
***

9.5 ***

0.9 ***

40.9
31.1
15.3
3.7
30.9

-14.4 ***
-19.4 ***
-3.7
-2.6
-9.0 ***

66.7
65.8
28.5
9.3
85.9

3 or Fewer Past Convictions
Transitional Job Search Difference
Jobs Group
Group
(Impact)

0.000

0.000
0.000
0.000
0.236
0.000

0.000
0.000
0.236
0.398
0.005

0.000
0.000
0.000
0.000
0.000

P-Valuea

18.2

2.2

72.3
72.9
42.3
32.4
92.2

16.4
26.0
26.1
25.6
47.1

65.9
64.6
22.0
8.9
84.9

10.3

1.3

32.7
37.6
30.3
27.1
56.2

32.9
37.3
30.5
27.3
55.6

0.0
0.0
0.0
0.0
0.0

***
***
***
***
***

***
***
***
*
***

7.8 ***

0.9 ***

39.6
35.3
12.0
5.3
36.1

-16.6 ***
-11.3 ***
-4.4
-1.8
-8.5 ***

65.9
64.6
22.0
8.9
84.9

More than 3 Past Convictions
Transitional Job Search Difference
Jobs Group
Group
(Impact)

Criminal History

One-Year Impacts on Employment and Earnings, by Criminal History: Full Sample

Appendix Table G.7

The Transitional Jobs Reentry Demonstration

0.001

0.000

0.000
0.000
0.000
0.088
0.000

0.000
0.000
0.156
0.553
0.010

0.000
0.000
0.000
0.000
0.000

P-Valuea

(continued)

†

††

Difference
Between
Subgroup
Impactsb

227
802
1495
1228
1137
4662

Total earningsd
Quarter 1 (random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)
467

204
495
870
1033
2601

Unsubsidized earnings
Quarter 1 (random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Sample size (total =1,765)

599
1,000
358
104
2,061

Transitional job earningsc
Quarter 1 (random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Earnings ($)

Outcome

455

314
944
1,036
957
3,252

310
936
1,032
954
3,232

0
0
0
0
0

***
***
***
***
***

488 ***
551 ***
191
180
1410 ***

-106 *
-441 ***
-162
78
-631 *

599
1,000
358
104
2,061

3 or Fewer Past Convictions
Transitional Job Search Difference
Jobs Group
Group
(Impact)

0.000
0.000
0.121
0.238
0.000

0.099
0.000
0.187
0.606
0.073

0.000
0.000
0.000
0.000
0.000

P-Valuea

420

764
1430
958
819
3971

180
464
639
678
1960

584
966
319
141
2,011

423

308
736
750
746
2,540

312
739
757
751
2,560

0
0
0
0
0

***
***
***
***
***

456 ***
693 ***
209
73
1431 ***

-132 **
-276 **
-119
-74
-600

584
966
319
141
2,011

More than 3 Past Convictions
Transitional Job Search Difference
Jobs Group
Group
(Impact)

Criminal History

Appendix Table G.7 (continued)

0.000
0.000
0.112
0.585
0.000

0.046
0.012
0.353
0.576
0.101

0.000
0.000
0.000
0.000
0.000

P-Valuea

(continued)

Difference
Between
Subgroup
Impactsb

228

NOTES: Statistical significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10 percent.
Results in this table are regression-adjusted for pre-random assignment characteristics.
aStandard errors are presented in this table for all impacts with a p-value of 0.000. Following are the standard errors for all impacts with a p-value of 0.000
(presented in the order in which they appear on the table beginning with the "3 or Fewer Past Convictions Group'" group). Employment: 2.141, 2.237, 2.142,
1.383, 1.625, 2.272, 2.327, 2.033, 1.376, 1.727, 2.717, 3.098, 2.921, 3.158, 2.911, 3.003, 3.184, 2.515, 3.115, 3.201, 3.274, 2.720, 0.083, 0.087, and 2.445.
Earnings: 29.555, 49.693, 37.768, 22.923, 94.71, 32.609, 52.258, 38.229, 25.988, 102.027, 98.023, 68.374, 101.758, 353.524, 73.775, 117.899, and 377.591.
bThe H-statistic is used to assess whether the difference in impacts between the subgroups is statistically significant. Significance levels are indicated as
follows: ††† = 1 percent, †† = 5 percent, and † = 10 percent.
cEarnings and employment from transitional job are based on payroll data from each of the sites.
dTotal earnings may not be equal to the sum of transitional job earnings plus unsubsidized job earnings due to rounding.

SOURCES: MDRC calculations using payroll data from each site and unemployment insurance data from each of the states in the demonstration (Michigan,
Minnesota, Wisconsin, and Illinois).

