Content uploaded by Lisa R Muftic
Author content
All content in this area was uploaded by Lisa R Muftic on May 15, 2015
Content may be subject to copyright.
http://tpj.sagepub.com
The Prison Journal
DOI: 10.1177/0032885507303741
2007; 87; 171 The Prison Journal
Jeffrey A. Bouffard and Lisa R. Muftic
Traditional Fines for Low-Level Offenders
The Effectiveness of Community Service Sentences Compared to
http://tpj.sagepub.com/cgi/content/abstract/87/2/171
The online version of this article can be found at:
Published by:
http://www.sagepublications.com
On behalf of:
Pennsylvania Prison Society
can be found at:The Prison Journal Additional services and information for
http://tpj.sagepub.com/cgi/alerts Email Alerts:
http://tpj.sagepub.com/subscriptions Subscriptions:
http://www.sagepub.com/journalsReprints.navReprints:
http://www.sagepub.com/journalsPermissions.navPermissions:
http://tpj.sagepub.com/cgi/content/refs/87/2/171
SAGE Journals Online and HighWire Press platforms):
(this article cites 7 articles hosted on the Citations
distribution.
© 2007 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at UNIV NORTH TEXAS LIBRARY on August 8, 2007 http://tpj.sagepub.comDownloaded from
The Effectiveness of
Community Service
Sentences Compared to
Traditional Fines for
Low-Level Offenders
Jeffrey A. Bouffard
Washington State University
Lisa R. Muftic
′
University of North Texas
Relatively little research in the United States has examined the effectiveness
of community service (CS) sentences among adult offenders, despite use of
this alternative sanction for nearly 40 years. What little research exists, pri-
marily from Europe, suggests that CS may not yield significant reductions in
recidivism when compared to incarceration; however, much of this research
suffers from important methodological limitations. This study employs a
more rigorous evaluation design and a more appropriate comparison sample of
offenders sentenced to pay traditional monetary fines. Results reveal that those
who participate in CS sentences are less likely to experience post-program
recidivism, controlling for several initial group differences. Limitations of
the design and suggestions for future research are also discussed.
Keywords: community service; alternative sanctions; recidivism
C
ommunity service (CS) emerged as an alternative sanction in the United
States in the 1960s (McDonald, 1986) and was initially designed to
meet the goal of providing an alternative to imprisonment or fines for less
serious types of offenders, such as those convicted of traffic violations, petty
theft, and other nonviolent offenses. By contrast, in Europe, CS sentences are
more likely to be used as an alternative to periods of incarceration
(Muliluvuori, 2001; Tonry, 1998). In the United States, CS can also be used
The Prison Journal
Volume 87 Number 2
June 2007 171-194
© 2007 Sage Publications
10.1177/0032885507303741
http://tpj.sagepub.com
hosted at
http://online.sagepub.com
171
distribution.
© 2007 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at UNIV NORTH TEXAS LIBRARY on August 8, 2007 http://tpj.sagepub.comDownloaded from
in combination with other sanctions, for instance as an add-on to traditional
probation or in addition to the imposition of fines. When compared to a tra-
ditional criminal fine, the use of CS sentences has the added benefit of pro-
viding direct service to the community through the provision of unpaid labor.
Although the process of sentencing less serious offenders to CS work
has been employed in the U.S. criminal justice system for nearly four
decades (McDonald, 1986), relatively little research has been conducted on
the operation (including frequency, prevalence, and severity of CS sanc-
tions) or effectiveness of such programs. It is estimated that in the late
1990s there were 547 CS and restitution programs being run nationwide
(Development Services Group, 2006). However, there is a paucity of
research that examines these programs. Delens-Ravier (2003) hypothesizes
that the primary reason for the lack of evaluation research on CS programs
is “because the objectives of CS are so varied and diverse” (p. 152), making
the exact criteria against which CS sentencing programs are to be measured
unclear. In fact, unlike programs in Europe and Asia, which tend to employ
CS as one component of a restorative justice system, community service
programs (CSPs) in the United States tend to derive from wide-ranging
goals, including retribution, rehabilitation, and skill building and function
as an alternative sanction (Delens-Ravier, 2003; Harris & Wing Lo, 2002;
Wing Lo & Harris, 2004). In addition, the lack of consistent theoretical
grounding for the goals of CS compounds the inability to determine what
the programs’ outcomes are or should be. As such, there are few studies that
evaluate the outcomes of CSPs, in terms either of the likelihood of suc-
cessfully completing a CS sentence or of post-program recidivism, and
those that do exist offer little guidance on the effectiveness of CS sentences
in the United States, especially when compared to other community-based
(nonincarceration) sanctions.
Although CS has been employed as a sanction for relatively less serious
offenders in this country for nearly 40 years, it is still a comparatively new
and under-researched form of correctional intervention. Even considering
the relatively recent implementation of this sanction type, the number of
studies examining its impact on recidivism, especially in the United States,
is even smaller than one might expect given its use during the past several
decades. Tonry (1998), noting the lack of sufficient research on the effec-
tiveness of CS, suggests that it may also be the most underused alternative
sanction in the United States. He notes that although CS is often simply
used as a condition of probation in the United States, its use could be
172 The Prison Journal
distribution.
© 2007 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at UNIV NORTH TEXAS LIBRARY on August 8, 2007 http://tpj.sagepub.comDownloaded from
expanded as a stand-alone sanction for both minor and some types of mod-
erate offenders, as it is in Europe. Tonry also anticipates that the public
might even appreciate CS as a “tough” sanction that can be readily scaled
to the severity of the crime, in addition to the fact that it is relatively inex-
pensive to administer and produces a valuable outcome in the form of
unpaid labor provided to the wider community.
Finally, Tonry (1998) points out that the existing European and U.S.
research (all of which compares CS-sentenced offenders to those sentenced
to some period of incarceration) suggests that CS does not reduce recidi-
vism when used in lieu of short-term prison sentences. This finding may
also suggest that some of these incarceration-bound offenders could be
safely managed in the community using a CS sentence (with no increase in
the likelihood of recidivism), thus reducing correctional crowding while
still protecting public safety. On the other hand, in the United States, incar-
ceration-bound offenders may not be an appropriate comparison group for
those receiving CS sentences because there is a greater tendency to use
incarceration as a criminal sanction, even for less serious offenses, than is
the case in many European countries (Langan & Farrington, 1998; Lynch,
2002; Tonry, 1998).
In recognition of this issue, the current study examines the recidivism rate
of offenders receiving CS sentences (n = 200) relative to that of a more appro-
priate comparison sample of offenders sentenced using a traditional commu-
nity-based sanction, criminal fines (n = 222). Specifically, this study expands
on previous research by asking three distinct questions. First, do CS offenders
recidivate at lower rates (post-sentencing) than do fined offenders? Second,
what factors are related to offending for offenders completing a CS sentence?
Third, what impact does sentence type have on post-program recidivism?
