Recidivism among Participants of a ReentryProgram for Prisoners Released without

SupervisionNora Wikoff, Donald M. Linhorst, and Nicole Morani

As higher numbers of individuals are released from prison and rejoin society, reentry pro-grams can help former offenders reintegrate into society without continuing to engage incrime. This quasi-experimental study examined whether participation in reentry program-ming was associated with reduced recidivism among offenders who were no longer undercriminal justice supervision. Offenders who completed their sentences in prison wereinvited to participate in Project Re-Connect (PRC), a six-month, voluntary prisonerreentry program. Following participants' release from prison, PRC provided case manage-ment and direct monetary support to participants for up to six months. Survival analysiswas used to compare recidivism rates between 122 PRC participants and 158 eligible non-participants. Cox regression coefficients indicated that program participation and having ahigh school diploma or its equivalent were associated with reduced likelihood of newconvictions, whereas substance abuse was associated with higher risk of subsequent con-victions. The implications for social work policy and practice are discussed.

KEY WORDS; case management; former offenders; prisoner reentry; program evaluation

T he U.S. prison population has grown ex-ponentially over the past 30 years, at greatcost to taxpayers and offenders alike.

Between 1980 and 2008, the prison populationexpanded by 475%, reaching 1,518,559 in 2008(Bureau of Justice Statistics, 2010). Policy changesfueled this rapid growth, as many states adoptedmandatory and determinant sentencing guidelinesthat resulted in more individuals serving longerprison terms. Meanwhile, stricter parole require-ments returned more ex-offenden to prison ontechnical parole violations (Seiter & Kadela, 2003;Zhang, Roberts, & CaUanan, 2006). Parole viola-tors who complete their sentences in prison are nolonger subject to supervision once released fromprison, thereby restricting society's ability tomonitor and assist these individuals during reentry(Braga, Piehl, & Hureau, 2009; O'Brien, 2009;Seiter & Kadela, 2003).

The number and the rate of inmates releasedwithout parole supervision have increased evenover the last decade, as the number of unsuper-vised releases grew from 118,886 in 2000 (20.4%of aU released) to 165,568 in 2008 (24.2% of allreleases) (Sabol, West, & Cooper, 2009; Seiter &Kadela, 2003). Former offenders commit crimes at

higher rates than the general population, so incombination with technical parole violations, manyex-offenders recidivate and return to prison withinthe fint few years of release (Braga et al., 2009). Asof 1994, more than two-thirds of state prisonerswere rearrested for one or more serious crimeswithin three yean of release. Almost half of thosereleased returned to prison during that time framefor parole violations or new convictions (Langan &Levin, 2002).

Prisoner reentry has become a critical topic ascommunities prepare to absorb increasing numbersof returning former offenders; 683,106 inmateswere released from state or federal prisons in 2008,an increase of nearly 20% over the number releasedin 2000 (Petersilia, 2003; Roman & Travis, 2006;Sabol et al., 2009; Seiter & Kadela, 2003; Wilson &Davis, 2006). Reentry programs have been dev-eloped nationwide to address offender needsand smooth the transition from prison into thecommunity.

RISK FACTORS FOR REENTRY, CURRENT

PRACTICE, AND EMPIRICAL EVIDENCE

Several risk factors increase the likeHhood thatex-offenden wül return to prison on new charges.

doi: 1O.1093/swr/svsO21 O 2012 Nationai Association of Sociai Woricers 289

These risk factors include age, gender, race, gangmembership, substance abuse, antisocial behavior,low social achievement, negadve peers, length ofprior criminal history, and the number of yean in-carcerated before release (Braga et al., 2009; Gen-dreau, Litde, & Goggin, 1996; Huebner, Varano,& Bynum, 2007; Langan & Levin, 2002; Listwan,2009; O'Brien, 2009; Seiter & Kadela, 2003;Wheeler & Patterson, 2008; Wilson & Davis,2006; Yahner & Visher, 2008). Economic difficul-ties also compromise offenders' abüities to reinte-grate into society successfully. For example,without access to necessities—such as food, cloth-ing, shelter, transportation, and personal identifica-tion—fomier inmates may see no other opdonthan to retum to illegal acdvities to meet theirneeds (La Vigne, Davies, Palmer, & Halberstadt,2008). In addition, ex-offenders often lack sufficienthuman and social capital to help them navigate Ufeoutside of prison (Wilson & Davis, 2006). Manylack a high school diploma or employable skills,and others stmggle with mental health or substanceabuse problems (Lewis, Garfinkel, & Gao, 2007).Likewise, many offenders in the communityexperience feelings of depression and disrupdonwhen transidoning from prison Ufe to Ufe outsideof prison, and these feeUngs may contribute to highrates of recidivism, especially among repeat offend-ers (Arrigo & Takahashi, 2008; Petenilia, 2003).

To add to these chaUenges, fomier inmatesgenerally retum to urban communities with con-centrated social, economic, and poUdcal Stressorssuch as high unemployment, active drug markets,Umited social services, high crime, endangeredpubUc health, and homelessness (Braga et al.,2009; Katel, 2009; O'Brien, 2009; Seiter &Kadela, 2003; Zhang et al., 2006). Braga et al.(2009) identified several community-level factorsthat affect successful transitions, including theavailability of housing, substance abuse treatment,behavioral and physical health services, and accessto educadon and employment opportunities.Ex-offenden also face legal barriers to receivingpubUc services, such as bans on pubUc assistancereceipt, pubUc housing restricdons and Umitedtransitional housing options, and difficulty obtain-ing state-issued idendficadon (Wheeler & Patter-son, 2008). Finally, ex-offenders experience stigmafrom family, friends, prospecdve employers, andothers because of their criminal backgrounds(Wilson & Davis, 2006).

