MENU TITLE: Work Release: Recidivism and Corrections Costs in Washington State.
Series: NIJ Research in Brief
Published: December 1996
30 pages
54,325 bytes

------------------------------

Figures, charts, forms, and tables are not included
in this ASCII plain-text file. To view this
document in its entirety, download the Adobe
Acrobat graphic file available from this Web site
or order a print copy from NCJRS at 800-851-3420.

------------------------------

U.S. Department of Justice
Office of Justice Programs
National Institute of Justice

National Institute of Justice
Research in Brief

Jeremy Travis, Director
December 1996

------------------------------

Issues and Findings

Discussed in this Brief: Two NIJ-sponsored
evaluations of Washington State's work release
program, conducted between 1991 and 1994. The first
study analyzed a cohort of 2,452 males released
from Washington prisons in 1990, nearly 40 percent
of whom spent a part of their sentences on work
release, to describe how work release operates and
how successfully inmates perform in the program.
The second compared the recidivism of 218
offenders; approximately half participated in work
release and half completed their sentences in
prison.

Key issues: In contrast to the national decline of
work release as a means of preparing imprisoned
offenders for reintegration into the community,
Washington has allocated more than one-third of its
community corrections budget to work release. The
releasees obtain daytime jobs, live in community
facilities, and contribute to their room and board.
Key issues explored in the studies were corrections
costs, the consequences of participants committing
infractions while on work release, and recidivism.

Key findings: The results of the evaluation were
mostly positive: 

o Nearly a quarter of all prisoners released in
Washington made a successful transition to the
community through work release.

o Less than 5 percent of the work releasees
committed new crimes while on work release, 99
percent of which were less serious property
offenses, such as forgery or theft.

However, with heightened supervision under strict
conditions, many work releasees incurred
infractions (most for rule violations and drug
possession), and a quarter returned to prison.
Thus, when one considers the reincarceration time
spent by work release "failures," the time under
correctional supervision was as long or longer for
work  release participants as for nonparticipants.
And there were no differences in corrections costs
between work releasees and inmates completing their
full terms in prison.

Other findings indicated:

o Middle-aged offenders and offenders convicted of
property crimes were most likely to participate in
work release. Hispanic offenders were less likely
to go to work release than white or black
offenders.

o Fifty-six percent of the 965 work releasees in
the cohort studied were termed "successful"; they
incurred no program infractions or arrests. Another
13.5 percent were "moderately successful"; their
infractions were not serious enough to return them
to prison. Almost 30 percent were "unsuccessful";
they returned to prison.

o Older offenders were more successful than younger
ones, and whites were more successful than either
Hispanics or blacks. Success was also associated
with having no prior criminal record.

Target audience: State and local officials and
legislators, judges, and researchers and
practitioners in community corrections. 

------------------------------

Work Release: Recidivism and Corrections Costs in
Washington State

by Susan Turner, Ph.D., and Joan Petersilia, Ph.D.

Each year U.S. prisons release more than 400,000
criminal offenders to their communities. Most of
those released will not remain crime free, and
national statistics show that within 3 years of
release, 40 percent will be returned to prison or
jail.(1) Experts debate the reasons for such high
recidivism rates, but all agree that the lack of
adequate job training and work opportunities is a
critical factor. Offenders often have few
marketable skills and training and, as a result,
have a difficult time securing legitimate
employment. With no legitimate income, many resort
to crime.

Since the early 1920s, corrections officials have
attempted to remedy the problem through prison work
release programs. Work release programs permit
selected prisoners nearing the end of their terms
to work in the community, returning to prison
facilities or community residential facilities in
nonworking hours. Such programs are designed to
prepare inmates to return to the community in a
relatively controlled environment, while they are
learning how to work productively. Work release
also allows inmates to earn income, reimburse the
State for part of their confinement costs, build up
savings for their eventual full release, and
acquire more positive living habits.

During the 1970s prison work release programs
expanded considerably, but they have declined in
recent years. Although 43 States have existing
statutes authorizing work release, only about
one-third of U.S. prisons report operating such
programs, and fewer than 3 percent of U.S. inmates
participate in them.(2)

Despite public disenchantment with work release
(see "Reasons for the Decline of Work Release"),
the State of Washington has maintained its
commitment to the program since initiating it in
1967. The Washington work release program permits
selected inmates to serve the final 4 to 6 months
of their prison sentences in privately run,
community residential facilities, where they are
required to be employed, submit to drug testing,
and abide by curfews and numerous program rules.

It is not that Washington has been immune to the
challenges faced by other States. Washington's
prison population has jumped 71 percent since 1980,
while the State's general population has grown just
13 percent. Citizens are frustrated with high
levels of crime and violence, as is evidenced by
their legislature's enactment of the Nation's first
"three strikes" law -- in part a reaction to a
felon's committing a rape while on parole.
Observers have wondered how the program can
continue to operate successfully when the emphasis
is on punitive policies and politics.