Previous Research on CS Outcomes
As noted previously, most if not all of the published outcome studies of
CS sentences examine their effectiveness compared to incarceration sen-
tences in European countries. A review of the published literature on the
effectiveness of CS sentences on recidivism revealed several such European
studies (Killisa, Aebi, & Ribeaud, 2000; Muliluvuori, 2001; Spaans, 1998).
Only one published study of a CS sentencing program in the United States
was located (McDonald, 1986), which also examined CS effectiveness
relative to incarceration.
Bouffard, Muftic´ / Community Service Sentences 173
distribution.
© 2007 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at UNIV NORTH TEXAS LIBRARY on August 8, 2007 http://tpj.sagepub.comDownloaded from
Among the European studies examining the impact of CS sentences on
reoffending, Muliluvuori (2001), using a quasiexperimental research
design, compared the recidivism rates of 342 offenders receiving CS sen-
tences to those for offenders given a prison sentence (n = 342) of up to 8
months in Finland. Although the CS group had a lower post-program
recidivism rate (62%) than did the prison group (72%), the difference was
not statistically significant. In another of the European studies, Killisa and
colleagues (2000) used a similar design to compare the recidivism rates of
84 offenders sentenced to CS and 39 offenders given a short prison sen-
tence (i.e., 14 days maximum prison stay) in Switzerland. Using follow-up
interviews and official rearrest and conviction data, these authors found no
differences in recidivism rates between the two types of offenders, although
the small sample size limits the statistical power of these analyses.
Finally, Spaans (1998), using matched samples of offenders, compared
the recidivism rate of those sentenced to CS to those given a suspended
short-term jail sentence in the Netherlands (N = 1,200). Using official data,
she found that offenders sentenced to CS recidivated at a lower rate (60%)
than did offenders given a suspended jail sentence (80%) during a 5-year
follow-up period. However, this difference was not statistically significant.
In addition, using “seriousness scores” (i.e., scales designed to measure the
severity of the current offense), Spaans found that more serious offenders
were being sentenced to jail, whereas less serious offenders were given CS.
These findings suggest that those offenders receiving CS sentences (i.e.,
less serious offenders) should have had a lower initial likelihood of recidi-
vating than the comparison offenders, thus biasing the study toward finding
a relative improvement in reoffending among the CS-sentenced offenders
(a finding that did not materialize).
CS Outcome Research in the United States
A review of the published literature revealed only one U.S. evaluation
designed to examine the impact of CS sentences on adult offenders’ recidi-
vism (McDonald, 1986). This study evaluated the Bronx Community
Service Sentencing Project in New York. Using a quasiexperimental design,
McDonald (1986) compared 494 adult offenders who received a CS sen-
tence to a sample of 417 who had received a jail sentence. McDonald
reports that although the two groups were not randomly selected from
among all court cases or randomly assigned to receive CS versus jail sentences,
“they resembled each other very closely and the differences that existed
between them were not substantial enough to invalidate the comparisons of
174 The Prison Journal
distribution.
© 2007 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at UNIV NORTH TEXAS LIBRARY on August 8, 2007 http://tpj.sagepub.comDownloaded from
rearrest rates” (p. 177), although no statistical tests of these differences
were presented to validate this claim.
Among offenders sentenced to CS, McDonald’s (1986) recidivism mea-
sure included rearrest within the first 6 months after being sentenced. As
such, for these CS-sentenced offenders, the dependent variable in his
analyses potentially includes both in-program recidivism (prior to com-
pletion of the CS sentence) and post-program (after CS sentence comple-
tion). He compares this “combined recidivism” measure to recidivism
during a 6-month period after release from jail for the comparison group.
He found that within 180 days after CS sentencing or jail release, 43% of
CS-sentenced offenders had been rearrested, whereas 41% of the offend-
ers sentenced to jail had been rearrested. Of these rearrested offenders in
both groups, 75% had been rearrested for a nonviolent property crime.
Unfortunately, McDonald does not report any statistical tests of the differ-
ence in recidivism for these two groups, concluding instead that “these
small differences (in percentages) were probably not significant and can be
partly attributable to error in measurements” (p. 180). Although McDonald
did not find convincing evidence that CS sentences had any impact on
recidivism, he did conclude that the program was relatively inexpensive to
administer and saved the criminal justice system money and jail bed space
by diverting some offenders out of potential incarceration. In addition,
McDonald advocated for the continued use of CS sentences with minor,
nonviolent offenders.
Contributions of the Current Study
The literature review presented above demonstrates the limitations of
what is known about the effectiveness of CS sentences in the existing
empirical research. Moreover, the research that does exist does not neces-
sarily generalize to the United States (most of it is from European studies)
and provides little guidance regarding the impact of CS compared with
other community-based sanctions.
1
More importantly, the one study that
has examined this type of sanction in the United States (McDonald, 1986)
suffers from several methodological limitations and other weaknesses that
reduce its usefulness. First, the Bronx program was evaluated nearly 20
years ago and compares CS sentences to jail terms, a potentially inappro-
priate comparison group. Second, the evaluation did not use any statistical
procedures to control for initial group differences, did not report tests of sta-
tistical significance for differences in recidivism outcomes, and combined in-
program and post-program recidivism for the CS-sentenced offenders.
Bouffard, Muftic´ / Community Service Sentences 175
distribution.
© 2007 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at UNIV NORTH TEXAS LIBRARY on August 8, 2007 http://tpj.sagepub.comDownloaded from
Given the paucity of outcome research on CS in the United States, one
of the central contributions of the current study is its examination of a more
recent CS sentencing program in the United States than was examined in
McDonald’s (1986) study. In addition, this study examines the impact of
CS sentences relative to another traditional and nonincarcerative sentence
type (criminal fines), which may be a more typically applied sanction for
comparable offenders in the United States. Although the available literature
would suggest that CS may not be effective compared to the use of short
periods of incarceration, this type of sentence may not be a realistic com-
parison for CS if more serious offenders are consistently sentenced to short
terms in prison or jail and are only rarely considered appropriate for CS
sentences. Thus, this study examines whether CS is more effective than a
sentence that is more directly comparable than jail or prison terms. If CS
sentences are as effective or more effective than the imposition of criminal
fines, some jurisdictions may wish to consider the additional community
benefits (e.g., unpaid labor) that accrue from the use of CS.
Research Questions
In terms of specific research questions, this study first asks whether
receiving a CS sentence is related to a lower probability of “any recidivism”
post-sentence relative to similar offenders who receive monetary fines.
Second, the study examines factors related to the likelihood of in-program
recidivism among a sample of offenders sentenced to complete CS sen-
tences. Finally, this study attempts to determine if receiving a CS sentence
reduces reoffending rates post-program participation relative to the use of
traditional fines, statistically controlling for initial differences between
these two groups of offenders.