RATIONALE FOR AND DEFINITION OF REENTRY

PROGRAMS

Compounding the reentry challenges that manyex-offenders face, most do not receive assistance—either pre- or postrelease—to prepare them for re-turning to the community. Although most offend-ers have minimal education or job skills, justone-third of all prisoners released receive vocadon-al or educadonal training whue in prison. Three-quarters of aU inmates abuse substances, but onlyone-fourth participate in substance-abuse program-ming whue incarcerated (O'Brien, 2009; Petersüia,2003). Given the associations between low educa-tional attainment and crime and between substanceabuse and crime, the lack of programming inprison means that most offenders leaving prison areUkely to recidivate quickly upon release.

Most correcdonal departments and communityorganizations now recognize that addressing ex-offender needs may reduce recidivism, leading or-ganizations across the United States to developprisoner reentry programs. The Second ChanceAct of 2007 (P.L. 110-199) provides addidonal in-centives for correctional agencies and community-based organizations to develop prisoner reentryprograms. Signed into law on April 9, 2008, theSecond Chance Act authorized federal grants togovemment agencies and nonprofit organizationsto provide employment assistance, substance abusetreatment, housing, family programming, mentor-ing, and other ser/ices to help reduce recidivism.

Most programs exist at the community level toserve a specific cUentele v/ithin the criminal jusdcesystem, so no simple definition exists to describe thefull range of prisoner reentry programming (PetersiUa,2004; Wilson & Davis, 2006). For example, PetersiUa(2004) and Wheeler and Patterson (2008) argued thatreentry includes all acdvides and programming thatprepare ex-convicts to retum safely to the communi-ty as law-abiding cidzens. Seiter and Kadela (2003)defined reentry more narrowly to include only pro-grams that specifically focus on the transidon fromprison to community or that inidate treatment in aprison setdng and Unk with a community programto provide continuity of care.

REENTRY PROGRAM DESIGN AND

EFFECTIVENESS

Reentry programs differ considerably in structure,services provided, and cUents served, though mostbegin working with offenders before they are

290 Social Work Research VOLUME 36, NUMBER 4 DECEMBER 2012

released (Katel, 2009; Wheeler & Patterson, 2008;Wilson & Davis, 2006). Community-based, prison-based, and parole-based programs exist, as doprograms that combine prison, parole, and com-munity services. In this way, even those releasedfrom jail or prison without parole supervision canreceive services. Some programs assist participantswith one specific need, such as employment,housing, or substance abuse education, whereasother programs offer multiple services to meetparticipants' self-identified needs. Program lengthalso varies widely, extending from weekend-longskills classes to intensive case management overseveral years (Wheeler & Patterson, 2008; Wilson& Davis, 2006).

Although variety compUcates comparisonsacross programs (PetersOia, 2004), evaluations haveshown mosdy positive outcomes. Successful pro-grams incorporate intensive behavioral and cogni-tive approaches to encourage prosocial behaviorwhile program staff provide support and encour-agement to reinforce offenders' changed Ufestyles(Gendreau et al., 1996). Zhang et al. (2006)argued that reentry programs are most successfralwhen they appropriately match services to offend-ers' needs, especially the needs of those who facethe highest risk of recidivism.

Seiter and Kadela (2003) reviewed 32 pubUshedstudies and concluded that vocational trainingand work release programs effectively reduced re-cidivism and improved job readiness skills, whuedrug treatment reduced dmg use, recidivism,drug-related crimes, and parole violations. Educa-tional programs increased educational achieve-ment scores but did not decrease recidivism.Halfway house programs reduced severe criminalbehavior, and prerelease programs reduced recidi-vism (Seiter & Kadela, 2003).

The Boston Reentry Initiative—which pro-vides mentoring, social service assistance, andvocational development to jailed violent adult of-fenders—significandy reduced overall and violentarrest failure rates by 30% (Braga et al., 2009).Bouffard and Bergeron (2006) evaluated the smallSerious and Violent Offender Reentry Initiativeprogram in the upper Midwest, comparinginmates receiving enhanced reentry services witha sample of similar prisoners receiving only tmdi-donal prison or parole services. The reentry program

provided more referrals to community-based servicesthan did the prison or parole services, and it alsoincreased drug-tesdng frequency during parole.Reentry participants were less Ukely to test positivefor drug use while on parole, and they were 60% lesslikely to be arrested after parole than were those inthe comparison sample. Parole revocadon rates werecomparable for each group, however (Bouffard &Bergeron, 2006).

FinaUy, in the late 1990s, California legislatorsfunded the Preventing Parolee Crime Program(PPCP), a statewide, community-based correc-tional program intended to reduce parolee recidi-vism (Zhang et al., 2006). Parolees receivedUteracy training, employment services, housing as-sistance, and substance abuse treatment. Zhanget al. (2006) found that PPCP modesdy reducedreincarceration and parole absconding, potentiallycreating substantial long-term cost savings for Cal-ifornia taxpayers.