This Research in Brief presents findings from two
studies of Washington's work release program,
conducted under a National Institute of Justice
(NIJ) grant between 1991 and 1994. In sponsoring
the studies, NIJ hoped that results might add to
the growing body of research on intermediate
sanctions and provide guidance to other States that
were looking for effective and less costly
corrections programs.(3)

Study 1 analyzed a cohort of all males released
(n=2,452) from Washington prisons in 1990. Data
pertaining to their criminal and social background
and their participation and performance in work
release were analyzed to describe how work release
was implemented in Washington and how well inmates
performed in the program. Estimated costs of work
release versus prison were based on analysis not
only of the cost of the program but on costs
incurred if the work release participant was
returned to prison for violation of program rules
or for committing a crime.

Study 2 evaluated the impact of work release on
recidivism and on corrections costs by comparing a
sample of inmates who participated in work release
with a comparable sample of inmates who completed
their sentences in prison. Investigators collected
information about program implementation and
recidivism at 6 and 12 months following the
inmates' assignment to the study.

Overall, the studies found that the program
achieved its most important goal: preparing inmates
for final release and facilitating their adjustment
to the community. The program did not cost the
State more than it would have if the releasees had
remained in prison. The public safety risks were
nearly nonexistent because almost no work releasee
committed new crimes, and when they committed rule
violations they were quickly returned to prison.

While in the program these inmates maintained
employment, paid room and board, reconnected with
their communities, and most remained drug free.
Less than 5 percent committed new crimes while on
work release, and 99 percent of those crimes were
less serious property offenses, such as forgery and
petty theft. Moreover, offenders who participated
in work release were somewhat less likely to be
rearrested, but the results were not statistically
significant. One could reasonably conclude from
these results that work release in Washington is a
program "that works."

How Washington's work release program operates

Washington's work release program, created by
legislative action in 1967, gave the State
permission to allow inmates to serve sentences in
the community for the purpose of work training and
experience. In 1970 the first community-based
program was established in Seattle under a contract
agreement with Pioneer Cooperative, a private,
nonprofit corporation (now Pioneer Human Services -- 
see "A Partnership With Private Industry").

Staffing and costs. The work release program is the
responsibility of the Division of Community
Corrections within the State Department of
Corrections (DOC). Division staff establish
guidelines governing the selection of offenders for
placement in work release, work with contractors to
provide food and shelter in a halfway house
arrangement, and supervise the correctional
officers who are assigned to work release
facilities.

The actual operation of the work release facilities
is done on a contractual basis, with contracts
renegotiated every 2 years. DOC contracts with
providers for the buildings, plus the day-to-day
activities, including staff, meals, shelter, inmate
sign-in and sign-out procedures, urinalysis, and
job checks. Contracts are negotiated on a per bed,
per day basis, regardless of whether the bed is
actually occupied. In 1992, at the time of the
study, the average work release contract costs were
between $32 and $35 per day, per bed.

DOC contracts with 15 residential work release
facilities, which house more than 350 offenders
(mostly adult males) on any single day throughout
the State. The facilities can handle a range of
approximately 15 to more than 100 residents, but
most accommodate between 20 and 40 inmates. Small
numbers of beds are available to female offenders
and mentally ill or developmentally disabled
offenders.

In the 1991-1993 biennium, approximately $115
million of the $312-million DOC budget was
allocated to community corrections. About 38
percent (or $43 million) of the community
corrections budget was for work release.(4) These
funds pay primarily for DOC staff to administer the
statewide program and monitor participants at the
work release centers.

Admission criteria and process. Washington's work
release program is governed by the stipulations set
forth in the Sentencing Reform Act (SRA) of 1981,
which significantly altered the State's criminal
sentencing practices as well as the guidelines for
work release. The SRA stipulates that an inmate may
not enter work release sooner than 6 months prior
to discharge. The exact time depends on the length
of the offender's original sentence. Inmates
generally apply for work release within 12 to 17
months prior to their release date. The application
is screened by the correctional counselor, the unit
supervisor, and finally by the current DOC facility
superintendent.(5)

Inmates can apply for work release only if:

o They have a minimum security status.

o They have less than 2 years to serve on their
minimum term, including anticipated good time
credits.

o They have not been convicted of rape in the first
degree or, if so, are beyond the first 3 years of
confinement.

o They have not been convicted of murder in the
first degree or, if so, have the written approval
of the Secretary of Corrections.

Inmates who meet these initial screening criteria
may be subsequently denied work release by DOC if
(1) they have exhibited assaultive behavior while
in prison, (2) the offender's victim resides in the
area, (3) the offender has made threats to the
victim during incarceration, or (4) the offender
has two or more prior work release failures during
the current commitment.

Once DOC judges an offender eligible for work
release, the work release facilities' Community
Screening Board, consisting of work release staff
and local citizens, must agree to accept the inmate
for admission to their local work release center.
In most cases the Community Screening Board accepts
DOC's recommendations.

Daily operations. After being accepted for a
particular residential facility, inmates receive a
set of standard rules. These rules specify that
they are to abide by their work plan, remain in the
work release facility at all times except those
approved for work and other appointments, remain
alcohol and drug free, be employed or have
resources in order to meet financial needs, report
all earnings to DOC community corrections officers
(CCOs), and obey all Federal, State, and local
laws.

