restorative justice, communities, and delinquency: whom do we reintegrate?

\\server05\productn\C\CPP\4-1\CPP106.txt unknown Seq: 1 19-JAN-05 14:50

RESTORATIVE JUSTICE, COMMUNITIES,AND DELINQUENCY: WHOM DO WEREINTEGRATE?*

NANCY RODRIGUEZArizona State University

Research Summary:Communities represent an important facet of restorative justice; yet

few studies have empirically evaluated the influence that communitycharacteristics have on reintegrative programs. In this study, I use offi-cial juvenile court data from an urban, metropolitan county in Arizonaand Census data to examine how individual- and community-level datainfluence the selection of offenders to a restorative justice program. Toexamine the effectiveness of the restorative justice program and todetermine if program impact varies by community, recidivism ismodeled with individual- and community-level data. Findings indicateindividual and community characteristics are important predictors ofrestorative justice program placement. Also, juveniles who successfullycompleted the restorative justice program were less likely to recidivatethan were juveniles in a comparison group. Recidivism results show novariation of program impact across communities.

Policy Implications:Study findings demonstrate the importance of incorporating commu-

nity-level data in studies of restorative justice. Restorative justice pro-grams must be sensitive to local, community characteristics and ensurethat reintegration efforts do not exclude offenders in need. The success-ful reintegration process experienced by juveniles who participated inthe restorative justice programs is clearly reflective of programeffectiveness.

KEYWORDS: Restorative Justice, Community Justice, Juvenile CourtDecisions

Criminal justice systems have created and implemented various pro-grams based on the ideals of restorative justice. A fundamental compo-nent of restorative justice is the community’s capacity to successfullyreintegrate offenders back into their communities (Bazemore, 1999; Clearand Karp, 1999, 2000; Karp, 2001). Within this theoretical framework,

* Acknowledgment : Special thanks to Todd R. Clear and the three reviewers fortheir insightful comments on an earlier draft of this article.

VOLUME 4 NUMBER 1 2005 PP 103–130 R


104 RODRIGUEZ

community characteristics become especially important given the directrole community members play in restorative justice programs. Althoughresearchers have proposed that structural characteristics have an impor-tant impact on restorative justice and community justice programs, noprior research has incorporated community-level measures to the study ofsuch programs. This lack of research is especially surprising given thatcommunity characteristics remain significant predictors of crime, delin-quency, and recidivism (Bursik and Grasmick, 1993; Eliott et al., 1996;Gottfredson et al., 1991; Hoffman, 2003; Rountree et al., 1994; Sampsonand Groves, 1989) and have been shown to be relevant in decision-makingprocesses within the justice system (Bortner et al., 2000; Bridges andCrutchfield, 1988; Bridges et al., 1993; Britt, 2000; Frazier and Lee, 1992;Hawkins, 1993; 1999; Myers and Talarico, 1987; Peterson and Hagan, 1984;Sampson and Laub, 1993; Secret and Johnson, 1997; Wilson, 1987, 1996).Given that restorative justice and community justice programs alike relyon the established relationships found within communities to bring “jus-tice” to neighborhoods, examining community factors that are reflective ofa community’s capacity to provide such justice is central to a comprehen-sive understanding of restorative justice program processes andeffectiveness.

This study relies on multilevel linear modeling to analyze how individualand community indicators influence restorative justice processes andimpact. Juvenile court data from a restorative justice program in a metro-politan county in Arizona are used to explore the individual and commu-nity characteristics (e.g., heterogeneity, percent unemployed, percentSpanish-speaking households, and crime rate) of juveniles selected forrestorative justice participation. To identify the impact of restorative jus-tice on offenders, this study also compares the levels of recidivism betweenjuveniles in the restorative justice program and juveniles on standardprobation.