The aforementioned programs predominatelyserved offenders on release from prison, but manyreentry programs occur in prison before offendersare released. Jensen and Kane (2010) evaluated anin-prison therapeutic community that helped pris-oners develop healthy frinctioning, skills, andvalues as well as improve their physical and emo-tional health. Comparing program completers toeUgible nonparticipants and dropouts, Jensen andKane (2010) found that participation in the thera-peutic community delayed time to first rearrest byup to two years foUowing release from prison.

Not all reentry programs achieve their statedgoals, however. Wilson and Davis (2006) evaluat-ed Project Greenlight, a short-term, prison-basedreentry program. Survival analysis showed thatparticipants performed significandy worse onmultiple recidivism measures after one year, andmultivariable analyses indicated that covariatesfailed to mediate the observed relationships. Thenegative outcomes might have resulted from im-plementation difficulties, faulty program design,or a mismatch between the targeted offender pop-ulation and the program. Existing reentry evalua-tions suggest that although reentry assistance oftenimproves reentry outcomes for participants,poorly designed or implemented programs mightnot reduce, and may even increase, recidivism risk(Wilson & Davis, 2006).

WIKOFF ET AL. / Recidivism among Participants of a Reentry Pro-am for Prisoners Released without Supervision 291

PROJECT RE-CONNECT

This article focuses on Project Re-Connect (PRC),a reentry program that assisted offenders returningto the city of St. Louis, a large city in Missouri. Forsix months following participants' release from acompleted state prison term, the program providedcase management and direct monetary support toindividuals no longer under criminal justice super-vision. Prospecdve PRC cHents had to have beenconvicted in the state circuit court Located inSt. Louis; have had a primary address in St. Louiswhen they entered prison; and not have been re-leased under parole supervision, having served theirmaximum prison sentence.

PRC casé managers worked with participantsas they finished their prison sentences to idendfypostrelease needs and develop reentry plans. Theprogram provided up to $5,000 in funding foreach client, $2,000 of which reimbursed agenciesfor providing case management. Clients receivedthe remaining $3,000 during the six-monthprogram, in the form of bus passes, gift cards togrocery or clothing stores, payments for subsidizedor transitional housing, substance abuse treatment,and job and skills training programs. Each partici-pant worked with a case manager to determinehow to allocate frinds, and participants couldapply for additional money beyond the inidal$3,000 allotment, pending approval from agencystaff. AddidonaL informadon about PRC can befound elsewhere (Morani, Wikoff, Linhorst, &Btatton, 201L).

The monetary stipends were a unique programelement not found in most reentry programs.OnLy two other evaLuated reentry programs.Living Insurance for Ex-Offenders (LIFE) and theTransidonaL Aid Research Project (TARP), pro-vided monetary compensation to released offend-en. Evaluations of these programs provide mixedsupport. Former prisoners in the LIFE experimentwho received financial aid were significandy lessLikely to be arrested for theft crimes than werethose who received only employment counselingand placement or no assistance at all (MaLlar &Thornton, L978). In a similar manner, cash assis-tance receipt reduced both property and non-property arrests in the TARP experiment, butthese reductions were offset by program condi-tions that created large work disincentives amongtreatment participants. The TARP experiment in-troduced high tax rates (between 25% and L00%)

on legitimate earnings for participants who re-ceived cash assistance. These tax rates reducedemployment and led to increased property andnonproperty arrests. The findings suggest that cashassistance may reduce recidivism as long as it doesnot reduce employment, but no recent researchhas examined how current programs might usemonetary assistance to promote successful reentry(Berk, Lenihan, & Rossi, 1980).

FOCUS OF THE CURRENT STUDY

This quasi-experimentaL study addressed three re-search questions; First, what was the recidivismrate of PRC cHents? Second, how did that ratecompare with that of offenders who were eLigiblefor assistance but chose not to participate? Third,what factors were associated with new convictionsfollowing release from prison? This article makestwo contributions to the Literature on prisonerreentry and recidivism. First, it examines whethera reentry intervention that included direct finan-ciaL assistance was associated with reduced recidi-vism among participants. Second, it examinesrecidivism predictors among maxed-out offenders,a segment of the reentering prison population thatmay differ from that of those reLeased on parolebut on which there has been htde prior research.

METHOD

SampleThis study included two groups of inmates re-Leased from Missouri Department of Corrections(MDOC) prisons. The PRC participant group in-cluded 122 inmates who entered the programafter its start on March 1, 2007, and who hadbeen released from prison by February 1, 2008.Most PRC clients (rt=L08) were released fromprison during this time frame, whereas the other14 participants were released between June 6,2006, and Januar)' 30, 2007. The nonparticipantgroup included 158 eligible offenders releasedfrom MDOC prisons between March 1, 2007,and February 1, 2008, who chose not to partici-pate in the program.

AH eligible offenders were included except oneNadve American, who was removed from theanaLysis to restrict the sample to Caucasians andAfrican Americans. Across both groups, mean agewas 37.2 years {SD = 9.8) and ranged from 22.0to 70.3 years. Ten percent were female, 73.2%

292 Social Work Research VOLUME 36, NUMBER 4 DECEMBER 2012

were African American, 47.3% did not have ahigh school diploma or its equivalent, and 35.6%had entered prison for violent offenses. The char-acteristics of both groups are detailed in Table 1.