Appendix Table G.7 (continued)

229
36.1
48.7

Total days incarcerated in prison

Arrested, convicted, or admitted to prison (%)

4.2
17.3
34.5
44.0
480

Incarcerated and employed
Incarcerated and not employed
Not incarcerated and employed
Not incarcerated and not employed

Sample size (total = 1,802)

Status in the last quarter of Year 1f (%)

23.5
10.9

21.0
10.2

469

4.1
22.3
30.8
42.8

51.2

47.6

36.2
4.0

32.5
3.8

3.6
2.5
2.4
3.4

11.3
7.2
3.6

29.0

Admitted to prisone (%)
Admitted to prison for a new crime
Admitted to prison for a technical
parole violation
Admitted to prison for other reason

1.1
2.1
3.9
2.1

Conviction categoriesd (%)
Convicted of a violent crime
Convicted of a property crime
Convicted of a drug crime
Convicted of a public order crime

32.4
9.0
6.3
1.9

Convicted of a crime (%)
Convicted of a felony
Convicted of a misdemeanor

Arrestedb (%)

Outcome

0.1
-5.0 *
3.7
1.2

-2.6

-11.5 **

-2.6
-0.7

-3.7
-0.2

-2.5 **
-0.4
1.4
-1.2

-2.3
-0.9
-1.8

3.4

0.930
0.055
0.232
0.692

0.408

0.020

0.312
0.689

0.201
0.882

0.010
0.656
0.224
0.250

0.246
0.568
0.101

0.248

424

4.8
21.1
27.4
46.7

58.4

43.7

23.7
7.7

37.6
8.5

3.4
7.5
3.8
5.3

18.2
11.9
6.8

45.1

429

4.7
20.7
22.5
52.1

62.1

49.1

24.8
8.3

37.2
6.0

2.0
7.7
3.7
5.2

18.6
9.0
6.6

43.0

0.1
0.3
4.9 *
-5.4

-3.6

-5.4

-1.1
-0.6

0.5
2.5

1.5
-0.2
0.1
0.1

-0.4
2.9
0.2

2.1

0.921
0.900
0.099
0.103

0.273

0.305

0.682
0.725

0.881
0.167

0.199
0.927
0.949
0.951

0.868
0.163
0.916

0.534

(continued)

†††

Difference
Criminal History
Between
3 or Fewer Past Convictions
More than 3 Past Convictions
Transitional Job Search Difference
Transitional Job Search Difference
Subgroup
Jobs Group
Group (Impact)
P-Value
Jobs Group
Group (Impact) P-Value Impactsa

One-Year Impacts on Recidivism, by Criminal History: Full Sample

Appendix Table G.8

The Transitional Jobs Reentry Demonstration

230

NOTES: Statistical significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10 percent.
Results in this table are regression-adjusted for pre-random assignment characteristics.
Subcategories may sum to more than the total due to multiple arrests, convictions or incarcerations per person during the follow-up period.
aThe H-statistic is used to assess whether the difference in impacts between the subgroups is statistically significant. Significance levels are indicated
as follows: ††† = 1 percent, †† = 5 percent, and † = 10 percent.
bEach arrest date is counted only as a single event. If there are multiple crimes or charges on the same date, only the most serious charge is recorded in
the analysis.
cEach conviction date is counted only as a single event. If there are multiple convictions on the same date, only the most serious conviction is recorded
in the analysis. Some convictions may have been associated with an arrest that occurred prior to random assignment.These convictions are counted in the
analysis as occurring after random assignment. Total includes convictions for felony, misdemeanor, and other crime classes.
dThe categorization of charges is based on definitions from Langan and Levin (2002).
eIncludes all reasons for prison admission such as sentences for new crimes, technical parole violations, and other reasons.
fIncarceration and employment status in Quarter 4 after random assignment.

Appendix Table G.8 (continued)

231
17.0

Employed in all 4 quarters (%)

75.8
76.1
42.2
31.0
94.2

Total employment (%)
Quarter 1 (random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)
2.3

14.1
21.4
25.2
24.5
45.4

Unsubsidized employment (%)
Quarter 1 (random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Number of quarters employed

72.5
69.7
22.6
8.4
90.2

Transitional employmentc (%)
Quarter 1 (random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Employment

Outcome

10.5

1.3

26.3
39.1
32.0
29.2
54.3

26.5
39.1
32.1
29.2
54.1

0.0
0.0
0.0
0.0
0.0

***

***
***
***

***
***
**
*
***

***
***
***
***
***

6.5 ***

1.0 ***

49.4
37.0
10.3
1.8
40.0

-12.5
-17.6
-6.9
-4.7
-8.7

72.5
69.7
22.6
8.4
90.2

0.002

0.000

0.000
0.000
0.000
0.518
0.000

0.000
0.000
0.011
0.079
0.003

0.000
0.000
0.000
0.000
0.000

25.8

2.4

66.1
73.8
57.5
43.2
90.7

20.6
33.6
38.2
36.5
58.6

56.2
58.4
30.2
10.0
77.5

13.0

1.6

40.1
46.0
37.6
34.4
67.8

40.3
45.8
37.4
34.3
67.1

0.0
0.0
0.0
0.0
0.0

***
***
***
***
***

***
***
***
**
***

12.8 ***

0.8 ***

26.0
27.9
19.9
8.8
23.0

-19.6 ***
-12.2 ***
0.8
2.2
-8.5 **

56.2
58.4
30.2
10.0
77.5

Exposure to Retention Bonuses
Not Exposed to Retention Bonuses
Exposed to Retention Bonuses
Transitional Job Search Difference
Transitional Job Search Difference
P-Valuea
Jobs Group
Group
(Impact)
Jobs Group
Group
(Impact)