Offenders must obtain gainful employment or
training and must pay about $10 a day for room and
board. Participants are ultimately responsible for
finding work in the community, although CCOs and
staff provide referrals. Residents also pay support
to their families and court-ordered restitution.
The average length of stay in the work release
facility is about 4 months but varies according to
the length of the original court-imposed sentence.

The CCOs work at the work release facilities a
minimum of 8 hours a day. Their primary role is to
provide case management services, which includes
providing informal advice and conducting intake
interviews. For example, they review each client's
progress as well as discuss personal problems, such
as substance abuse. CCOs can require offenders to
participate in special treatment or may initiate
procedures for sending them back to prison.
Residents fill out "Offender Schedule Plans"
listing place of employment, immediate supervisor,
and rate of pay; CCOs check with employers to
verify these facts.

Study 1: A statewide analysis of work release

Study 1 was designed to answer the following
research questions:

o What are the characteristics of prisoners placed
on work release? How have these characteristics
differed from those of the general prison
population?

o How many offenders placed on work release have
successfully completed their programs? Why have
some not succeeded?

o What characteristics distinguish those who have
been successful from those who have been
unsuccessful on work release?

o What are the costs associated with a prison term
that includes a work release sentence compared to
those of a prison sentence without work release?

The data for study 1 were supplied by DOC research
staff. They identified all male inmates released
from prison (to the streets) during calendar year
1990.(6) For each of the 2,452 offenders in this
cohort, background and criminal record information
was obtained from DOC's computerized Offender-Based
Tracking System (OBTS). OBTS lists the inmate's
age, race, prior criminal record, most serious
conviction, sentence length imposed, and date
admitted and discharged from DOC. It also records
transfers to jail, prerelease, and work release and
work release application and performance
information. To this information was added more
detailed data from DOC files on the exact reason
for return (e.g., new crime, drug possession) for
work releasees who had been returned to custody.

Study 1 findings

The data reveal that almost half (48.8 percent) of
the inmates in the 1990 release cohort applied for
work release during their prison term, and 39.4
percent were placed in work release facilities at
some point during their sentence.

An average of 7.1 months elapsed between the time
an inmate was admitted to prison and when he
applied for work release. Once the application was
submitted, it took just over 3 weeks before a
decision was made to accept or reject him for the
program. After an additional 56 days, the inmate
actually entered a work release facility; this time
was used to prepare the inmate for work release and
to locate a suitable and available residential
placement. These inmates served an average sentence
of 15 months. With an average application process
of 10 months, the inmates spent approximately 5
months in work release.

Participant characteristics. One would expect work
release participants to differ from the general
prison population since the law and DOC policies
indicate that only certain offenses and offenders
(generally the less serious) are eligible. After
meeting these legal standards, inmates are free to
decide if they want to apply for work release.
Incentives to participate include the opportunity
to live in the community, be closer to family and
friends, and obtain some practical work experience.
But some inmates do not plan to be released to a
location where a work facility exists, and some
inmates judge the requirements and closer
supervision of work release too bothersome. They
choose instead to serve out their full terms in
prison.

Several background characteristics distinguished
inmates who did not choose work release from those
who did:

o Younger and older inmates were less likely than
middle-aged inmates to go to work release.

o Hispanics were less likely to go to work release
than either whites or blacks.

o Inmates with no prior criminal record were less
likely to go to work release.

o Inmates convicted of property offenses (burglary,
theft, and forgery) were the most likely to go to
work release.

At least one of these differences may reflect the
screening procedures for work releases. For
example, policies deny work release to offenders
convicted of certain violent offenses -- hence their
lower representation in the work release sample.
There could be another explanation for the finding
that inmates with no prior criminal records were
less likely to participate in work release. These
lower risk inmates had other program options for
decreasing their prison length of stay (such as a
"work ethic" camp for nonviolent offenders).

Although Hispanics were disproportionately
underrepresented in the work release cohort, this
did not appear to result from their having fewer of
the characteristics that would have made them
eligible. In fact, Hispanics were overrepresented
as drug offenders -- which indicates their
representation among work releasees should have
been much higher than observed. DOC staff suggested
one reason for fewer-than-expected Hispanics is
that some may have had Immigration and
Naturalization Service detainers, making them
ineligible for work release.

Who was successful? Each inmate was classified
according to performance:

o Successful inmates were those who completed work
release without any type of rule infraction or new
crimes having been noted in their official
corrections record and who returned directly into
the community from the work release facility.

o Moderately successful inmates were those who had
committed an infraction but one not judged serious
enough to remove them permanently from work
release.

o Unsuccessful inmates committed some type of
infraction, either a rule violation or new criminal
conduct, and were returned to prison to serve out
the remainder of their term.

Of the 965 work releasees in the cohort, 544 (or 56
percent of participants) were judged successful in
work release. If one considers that nearly 40
percent of the original cohort were placed on work
release to begin with, and that about two-thirds
succeeded, it is clear that almost one out of four
inmates in Washington made a successful transition
to the community through work release.

An additional 131 inmates (or 13.5 percent of
participants) were moderately successful, and 290
inmates (or 30 percent of participants) were judged
unsuccessful in work release and were returned to
prison to complete their terms.