Method
The current study utilizes a quasiexperimental research design to com-
pare the effectiveness of two types of community-based sanctions imposed
on a sample of misdemeanants during calendar year 2003. Specifically, the
recidivism rate for a sample of 200 offenders who received a CS sentence
is compared to that for a sample of 222 offenders sentenced to pay a mon-
etary fine after being convicted of a first-time driving under the influence
(DUI) offense. To answer the central question posed in this study, data are
analyzed using multivariate logistic regression models to determine whether
there is evidence to demonstrate that being sentenced to CS affects the
176 The Prison Journal
distribution.
© 2007 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at UNIV NORTH TEXAS LIBRARY on August 8, 2007 http://tpj.sagepub.comDownloaded from
likelihood of any recidivism (and, in later models, post-program recidivism)
relative to receiving a traditional monetary fine. Similar multivariate analy-
ses are also used to examine the factors related to in-program recidivism
among CS participants only.
Sample
As part of a larger program evaluation, data were collected on a sample of
offenders (N = 810) monitored by RESTORE, Inc., a nonprofit, community-
based corrections agency in the small metropolitan area (population of roughly
100,000) of Fargo, North Dakota. The agency’s records contained information
on 810 adult offenders who had been sentenced to complete CS hours during
the period January 1, 2003, to December 31, 2003. This group of 810 offend-
ers had engaged in more than 46,295 hours of CS during the 1-year study
period, contributing an estimated total value of $296,000 worth of unpaid
labor to local businesses and community agencies. During this period, 560
offenders (69.1%) successfully completed their CS sentence. Male offenders
constituted almost two thirds of the total sample (65.1%), and the majority of
these CS offenders were White (70.7%). Many of these CS-sentenced offend-
ers had at least a high school education or higher (89.9%), with an average age
of 23 years. Approximately half (51.0%) of the CS sample had a prior arrest
record, with the majority sentenced to CS for a drug- or alcohol-related
offense (74.2%). Nearly three fourths (74.3%) were referred from the local
municipal court, with the remainder being sentenced by the county district
court. On average, these offenders were ordered to complete 57.6 CS hours
and had been given an average of 93 days to complete their CS sentence.
Subsample of CS offenders for recidivism analyses. The data utilized in
the recidivism analyses presented in this study were collected by a single
police officer from the local department who volunteered personal time to
gather this information. Because of limitations on the amount of time avail-
able for such data collection, a smaller sample of 200 offenders was ran-
domly selected from among the entire group of 810 CS-sentenced offenders
for use in the recidivism analyses. This sample of 200 offenders sentenced
to CS included 100 offenders randomly selected from among the 560 who
had successfully completed their sentence and another 100 offenders ran-
domly selected from among those 250 who were terminated from the CS
program (either for violation of program rules or for not working all of the
required hours within the allotted time). In general, these randomly selected
samples represented the larger groups of CS-sentenced offenders from
which they were drawn.
2
Bouffard, Muftic´ / Community Service Sentences 177
distribution.
© 2007 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at UNIV NORTH TEXAS LIBRARY on August 8, 2007 http://tpj.sagepub.comDownloaded from
Fined-offenders sample. Data were collected on a comparison group of
222 first-time DUI offenders from the same jurisdiction as the CS sample.
First-time DUI is classified as a Class B misdemeanor, as are the offenses
included in the CS sample. These 222 DUI offenders were mandated by the
municipal court to pay a monetary fine in calendar year 2003. The sentenc-
ing options available to the municipal court include CS, monetary fines, eval-
uation for substance use or domestic violence treatment needs, and/or jail.
Probation is not a sentencing option for municipal court. As such, no offenders
processed through the municipal court are supervised by probation.
This group of fined offenders paid a total of $107,100 in monetary fines
to the municipal court during the course of 12 months. Offenders receiving
a fine were overwhelming male (70.3%) and White (96.4%). The average
age of fined offenders was 21.3 years (ranging from 18 to 25). Just more
than one third of these first-time DUI offenders had a prior arrest record
(35.6%). Almost all of these fined offenders (90.5%) paid their respective
fine (usually at the time of imposition), with an average fine amount of
$532 (range = $250 to $800).
Materials and Data Sources
CS program data. Information regarding demographic characteristics
and program services received by the offenders ordered to this community
corrections agency during the evaluation period was collected from the
agency’s electronic and paper files, including age, gender, race, employ-
ment status, education level, previous CS sentences, the number of hours to
be completed under the CS sentence, the number of days given to complete
the sentence, the referring agency (e.g., municipal court), the primary
offense charge, and whether the offender successfully completed the CS
sentence. To understand the operation of the CS program, interviews were
conducted with the program director, all of the agency’s case managers, and
several other local criminal justice system officials (e.g., state’s attorney).
Fined-offenders’ data from the municipal court. Data containing infor-
mation on 222 first-time DUI offenders who were sentenced by the court to
pay a monetary fine were collected from the municipal court clerk’s office.
These data included the name of offender, date of offense, offense type (i.e.,
DUI), date of sentence, amount of fine, and amount of fine paid (additional
data, such as education level and current employment status, that were
available from the community corrections agency’s files were not available
178 The Prison Journal
distribution.
© 2007 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at UNIV NORTH TEXAS LIBRARY on August 8, 2007 http://tpj.sagepub.comDownloaded from
from the court’s database for the comparison sample). An interview was
also conducted with the court clerk to understand the operation of the mon-
etary fine process. To augment the data collected from the court clerk, data
regarding each offender’s gender, date of birth (used to calculate offender’s
age at time of sentence), and race were retrieved from the local police
department’s database.
Criminal histories and recidivism data. Officially recorded arrest histories
and rearrest information were collected from the local police department on
the 200 CS-sentenced offenders and 222 fined offenders described above.
Each offender’s name and date of birth were used to search various local,
state, and regional criminal history databases, including the local police
department’s database (reflecting arrests in several local cities and counties)
and the local states attorney’s database. All arrest data were cross-referenced
between each of these databases to ensure no arrest incident was counted
more than once for each offender. The highest-level (most serious) offense
was used in cases where there were multiple charges in any given arrest event.
An offender was considered to have a prior arrest record if he or she had
one or more arrests that occurred before his or her current court date. A CS
offender was considered to have recidivated post-program if he or she was
rearrested after either his or her date of successful completion or his or her
date of termination from the CSP. Fined offenders were considered to have
recidivated post-program if they had another arrest after the date that their
fine was imposed by the court. In-program recidivism for CS offenders was
recorded as those instances in which the person’s first rearrest after the cur-
rent CS sentence date occurred prior to the date he or she was terminated
from or completed the CS sentence according to the CS agency’s records.
Information was also collected on the date the rearrest occurred.