0.000

0.000
0.000
0.000
0.017
0.000

0.000
0.001
0.829
0.547
0.022

0.000
0.000
0.000
0.000
0.000

P-Valuea

One-Year Impacts on Employment and Earnings, by Exposure to Retention Bonuses: Full Sample

Appendix Table G.9

The Transitional Jobs Reentry Demonstration

(continued)

†

†††

†††
††
††

†

†††

†††
†††
††

Difference
Between
Subgroup
Impactsb

232
775
1,389
958
838
3,961

Total earningsd
Quarter 1 (random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)
551

142
403
641
702
1,888

Unsubsidized earnings
Quarter 1 (random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Sample size (total = 1,775)

633
986
317
136
2,072

Transitional job earningsc
Quarter 1 (random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Earnings ($)

Outcome

545

268
775
824
824
2,691

271
776
826
826
2,699

0
0
0
0
0

***

**
***
*

***
***
***
***
***

507 ***
614 ***
134
14
1270 ***

-129
-373
-185
-124
-811

633
986
317
136
2,072

0.000
0.000
0.202
0.914
0.000

0.017
0.000
0.071
0.327
0.008

0.000
0.000
0.000
0.000
0.000

342

815
1,618
1,359
1,231
5,022

279
616
971
1,131
2,997

536
1,001
387
100
2,025

336

366
931
998
906
3,202

373
936
998
909
3,216

0
0
0
0
0

***
***
***
***
***

449
686
361
324
1820

***
***
**
*
***

-94
-320 **
-27
222
-219

536
1,001
387
100
2,025

Appendix Table G.9 (continued)

0.000
0.000
0.026
0.054
0.000

0.259
0.012
0.866
0.187
0.622

0.000
0.000
0.000
0.000
0.000

P-Valuea

(continued)

†

††

Difference
Between
Subgroup
Impactsb

233

NOTES: In this table, the "Exposed to Retention Bonuses" subgroup includes all sample members in Milwaukee and the sample members from St. Paul who
were eligible for the bonus (randomly assigned after November 2007). The "Not Exposed to Retention Bonuses" subgroup includes sample members from St.
Paul who were not eligible for the bonus (randomly assigned before December 2007) and all sample members in Chicago and Detroit.
Significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10 percent.
aStandard errors are presented in this table for all impacts with a p-value of 0.000. Following are the standard errors for all impacts with a p-value of 0.000
(presented in the order in which they appear in the table beginning with the "Not Exposed to Retention Bonuses'"group). Employment: 1.923, 1.983, 1.793,
1.205, 1.281, 2.556, 2.742, 2.562, 1.660, 2.276, 2.361, 2.714, 3.539, 2.557, 2.734, 2.856, 2.303, 3.633, 3.636, 3.782, 3.045, 0.074, 0.103, and 2.988. Earnings:
27.95, 44.05, 35.125, 24.355, 93.99, 34.780, 62.411, 42.722, 21.652, 101.876, 89.867, 60.287, 95.662, 312.307, 86.600, 132.478, and 448.714.
bThe H-statistic is used to assess whether the difference in impacts between the subgroups is statistically significant. Significance levels are indicated as
follows: ††† = 1 percent; †† = 5 percent; † = 10 percent.
cEarnings and employment from transitional jobs are based on payroll data from each of the sites.
dTotal earnings may not be equal to the sum of transitional job earnings plus unsubsidized job earnings due to rounding.

SOURCES: MDRC calculations using payroll data from each site and unemployment insurance data from each state in the demonstration (Michigan,
Minnesota, Wisconsin, Illinois).

Appendix Table G.9 (continued)

234
32.1
50.8

Total days incarcerated in prison

Arrested, convicted, or admitted to prison (%)

3.2
15.9
27.7
53.1
567

Incarcerated and employed
Incarcerated and not employed
Not incarcerated and employed
Not incarcerated and not employed

Sample size (total = 1,808)

Status in the last quarter of Year 1 (%)

20.6
0.0

20.3
0.2

562

2.6
17.1
26.5
53.8

50.4

37.2

25.4
5.4

2.8
5.4
3.6
3.9

27.3
7.3

2.6
5.9
4.0
4.3

15.4
9.4
3.7

37.3

Admitted to prison (%)
Admitted to prison for a new crime
Admitted to prison for a technical
parole violation
Admitted to prison for other reason

Conviction categories (%)
Convicted of a violent crime
Convicted of a property crime
Convicted of a drug crime
Convicted of a public order crime

15.7
10.0
5.7

Convicted of a crime (%)
Convicted of a felony
Convicted of a misdemeanor

41.5

Arrested (%)