The most frequent reasons for return to prison from
work release were failure to abide by curfews or
absconding from the program, drug possession, and
other program rule infractions. New crimes or law
violations accounted for a very small percentage
(3.6 percent) of those returned. In fact, in the
entire cohort of 965 work releasees studied,
records indicate that no offender committed a
violent felony while in the program: the few crimes
committed were thefts and forgeries. These findings
suggest that Washington's work release program
poses very little risk to the surrounding
community.

Predicting success. Variables indicating race,
education level, occupation before imprisonment,
marital status, employment status at arrest,
previous work stability, conviction offense type,
prior criminal record, substance abuse dependency,
and length of current sentence were each
cross-tabulated with the three success-level
variables, revealing that:

o Older offenders were more successful at work
release than younger offenders.

o Whites were more successful than Hispanics and
blacks. More than 40 percent of Hispanic and black
work releasees were returned to prison compared
with 25 percent of white offenders.

o Offenders with no prior criminal record were more
successful than those with a prior record. Almost
two-thirds of inmates with no prior record were
successful compared with fewer than half the
offenders with a prior conviction.

o Offenders convicted of person crimes (e.g.,
robbery and assault) were more successful than
offenders convicted of property or drug crimes.(7)

Comparing work release costs with prison costs. A
recent survey of prison administrators reported
that, while they recognized the rehabilitative
potential of work release, they were most
interested in using work release to reduce prison
crowding and associated costs.(8) Community-based
facilities are much less expensive to operate than
prisons, and so allowing inmates to serve the last
several months of their prison sentence in the
community can reduce corrections costs. According
to national estimates, for example, work release
costs about $34 a day, and prison costs $54 a day
(operational costs only).(9) In addition, inmates
who work can be required to help pay for their room
and board, support their families, and pay taxes.

Unfortunately, the cost savings accrued through
work release and other community-based programs
have not been as substantial as proponents had
hoped. For example, Minnesota recently implemented
an Intensive Supervision Program designed to
release prisoners early from prison on the
condition that they participate in an enhanced
community supervision program. Fewer offenders than
expected actually participated in early release,
reducing the potential for large cost savings.
Similar results have been seen in New Jersey,
Florida, Arizona, Texas, and Oregon.(10)

It is therefore clear that community-based
sanctions are not necessarily less expensive than
prison. It all depends on how offenders behave in
the program and how program administrators choose
to punish infractions (particularly drug use). If
infractions and rule violations are punished with
incarceration, initial cost savings are reduced or
eliminated as a result of the additional dollars
required to incarcerate program failures. In
effect, close surveillance may end up generating
many program failures who are eventually returned
to prison. In that case, the State incurs the cost
of their work release program in addition to
eventual reuse of a prison cell. The
community-based sanction may actually end up
costing more than a single term of incarceration
uninterrupted by work release. Previous work
release evaluations have not provided a complete
accounting of total costs, which at a minimum must
include the cost of rehousing unsuccessful work
release offenders in jail or prison.

For each of the 2,452 inmates in the work release
cohort, files on their movements since coming to
prison were available. The data made it possible to
record how much time each inmate spent in work
release and nonwork release institutions as well as
in local jails.

DOC then provided statewide estimates of the daily
cost of different sanctions; by combining this
information, researchers were able to calculate the
average cost of a prison sentence that included
work release and one that did not.

When all incarceration time served is taken into
account, inmates who participated in work release
served, on average, about 4 months less time in a
prison or prerelease facility than inmates who did
not participate. However, for every 3 days in work
release, offenders spend about 1 day back in an
institution -- either prison or a prerelease center.

Inmates who participated in work release cost the
State about $4,000 less than inmates who did not.
However, the cost comparison does not account for
crime and background differences between work
releasees and nonwork releasees, which have already
been shown to be considerable. It is known that
nonwork releasees were, on average, convicted of
more serious crimes. As such, one would expect that
they would serve longer prison terms and cost the
State more as a result. To make credible cost
comparisons among sanctions, one must compare
offenders who are similar in background
characteristics and conviction crimes. Study 2 was
designed to do exactly that.

Study 2: The impact of work release on recidivism
and on corrections costs

This evaluation of the effects of work release was
designed as an experiment in which eligible
offenders were to be randomly assigned to
participate or not to participate in work release.
In a randomized experiment, one has more confidence
that any observed differences result from
participating in work release rather than from
preexisting background differences.

The experiment was conducted in Seattle, which is
Washington's largest city and home to more than 50
percent of the work release offenders. The study
had the full cooperation of the Washington DOC, the
DOC Community Corrections Division, and the various
managers of the six individual work release
facilities serving males in Seattle.

Identifying the sample. The study required a fairly
large sample size in order to detect the relatively
small outcome differences that were expected
between experimental and control offenders. Based
on prior evaluations, it was estimated that the
work release effects (e.g., recidivism reduction)
would not be large, with perhaps 10- to 20-percent
differences between experimental and controls.
Procedures were devised that would yield expected
sample sizes of several hundred offenders during
the time frame of the study.

Offenders who were deemed eligible for work release
by DOC were randomly assigned to either the
experimental group (who progressed toward work
release using normal procedures) or the control
group (who were removed from the work release
eligibility list and remained in prison to complete
their full prison terms).