Description of the CS process. As previously mentioned, interviews with
case managers and the program director and a review of written program
materials (e.g., policy and procedure manuals, client forms) were used to
document the operation of the CS program being studied. This community
corrections program serves both adult and juvenile offenders (only adults
are examined in this study because of restrictions on accessing juvenile
data), primarily accepting referrals from the local municipal court, with
smaller numbers also coming from the district court (including the juvenile
court) and the local office of the state’s adult parole and probation agency.
On average, the program conducts 84 intake interviews with adult offend-
ers referred from various local courts (66.7% from municipal court, 15.5%
Bouffard, Muftic´ / Community Service Sentences 179
distribution.
© 2007 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at UNIV NORTH TEXAS LIBRARY on August 8, 2007 http://tpj.sagepub.comDownloaded from
from District court) and/or parole or probation agency (17.8%) during a
typical month. During these in-person interviews, case managers screen the
offender for program eligibility and evaluate the offender’s skills, interests,
abilities or disabilities, and circumstances that may interfere with work site
placement. In addition, the case manager familiarizes the offender with the
program’s policies, procedures, and expectations, and offenders sign a
program contract or agreement and medical release form.
Program guidelines instruct case managers to place offenders in various
work site placements after considering the court’s objectives, the offender’s
characteristics (i.e., offender interests, skills, and abilities, geographic loca-
tion of work site), and the work site agency’s needs. Successful work site
placements are described by the agency as those that match the offender’s
interests and skills while offering the offender a challenging and rewarding
experience. The case manager also involves the work site in the process of
assigning workers and communicates program expectations to the relevant
work site supervisor, who then oversees the CS offender while working
onsite. The offender is notified during the intake interview process of the
date, time, and address where he or she is to report for CS work.
Throughout the duration of the offender’s sentence, the case manager is
responsible for monitoring the offender’s progress, including evaluating the
offender’s performance, recording of the number of CS hours successfully
completed, and quickly intervening if any problems develop. However, the
offender is directly supervised by an individual at the workplace, not by the
case manager. It is important to note that additional monitoring activities
such as verification of employment or stable housing, tracking the
offender’s location throughout the day, or drug testing (as might occur
among offenders sentenced to probation supervision) are not conducted on
these CS-sentenced offenders. On average, the case manager contacts the
work site supervisor by telephone once a month to monitor the offender’s
progress. An offender who completes the required number of CS hours will
be recorded as a “successful completion” of the CS sentence.
Those offenders who repeatedly fail to comply with the terms of the work
site contract or court order are terminated from the CS program as “unsuc-
cessful.” In either case, the referring agency is provided written notification
of the outcome for each offender. If the offender is cited for what is termed
“incidental behavior” (e.g., minor infractions of program rules), the primary
goal is to bring the offender in compliance rather than to immediately termi-
nate the offender and return the case to the referring agency, although
repeated instances of incidental behavior do result in the return of the case to
the referring agency as an unsuccessful termination from the program.
180 The Prison Journal
distribution.
© 2007 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at UNIV NORTH TEXAS LIBRARY on August 8, 2007 http://tpj.sagepub.comDownloaded from
Examples of incidental misbehavior while in the CS program can include
marginal performance, personality conflicts with staff, or occasional lateness
or absenteeism because of transportation problems (note that the subsample
of 200 CS offenders used in the recidivism analyses presented here included
100 such unsuccessful offenders, randomly selected from among the 250
who were recorded as unsuccessful participants during 2003).
Overview of the fine process. In the local jurisdiction, adults arrested on
a DUI charge are taken to the local jail. On posting bond (typically within
24 hours), arrestees are given a date to appear before a municipal court
judge (typically 3 weeks following arrest). Convicted first-time DUI
offenders in this jurisdiction are sentenced to pay a monetary fine (average
for this sample = $532), although the judge also has the discretion to sen-
tence offenders to lower (a mandatory minimum of $250 is imposed by
state law for first-time offenders) or higher fine amounts. Although the
majority of offenders pay their fine at the time of sentencing, fines may also
be paid in installments, again at the discretion of the sentencing judge.
Results
Comparison of Offender Samples
As depicted in Table 1, the two samples of offenders utilized in the current
study appear to have some important differences in individual characteristics;
however, both groups share some important similarities as well (unless
otherwise noted, all tests of statistical significance are reported at the one-
tailed level). For instance, both samples are predominantly White (90.5% of
CS offenders, 96.4% of fined offenders) and male (60.5% of CS offenders,
70.3% of fined offenders). The racial homogeneity of both samples reflects
that of the wider community in which this study was conducted.
3
In terms of
other individual characteristics, CS offenders and fined offenders exhibit
some notable differences. For instance, offenders from the CS sample were,
on average, 3 years older than offenders from the fined sample (24.3 years
and 21.3 years, respectively; t = –4.918, p < .001). CS offenders also appear
to have more serious criminal histories (51.0% with a prior arrest) compared
to fined offenders (35.6%; χ
2
= 10.206, p < .001). The majority of the CS
offenders were arrested on a drug or alcohol offense (85.3%), whereas all
offenders in the fined sample were charged with a current (first-time) DUI
offense (100%; χ
2
= 34.897, p < .001). Finally, the time at risk (i.e., number
Bouffard, Muftic´ / Community Service Sentences 181
distribution.
© 2007 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at UNIV NORTH TEXAS LIBRARY on August 8, 2007 http://tpj.sagepub.comDownloaded from
of months from date of sentence to the end of the data-collection period) for
each group of offenders was significantly different (t = 21.464, p < .001). The
follow-up time for fined offenders was, on average, almost 8 months longer
than CS offenders (18.0 months vs. 10.5 months).
In general, these groups share important similarities in terms of race,
gender, and current offense level (all misdemeanants), although the CS
sample appears to be composed of somewhat more serious offenders
(higher likelihood of prior arrests and age)
4
and to have experienced less
time at risk than the fined sample. In light of these differences between the
samples, multivariate models presented below include controls for the indi-
vidual characteristics on which the two samples vary.
5
Predictors of Any Recidivism
Bivariate analyses. Bivariate analyses reveal that several independent
variables were related to the probability of any recidivism after the date
that the CS sentence or fine was imposed, including race and criminal
history (see Table 2). For instance, non-White offenders were more likely
to reoffend compared to White offenders (63.0% vs. 33.0%; χ
2
= 9.986,
182 The Prison Journal
Table 1
Description of Community Service (CS) Participants
(Subsample) and Fined Sample
Variable CS
a
Fined
b
Test Statistic
Age in years (M)*** 24.3 21.3 –4.918
Percentage male 60.5 70.3 4.453
Percentage White 90.5 96.4 6.108
Percentage with prior 51.0 35.6 10.206
arrest record***
Current offense type*** 34.897
Percentage alcohol or drug 85.3 100.0
Percentage violent 3.8 —
Percentage nonviolent 10.9 —
Percentage completed all 50.0 90.5 84.552
CS hours or paid fine***
Time at risk (# of months)*** 10.5 18.0 21.464
Post-program recidivism** 23.4 35.6 6.803
a. n = 200.
b. n = 222.