Outcome

0.6
-1.1
1.2
-0.6

0.4

-5.0

-0.2
0.2

1.9
1.9

-0.2
0.5
0.4
0.4

0.2
0.6
1.9

4.2

0.565
0.609
0.655
0.826

0.898

0.213

0.918
0.387

0.441
0.198

0.855
0.698
0.706
0.712

0.911
0.753
0.128

0.139

342

6.7
24.9
36.5
32.0

58.3

53.1

25.5
23.9

48.2
4.2

1.8
2.8
3.3
2.8

10.2
7.5
2.2

34.1

337

6.9
27.7
27.5
38.0

65.4

65.2

30.0
25.3

54.4
3.8

2.6
4.0
2.3
4.5

12.9
5.5
6.7

32.0

-0.2
-2.8
9.0 **
-6.0

-7.1 *

-12.2 *

-4.5
-1.4

-6.2
0.4

-0.8
-1.1
0.9
-1.7

-2.7
1.9
-4.6 ***

2.1

0.925
0.402
0.012
0.102

0.055

0.072

0.180
0.656

0.107
0.794

0.505
0.405
0.470
0.250

0.276
0.305
0.005

0.549

(continued)

†

†††

Difference
Exposure to Retention Bonuses
Between
Not Exposed to Retention Bonuses
Exposed to Retention Bonuses
Transitional Job Search Difference
Transitional Job Search Difference
Subgroup
a
Jobs Group
Group (Impact) P-Value Jobs Group
Group (Impact)
P-Value Impacts

One-Year Impacts on Recidivism, by Exposure to Retention Bonuses: Full Sample

Appendix Table G.10

The Transitional Jobs Reentry Demonstration

235

NOTES: In this table, the "Exposed to Retention Bonuses" subgroup includes all sample members in Milwaukee and the sample members from St. Paul who
were eligible for the bonus (randomly assigned after November 2007). The "Not Exposed to Retention Bonuses" subgroup includes sample members from St.
Paul who were not eligible for the bonus (randomly assigned before December 2007) and all sample members in Chicago and Detroit.
Statistical significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10 percent.
Subcategories may sum to more than the total due to multiple arrests, convictions, or incarcerations per person during the follow-up period.
aThe H-statistic is used to assess whether the difference in impacts between the subgroups is statistically significant. Significance levels are indicated as
follows: ††† = 1 percent, †† = 5 percent, and † = 10 percent.
bEach arrest date is counted only as a single event. If there are multiple crimes or charges on the same date, only the most serious charge is recorded in the
analysis.
cEach conviction date is counted only as a single event. If there are multiple convictions on the same date, only the most serious conviction is recorded in
the analysis. Some convictions may have been associated with an arrest that occurred prior to random assignment.These convictions are counted in the analysis
as occurring after random assignment.Total includes convictions for felony, misdemeanor, and other crime classes.
dThe categorization of charges is based on definitions from Langan and Levin (2002).
eIncludes all reasons for prison admission such as sentences for new crimes, technical parole violations, and other reasons.
fIncarceration and employment status in Quarter 4 after random assignment.

Appendix Table G.10 (continued)

236
22.7

Employed in all 4 quarters (%)

73.4
75.8
51.5
39.2
93.4

Total employment (%)
Quarter 1 (random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)
2.4

17.8
28.7
34.1
33.7
54.5

Unsubsidized employment (%)
Quarter 1 (random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Number of quarters employed

67.4
65.0
25.8
7.9
85.5

Transitional employment c (%)
Quarter 1 (random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Employment

Outcome

14.1

1.5

35.5
45.8
37.3
35.6
63.1

35.7
45.7
37.4
35.7
62.8

0.0
0.0
0.0
0.0
0.0

***
***
***
***
***

***

***
***
***

8.6 ***

0.9 ***

37.9
30.1
14.2
3.6
30.3

-17.9 ***
-17.0 ***
-3.2
-2.0
-8.3 ***

67.7
65.0
25.9
8.0
85.5

0.000

0.000
0.000
0.000
0.186
0.000

0.000
0.000
0.230
0.463
0.002

0.000
0.000
0.000
0.000
0.000

15.9

2.1

71.3
73.6
41.2
27.4
91.9

13.0
19.0
22.4
18.7
40.0

67.2
66.8
24.6
10.2
87.3

5.7

1.0

21.3
32.4
27.3
22.4
50.4

22.3
33.1
27.3
22.5
50.8

0.0
0.0
0.0
0.0
0.0

***
***
***
***
***

***

***
***
***

10.1 ***

1.1 ***

49.9
41.2
13.9
5.1
41.5

-9.3 **
-14.1 ***
-4.9
-3.8
-10.8 **

68.5
67.7
24.6
10.3
88.1

High School Diploma or GED Certificate
Received
Did Not Receive
Transitional Job Search Difference
Transitional Job Search Difference
P-Valuea Jobs Group
Jobs Group
Group
(Impact)
Group
(Impact)

One-Year Impacts on Employment and Earnings, by Educational Attainment: Full Sample

Appendix Table G.11

The Transitional Jobs Reentry Demonstration

0.001

0.000

0.000
0.000
0.003
0.244
0.000

0.015
0.002
0.259
0.355
0.033

0.000
0.000
0.000
0.000
0.000

(continued)

†

††

††
††

†

Difference
Between
Subgroup
P-Valuea Impactsb

237
798
1,570
1226
1160
4,754

Total earningsd
Quarter 1 (random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)
639