To avoid seriously disrupting the flow of inmates
into Seattle work release facilities, investigators
assigned 1 out of every 10 eligible inmates to the
control group and 1 to the experimental group. The
remaining 8 eligible offenders did not participate
in the study but instead maintained their status on
the waiting list and were handled according to
normal DOC operating procedures. Based on analysis
of available DOC data, these procedures were
estimated to yield approximately 200 study
offenders a year.

However, the flow of offenders entering the DOC
pool of work release "eligibles" was too slow to
permit reaching an adequate sample size during the
study's assignment period. A year after the random
assignment procedures were put in place, only 125
offenders had been assigned to the study. It is
unclear why the inmate flow did not meet
expectations, but DOC officials suggest it resulted
from fewer applications to the program (not from
rejecting more applicants).

To increase the study's sample size, researchers
supplemented the randomized assignment with
offenders who during the study period had been part
of the 8 out of every 10 offenders who were not
included in the initial randomized assignment
procedures. Forty-eight offenders were randomly
selected from this initial pool of work release
eligibles and placed in the experimental study
group. In addition, DOC identified 45 additional
control offenders who for reasons unrelated to
eligibility criteria had not actually participated
in work release prior to their release from prison,
primarily because they had "maxed out" on their
sentences before all the administrative procedures
had been completed or work release beds had become
available. These offenders were used to supplement
the control group. Thus, the final sample totaled
218 offenders, of whom 125 had been randomly
assigned to the study and an additional 93 had been
chosen as a matched comparison group.(11)

Data collection. For each of the 218 study
participants, a number of data collection forms
were completed by the researchers' onsite staff.
The background review captured information about
the inmate's demographics (e.g., age, race,
education, marital status), prior employment
history, drug use information, current offense
information, and prior criminal history.

At 6 months and again at 12 months after
assignment, a series of other forms were completed
that recorded such information as time spent in
different institutions and work release centers,
contacts made and services received, and the number
of days on postdischarge status. Finally,
recidivism information was obtained for each
offender based on State-level criminal history "rap
sheets." The date, nature of arrest, disposition,
and sentence were recorded for each offense
occurring during the 1-year followup period after
study assignment.

Characteristics of the Seattle sample

The Seattle sample closely resembled the statewide
sample, which was to be expected since about
one-third of Washington's DOC population comes from
the Seattle area. The study work release offenders
averaged about 30 years of age. About half of the
offenders in the sample were white, and 43 percent
were black. More than half of the sample were
convicted of drug offenses, and 84 percent were
classified as dependent on alcohol, with almost
three-quarters dependent on cocaine. About
two-thirds of the offenders in the sample had never
worked or had worked only occasionally (6.2 percent
and almost 60 percent, respectively).

One can see that the characteristics of work
release participants were not significantly
different from those of nonwork release
participants, except in three instances: number of
prior arrests, occupational history, and type of
current offense. This suggests that the
randomization and matching strategy used was
successful in that it was desired that the two
groups be similar in background characteristics so
that any observed differences in recidivism could
be attributed to participation in work release.(12)

Work release program activities

The investigators sought to learn what actually
transpired when offenders were placed in work
release. What kinds of jobs did they get, and what
types of monitoring activities were imposed? How
well did inmates perform, and how long did they
stay in the program? To answer those questions, the
work release offenders were "followed" into their
individual site placements, and information was
coded about the kinds of jobs they obtained, how
much money they earned, the number of times they
were seen by CCOs, how often they were tested for
drugs, and the kinds of rehabilitation programs
they participated in.

As expected, most of the program's focus was on
obtaining a job. In fact, if offenders did not have
a job within the first few weeks, they might be
returned to the institution. Generally, offenders
had one or two jobs during their 3- to 4-month work
release stay. Many more interviews were attempted
with employers than were actually completed. The
analysis of offenders' first jobs showed that most
were either in the restaurant or construction
industries, with a median pay of $8 an hour. More
than half of the offenders were still working at
their first jobs when they left work release
(discharged from DOC or placed on postrelease
supervision) or when the review of their files was
completed.

Work releasees were tested for drugs at the work
release facility an average of once a month. The
tests revealed very low percentages of drug use.
Only 8 percent of work release participants tested
positive for drugs, usually for cocaine. On
average, work releasees participated in outpatient
drug and alcohol counseling sessions (usually
Alcoholics Anonymous and Narcotics Anonymous
meetings) about once every 2 months. In addition,
they had face-to-face meetings with their CCO about
once a week.

Recidivism outcomes during 12-month followup

It is generally believed that recidivism rates are
the key factor in determining the effectiveness of
work release. This study included a 1-year followup
period, which began at the point of random
assignment. Work releasees were not immediately
placed in the community following random
assignment, since several weeks often passed
between approval for the program and actual
transfer to the work release facility. Thus, the
12-month followup included an average of 2 months
initial time in an institution, followed by an
average of 10 months in the community for work
release study offenders. During the followup
period, control offenders were "on the street" an
average of just under 7 months after being released
from prison.