RESTORATIVE JUSTICE AND JUVENILEDELINQUENCY

Restorative justice can be characterized by dialogue (i.e., among a vic-tim, an offender, and community members), relationship building, and thecommunication of moral values (Presser and Van Voorhis, 2002). The dia-logue consists of open discussions on the harm caused by the offense andthe designation of an appropriate resolution that reflects the values of thecommunity (Clear and Karp, 1999). The inclusion of community membersin restorative justice processes empowers the entire community in that itallows local citizens to represent their neighborhood’s values and norms(Clear and Karp, 1999, 2000; Karp, 2001). Consistent with the ideals of


RESTORATIVE JUSTICE & COMMUNITIES 105

community justice, restorative justice programs must ensure that commu-nities have the capacity to recommend and provide resolutions for crimi-nal behavior. Existing ties between individuals and the ability to developwell-defined skills that offenders need to successfully reintegrate into theircommunities are some indicators of this capacity (Clear and Karp, 1999;Karp and Clear, 2002; Karp et al., 2002; Morris, 2002). In theory, the col-laborative effort between community members and criminal justice agen-cies produces an effective mechanism of crime reduction that may alsolead to the solving of other community/neighborhood problems.

Within juvenile courts, restorative justice programs hold juvenilesaccountable for their delinquent acts, aim to develop juvenile competen-cies, and protect the community (Bazemore and Griffiths, 1997; Bazemoreand Maloney, 1994; Hayes and Daly, 2003; Umbreit and Stacy, 1996;Umbreit and Zehr, 1996). Juveniles are held responsible for their acts andreleased back into their communities having received needed services suchas counseling, educational, and/or vocational training (Bazemore, 1992;Bazemore and Umbreit, 1995). Juveniles are then reintegrated into com-munities ready to contribute to the overall well-being of their community.

Juvenile delinquency cases have been processed through several typesof restorative justice programs including victim-offender mediation pro-grams, community reparative boards, circle sentencing, and family groupconferencing (Bazemore and Griffiths, 1997; Bazemore and Umbreit,2001; Maxwell and Morris, 1993; McElrea, 1996; Melton, 1995; Umbreit,2000; Umbreit et al., 2001). Although these programs represent the mostcommon restorative justice programs for juveniles, modifications to pro-grams have been encouraged to ensure that restorative justice meets theunique needs of particular communities (Bazemore and Umbreit, 2001).The various types of programs, share the common features of communityparticipation and the localization of the resolution process (Bazemore andGriffiths, 1997; Bazemore and Umbreit, 2001; Melton, 1995; Umbreit,1994; 1995; Umbreit and Coates, 1993; Umbreit et al., 2001).

To date, studies have produced mixed findings on the impact of restora-tive justice programs on recidivism.1 While some studies show lower ratesof recidivism among juveniles in restorative justice programs than offend-ers in comparison groups, such differences fail to achieve statistically sig-nificant levels (McCold and Wachtel, 1998; Niemeyer and Shichor, 1996;Umbreit, 1994; 1993). Other studies fail altogether in showing lower ratesof recidivism among juveniles in victim-offender programs (Roy, 1993).More recent studies have produced more favorable findings regarding theimpact of restorative justice programs on recidivism (Hayes and Daly,

1. For a comprehensive review of restorative justice studies, see Umbreit et al.,(2002).


106 RODRIGUEZ

2003; Luke and Lind, 2002; Maxwell and Morris, 2001; McGarrell, 2001;Nugent and Paddock, 1995; Nugent et al., 2001; Sherman et al., 2000). Forexample, an extensive meta-analysis of 35 restorative justice programs(i.e., 27 victim-offender meditation programs and 8 conferencing pro-grams) revealed that restorative justice programs were far more effectivein reducing crime than were traditional correctional supervision mecha-nisms (Latimer et al., 2001). Similarly, Nugent et al., (2001) reanalyzeddata from four studies of victim-offender mediation programs and foundthat juveniles who participated in such program were far less likely torecidivate than were juveniles in the comparison groups.

Restorative justice assumes a community responsibility for criminalactivity, while addressing the harm caused by the offender. However,community aspects can facilitate or hinder offenders’ reintegration pro-cess. Community characteristics such as crime, racial/ethnic composition,and economic resources can directly influence this reintegration process.For example, a juvenile offender from a community characterized by simi-lar norms and values may have an easier time reintegrating (e.g., remain-ing crime free) in their community than an offender from a sociallydisorganized community (Braithwaite, 1989; McElrea, 1996). Unfortu-nately, little is known regarding how community aspects influence restora-tive justice processes or how they impact future crime and delinquency.