Data SourcesThis study used data obtained from MDOC andthe PRC coordinating agency. MDOC provideddemographic, assessment, and reconviction datafor PRC chents and nonparticipants from its in-temal prisoner database. PRC staff provided infor-mation about participants' entry and exit datesfrom the program. MDOC data were missing for12 inmates for selected variables. This includedone to four missing variables for four nonpartici-pants and one to two missing variables for eightparticipants. Bivariate and multiple regressionanalyses determined that observadons weremissing at random, so we used listwise deletion toremove observations with missing data. Prior toobtaining confidential and sensitive data, theprogram evaluation team received approval forthe study from the Saint Louis University institu-tional review board and the participating agencies.

VariablesDependent Variable. Using MDOC convictiondata, recidivism measured whether offenders wereconvicted of a state-level crime that resulted in anew sentence of probation or incarceration inprison by the end of the period of observation,October 16, 2009. New probation or convictionsentence was coded as 1 for respondents whowere convicted of a new prison or probationoffense between time of release from prisonthrough October 16, 2009, and as 0 for offenderswho were not convicted during that same timeframe. This operational definition excluded con-victions for offenses that resulted in fines or jaüterms, thus undercounting the number of offensesresulting in convictions. We attempted unsuccess-fully to obtain rearrest or complete convictiondata from other sources, so the MDOC data pro-vided the most reliable recidivism data available.

Independent Variables. Participation measuredwhether individuals received any form of programassistance upon release (1 = participant, 0 =eligible nonparticipant). Most participants (71.3%)received the full six months of case management,with 3.3% receiving five months, 8.2% receivingfour months, 9.8% receiving three months, 4.9%

Table 1: Participant and NonparticipantCharacteristics

Age (in years) at time ofrelease*M

SDGender (%)"

MaleFemale

Race (%)African AmericanCaucasian

Education (%)* (missingdata = 3)Less than HS

diploma/no GEDHS diploma or GED

Prior offense (%)(missing data = 2)ViolentNonviolent

DOC institutional riskscore (%) (missingdata = 6)Acceptable

institutionaladjustment

Some conductviolations

Moderate number ofviolations

Significant numberof violations

One or more veryserious violations

M

SDDOC mental health

score (%) (missingdata = 6)

No current MHtreatment needs

Mild level of MHtreatment needs

Moderate level ofMH treatmentneeds

Serious impairmentdue to MHdisorderM

SDDOC substance abuse

score (%) (missingdata =5)No apparent

substance abusetreatment needs

[PärticipantsINonparticipants;|

38.960.84

82.4617.54

72.81

27.19

38.6061.40

40.3559.65

63.16

14.04

10.53

6.14

6.141.780.11

57.89

29.82

12.28

0.001.540.06

6.14

35.930.83

94.81

5.19

72.7327.27

53.2546.75

31.1768.83

56.49

12.34

10.39

7.14

13.642.090.12

68.83

18.18

11.69

1.30

1.450.06

5.19

WiKOFF ET AL. / Recidivism among Participants of a Reentry Program fir Prisoners Released without Supervision 293

Table 1: Continued

LVariabLe^^H^H; Episodic suhstance

abuse

; Mild dependence orsubstance abuse

Moderate dependenceor substance abuse

Severe dependence orsubstance abuseM

SD

KarticípantsJ

8.77

24.56

54.39

6.14

3.46

0.09

[Nonparticipants

9.74

30.52

41.56

12.993.470.08

Notes: A/=280. HS = high schcoi; GED = gener3i equivatency diploma; DOC = Departmentof Corrections; MH = mentai heaith.*p<.05. * *p<.01.

receiving two months, and 2.5% receiving onemonth. Demographic vadables included gender(1 = male, 0 = female), race (1 = African Ameri-can, 0 = Caucasian), education (1 = less than highschool diploma, 0 = high school diploma or itsequivalent), and age at time of release fromprison.

The severity of the crime for which participantsserved the complete sentence, and for which theywere released at the start of this study, was codedas 1 = violent, 0 = nonviolent. Three risk scoresmeasured during imprisonment captured institu-tional risk, substance abuse severity, and mentalhealth needs. Institutional tisk scores ranged from1 for acceptable institutional adjustment to 5 formajor conduct violations. Mental health scoresranged from 1 for no current mental health needsto 5 for severe functional impairment due tomental health needs. Finally, substance abusescores ranged from 1 for no apparent substanceabuse needs to 5 for severe dependence or sub-stance abuse. Mean scores for institutional prob-lems, substance abuse problems, and mentalhealth problems were 2.0 (SD= 1.4), 3.5(SD= 1.0), and 1.5 (SD = 0.7), respectively.

Data AnalysisWe fint used chi-squares and t tests to determinewhether program clients and nonparticipants dif-fered significantly across control variables. Second,we used chi-square analysis to determine differ-ences in reconviction rates between program par-ticipants and nonparticipants. Finally, we usedsurvival analysis (SPSS Cox regression) to examinewhether program participation was associatedwith reduced recidivism among participants, whenother variables in the model were controlled for.

Cox regression is similar to logistic regression inthat it calculates the odds that an event wül occurwhue also accounting for the differing lengths oftime that individuals are exposed to the risk. Indi-viduals who expetience the event during the ob-servation pedod are determined to be failures,whereas those who do not expedence the eventduring that time pedod are censored. Censoringindicates that the event may stUl occur for those in-dividuals, even though it did not occur dudng theobservation pedod.