213
565
908
1046
2,733

Unsubsidized earnings
Quarter 1 (random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Sample size (total =1,774)

585
1,005
318
113
2,021

Transitional job earningsc
Quarter 1 (random assignment)
Quarter 2
Quarter 3
Quarter 4
Year 1 (Q1 - Q4)

Earnings ($)

Outcome

604

349
983
1,074
1,039
3,445

351
981
1,076
1,042
3,448

0
0
0
0
0

***
***
***
***
***

450 ***
587 ***
152
121
1,309 ***

-137 **
-416 ***
-167
5
-716 **

587
1,002
319
116
2,025

0.000
0.000
0.193
0.369
0.000

0.020
0.000
0.150
0.972
0.032

0.000
0.000
0.000
0.000
0.000

208

774
1,288
870
641
3,573

159
317
469
487
1,433

615
971
401
154
2,140

210

218
520
485
456
1,678

218
523
481
463
1,684

0
0
0
0
0

***
***
***
***
***

556
768
385
185
1,895

***

***
***
***

-58
-205 *
-11
24
-252

615
973
396
162
2,146

High School Diploma or GED Certificate
Received
Did Not Receive
Transitional Job Search Difference
Transitional Job Search Difference
P-Valuea
Jobs Group
Group (Impact)
Jobs Group
Group (Impact)

Appendix Table G.11 (continued)

0.000
0.000
0.008
0.215
0.000

0.478
0.061
0.932
0.871
0.523

0.000
0.000
0.000
0.000
0.000

(continued)

††

Difference
Between
Subgroup
P-Valuea Impactsb

238

NOTES: Statistical significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10 percent.
Results in this table are regression-adjusted for pre-random assignment characteristics.
aStandard errors are presented in this table for all impacts with a p-value of 0.000. Following are the standard errors for all impacts with a p-value of 0.000
(presented in the order in which they appear in the table, beginning with the "Age 26 or Younger" group). Employment: 1.853, 1.949, 1.778, 1.105, 1.416,
3.282, 3.412, 3.099, 2.152, 2.344, 2.432, 2.665, 2.704, 2.556, 2.594, 2.744, 2.124, 4.306, 4.733, 4.126, 0.073, 0.121, and 2.172. Earnings: 25.46, 43.63, 30.49,
19.74, 81.61, 48.977, 77.136, 64.709, 41.719, 163.480, 96.889, 63.401, 101.106, 336.306, 93.369, 126.282, and 422.444.
bThe H-statistic is used to assess whether the difference in impacts between the subgroups is statistically significant. Significance levels are indicated as
follows: ††† = 1 percent; †† = 5 percent; and † = 10 percent.
cEarnings and employment from transitional jobs are based on payroll data from each of the sites.
dTotal earnings may not be equal to the sum of transitional job earnings plus unsubsidized job earnings due to rounding.

SOURCES: MDRC calculations using payroll data from each site and unemployment insurance data from each of the states in the demonstration (Michigan,
Minnesota, Wisconsin, and Illinois).

Appendix Table G.11 (continued)

239
48.2

33.8
5.4
21.6
9.5
37.2
51.0

Admitted to prisone (%)
Admitted to prison for a new crime
Admitted to prison for a technical parole violation
Admitted to prison for other reason

Total days incarcerated in prison

Arrested, convicted, or admitted to prison (%)

4.8
17.1
34.5
43.7
649

Incarcerated and employed
Incarcerated and not employed
Not incarcerated and employed
Not incarcerated and not employed

Sample size (total = 1,808)

Status in the last quarter of Year 1f (%)

36.3
5.0
24.4
9.1

2.5
4.7
3.3
3.6

Conviction categoriesd (%)
Convicted of a violent crime
Convicted of a property crime
Convicted of a drug crime
Convicted of a public order crime

619

5.3
21.1
30.4
43.3

54.8

3.0
4.8
3.1
4.4

14.8
8.5
4.9

13.0
8.1
4.9

Convicted of a crimec (%)
Convicted of a felony
Convicted of a misdemeanor

34.5

36.1

Arrestedb (%)

Outcome

-0.5
-4.0 *
4.1
0.4

-3.8

-11.1 ***

-2.4
0.5
-2.8
0.5

-0.5
-0.1
0.2
-0.8

-1.8
-0.4
0.1

1.6

0.690
0.073
0.125
0.877

0.157

0.009

0.337
0.707
0.211
0.728

0.596
0.920
0.847
0.470

0.346
0.818
0.950

0.540

212

4.5
25.0
23.1
47.4

61.3

48.8

37.8
9.0
24.2
6.7

2.1
5.8
5.4
4.1

17.0
13.0
2.4

47.9

213

1.8
24.0
20.2
54.0

62.1

48.6

37.9
5.1
24.8
10.2

2.7
4.6
4.1
4.8

16.1
8.3
5.6

40.1

2.7
1.0
2.9
-6.6

-0.7

0.2

-0.1
3.8
-0.6
-3.5

-0.6
1.2
1.3
-0.7

0.8
4.7
-3.1

7.9 *

High School Diploma or GED
Received
Did not Receive
Transitional Job Search Difference
Transitional Job Search Difference
Jobs Group
Group
(Impact)
P-Value Jobs Group
Group
(Impact)