THE INFLUENCE OF COMMUNITIES IN DECISION-MAKING PROCESSES, DELINQUENCY, AND

COMMUNITY JUSTICE

Community characteristics have been widely used by researchers instudies of adult and juvenile courts to explain the decision-makingprocesses of court officials (Bortner et al., 2000; Bridges and Crutchfield,1988; Britt, 2000; Myers and Talarico, 1987; Peterson and Hagan, 1984;Secret and Johnson, 1997). Most of these studies have incorporated com-munity-level data in an effort to explain the more severe punishmentreceived by minority offenders. For example, the perceived “dangerous-ness” of racial/ethnic groups who are viewed responsible for street-levelcrime may lead to more severe punishment (or formal social control) ofminority offenders, especially when their communities are characterizedby high levels of unemployment, crime, and racial/ethnic heterogeneity(Blalock, 1957; Bridges and Crutchfield, 1988; Frisbie and Neidert, 1977).In studies of juvenile court processing, community-level data such ascrime, unemployment, and poverty rate have been shown to significantlyinfluence detention, adjudication, and disposition decisions of white andnon-white juvenile offenders (Bridges et al., 1993; Frazier and Lee, 1992;Sampson and Laub, 1993; Secret and Johnson, 1997).



An extensive body of research has also examined the role of communitycharacteristics on crime and delinquency. Community characteristics haveprimarily explained the connection between race and crime, as well asoffending differences within and between communities. Researchers havefound direct and indirect relationships between crime and communitycomponents, including residential instability, family disruption, ethnic het-erogeneity, poverty, and urbanism (Bursik, 1988; Hagan, 1992; Hawkins,1993, 1999; Osgood and Chambers, 2000; Wilson, 1987, 1996). Character-ized as features of social or community disorganization, these structuraldimensions have been shown to mediate the relationship between individ-ual offenders and criminal activity.

Given the important role of community characteristics in court decision-making processes and crime, the inclusion of community factors in evalu-ating a restorative justice framework is appropriate. In fact, proponents ofrestorative justice and community justice have identified how communitycharacteristics influence the reintegration processes. Braithwaite’s (1989)indicates that “systematically blocked legitimate opportunities” makestructural economic dimensions (e.g., urbanization and high residentialmobility) instrumental in the shaming process (p. 97). Braithwaiteacknowledges the possible use of measures at the individual and commu-nity level in studies of offender reintegration. In fact, he notes it is appro-priate to create and use aggregate (community) measures from individual-level data in such studies.

Clear and Karp’s (1999) notion of community context within a commu-nity justice framework emphasizes the role of social class (i.e., disadvan-taged neighborhoods vs. economically and socially prosperouscommunities) in the reintegration process. Within this framework, it isrecognized that community context (e.g., high rates of poverty, unemploy-ment, single-parent households) cannot be ignored given the particularassociation such measures have with crime. Clear and Karp identify urbanchange and segregation as the two most important structural factors thataffect the quality of life. Further, they argue that community justice willnot succeed unless such community dimensions are addressed. It is possi-ble that those communities plagued by structural factors that limit thelegitimate opportunities of community members are the areas where com-munity justice may have the most impact on the overall quality of life(Clear and Karp, 1999).

LOCALIZING RESTORATIVE JUSTICE: THEMARICOPA COUNTY COMMUNITY

JUSTICE COMMITTEES

Consistent with the efforts of other criminal justice agencies to develop


108 RODRIGUEZ

restorative justice programs, the Maricopa County Arizona Juvenile Pro-bation Department created Community Justice Committees (CJC) in 1995to deal with an increasing juvenile crime rate. The purpose of this pro-gram is to divert juvenile offenders from formal court processing and bringthose cases to communities for resolution. A specific intent of the com-mittees is to work with the juveniles, family members, and the communityin an attempt to develop juveniles’ skills, restore a sense of communitythat was destroyed by the delinquent offense, and ultimately, holdjuveniles accountable for their actions. The committees follow a familygroup conferencing model in that victims, family members, and communi-ties actively respond to the delinquent offense in search of appropriateresolutions for offenders (Bazemore and Umbreit, 2001; Maxwell andMorris, 1993; 1997; McElrea, 1996; Morris et al., 1993; Umbreit, 2000).