In this study, â̂i'/Mre meant any new convictiondudng the pedod of observation, including pdsonand probation offenses. The time at dsk of recon-viction extended from the date that offendenwere released from pdson untu the end of the ob-servation pedod on October 16, 2009. Using thedichotomous vadable for any new probation orpdson sentence, we calculated the time at dsk asthe difference between date of new convictionand previous date of release from pdson for thosewho were convicted. Individuals who were notconvicted of any new charges by the end of theobservation pedod were given October 16, 2009,as an end date.

Offenden were released from June 6, 2006,untU February 2, 2008, so the pedod of observa-tion was longer for some individuals than forother individuals; despite this, the pedod of obser-vation for each individual did not begin until thedate that they were released, so none of the obser-vations were left-censored. Right-censodng didoccur for individuals who had not been convictedon new charges before the end of the observation.None of the individuals in the study wereremoved from the sample dudng the study timepedod for reasons other than new conviction.Cox regression analysis assumes that the hazardratio remains constant over time (the proportion-ality assumption). We used the Kaplan-Meierprocedure to confirm that the survival plots metthe proportionality assumption.

RESULTS

Descriptive Comparison of Participantsand NonparticipantsProgram participants were on average older thannonparticipants (39.0 and 35.9 yean, respectively),were much more likely to be female than werenonparticipants (17.5% and 5.2%, respectively).

294 Social Work Research VOLUME 36, NUMBER 4 DECEMBER 2012

and were more Ukely to have received a highschool diploma or its equivalent than were non-participants (61.4% and 46.8%, respectively).There were no statistically significant differencesbetween the two groups for race; committingoffense; or institutional risk; mental health, andsubstance abuse scores. A comparison of charac-teristics for PRC participants and nonparticipantsis presented in Table 1.

Recidivism Outcomes Comparison ofParticipants and NonparticipantsBy the end of the observation period, 20.3% ofnonparticipants and 7.4% of participants had beenconvicted of new charges [x^{l, A/=280) = 9.1,p= .003]. The higher rate of recidivism amongnonparticipants held even when we controlled forother factors in the Cox regression analysis. Theoverall regression model was stadsdcaUy significant[X^{9, N = 268) = 21.1, p=.O12]. Participationwas associated with a 42.2% reducdon in the overallconvicdon hazard rate {p = .038).

Factors Associated with RecidivismIn bivariate analyses, offenders convicted of newcharges during the study were less Ukely to haveeamed a high school cUploma or graduate equiva-lency diploma (36.8% and 55.7%, respectively),less Ukely to have an acceptable institutionaladjustment (47.4% and 61.3%, respectively), andmore Uke to have a severe substance dependenceor abuse problem (23.7% and 7.8%, respectively).Detailed bivariate findings are presented inTable 2.

In the Cox regression, in addition to programparticipation, higher substance abuse risk scoreswere associated with increased UkeUhood of re-conviction. For each increase in level of substanceabuse severity, the subsequent overall convictionhazard rate increased by 46.6% [p = .049). Finally,education was found to be marginally significant,as the new conviction hazard rate was 104.8%higher for those with less than a high schooldiploma {p = .056). Complete results of the Coxregression analysis are shown in Table 3.

DISCUSSION

The results showed that program participants wereconvicted on new charges at lower rates thannonparticipants, even when we controlled for

Table 2: Factors Associated with NewConviction

NoVariableParticipation status**

ParticipantNonparticipant

Age at time of releaseM

SD

Gender (%)

MaleFemale

Race (%)African AmericanCaucasian

Education (%)* (missingdata = 3)Less than HS diploma/no

GEDHS diploma or GED

Prior offense (%)ViolentNonviolent

DOC institutional risk score(%)* (missing data = 6)Acceptable institutional

adjtistmentSome conduct violationsModerate number of

violations

Significant number ofviolations

One or more very seriousviolationsM

SD

DOC mental health score(%) (missing data = 6)No current MH treatment

needsMild level of MH

treatment needsModerate level of MH

treatment needsSerious impairment due to

MH disorderM

SD

DOC substance abuse score(%)* (missing data = 5)

No apparent substanceabtise treatment needs

Episodic substance abuseMild dependence or

substance abuse

Conviction

7.38

20.25

35.891.86

97.372.63

28.95

71.05

63.1636.84

23.6876.32

47.37

15.79

23.68

2.63

10.532.130.22

71.05

18.42

7.89

2.631.420.12

0.0013.16

21.05

Conviction

92.62

79.75

37.44

0.63

88.2611.74

26.9673.04

44.3555.65

36.9663.04

61.3012.61

8.26

7.39

10.431.930.09

63.04

23.91

12.61

0.431.50

0.05

6.528.70

29.13

WiKOFF ET AL. / Recidivism among Participants of a Reentry Program for Prisoners Released without Supervision 295

Table 2: ContinuedNo

VariableModerate dependence or

substance abuseSevere dependence or

substance abuseATSD

Conviction

42.11

23.683.760.16

Conviction

47.83

7.833.42

0.06

of Corrections; MH = mental health.*p<.os.*^p<.o^.