One-Year Impacts on Recidivism, by Educational Attainment: Full Sample

Appendix Table G.12

The Transitional Jobs Reentry Demonstration

0.137
0.813
0.483
0.189

0.882

0.980

0.986
0.138
0.878
0.151

0.694
0.589
0.533
0.732

0.816
0.127
0.115

0.096

(continued)

Difference
Between
Subgroup
P-Value Impactsa

240

NOTES: Statistical significance levels are indicated as follows: *** = 1 percent; ** = 5 percent; * = 10 percent.
Subcategories may sum to more than the total due to multiple arrests, convictions, or incarcerations per person during the follow-up period.
aThe H-statistic is used to assess whether the difference in impacts between the subgroups is statistically significant. Significance levels are indicated as follows:
††† = 1 percent, †† = 5 percent, and † = 10 percent.
bEach arrest date is counted only as a single event. If there are multiple crimes or charges on the same date, only the most serious charge is recorded in the
analysis.
cEach conviction date is counted only as a single event. If there are multiple convictions on the same date, only the most serious conviction is recorded in the
analysis. Some convictions may have been associated with an arrest that occurred prior to random assignment. These convictions are counted in the analysis as
occurring after random assignment. Total includes convictions for felony, misdemeanor, and other crime classes.
dThe categorization of charges is based on definitions from Langan and Levin (2002).
eIncludes all reasons for prison admission such as sentences for new crimes, technical parole violations, and other reasons.
fIncarceration and employment status in Quarter 4 after random assignment.

Appendix Table G.12 (continued)

References
Aos, Steve, Polly Phipps, Robert Barnoski, and Roxanne Lieb. 2001. The Comparative Costs and
Benefits of Programs to Reduce Crime, Version 4.0. Olympia, WA: Washington State Institute
for Public Policy.
Berlin, Gordon L. 2000. Encouraging Work, Reducing Poverty: The Impact of Work Incentive
Programs. New York: MDRC.
Bloom, Dan. 2009. “The Joyce Foundation’s Transitional Jobs Reentry Demonstration: Testing
Strategies to Help Former Prisoners Find and Keep Jobs and Stay Out of Prison.” Policy Brief.
Chicago: The Joyce Foundation.
Bloom, Dan. 2010. Transitional Jobs: Background, Program Models, and Evaluation Evidence.
New York: MDRC.
Bloom, Dan, Cindy Redcross, Janine Zweig, and Gilda Azurdia. 2007. “Transitional Jobs for ExPrisoners: Early Impacts from a Random Assignment Evaluation of the Center for Employment
Opportunities (CEO) Prisoner Reentry Program.” Working Paper. New York: MDRC.
Bloom, Howard S., Carolyn J. Hill, and James Riccio. 2001. Modeling the Performance of Welfareto-Work Programs: The Effects of Program Management and Services, Economic Environment, and Client Characteristics. New York: MDRC.
Bonczar, Thomas. 2008. “Characteristics of State Parole Supervising Agencies, 2006.” Washington,
DC: U.S. Department of Justice, Bureau of Justice Statistics.
Bos, Johannes, Aletha C. Huston, Robert C. Granger, Greg J. Duncan, Thomas W. Brock, and
Vonnie McLoyd. 1999. New Hope for People with Low Incomes: Two-Year Results of a Program to Reduce Poverty and Reform Welfare. New York: MDRC.
Bureau of Justice Statistics. 2008. “Expenditures/Employment.”
Web site: http://bjs.ojp.usdoj.gov/index.cfm?ty=pbdetail&iid=1022.
Bureau of Labor Statistics. n.d. Databases & Tables. Local Area Unemployment Statistics. Washington, DC: U.S. Department of Labor.
Web site: http://data.bls.gov/cgi-bin/dsrv?la.
Bureau of Labor Statistics. 2010. “Current Unemployment Rates for States and Historical
Highs/Lows.” Databases & Tables. Local Area Unemployment Statistics. Washington, DC:
U.S. Department of Labor.
Crime Information Bureau. n.d. “Removal of Arrest Information: Can Arrest Information Be
Removed from My Criminal Record?” Madison, WI: Wisconsin Department of Justice.
Web site: www.doj.state.wi.us/dles/cib/forms/brochures/remove_arrest.pdf.