Juvenile probation officers are responsible for overseeing the commit-tees that include two to four community members (i.e., volunteers), thevictim(s), family members, and the juvenile offender. The recruitment ofvolunteers is handled by the Maricopa County Juvenile Probation Depart-ment. The Department solicits volunteers who are then assigned to one ofthe CJCs and work alongside the CJC probation officer. Volunteers takepart in training that includes a review of the restorative justice model andthe “balanced approach,” which has been adopted by the juvenile court.Volunteers are responsible for reviewing the case file of participatingjuveniles with the juvenile and parents/guardians. During the conference,volunteers discuss the case with other panel members and determine anappropriate resolution to be met by the juvenile. The resolutions may beto include any of the following tasks: (1) restitution to the victim, (2) com-munity service, (3) fine, (4) counseling, and (5) educational sessions. Juve-nile offenders are given between 60 and 90 days to successfully completethe terms recommended by the committees.

Volunteers are also responsible for representing their neighborhoods inthe conference process. Most often, volunteers (i.e., community members)may reside in the same neighborhoods as the juvenile offenders. Toensure that the restorative justice program reaches varying neighborhoodsand communities, conferences are held in different locations throughoutthe county.

The placement of juveniles into the restorative justice program is basedon a decision-making process that involves juvenile probation staff and theMaricopa County Attorney’s Office. Officials from both agencies reviewall juvenile referrals and select those cases in which formal prosecutionwill be deferred if the juvenile is willing to participate in the diversionprogram. Upon selection, juveniles must accept responsibility for thedelinquent offense and agree to have their cases processed through therestorative justice program. Juvenile court personnel attempt to target



first- or second-time offenders and exclude sex offenders and violent fel-ony offenders from program participation.2 In cases in which a juvenilefails to successfully complete the terms designated by the resolution, a for-mal petition for the original delinquent offense is filed by the CountyAttorney’s Office.

THE PRESENT STUDY

Using 1999 through 2001 official juvenile court data, I examine theselection process and the impact of the restorative justice program onrecidivism in Maricopa County, Arizona. The lack of studies that haveincorporated individual- and community-based characteristics of offendersto examine restorative justice programs has been previously documented(Kurki, 2000). Virtually no information exists on the types of communitieswhere restorative justice programs have made the greatest impact. Bycapturing how community characteristics such as racial/ethnic heterogene-ity, unemployment, and delinquency rates impact the reintegration ofjuvenile offenders, this study will be able to empirically establish whetherrestorative justice efforts are most often localized in particular communi-ties and whether recidivism of program participants varies by communitytype.

Prior research has documented the relationship between communitiesand juvenile court processing. For example, geographic location of juve-nile courts has been shown to have an impact on the assessment ofjuveniles’ needs (Sanborn, 1994). Researchers have proposed that studiesof juvenile courts should also examine the economic and social conditionsthat surround juvenile courts (Crutchfield et al., 1994; Feld, 1999; Hagan,1994). In fact, some researchers have noted that the incorporation of suchindicators may be most appropriate in early juvenile court processes(Bortner et al., 2000). An examination of deferred prosecution (an initialdecision-making process) that includes individual and community charac-teristics is consistent with the recommendations presented in prior studiesof juvenile court processes. To date, no prior study has examined theselection process of juveniles in restorative justice programs. Interest-ingly, the selection process serves as an indicator of what particularjuveniles are deemed most appropriate by court officials for reintegration.Identifying how individual and community indicators influence restorativejustice processes can provide an empirical test of claims that propose these

2. Violent felonies include the following offenses: first-degree murder, second-degree murder, forcible sexual assault, armed robbery, any offense committed by achronic felony offender, aggravated assault, aggravated assault involving the use of adeadly weapon, drive-by shooting, and discharging a firearm.


110 RODRIGUEZ

programs select offenders least in need of community support and treat-ment (Kurki, 2000).