Investigators obtained each offender's State rap
sheet and recorded all arrests (felony and
misdemeanor), convictions, and incarcerations
occurring during the followup period. Information
on infractions during work release and prison was
obtained from the DOC OBTS computer and work
release folders. Events that occurred while the
offender was in prison or on work release were
separated from events that occurred afterward
(including postdischarge).

The results show significant differences.
Infractions (mostly rule violations and drug
possession and use) were recorded for 58 percent of
work releasees but for only 4.7 percent of nonwork
releasees. As a result, about a quarter of all work
releasees were returned to prison, compared to 1
percent of nonwork releasees. However, when looking
at infractions while offenders were in an
institution, one finds that nonwork releasees
incurred significantly more infractions during
their time in prison. When one adds together
infractions during work release and while in
prison, work release offenders were twice as likely
to incur a violation when compared to nonwork
releasees (67.3 percent versus 33.6 percent).

In terms of new crimes, although work releasees
were less likely to be arrested during the followup
period, the difference was not statistically
significant. As in study 1, a very small percentage
of work releasees were arrested for new crimes
while on work release (less than 3 percent). By the
end of the 1-year followup period, which included
an average of about 3 months after discharge from
DOC, 30 percent of nonwork releasees (controls) had
been arrested, compared to 22 percent of work
releasees (experimentals). As a result, about 4
percent of both groups were returned to prison for
a new crime during the year.

Overall, if one combines all the returns to prison
(for either a rule infraction or new crime), one
finds that 29.5 percent of work releasees were
returned to prison during the 1-year followup,
compared to 5.7 percent for nonwork release
participants.

The returns to prison were mostly for short stays,
however. By the end of the 1-year followup period,
71.4 percent of control offenders and 52.7 percent
of experimental offenders had been discharged. Nine
percent of work release participants were
institutionalized, contrasted with 2.9 percent of
controls, a marginally significant difference.
These differences suggest that the length of time
under correctional supervision may actually have
been longer for those participating in work
release.

Costs of work release versus prison

Is work release less expensive than serving a
complete term in prison? To answer this question,
one must determine the relevant period of time for
which costs are to be estimated. Since about 26
percent of work releasees were returned to prison
as a result of a work release infraction, the costs
of rehousing them in prison needed to be considered
in the overall cost calculation.

If one considers only costs before the inmates
either leave prison or work release for the first
time, one ignores "reprocessing" for offenders who
are returned to prison from work release facilities
or the community. However, if one wants to include
costs following first release, one must make some
decision about how far in the future to extend the
window for analysis. Not all the study offenders
were discharged during the study time period; thus,
one cannot estimate costs for the entire sentence
served in the institution plus time in the
community before discharge for all offenders. The
analysis reflects the costs associated with
completed sentences for 70 per-cent of controls and
50 percent of experimentals.

Washington corrections officials furnished the
daily cost of each of the correctional programs,
making it possible to compute the total costs of
supervising each inmate during the initial sentence
and the followup period. By averaging these cost
estimates across inmates in the control and
experimental groups, one can compare the costs of a
prison term that includes work release to one that
does not.

The analysis shows basically no difference in costs
between work releasees and inmates completing their
full terms in prison. If one considers the costs
associated with work release from the time an
inmate was admitted to prison until his discharge,
the estimated cost would average $25,883 per
inmate. This is in contrast to the estimated
$25,494 it would have cost per inmate, on average,
to serve out his time in prison.

The two studies thus produced similar findings
regarding the costs of work release. Other findings
related to success, failure, and recidivism lead to
several conclusions.

Who is successful on work release?

Although work release might not reduce the overall
recidivism rates of participants, there may be
subgroups for whom the program could produce more
successful outcomes. Similar to the analysis in
study 1, an attempt was made to identify factors
that were related to three recidivism outcomes: any
infraction during work release; any arrest during
the 1-year followup; and any return to prison (as a
result of infraction or arrest) during the 1-year
followup. A number of background characteristics of
offenders were considered, such as age, race,
education, prior employment, drug involvement, and
current offense.

It was not possible to find characteristics of
offenders, or a profile of an offender, who
appeared to do better on work release than in
prison. However, a few characteristics were related
to success, regardless of condition. For example,
offenders with a prior history of cocaine or crack
dependence were more likely to have committed an
infraction than others; those whose most serious
current offense was for a theft charge were more
likely to be rearrested; and white offenders were
less likely than others to commit an infraction.

Summary and conclusions

Washington's work release program has been
successful on several fronts: nearly a quarter of
all prisoners released in Washington under current
statutes made a successful transition to the
community through work release. While in the
program, these inmates maintained employment,
reconnected with their local communities, paid for
their room and board, and most remained drug free.
Given that in most States fewer than 5 percent of
prisoners even participate in work release, this is
quite an accomplishment.

However, the work release program did not reduce
offender recidivism rates or corrections costs.
Critics of community corrections often argue that
such programs should deliver all of the above
services while showing a reduction in recidivism
and costs. Such expectations are unrealistic.
Prison programs should not assume the goals and
functions of other social institutions such as
schools and welfare agencies.(13)

Realistic measures of corrections programs'
effectiveness should account for daily activities
and the constraints under which the programs
operate. 