*p < .05, one-tailed. **p < .01, one-tailed. ***p < .001, one-tailed.
distribution.
© 2007 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at UNIV NORTH TEXAS LIBRARY on August 8, 2007 http://tpj.sagepub.comDownloaded from
p < .001). As expected, offenders with a prior criminal record were signifi-
cantly more likely to recidivate than were offenders with no prior criminal
record (47.0% vs. 25.8%; χ
2
= 20.268, p < .001). Variables that were not
found to be significantly related to recidivism include age, gender, offense
type (i.e., alcohol- or drug-related offense vs. violent or property offense),
sentence severity (i.e., being at or above the median fine or CS hours
amount), sentence type (CS or fine), and time at risk (i.e., number of
months from date of sentence to the end of the data-collection period).
Multivariate analyses. A logistic regression model was computed to
examine whether any of these variables are significantly related to the prob-
ability of an offender recidivating at any time after the imposition of the
CS sentence or fine (see Table 3). Results of the model reveal that three
Bouffard, Muftic´ / Community Service Sentences 183
Table 2
Bivariate Analyses: Predictors of Any Recidivism
Variable No Recidivism
a
Recidivism
b
Test Statistic
Age in years (M) 22.4 23.3 –1.421
Gender 0.241
Percentage male 64.3 35.7
Percentage female 66.7 33.3
Race*** 9.986
Percentage White 67.0 33.0
Percentage non-White 37.0 63.0
Criminal history*** 20.268
Percentage with prior arrest record 53.0 47.0
Percentage with no prior arrest record 74.2 25.8
Current offense type 2.614
Percentage alcohol or drug 62.8 37.2
Percentage violent 80.0 20.0
Percentage nonviolent 71.4 28.6
Sentence severity 67.8 69.7 0.160
Sentence type 0.092
Community service 65.8 34.2
Fined 64.4 35.6
Time at risk (# of months) 14.2 14.9 –1.492
Note: N = 421.
a. n = 274.
b. n = 147.
*p < .05, one-tailed. **p < .01, one-tailed. ***p < .001, one-tailed.
distribution.
© 2007 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at UNIV NORTH TEXAS LIBRARY on August 8, 2007 http://tpj.sagepub.comDownloaded from
variables were significantly related to any post-sentence recidivism, includ-
ing race, prior criminal history, and offense type (i.e., alcohol- or drug-
related offense vs. violent or property offense). Consistent with the results
of the bivariate analyses, non-White offenders have a significantly higher
likelihood of reoffending (B = 1.288, p < .05), as do offenders with a prior
criminal history (B = 0.953, p < .001). Interestingly, offenders charged with
a violent or property crime (vs. a drug- or alcohol-related offense) are less
likely to reoffend at any point after receiving their sentence (B = –1.076,
p < .05). Variables not found to be significantly related to the probability of
recidivism at the multivariate level include gender, age, sentence severity
(being at or above the median fine or CS hours amount), sentence type (CS
or fine), and time at risk (number of months from date of sentence to the
end of the data collection period).
Additional Recidivism Analyses
McDonald (1986), in his evaluation of a CS program in New York City,
examined the impact of a CS sentence on recidivism from the date of sen-
tence imposition onward, including in his outcome measure any recidivism
184 The Prison Journal
Table 3
Logistic Regression: Any Recidivism for Community
Service and Fined Offenders
Independent Variable Recidivism
B ExpB SE
Prior criminal history*** 0.953 2.593 0.227
Time at risk 0.031 0.324 1.032
Male 0.038 1.038 0.234
Non-White** 1.288 3.625 0.447
Age 0.008 1.008 0.019
Non–drug or alcohol offense* –1.076 0.341 0.227
Sentence severity –0.373 0.689 0.328
Community service sentence –0.053 0.948 0.412
Constant* –1.408 0.245 0.735
Note: R
2
= .12, X
2
(8, N = 394 ) = 34.606, p < .001
*p < .05, one-tailed. **p < .01, one-tailed. ***p < .001, one-tailed.
distribution.
© 2007 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at UNIV NORTH TEXAS LIBRARY on August 8, 2007 http://tpj.sagepub.comDownloaded from
that occurred both while the offenders were completing their CS hours and
several months after they completed their sentence. He compared this to the
recidivism rate of offenders sentenced to jail, after they had been released
to the community. As presented above, the current study replicates this type
of analysis and produces similar results. The next set of analyses attempts
to examine the impact of sentence type on post-participation recidivism
while also controlling for whether or not the offender completed his or her
sentence (worked all his or her CS hours or paid his or her fine in full).
To examine the impact of successfully completing either type of sentence,
a dependent variable that measures only recidivism that occurs after the
offender participates in the intervention (whether successfully completed or
not) must be employed to ensure correct temporal ordering. This is necessary
because in-program recidivism (among the CS participants) would negatively
affect the likelihood of successfully completing an ongoing intervention (e.g.,
CS), leading to faulty conclusions about the relationship between these two
variables. For fined offenders, post-intervention recidivism is simply mea-
sured as any recidivism after the date the fine is imposed because the deter-
rent impact of a fine may be expected to arise from the time of its imposition
onward. On the other hand, among the CS offenders in this sample, post-
participation recidivism means any recidivism either after the offender is ter-
minated from the program (for violation of program rules or failing to
complete all the hours as ordered, typically 3 months) or after the offender
successfully completes all CS hours as ordered.
6
In-program recidivism. Before examining the impact of CS sentences on
post-participation recidivism, a multivariate regression model is used to
examine the predictors of in-program recidivism among only those offend-
ers receiving a CS sentence. This analysis is useful for examining the types
of CS offenders who are more or less likely to succeed while in the com-
munity completing a CS sentence and thus contributes to an understanding
of whether more serious offenders can in fact be equally well managed with
CS programs. On average, CS offenders were given 3 months (93 days) in
which to complete their CS order. Of the 200 CS-sentenced offenders for
which arrest records were collected, 28 (14.0%) were rearrested prior to
completing or being terminated from the CSP. A logistic regression model
was computed to analyze which independent variables were related to in-
program recidivism among these CS offenders (see Table 4).
Although few independent variables exhibited significant effects on the
likelihood of in-program recidivism, those that did are instructive in rela-
tion to calls for increased use of CS sentences for more serious types of
Bouffard, Muftic´ / Community Service Sentences 185
distribution.