Table 3: Cox Regression Analysis: FactorsAssociated with Recidivism

VariableHazardratio SE

Client statusAge at releaseMaleAfrican

American

Less than HSdiploma

Violentoffense

Institutionalrisk score

Substanceabuse score

Mentalhealth score

0.42

1.00

3.35

0.66

2.05

0.71

1.16

1.47

0.90

0.180.02

3.51

0.26

0.77

0.28

0.14

0.29

0.23

-2.08*

0.051.16

-1.07

1.91

-0.88

1.20

1.96*

-0.41

0.19,0.95 21.11*0.96, 1.040.43, 26.07

0.31, 1.42

0.98, 4.28

0.32, 1.54

0.91, 1.47

1.00,2.15

0.55, 1.49Note: A'=268. CI = confidence interval; HS = high school.•p<.05.

significant differences in observed baseUne charac-teristics of participants and nonparticipants. Thequasi-experimental research design, and resultingsignificant differences between groups, means thatwe cannot rule out the influence of penonal mo-tivation on reentry outcomes, but the results dosuggest that reentry assistance may significandyreduce recidivism among offenders who completetheir sentences in prison. The personalized casemanagement and cash assistance may havereduced recidivism among participants by helpingthem navigate the reentry process. This repUcatesfindings from other reentry evaluations that haveshown that reentry assistance reduces recidivismamong participants (Bouffard & Bergeron, 2006;Braga et al., 2009; Seiter & Kadela, 2003).

This analysis also found that substance abuse wasassociated with increased risk of reconviction atthe bivariate and multivariable levels, as individuals

who had some history of substance abuse weremore Ukely to be convicted on new charges. Noneof the non-substance-abusing individuals wereconvicted of new crimes, whereas higher substanceabuse scores were associated with receiving a newconviction during the study period. When wecontrolled for other variables, the rate of new con-viction was elevated for individuals with a historyof substance abuse. Although üUcit substance useitself could explain offenders' increased risk of newconvictions, substance abuse might also have fos-tered criminal engagement. For example,substance-abusing offenders might have been moreUkely to commit crimes while under the influenceof, or as a strategy by which to obtain, drugs(French, Fang, & Fretz, 2010; Lipton, 1995).

At the bivariate level, this analysis found thatindividuals with less education and those withhigher institutional risk or substance abuse scoreswere more likely to be convicted on new charges.Those with less than high school education werenearly 1.5 times as Ukely to be convicted on a newcharge during the study time frame as those with atleast a high school diploma or its equivalent. Thisassociation remained marginally significant whenother variables were controlled for. It may havebeen that in contrast with less educated offenders,those who had higher educadon enjoyed morelegitimate employment opportunities upon release,perhaps weakening their attachment to illicit activi-ries (Harrison & Schehr, 2004).

A higher percentage of those who were notconvicted had adjusted appropriately to institu-tional conditions, whereas those with a moderatenumber of conduct violations were much moreUkely to be convicted again during the observa-tion period. Individuals who had higher institu-tional risk assessment scores may have been moreUkely to engage in ongoing antisocial behaviorafter release, thereby impeding their reentry pros-pects (MiUer, 2006).

Limitations and Areas for Future ResearchThis quasi-experimental study design compared el-igible participants with eUgible nonpardcipants, sowe cannot rule out the possibüity that self-selectioninto treatment explains the lower rate of new con-victions among participants. Future studies, whenpossible, should use randomized or matched re-search designs to minimize the chance that self-selection bias wül result in differences between the

296 Social Work Research VOLUME 36, NUMBER 4 DECEMBER 2012

treatment and control groups. Further research isalso needed to address gender-specific reentryneeds, as women were overrepresented in the par-ticipant group. At state and national levels, womenhave increasingly entered the criminal justicesystem over the last two decades. Not only is thesize of the female prison population growing at afaster rate than that of the male prison population,the percentage of female offenders returning toprison on technical or new violations is increasingat a higher rate than the percentage of new admis-sions for women (MDOC, 2009). More research isneeded to determine whether adequate programsexist for female offenders, whether pardcularprogram components are equally effecdve forwomen and men, and whether gender differencesexist for factors associated with program participa-tion, successflil program completion, and criminalrecidivism (Scroggins & MaLley, 20L0; Spjeldnes &Goodkind, 2009).

This program included monetary assistance andindividualized case management, components notfound in most reentry programs. Unfortunately,we couLd not test the effects of these two programcomponents, as both variables were higLiLycoOinear with participation status: OnLy 29% ofparticipants did not take advantage of the fiiLl sixmonths of case management, and more than haLfused at Least $2,850 of the $3,000 provided forthem. Even when the sampLe was limited toparticipants, the low rate of convictions for partic-ipants prevented us from examining the uniqueeffects of these program components on reentryoutcomes.

These results may not be generalizable to thebroader prison population, as a result of P R C eH-gibüity restrictions and demographics of the Mis-souri prison population. Fint, this study includedonly African Americans and Caucasians, as therewas only one Native American prisoner in thesample, so these results may not be generaHzableto prisoners from other racial or ethnic back-grounds. Second, participation in P R C was re-stricted to individuals who had completed theirsentences in prison, so these results may not applyto prisoners released from prison on parole. Wecould not find state- or nationaL-LeveL statistics onthe characteristics or recidivism rates of prisonenwho complete their sentences in prison. More re-search is needed to examine whether offenderswho complete their maximum sentence in prison

exhibit unique characteristics from other offendersor face distinct reentry chaUenges.

Finally, we had limited access to data for non-participants following release. Results may havebeen different had data been collected on theseprisoners at the time of release from prison.Future reentry evaluations should include variablesthat capture changes in participants and nonpartic-ipants following release (Draine, Wolff, Jacoby,HartwelL, & Duelos, 2005).