Realism, however, does not mean easy to achieve.
Most participants in Washington's work release
program had lengthy criminal histories, serious
substance abuse problems, and possessed limited
education and job skills. Yet, when supervised in
this work release program, they found jobs, paid
rent, and refrained from crime.

Although most corrections evaluations adopt
recidivism as their primary outcome measure, few
corrections officials believe that what they do
chiefly determines recidivism rates. As John J.
DiIulio, Jr., recently wrote: "Most justice
practitioners understand that they can rarely do
for their clients what parents, teachers, friends,
neighbors, clergy...or economic opportunities may
have failed to do."(14) Adopting more realistic
outcome measures may make it more possible to
bridge the wide gap between public expectations for
the justice system and what most practitioners
recognize as the system's actual capability to
control crime. By documenting what corrections
programs can accomplish, we can move toward
integrating programs like work release into a more
balanced corrections strategy. Such a strategy
would successfully return low-risk inmates to the
community, thereby making room to incarcerate the
truly violent offenders.

------------------------------

Notes

1. Bureau of Justice Statistics, Correctional
Populations in the United States, 1992, Washington
D.C.: U.S. Department of Justice, Bureau of Justice
Statistics, 1995, Table 5.12a.

2. Bureau of Justice Statistics, Census of State
and Federal Correctional Facilities, Washington,
D.C.: U.S. Department of Justice, Bureau of Justice
Statistics, 1990:1.

3. The few previous studies of work release had
shown some positive results -- mostly on employment
rather than recidivism. These studies relied
primarily on quasi-experimental methods that failed
to control adequately for preexisting differences
between study groups. Thus, one could not determine
whether outcome differences were due to the effect
of the intervention (work release) or to
characteristics on which work release and
comparison groups differed.

4. The work release budget includes the
institution's phase referred to as "prerelease," in
which inmates prepare for their placement in the
community work release facilities, as well as
actual placement in the community work release
site.

5. The offender can also change his mind about
participating in work release during the
application process. Of those who were denied work
release in 1989 by DOC, 30 percent were offenders
being held on detainers, 17 percent were offenders
with disciplinary problems, 9 percent had escape
infractions, and 17 percent refused to participate
in work release.

6. Only male inmates who were sentenced and
released under the Sentencing Reform Act (SRA) of
1981 were included in the sample. DOC estimates
that about 80 percent of offenders released in 1990
were SRA offenders and 90 percent were male.

7. The recidivism predictors are nearly identical
to those identified in a national sample of
released prisoners (Beck, Allen, and Bernard
Shipley, Recidivism of Prisoners Released in 1983,
Washington, D.C.: U.S. Department of Justice,
Bureau of Justice Statistics, 1989).

8. Simon, Senator Paul, "Wardens Call for Smarter
Sentencing, Alternatives to Incarceration, and
Prevention Programs," U.S. Senate Press Release,
Washington D.C., December 21, 1994.

9. Maguire, Kathleen, and Ann L. Pastore, eds.,
Sourcebook of Criminal Justice Statistics 1993,
Washington, D.C.: U.S. Department of Justice,
Bureau of Justice Statistics, 1994.

10. Clear, Todd, and Anthony Braga, "Community
Corrections," in Crime, eds. James Q. Wilson and
Joan Petersilia, San Francisco, California: ICS
Press, 1995.

11. Comparison of random versus nonrandom cases on
a number of background characteristics, including
age, race, current offense, length of prison term,
multiple measures of prior criminal record, and
legal financial obligations revealed no
statistically significant differences between the
two groups on any measure except the number of
prior arrests and the number of prior parole
revocations. Of these characteristics, nonrandom
offenders had fewer prior arrests and fewer prior
parole revocations than randomly assigned
offenders.

With any random-assignment protocol, the
possibility exists that the actual placement may be
inconsistent with the designated assignment. In the
current study, several factors affected actual
placement after assignment. Although experimental
offenders were to be placed in work release
facilities soon after the assignment call was made
to the researchers, in some instances offenders
reached the end of their terms before placement.
The potential for changing control offender
assignments was more often related to offender
dissatisfaction with remaining in prison rather
than with being placed in work release. Although
potential challenges to control assignments were
anticipated by RAND Corporation and DOC staff and
measures were taken to deal with them (i.e.,
reminding offenders that work release is a
privilege and not a right), a few control offenders
were placed in work release when it appeared their
needs outweighed the study assignment. When one
examines the actual placements versus those
assigned for the entire study sample, 91 percent of
controls remained in prison; 86 members of the
control group were placed in work release.

The study's main analysis of outcomes considers
cases as assigned, not as actually placed (thus one
sees infractions during work release for a smaller
percentage of control offenders). Although this
procedure results in diluted estimates of treatment
effects, it is the recommended strategy to ensure
unbiased estimates of program impact. See Garner,
Joel H., and Christy A. Visher, "Policy Experiments
Come of Age," NIJ Reports, Washington D.C.: U.S.
Department of Justice, National Institute of
Justice, September/October 1988.