© 2007 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at UNIV NORTH TEXAS LIBRARY on August 8, 2007 http://tpj.sagepub.comDownloaded from
offenders (e.g., Tonry, 1998). First, offenders with a prior arrest record
(B = 1.373, p < .05) were more likely to recidivate while in the process of
completing their CS order. Severity of sentence was also found to be sig-
nificantly related to in-program recidivism, with offenders who received a
more severe sentence (i.e., a sentence that was at or above the median
number of CS hours) having a higher likelihood of reoffending (B = 1.272,
p < .05). Finally, age was found to be significantly related to in-program
recidivism, with older offenders having a higher likelihood of reoffending
(B = 0.050, p < .05). Thus, controlling for all other variables in the model,
more serious offenders (i.e., older offenders who had a prior arrest record
and were given more CS hours to complete) have a higher likelihood of
recidivating before completion of their CS sentence. Conversely, another
potential indicator of offense severity (those who committed misdemeanor
property or violent crimes) did not increase the likelihood of recidivating
during participation in the CS program. Overall, the pattern of results pre-
sented thus far suggests that those with criminal histories (even among
those receiving a CS sentence and not necessarily prison or jail bound) may
have difficulty successfully avoiding reoffending during their participation
in such CS programs. However, those with non drug or alcohol offenses are
not more likely to reoffend in program. It may be possible to improve the
186 The Prison Journal
Table 4
Logistic Regression: In-Program Recidivism Among Community
Service Offenders Only
Independent Variable Recidivism
B ExpB SE
Prior criminal history* 1.373 3.947 0.645
Time at risk –0.087 0.916 0.078
Male –0.707 0.493 0.536
Non-White 0.040 1.041 0.701
Age* 0.050 1.051 0.028
Non–drug or alcohol offense –0.568 0.566 0.876
Sentence severity* 1.272 3.569 0.587
Constant** –2.882 0.056 1.139
R
2
.33
χ
2
(7, N = 126) 28.162
p < .001
Note: N = 126.
*p < .05, one-tailed. **p < .01, one-tailed. ***p < .001, one-tailed.
distribution.
© 2007 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at UNIV NORTH TEXAS LIBRARY on August 8, 2007 http://tpj.sagepub.comDownloaded from
effectiveness or “appropriateness” of CS sentences for more serious offend-
ers, however, through the use of additional supervision of these more seri-
ous offenders while they are serving a CS sentence in the community.
Post-program recidivism. In the following section, a logistic regression
model predicting recidivism after program participation is presented. To
reiterate, the post-intervention period for fined offenders is defined as any
recidivism after the date the fine is imposed, whereas for CS offenders it is
defined as any recidivism after the offender either completes or is termi-
nated from the CS program. Again, in this model, CS cases in which the
person recidivated during the time in the CS program are omitted. Although
the examination of post-program recidivism is necessary to include a con-
trol for program completion (an important potential measure of the
offender’s inherent motivation), the removal of CS in-program recidivists
could result in a comparison that is biased toward finding a crime-reducing
effect for CS sentences, if those CS offenders who remain in the sample are
less serious offenders. In this case, that does not seem to be the result. For
instance, those CS offenders who remain in the sample (those not recidi-
vating during the program) are still more likely to have a prior arrest
(45.3%) than are the fined offenders (35.6%; χ
2
= 3.854, p < .05) examined
here. In addition, those CS offenders who recidivated in program reof-
fended 3 times more quickly (an average of 19.7 days after sentencing) than
did even the group of “unsuccessful” offenders who were either terminated
from the CS program or who failed to pay their fines (an average of 59.3
days after sentencing), whereas offenders who eventually completed either
type of sentence lasted an average of 90.2 days from sentencing to first rear-
rest (F = 52.67, p < .01). This result for time to rearrest in particular sug-
gests that these in-program recidivist CS offenders may be qualitatively
different from those who do not recidivate while completing their CS sen-
tence and may be fairly removed from the comparison. Note that despite the
removal of the in-program recidivists from the CS sample, this comparison
is still essentially a conservative estimate of the impact of CS on post-
program recidivism relative to fines because the CS sample is still com-
posed of offenders who are more likely to have a prior arrest.
A logistic regression model was computed to determine which indepen-
dent variables were statistically significant predictors of post-program
recidivism (excluding those CSP offenders who recidivated during the
program), when controlling for a number of other individual-level vari-
ables. The other control variables used in this model include prior arrests
(0 = no prior arrest,1 = prior arrest), sentence severity (0 = below median,
Bouffard, Muftic´ / Community Service Sentences 187
distribution.
© 2007 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at UNIV NORTH TEXAS LIBRARY on August 8, 2007 http://tpj.sagepub.comDownloaded from
1 = at or above the median fine or amount of CS hours), time at risk (in
months), offense type (0 = non drug or alcohol offense,1 = drug or alco-
hol offense), sentence completion (0 = did not complete all CS hours or did
not pay fine in full,1 = all CS hours completed or fine paid in full), sen-
tence type (0 = fine,1 = CS), and other demographic factors, including age,
gender (0 = female,1 = male), and race (0 = White,1 = non-White). The
dependent variable is the likelihood of recidivism among the sample of 172
CSP offenders and 222 fined offenders (because of missing data for 23
cases, some from each group were not included in the final model).
Results presented in Table 5 reveal that the likelihood of rearrest was
again higher among offenders with a prior arrest record (B = 0.861, p <
.001). Sentence severity appears to result in a deterrent effect, with offend-
ers receiving sentences at or above their respective medians (B = –0.902,
p < .01) recidivating at significantly lower rates than those receiving more
lenient CS sentences or fines. Of most interest in this study, sentence type
was also found to be significantly related to recidivism, with offenders sen-
tenced to complete CS work less likely to reoffend, controlling for all other
variables in the model (B = –0.926, p < .05). Successfully completing either
type of sentence was also related to a lower likelihood of repeat offending
(B = –0.749, p < .01), suggesting at least some effect for offender motiva-
tion in addition to the “intervention” effect seen for CS sentences.
Discussion
Overall, the results for post-program recidivism presented in this study
support the use of CS over another community-based sanction typically
used with comparable types of less serious offenders, specifically tradi-
tional monetary fines. This conclusion is supported in both bivariate results
and in multivariate analyses controlling for initial group differences and the
successful completion of either sentence type. These results are especially
noteworthy given that there are at least some indications that the CS sam-
ple (even when the in-program recidivists were removed) was composed of
more serious offenders (more likely to have prior arrests), which would
make this a relatively conservative test of the effectiveness of CS sentences
compared to traditional fines at reducing post-program recidivism.
Although the statistical analyses of any recidivism fail to show significant
reductions among CS-sentenced offenders, these analyses do not include con-
trols for program completion (a potentially important measure of offender
motivation). On the other hand, even in this model (which might arguably be
188 The Prison Journal
distribution.
© 2007 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at UNIV NORTH TEXAS LIBRARY on August 8, 2007 http://tpj.sagepub.comDownloaded from
seen as including less selection bias regarding the CS sample), the finding of
no effect may suggest that CS sentences are at least as effective as traditional
fines in terms of any post-sentencing recidivism. Given that the use of CS
sentences also likely results in other benefits to the community that do not
materialize with the use of fines, correctional officials may wish to consider
the expanded use of this type of alternative sanction, at least for some types
of low-level offenders. For instance, even in this small community, more than
46,000 hours of unpaid labor, valued at nearly $300,000, was provided back
to local service agencies and other businesses.