Implications for Social WorkThese results have implications for social workpractice at the micro and macro levels. At thedirect service level, many offenders have multipleservice needs, including transportadon, clothing,food, identification documents, housing, educa-tion, employment, heath, mental health and sub-stance abuse treatment, and support systems(Latdmore et al., 2012; La Vigne et al., 2008).SociaL workers who provide case management tooffenders play a crucial role in assessing serviceneeds and Linking offenders to services, especiallyfor offenders who are released from prisonwithout the supervision and assistance of paroleofficers. At the community level, social workerscan develop prison reentry programs and helpcreate services in communities that have the great-est need (Travis, 2005)

Reentry programs have implications for socialworkers at the policy level as weLl. With passageof the Second Chance Act, sociaL workers shouldhelp develop reentry programs eligible for thisfrinding while actively working at the state andfederal levels to rescind policies that restrict of-fenden' access to needed services. Such restric-tions exist in public and private housing, publicassistance programs, education, employment, andvoting (National Governors Association, 2005;Petersilia, 2003; Pogorzelski, Wolff, Pan, & BLitz,2005; Travis, 2005).

The resuLts presented in this article suggest thata combination of personalized case managementand financial assistance can help offenders reinte-grate into society and avoid returning to crime.Predictor variables that are commonly associatedwith recidivism but were not so associated in thisstudy suggest that maxed-out offenders may displayunique risk characteristics distinguishing them fromother offenders returning home from prison. Socialworkers and othen who work with reentering

WIKOFF ET AL. / Recidivism among Participants of a Reentry Program for Prisonen Released without Supervision 297

offenden should carefiiUy consider the uniqueneeds of the clientele they serve when designingreentry programs. SSS3

REFERENCESArrigo, B. A., & Takahashi, Y. (2008). Theorizing com-

munity reentry for male incarcerates and confinedmothers: Lessons leamed ftom housing the homeless.Journal of Offender Rehabilitation, 46(1/2), 133-162.

Berk, R. A., Lenihan, K.J., & Rossi, P. H. (1980). Crimeand poverty: Some experimental evidence fromex-offenders. American Sociological Review, 45,766-786.

Bouffard, J. A., & Bergeron, L. E. (2006). Reentry works:The implementation and effectiveness of a seriousand violent offender reentry initiative. JoMma/ of Of-fender Rehabilitation, 44(2/3), 1-29.

Braga, A. A., Piehl, A. M., & Hureau, D. (2009). Control-ling violent offenders released to the community: Anevaluation ofthe Boston Reentry Initiative. Jowma/ ofResearch in Crime and Delinquency, 46, 411—436.

Bureau of Justice Statistics. (2010). Key facts at a glance: Cor-rectional populations. Retrieved ftom http://bjs.ojp.usdoj.gOv/content/glance/tables/con:2tab.cfm

Draine, J., Wolff, N., Jacoby, J. E., Hartwell, S., & Duelos,C. (2005). Understanding community re-entry offormer prisoners with mental illness: A conceptualmodel to guide new research. Behavioral Sciences andthe Law, 23, 689-707.

French, M. T., Fang, H., & Fretz, R. (2010). Economicevaluation of a prerelease substance abuse treatmentprogram for repeat criminal offenders. JoHma/ of Sub-stance Abuse Treatment, 38{\), 31-41.

Gendreau, P., Litde, T., & Goggin, C. (1996). A meta-analysis ofthe predictors of adult offender recidivism:What works! Criminology, 34, 575-607.

Harrison, B., & Schehr, R. C. (2004). Offenders and post-release jobs: Variables influencing success and failure.Journal of Offender Rehabilitation, 39(3), 35-68.

Huebner, B. M., Varano, S. P., & Bynum, T. S. (2007).Gangs, guns, and drugs: Recidivism among serious,young offenders. Criminology & Public Policy, 6,187-221.

Jensen, E. L., & Kane, S. L. (2010). The effect oftherapeutic community on time to first rearrest: Asurvival anaiyui. Journal of Offender Rehabilitation, 49,200-209.

Katel, P. (2009, December 4). Prisoner reentry. CQ Re-searcher, 19, 1005-1028.

Langan, P., & Levin, D. (2002). Recidivism of prisonersreleased in 1994. Bureau of Justice Statistics SpecialReport. Washington, DC: U.S. Department of Justice,Bureau of Justice Statistics.

La Vigne, N., Davies, E., Palmer, T., & Halberstadt, R.(2008, September). Release planning for successfulreentry: A guide for corrections, service providers, andcommunity groups (Urban Institute Justice PolicyCenter Research Report). Washington, DC:Urban Institute.

Lattimore, P. K., Barrick, K., Cowell, A., Dawes, D.,Steffey, D., TueUer, S., & Visher, C. A. (2012, April).Prisoner reentry services: What worked for SVORJ evalua-tion participants? Final report. Retrieved ftom https://www.ncj rs.gov/pdffilesl/nij/grants/238214.pdf

Lewis, C. E., Jr., Garfinkel, I., & Gao, Q. (2007). Incarcer-ation and unwed fathers in fragile families. Journal ofSociology & Social Welfare, 34(3), 77-94.

Lipton, D. S. (1995). The effectiveness of treatment for drugabusers under criminal justice supervision (National

Institute of Justice Report 157642). Retrieved fromhttp://www.ncjrs.gov/pdffiles/drugsupr.pdf

Listwan, S. J. (2009). Reentry for serious and violent of-fenders: An analysis of program attrition. CriminalJustice Policy Review, 20, 154-169.