241

Division of Community Corrections. 2003. Operations Manual. Rev. 2000 (updated 2003). Madison, WI: State of Wisconsin Department of Corrections.
Web site: www.wi-doc.com/index_community.htm.
Greenberg, David H., Charles Michalopoulos, and Philip K. Robins. 2001. A Meta-Analysis of
Government-Sponsored Training Programs. New York: MDRC.
Herbst, Chris. 2008. “Do Social Policy Reforms Have Different Impacts on Employment and
Welfare Use as Economic Conditions Change?” Journal of Policy Analysis and Management,
27, 4: 867-894.
Holzer, Harry, Steven Raphael, and Michael Stoll. 2007. “The Effect of an Applicant’s Criminal
History on Employer Hiring Decisions and Screening Practices: Evidence from Los Angeles.”
In Shawn Bushway, Michael Stoll, and David Weiman (eds.), Barriers to Reentry. New York:
Russell Sage Foundation.
Illinois Department of Corrections. 2008. IDOC 2008 Annual Report. Springfield, IL: Illinois
Department of Corrections.
Jenks, Stephen, Lois MacGilliveray, and Karen Needels. 2006. Youth Offender Demonstration
Project Evaluation Final Report: Volume One. Washington, DC: Research and Evaluation Associates.
Kirby, Gretchen, Heather Hill, LaDonna Pavetti, Jon Jacobson, Michelle Derr, and Pamela Winston.
2002. Transitional Jobs: Stepping Stones to Unsubsidized Employment. Princeton: Mathematica Policy Research, Inc.
Landenberger, Nana, and Mark Lipsey. 2005. “The Positive Effects of Cognitive-Behavioral
Programs for Offenders: A Meta-Analysis of Factors Associated with Effective Treatment.”
Journal of Experimental Criminology 1, 4: 451-476.
Langan, Patrick, and David Levin. 2002. “Recidivism of Prisoners Released in 1994.” Bureau of
Justice Statistics Special Report. Washington, DC: U.S. Department of Justice, Bureau of Justice Statistics.
Lattimore, Pamela, and Christy Visher. 2009. The Multi-site Evaluation of the Serious and Violent
Offender Reentry Initiative. Washington, DC: The Urban Institute.
Martinson, Karin, and Richard Hendra. 2006. The Employment Retention and Advancement Project:
Results from the Texas ERA Site. New York: MDRC.
Michalopoulos, Charles. 2005. Does Making Work Pay Still Pay? An Update on the Effects of Four
Earnings Supplement Programs on Employment, Earnings, and Income. New York: MDRC.
Michigan Legislative Council. 2009. “Bureau of Criminal Identification and Records, Act 289 of
1925.” Lansing, MI: Michigan Legislature.
Web site: www.legislature.mi.gov.

242

Minnesota Department of Corrections. 2008a. 2007 Probation Survey. April. St. Paul, MN: Minnesota Department of Corrections.
Web site: www.doc.state.mn.us/publications/documents/04-082007ProbationSurvey.pdf
Minnesota Department of Corrections. 2008b. “Intensive Supervised Release.” DOC Backgrounder
(December).
Web site: www.doc.state.mn.us.
Minnesota Department of Corrections. 2009. Review of Guidelines for Revocation of Parole and
Supervised Released: 2009 Report to the Minnesota Legislature. St. Paul, MN: Minnesota Department of Corrections.
Web site: www.doc.state.mn.us/publications/legislativereports.
Minnesota Judicial Branch. n.d. “Basics on Criminal Expungement.” Self Help Center. St. Paul,
MN: Minnesota Judicial Branch.
Web site: www.courts.state.mn.us/selfhelp/?page=328.
Minnesota Sentencing Guidelines Commission. 2009. “Minnesota Sentencing Guidelines and
Commentary.” St. Paul, MN: Minnesota Sentencing Guidelines Commission.
Web site: www.msgc.state.mn.us/guidelines/guide09.pdf.
Mitchell, Ojmarrh, David Wilson, and Doris MacKenzie. 2007. “Does Incarceration-Based Drug
Treatment Reduce Recidivism? A Meta-Analytic Synthesis of the Literature.” Journal of Experimental Criminology 3: 353-375.
National Bureau of Economic Research. 2008. “Determination of the December 2007 Peak in
Economic Activity.” Cambridge, MA: National Bureau of Economic Research.
National Transitional Jobs Network. 2009. “Questions About the Transitional Jobs Strategy.”
Washington, DC: U.S. Department of Health and Human Services, Administration for Children
and Families, Welfare Peer Technical Assistance Network.
Office of the State Appellate Defender. 2010. “How to Clear Your Illinois Criminal Record.”
Chicago: Office of the State Appellate Defender.
Pager, Devah. 2007. “Two Strikes and You’re Out: The Intensification of Racial and Criminal
Stigma.” In Shawn Bushway, Michael Stoll, and David Weiman (eds.), Barriers to Reentry.
New York: Russell Sage Foundation.
Petersilia, Joan. 2003. When Prisoners Come Home: Parole and Prisoner Reentry. New York:
Oxford University Press.
Raphael, Steven, and Michael A. Stoll. 2007. “Why Are So Many Americans in Prison?” Institute
for Research on Poverty Discussion Paper No. 1328-07. Madison, WI: Institute for Research on
Poverty.
Redcross, Cindy, Dan Bloom, Gilda Azurdia, Janine Zweig, and Nancy Pindus. 2009. Transitional
Jobs for Ex-Prisoners: Implementation, Two-Year Impacts, and Costs of the Center for Employment Opportunities (CEO) Prisoner Reentry Program. New York: MDRC.