More importantly, the positive effects demonstrated by this study were
generated using a design and methodology that is a substantial improve-
ment over the existing literature on CS effectiveness, most of which either
lacks generalizability because it is outdated or was conducted overseas or
lacks internal validity because it compares CS to incarceration or employs
inadequate research methods. In fact, the current study’s design is consid-
erably stronger than the one previously published U.S. study of recidivism
among those sentenced to CS (McDonald, 1986) in that it employs an
Bouffard, Muftic´ / Community Service Sentences 189
Table 5
Logistic Regression: Post-Program Recidivism for
Community Service (CS) and Fined Offenders
Recidivism
Independent Variable B ExpB SE
Prior criminal history*** 0.861 2.365 0.242
Time at risk 0.043 1.044 0.034
Male 0.278 1.321 0.256
Non-White 1.321 3.746 0.504
Age –0.020 0.980 0.026
Non–drug or alcohol offense –1.071 0.343 0.673
Successful completion of sentence**
a
–0.749 0.473 0.315
Sentence severity** –0.902 0.406 0.382
CS sentence* –0.926 0.396 0.485
Constant 0.009 1.009 0.901
R
2
.14
χ
2
(9, N = 371) 39.515
p < .001
Note: N = 371.
a. The proportion of offenders who successfully completed their CS sentence was 50%.
*p < .05, one-tailed. **p < .01, one-tailed. ***p < .001, one-tailed.
distribution.
© 2007 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at UNIV NORTH TEXAS LIBRARY on August 8, 2007 http://tpj.sagepub.comDownloaded from
appropriate comparison sample of offenders receiving community-based
sanctions, controls for initial group differences, and employs statistical test-
ing to examine differences in reoffending outcomes using a post-program
measure of recidivism.
This study’s results also provide some interesting, if preliminary, sup-
port for the claims of restorative justice scholars (e.g., Bazemore &
Maloney, 1994), who suggest that CSPs could be designed to incorporate
restorative principles (e.g., matching offenders to appropriate CS work
experiences, as was done in this program) and that doing so would lead to
positive program outcomes. These results also support the “principles of
effective treatment,” as outlined by correctional rehabilitation scholars
(Andrews et al., 1990), especially in regard to the concept of “matching
offenders’ risks or needs,” as is done in the program studied here.
McDonald (1986) also advocated for the use of CS with low-level offend-
ers, underlining the importance of matching offenders’ risk or need level to
the intensity of the intervention to which they are sentenced; however, the
Bronx program he studied did not actually individualize work placements
for offenders. Instead, all offenders were assigned to complete a fixed
number of hours within a limited number of work types.
On the other hand, Gelsthrope and Rex (2004) conclude from their study
of CS-sentenced offenders that for offenders to benefit from this type
of sanction (i.e., reductions in recidivism, improvements in attitudes, self-
perceived problems), offenders should be screened by CS program staff and
matched to work experiences based on their needs (Gelsthrope & Rex,
2004). This concept of matching offenders with appropriate work experi-
ences then is also consistent with both the “principles of effective interven-
tion” (Andrews et al., 1990) and one of the central restorative justice
suggestions made by Bazemore and Maloney (1994). Given these sugges-
tive findings, future research more directly comparing different styles of
delivering CS sentences also seems warranted.
At a broader level, the apparent success of this CS program at reducing
recidivism underlines the utility of innovative, alternative sanctioning
efforts in general, in contrast to the reliance on incarcerative sentences,
which is often the predominant response to offenders of all levels of seri-
ousness in the United States. If, as some authors (see Tonry, 1998) have
suggested, community correctional programs can be relatively effective at
reducing recidivism and maintaining public safety, then their increased
usage as an appropriate and effective response to certain types of (non-
violent) offenders seems warranted. This is especially true if these commu-
nity-based programs, such as CS, provide other additional benefits to the
190 The Prison Journal
distribution.
© 2007 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at UNIV NORTH TEXAS LIBRARY on August 8, 2007 http://tpj.sagepub.comDownloaded from
larger community (e.g., unpaid labor) that do not materialize as a result of
placing offenders in jail or prison.
Limitations and Future Research
Although these results appear quite promising relative to the potential
impact of CS sentences on reoffending, the research design employed here
is not without limitations. For instance, although the offenders in both the
CS and fined groups studied here are comparable in terms of the legal
severity of their offenses (all misdemeanors), and although statistical pro-
cedures were used to control for initial group differences (e.g., prior arrest
histories), the use of a true experimental design was not possible in this
jurisdiction. As such, the possibility cannot be ruled out that other initial
group differences actually accounted for recidivism differences but were
not controlled with the data available for this study.
In addition, although these results were generated with a design that
improved on the internal validity of the study, they may not generalize to
other programs in other jurisdictions. Specifically, the community under
examination here is in a relatively small urban area, with its own municipal
court system and a relatively homogeneous populace in terms of racial com-
position. Similarly, although the evaluation of a CS program that incorporates
at least some restorative justice principles suggests support for some of the
claims about the potential benefits of this type of approach, no direct com-
parison to a more traditionally delivered CS program was included. Thus, it
is not clear if the positive results seen in this program are from the restorative
components included in the program, from the use of a similar sample of
offenders sentenced to another community-based sanction (in contrast to
what is seen in the existing literature finding no effects), or from some other
unique but unaccounted for aspect of this particular CS program.
In terms of future research, the authors hope that this study, which sug-
gests that CS sentences can be effective at reducing recidivism among rel-
atively less serious offenders, will spur additional research into this type of
alternative sanction. Specifically, future research should attempt to repli-
cate these positive results among CS programs in more diverse jurisdictions
(racially and in terms of population size), among CS programs that serve
offenders from other than primarily municipal courts, and among other
types of community-based sanctions (e.g., day-reporting centers, home
confinement, or misdemeanor probation). As is the case in many correc-
tional program evaluations, additional research should attempt to employ
Bouffard, Muftic´ / Community Service Sentences 191
distribution.
© 2007 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at UNIV NORTH TEXAS LIBRARY on August 8, 2007 http://tpj.sagepub.comDownloaded from
even more rigorous (i.e., experimental) designs to continue to examine the
impacts of CS sentences on reoffending.
Finally, given that one of the aspects of this CS program that may have
accounted for its positive results is the focus on implementing at least some
restorative justice concepts, additional research directly comparing CS
programs that are delivered in different styles (traditional vs. restorative) is
also called for. Overall, the authors agree with other scholars writing about
the underuse of CS as a viable alternative sanction (e.g., Tonry, 1998), espe-
cially in light of the benefits that are provided back to the wider community
from the use of this type of sanction, as opposed to requiring offenders to
simply pay a fine to the court. In the community studied here, offenders
sentenced to complete CS hours not only were significantly less likely to
recidivate (post-program, regardless of successful completion) but also
annually provided thousands of hours and hundreds of thousands of dollars
in unpaid work to their local communities. Hopefully, recognition of these
types of benefits, along with the results of this initial study of the impact of
CS sentences’ impact on recidivism, will lead to more empirical and prac-
tical attention to this type of sanction.