12. Of the 25 characteristics on which experimental
and control offenders were compared, one would
expect at least one comparison to be significant by
chance. Nonetheless, the three significant
differences found indicated that work releasees
might have been slightly less serious offenders
than their control group counterparts. To the
extent these characteristics are related to
recidivism, one would tend to find more positive
outcomes for the work releasees. However, the study
found that work releasees were actually more likely
to commit infractions. To address the concern of
differences in background characteristics on the
"any arrest" outcome, a multiple logistic
regression analysis was conducted in which "any
arrest" was regressed on controls versus
experimentals, age, race, high school graduate,
employed prior to prison, dependence on cocaine or
crack, whether postrelease supervision had been
ordered, whether the inmate was referred to the
study from a prerelease center or an institution,
current offense type, and number of prior arrests.
The results -- work releasees were less likely,
although not significantly so, to be arrested
during the followup period. In other words, when
background factors are controlled, conclusions are
not changed.

13. Charles Logan, a corrections expert, has
written on such matters for the Department of
Justice and recommended that realistic goals for
prison programs should be fairly narrow,
achievable, and measurable within the policies and
programs operated by the corrections department. He
says that prison programs should be held
accountable for keeping prisoners safe, in line
(not committing crimes), healthy, and busy, and
they should try to do all of this as efficiently as
possible and without causing undue suffering. If
those outcome measures are adopted (they were
recently endorsed by the General Accounting
Office), Washington's work release program is
certainly successful. See Logan, Charles H.,
"Criminal Justice Performance Measures for
Prisons," Performance Measures for the Criminal
Justice System, Washington D.C.: U.S. Department of
Justice, Bureau of Justice Statistics, 1993.

14. DiIulio, John J., Jr., "Rethinking the Criminal
Justice System: Toward a New Paradigm," in
Performance Measures for the Criminal Justice
System, Washington D.C.: U.S. Department of
Justice, Bureau of Justice Statistics, 1993:1-2.

------------------------------

Susan Turner, Ph.D., is a senior researcher with
the Criminal Justice Program, RAND Corporation, and
Joan Petersilia, Ph.D., is professor of
criminology, Social Ecology Department, University
of California, Irvine. This project was conducted
in collaboration with the Washington   Department
of Corrections and Pioneer Human Services. The
authors wish to thank the following individuals who
assisted with the project: Chase Riveland,
Secretary, Washington Department of Corrections;
Dave Savage,  Director, Division of Community
Corrections; Peggy Smith, Ph.D., Planning and
Research Manager; and Gary Mulhair, President,
Pioneer Human Services. This study was conducted
under NIJ award 90-DD-CX-0056.

------------------------------

Findings and conclusions of the research reported
here are those of the authors and do not
necessarily reflect the official position or
policies of the U.S. Department of Justice. 

------------------------------

The National Institute of Justice is a component of
the Office of Justice Programs, which also includes
the Bureau of Justice Assistance, Bureau of Justice
Statistics, Office of Juvenile Justice and
Delinquency Prevention, and the Office for Victims
of Crime.

NCJ 163706

------------------------------

Reasons for the Decline of Work Release

One reason work release programming has declined
pertains to funding. Many work release programs
begun in the 1970s were paid for by the Federal
Government using funds from the Law Enforcement
Assistance Administration. When Federal funding
ceased, many programs were discontinued. And as the
rehabilitation ideal -- of which work release was
very much a part -- started to fade, the public
embraced imprisonment as the only sure way to
forestall crime. Programs that focused on
rehabilitation, job training, and transitional
services seemed hopelessly out of touch with the
public.

Moreover, a few highly publicized and sensational
failures convinced the public that early-release
programs, such as work release or furlough,
threatened public safety. The most extreme example
was Willie Horton, the Massachusetts inmate who
absconded and committed serious and violent crimes
while on furlough. (Horton was actually on work
furlough, not work release, but the two terms are
often used interchangeably.) The case became an
issue in the 1988 presidential campaign, and the
negative publicity helped further erode community
support for work release programs.

------------------------------

A Partnership With Private Industry

Washington's work release program has benefited
enormously from a particularly close working
partnership with private industry, particularly
Pioneer Human Services (PHS). PHS has been
contracting with DOC to operate work release
facilities since its inception nearly 30 years ago.
It now operates 4 such facilities, housing about
1,200 work releasees annually (or about one-third
of all Washington's work releasees). Over the years
PHS has grown to a full-service organization,
providing job training at a manufacturing facility
it runs, prerelease and postrelease employment in a
food service business it founded, housing for
offenders with special needs, and electronic
monitoring of State and Federal offenders. PHS is
highly regarded nationally and in the State and was
recognized by former President Bush in 1992 as one
of the Nation's "Points of Light."

------------------------------

Quick Access to NIJ Publication News
For news about NIJ's most recent publications,
including solicitations for grant applications,
subscribe to JUSTINFO, the bimonthly newsletter
sent to you via e-mail. Here's how: 

o Send an e-mail to listproc@ncjrs.orgo Leave the
subject line blank

o Type subscribe justinfo your name  (e.g.,
subscribe justinfo Jane Doe) in the body of the
message

Or check out the "What's New" section at the
Justice Information Center homepage:
http://www.ncjrs.org