243

Sabol, William, Heather West, and Matthew Cooper. (2009). “Prisoners in 2008.” Washington, DC:
U.S. Department of Justice, Office of Justice Programs, Bureau of Justice Statistics.
Solomon, Amy, and Gillian Thompson. 2004. Prisoner Reentry in Michigan. Washington, DC: The
Urban Institute.
Solomon, Amy, Kelly Johnson, Jeremy Travis, and Elizabeth McBride. 2004. From Prison to Work:
The Employment Dimensions of Prisoner Reentry. Washington, DC: Urban Institute.
Uggen, Christopher. 2000. “Work as a Turning Point in the Life Course of Criminals: A Duration
Model of Wage, Employment, and Recidivism.” American Sociological Review 67: 529-546.
Uggen, Christopher, and Jeremy Staff. 2001. “Work as a Turning Point for Criminal Offenders.”
Corrections Management Quarterly 5, 4: 1-16.
Van Stelle, Kit, and Janae Goodrich. 2009. 2008/2009 Study of Probation/Parole Revocation.
Madison, WI: University of Wisconsin Population Health Institute.
Visher, Christy, Laura Winterfield, and Mark Coggeshall. 2005. “Ex-Offender Employment Programs
and Recidivism: A Meta-Analysis.” Journal of Experimental Criminology 1: 295-315.
West, Heather C. 2010. “Prison Inmates at Midyear 2009  Statistical Tables.” NCJ 230113.
Washington, DC: Bureau of Justice Statistics.
Web site: http://bjs.ojp.usdoj.gov/index.cfm?ty=pbdetail&iid=2200.
Western, Bruce, Jeffrey Kling, and David Weiman. 2001. “The Labor Market Consequences of
Incarceration.” Working Paper No. 450. Princeton: Princeton University Industrial Relations
Section.
Wilson, David, Ojmarrh Mitchell, and Doris MacKenzie. 2006. “A Systematic Review of Drug
court Effects on Recidivism.” Journal of Experimental Criminology 2, 4: 459-487.
Wyckoff, Laura, and Carol Clymer. 2005. Job Development Essentials: A Guide for Job Developers. New York: Public/ Private Ventures.
Zweig, Janine, Jennifer Yahner, and Cindy Redcross. Forthcoming. For Whom Does a Transitional
Jobs Program Work? Examining Recidivism Effects of the Center for Employment Opportunities Program on Former Prisoners at High, Medium, and Low Risk of Re-Offending (working
title). New York: MDRC.

244

EARLIER MDRC PUBLICATIONS ON TRANSITIONAL JOBS
Transitional Jobs
Background, Program Models, and Evaluation Evidence
2010. Dan Bloom.
Alternative Welfare-to-Work Strategies for the Hard-to-Employ
Testing Transitional Jobs and Pre-Employment Services in Philadelphia
2009. Dan Bloom, Sarah Rich, Cindy Redcross, Erin Jacobs, Jennifer Yahner, Nancy Pindus.
Transitional Jobs for Ex-Prisoners
Implementation, Two-Year Impacts, and Costs of the Center for Employment Opportunities
(CEO) Prisoner Reentry Program
2009. Cindy Redcross, Dan Bloom, Gilda Azurdia, Janine Zweig, and Nancy Pindus.
The Joyce Foundation’s Transitional Jobs Reentry Demonstration
Testing Strategies to Help Former Prisoners Find and Keep Jobs and Stay Out of Prison.
2009. Dan Bloom.

____________________________
NOTE: All MDRC publications are available for free download at www.mdrc.org.

245

About MDRC
MDRC is a nonprofit, nonpartisan social policy research organization dedicated to learning
what works to improve the well-being of low-income people. Through its research and the
active communication of its findings, MDRC seeks to enhance the effectiveness of social and
education policies and programs.
Founded in 1974 and located in New York City and Oakland, California, MDRC is best known
for mounting rigorous, large-scale, real-world tests of new and existing policies and programs.
Its projects are a mix of demonstrations (field tests of promising new program approaches) and
evaluations of ongoing government and community initiatives. MDRC’s staff bring an unusual
combination of research and organizational experience to their work, providing expertise on the
latest in qualitative and quantitative methods and on program design, development, implementation, and management. MDRC seeks to learn not just whether a program is effective but also
how and why the program’s effects occur. In addition, it tries to place each project’s findings in
the broader context of related research — in order to build knowledge about what works across
the social and education policy fields. MDRC’s findings, lessons, and best practices are proactively shared with a broad audience in the policy and practitioner community as well as with the
general public and the media.
Over the years, MDRC has brought its unique approach to an ever-growing range of policy
areas and target populations. Once known primarily for evaluations of state welfare-to-work
programs, today MDRC is also studying public school reforms, employment programs for exoffenders and people with disabilities, and programs to help low-income students succeed in
college. MDRC’s projects are organized into five areas:


Promoting Family Well-Being and Child Development



Improving Public Education



Promoting Successful Transitions to Adulthood



Supporting Low-Wage Workers and Communities



Overcoming Barriers to Employment

Working in almost every state, all of the nation’s largest cities, and Canada and the United
Kingdom, MDRC conducts its projects in partnership with national, state, and local governments, public school systems, community organizations, and numerous private philanthropies.