The inclusion of community-based data to assess recidivism of juvenilesin a restorative justice program is theoretically appropriate given the criti-cal role of communities in the reintegration process. Moreover, commu-nity-level measures such as urbanism, poverty level, and ethniccomposition of communities have been shown to be important predictorsof crime and delinquency (Bursik, 1988; Hawkins, 1993, 1999; Osgood andChambers, 2000; Wilson, 1987, 1996). Although the role of communitycharacteristics in the reintegration process is certainly essential, studies inthis area have yet to empirically analyze such measures. An examinationof community characteristics can reveal whether restorative justice pro-grams tend to be concentrated in specific communities (e.g., predomi-nantly white versus racially/ethnic heterogeneous communities) andwhether such communities have the “capacity” (e.g., high unemploymentversus low unemployment areas) to successfully reintegrate juvenileoffenders. One possible assumption is that the more homogeneous a com-munity, the easier the reintegration of offenders given the common valuesand norms of community members. Conversely, the more heterogeneous(e.g., economically, culturally, and racially/ethnically) a community, themore challenging the reintegration process given such differences.

The incorporation of community characteristics to the study of restora-tive justice has both pragmatic and theoretical relevance. Such an inclu-sion will not only reveal how juvenile court officials use community factorsin the selection of juveniles for reintegration but also whether programimpact varies across communities. Consequently, this study examines thefollowing research questions:

1. Is there as relationship among individual-level characteristics (e.g.,race/ethnicity, gender), community-level characteristics (e.g., eth-nic heterogeneity, unemployment rate, juvenile crime rate), andthe decision to select juvenile offenders for participation in arestorative justice program?

2. Is there a relationship between participating in a restorative justiceprogram and recidivism, net the effects of individual- and commu-nity-level characteristics?

DATA AND METHODOLOGY

Data for this study come from the Maricopa County Juvenile On-LineTracking System database. These data capture information on juveniles’court processing from the time the juvenile is referred to the juvenile court



until court disposition. To examine the selection of juveniles to the restor-ative justice program and program impact on recidivism, all juvenile refer-rals eligible for diversion from January 1999 through June 2001 wereexamined (N = 7,264). The first dependant variable, CJC (CJC = 1; otherdiversion program = 0), is measured by comparing those juveniles selectedfor the CJC program with offenders not selected for the program.Juveniles not selected for participation in the CJC program have theircases diverted through standard diversion mechanisms (e.g., paper cita-tion, standard diversion program).3 The second dependent variable ofinterest, recidivism (yes = 1; no = 0), was constructed by following offend-ers in both groups for a 24-month period after successfully completing pro-gram requirements. Any formal petition filed by the County Attorney’sOffice within the 24-month follow-up period constituted reoffending.4The data in this particular study are limited to those juveniles that com-pleted the diversion programs and exclude those who failed to complywith the terms of diversion (noncompleters).

At the individual level, both extralegal and legal variables are includedin the analyses. Demographic indicators consist of gender (boys = 1; girls= 0), race/ethnicity (dummy coded variables for Hispanic/Latinos andblacks with whites as the omitted category), and age at time of court refer-ral. Legal variables include the most serious offense at referral (i.e., prop-erty, status, and public order with person as the omitted category)5 andthe number of prior offenses in juveniles’ official court records. Controlsfor school status (enrolled in school = 1; not enrolled in school = 0) at the

3. Based on informal discussions with juvenile court personnel, regardless ofdiversion program, juveniles often receive similar treatment services. Technically, theonly difference in case processing between both groups is the presence of a committee.Thus, juveniles in the other diversion group had their terms of diversion developed byjuvenile probation officers rather than by committee members.

4. In Maricopa County, the County Attorney files a petition (i.e., referral) againsta juvenile who has allegedly committed a delinquent and/or incorrigible act.