Notes
1. The generalizability of the aforementioned European studies is limited largely because
the context in which community service (CS) is implemented in Europe is markedly different
than the context in which CS is delivered in the United States. Largely, CS is one of many
alternative sanctions imposed in European countries, while at the same time these nations
often rely less on the use of incarceration for minor and first-time offenders than does the
United States (Lynch, 2002).
2. Bivariate statistics (e.g., χ
2
, t tests) were used to examine any differences in these ran-
domly selected samples relative to their respective groups of CS completers and terminated
offenders, including age, race, gender, education level, current offense type, previous experi-
ence with CS, and days given to complete the sentence. These results generally demonstrate
that the randomly selected sample of 100 completers represented the entire sample of 560
completers, with the exception that the subsample of 100 appeared to include significantly
more non-Whites (9%) than did the entire sample of 560 (2.9%, χ
2
= 8.729, p < .01). No sig-
nificant differences were found between the random sample of 100 terminated CS offenders
and the entire sample of 250 terminated CS offenders.
3. The population of the local community from which the samples included in this study
were drawn is predominately White (94.2%).
4. In the CS sample, age is correlated with offense type. Specifically, younger CS offenders
were more likely to have been charged with less serious offenses (i.e., minor in possession),
whereas older offenders were more likely to have been charged with property and violent
offenses.
192 The Prison Journal
distribution.
© 2007 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at UNIV NORTH TEXAS LIBRARY on August 8, 2007 http://tpj.sagepub.comDownloaded from
5. Although the authors recognize that these are not perfectly matched samples, in con-
versations with the clerks from the municipal court, these fined, first-time driving under the
influence offenders were likely the most similar to the CS offenders who were available from
within this court’s jurisdiction.
6. One drawback of this type of comparison of post-intervention recidivism rates is the
exclusion of those CS cases in which the offender recidivated during the program. This pro-
cedure might, for instance, result in “creaming off” those CS offenders who are most likely to
fail in such a program, biasing the subsequent comparison of (some) CS to (all) fined offend-
ers. To examine the potential for selection bias in this modeling procedure, a similar regres-
sion model was computed to analyze the impact of sentence type on any recidivism (including
in-program recidivism among CS offenders). However, this supplemental model uses a three-
category completion variable to control for whether the offender (a) recidivated during the
program (coded 0), (b) avoided in-program recidivism (CS only) but did not successfully
complete all of his or her CS hours or pay his or her fine (coded 1), or (c) successfully com-
pleted either sentence without in-program recidivism (coded 2). Results for this model do not
change the conclusions presented here that participation in the CS and completion of either
sentence reduce the likelihood of post-program recidivism, relative to those CS offenders who
recidivated during the program. Results are available from the authors on request.
References
Andrews, D. A., Zinger, I., Hoge, R. D., Bonta, J., Gendreau, P., & Cullen, F. T. (1990). Does
correctional treatment work? A clinically relevant and psychologically informed meta-
analysis. Criminology, 28(3), 369-404.
Bazemore, G., & Maloney, D. (1994). Rehabilitating community service: Toward restorative
justice sanctions in a balanced justice system. Federal Probation, 58(1), 24-35.
Delens-Ravier, I. (2003). Juvenile offenders’ perceptions of community service. In L.
Walgrave (Ed.), Repositioning restorative justice (pp. 149-166). Portland, OR: Willan.
Development Services Group. (2006). Restitution/community service. Retrieved November 13,
2006, from http://www.dsgonline.com/mpg_non_flash/restitution_community_ service.htm
Gelsthrope, L., & Rex, S. (2004). Community service as reintegration: Exploring the poten-
tial. In G. Mair (Ed.), What matters in probation (pp. 229-254). Portland, OR: Willan.
Harris, R. J., & Wing Lo, T. (2002). Community service: Its use in criminal justice.
International Journal of Offender Therapy and Comparative Criminology, 46, 427-444.
Killisa, M., Aebi, M., & Ribeaud, D. (2000). Does community service rehabilitate better than
short-term imprisonment? Results of a controlled experiment. Howard Journal, 39(1), 40-57.
Langan, P. A., & Farrington, D. P. (1998). Crime and justice in the United States and in
England and Wales, 1981-1996: Executive summary. Washington, DC: Bureau of Justice
Statistics.
Lynch, J. (2002). Crime in international perspective. In J. Q. Wilson & J. Petersilia (Eds.),
Crime: Public policies for crime control (pp. 5-41). Oakland, CA: ICS Press.
McDonald, D. C. (1986). Punishment without walls: Community service sentences in New
York City. New Brunswick, NJ: Rutgers University Press.
Muliluvuori, M. L. (2001). Recidivism among people sentenced to community service in
Finland. Journal of Scandinavian Studies in Criminology and Crime Prevention, 2(1), 72-82.
Bouffard, Muftic´ / Community Service Sentences 193
distribution.
© 2007 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at UNIV NORTH TEXAS LIBRARY on August 8, 2007 http://tpj.sagepub.comDownloaded from
Spaans, E. C. (1998). Community service in the Netherlands: Its effects on recidivism and
net-widening. International Criminal Justice Review, 8, 1-14.
Tonry, M. (1998). Sentencing matters. Oxford, UK: Oxford University Press.
Wing Lo, T., & Harris, R. J. (2004). Community service orders in Hong Kong, England, and
Wales: Twins or cousins. International Journal of Offender Therapy and Comparative
Criminology, 48, 373-388.
Jeffrey A. Bouffard is an assistant professor in the Department of Political Science, Criminal
Justice Program at Washington State University. His research interests include community cor-
rections, offender rehabilitation, crime and delinquency prevention, criminological theory, and
program evaluation methods. He has published numerous peer-reviewed articles in Justice
Quarterly, Journal of Criminal Justice, The Prison Journal, Journal of Offender Rehabilitation,
and Journal of Drug Issues.
Lisa R. Muftic
′′
is an assistant professor in the Department of Criminal Justice at the
University of North Texas. She is working on research related to intimate partner violence,
community corrections, and criminological theory. Her additional research interests include
quantitative research methodology and comparative criminology. She has published articles in
Violence Against Women, Journal of Interpersonal Violence, Journal of Family Violence, Journal
of Offender Rehabilitation, International Journal of Offender Therapy and Comparative
Criminology, and Western Criminology Review.
194 The Prison Journal
distribution.
© 2007 SAGE Publications. All rights reserved. Not for commercial use or unauthorized
at UNIV NORTH TEXAS LIBRARY on August 8, 2007 http://tpj.sagepub.comDownloaded from