Mallar, C. D., & Thornton, V. D. (1978). Transitional aidfor released prisoners: Evidence ftom the Hfe experi-ment. Journa/ of Human Resources, 13, 208-236.

Miller, H. A. (2006). A dynamic assessment of offenderrisk, needs, and strengths in a sample of pre-releasegeneral offenders. Behavioral Sciences and the Law, 24,767-782.

Missouri Department of Corrections. (2009). A profile oftheinstitutional and supervised offender population on June 30,2009. Retrieved from http://doc.mo.gov/documents/publications/Offender%20Profile%20FY09.pdf

Morani, N. M., Wikoff, N., Linhorst, D. M., & Bratton, S.(2011). A description ofthe self-identified needs,service expenditures, and social outcomes of partici-pants of a prisoner reentry program. Prison Journal, 91,347-365.

National Governors Association. (2005, September). Issuebrief: Improving prisoner reentry through strategic policy in-novations. Washington, DC: National Governors As-sociation Center for Best Practices.

O'Brien, P. (2009). Reentry in the twenty-first century.In T. Maschi, C. Bradley, & K. Ward (Eds.), Forensicsocial work: Psychosocial and legal issues in diverse practicesettings (pp. 275—287). New York: Springer.

Petersilia, J. (2003). When prisoners come home: Parole andprisoner reentry. New York: Oxford University Press.

Petersilia, J. (2004). What works in prisoner reentry? Re-viewing and questioning the evidence. Federal Proba-tion, 68, 4-8.

Pogorzeiski, W., Wolff, N., Pan, K. Y., & BUtz, C. L.(2005). Behavioral health problems, ex-offenderreentry policies, and the "Second Chance Act."American Journal of Public Health, 95, 1718-1724.

Roman, C. G., & Travis, J. (2006). Where will I sleep to-morrow? Housing, homelessness, and the returningprisoner. Housing Policy Debate, 17, 389-418.

Sabol, W. J., West, H. C , & Cooper, M. (2009, Decem-ber). Prisoners in 2008 (Bureau of Justice StatisticsBulletin, NCJ 228417). Washington, DC: U.S.Department of Justice, Office of Justice Programs.

Scroggins, J. R., & Malley, S. (2010). Reentry and the(unmet) needs of women. JoMma/ of Offender Rehabili-tation, 49, 146-163.

Second Chance Act of 2007, P.L. 110-199, 122 Stat. 657(2008).

Seiter, R. P., & Kadela, K. R. (2003). Prisoner reentry:What works, what does not, and what is promising.Crime and Delinquency, 49, 360-388.

Spjeldnes, S., & Goodkind, S. (2009). Gender differencesand offender reentry: A review ofthe literature.Journal of Offender Rehabilitation, 48, 314-335.

Travis, J. (2005). But they all come back: Facing the challengesof prisoner reentry. Washington, DC: Urban InstitutePress.

Wheeler, D. P., & Patterson, G. (2008). Prisoner reentry.[National Health Line] Health & Social Work, 33,145-147.

Wilson, J. A., & Davis, R. C. (2006). Good intentionsmeet hard realities: An evaluation ofthe ProjectGreenlight Reentry Program. Criminology & PublicPolicy, 5, 303-338.

Yahner, J., & Visher, C. (2008, August). Illinois prisoners'reentry success three years after release. Washington, DC:Urban Institute Justice Policy Center.

Zhang, S. X., Roberts, E. R., & CaUanan, V. J. (2006).Preventing parolees from returning to pdson through

298 Social Work Research VOLUME 36, NUMBER 4 DECEMBER 2012

community-based réintégration. Crime & Delinquency,52,551-571.

Nora Wtkoff, MSLS, MSW, is a doctoral student, GeorgeWarren Brown School of Social Work, Washington Universityin St. Louis. Donald M. Linhorst, PhD, MSW, is profes-sor and director. School of Sodal Work, Saint Louis University.Nicole Morani, MSW, LMSW, is assertive communitytreatment team leader. Places for People, St. Louis. Addresscorrespondence to Nora Wikoff, George Warren Brown Schoolof Social Work, Washington University in St. Louis, CampusBox 1196, One Brookings Drive, St. Louis, MO 63130-4899; e-mail: nwikoff@go.wustl.edu.

Originai manuscript received Juiy 29. 2009Finai revision received November 23. 2010Accepted January 3, 2011Advance Access Publication December 27. 2012

Customize the Power of YourNASW Membership

CHILD WEUFAREecialty Practice Section

As a social worker in the field of child welfare,current informatioii on trends and policy issuesaffecting practice and service delivery is a must.NASW's Child Welfare Specialty Practice Section(SPS) links you with the key information,resources, and expertise you need to stay at theforefront of your practice specialty:

• Specialized newsletters• Practice updates• Opportunity for FREE CE credits by reading

Intersections in Practice, the SPS annualbulletin

• Members-only Web-based productsdesigned to enhance your professionaldevelopment within the child welfarecommunity

JOIN TODAY!

Go to www.socialworkers.org/sections or call202.408.8600 ext. 476.

*You must be o current NASW member to join o SpecialtyPractice Section.

#NASW.spsNational Association of Social WorkersSpeciolty Prociice Sections

www.socialworkers.org/sections202.408.8600 ext. 476

WIKOFF ET AL. / Recidivism among Participants of a Reentry Program for Prisoners Released without Supervision 299

Copyright of Social Work Research is the property of National Association of Social Workers and its content

may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express

written permission. However, users may print, download, or email articles for individual use.