5. Offenses were categorized in the following manner: assault, domestic violenceassault, and robbery, as person offenses; burglary, possession stolen property, theft, andmotor vehicle theft as property offenses; running away, truancy, incorrigibility, and cur-few violations as status offenses; and loitering, resisting arrest, obstruction, disorderlyconduct, alcohol possession, weapon possession as public order offenses. Specific pro-grams (e.g., drug court and a drug diversion program) have been created to deal withdrug offenders in this jurisdiction. Because most drug cases are processed in these drugdiversion programs, drug cases have been excluded from these analyses. To addressone reviewer’s concern regarding the possible varying effects of the four propertyoffenses (i.e., theft, possession stolen property, burglary, and motor vehicle theft),dummy variables were created and separate analyses were performed containing thedummy variables for each property offense versus a model containing offenses typecategories. The effects of the four offenses were in the same direction as the propertyoffense category. For the sake of simplicity, a property offenses category is used insteadof the property specific offenses.


112 RODRIGUEZ

time of court referral and the year when the case was processed are alsoincluded.6

To incorporate Braithwaite’s (1989) and Clear and Karp’s (1999) ideasregarding the role of community and structural factors in reintegrativeprocesses, four community measures are analyzed by taking juveniles’ resi-dential zip codes at time of referral and linking them to zip code data fromthe 2000 Census (U.S. Bureau of the Census, 2000, Summary Tape File 1and 3) (N = 110).7 A measure of racial and ethnic heterogeneity8 (Blau,1977), where higher values indicate greater racial/ethnic heterogeneity,was computed based on racial/ethnic proportions of the population foreach zip code:Racial and Ethnic Heterogeneity = 1−[(PWhite)2 + (PHispanic)2 + (PBlack)2 +

(PAmerican Indian)2 + (PHawaiian Pacific Islander)2

+ (PAsian)2 + (PMulti-racial/ethnic)2 +(POther)2]

Unemployment includes the percent of all residents in the labor force whoreported being unemployed. A cultural indicator, Spanish-speakinghouseholds, was also analyzed. Spanish-speaking households representthe percent of households who reported Spanish as the primary languagespoken in the home. This measure enables a test of whether languagebarriers present unique challenges to restorative justice and, more directly,to juveniles from predominately Spanish-speaking neighborhoods. Delin-quency consists of the number of all delinquency referrals within each zipcode reported to the juvenile court during the year 2000. These data wereprovided by the Maricopa County Juvenile Probation Department.

Given the nested nature of the data (i.e., individuals within zip codes),

6. This study does not incorporate all possible factors that may influence juvenilecourt decision-making processes and reoffending. For example, risk score may be animportant control measure. However, during informal discussions with juvenile courtpersonnel, they indicated they do not always administer the tool to assess “risk-needs”of juveniles in diversion and probation cases. They argue such a tool is a “poor” mea-sure of reoffending.

7. Prior research has empirically measured community-level data of varying geo-graphic areas (e.g., counties, zip codes, and census tracts). In this study, zip code area isused as a proxy for “community.” As no prior community justice research has empiri-cally tested community-level measures within or across geographic areas, zip code isbelieved to serve as an appropriate measure given the study focus is in one jurisdiction(i.e., county).

8. The racial/ethnic composition measure captures majority/minority relations.These relations are not based on particular majority/minority racial/ethnic groups (e.g.,whites vs. blacks or whites vs. Hispanics) but rather on general intergroup relations(which tend to be white-black relations in most research). Blau’s measure of heteroge-neity takes into account those communities that are 100% black, white, or Hispanic(i.e., “0” if 100% homogeneous). Thus, the effects of this measure can be generalizedto a homogeneous and a heterogeneous community.



hierarchical linear modeling is used to analyze the data. As both depen-dent variables in this study are dichotomous, hierarchical generalized lin-ear models (HGLM) are used to estimate the impact of individual- (Level1) and community-level (Level 2) data on the outcomes (Raudenbush andBryk, 2002). Appendix A contains a full review of the Level 1 and 2 mod-els. Table 1 presents the dependent and independent variables used in theanalyses along with the corresponding coding scheme.

TABLE 1. DESCRIPTION OF VARIABLES

Variable Code

Independent VariablesIndividual-Level Data (N = 7,264)

Sex Boys = 1; Girls = 0Race/ethnicity Separate dummy variables for race/ethnicity; Whites

are reference categoryWhiteBlackHispanic/Latino

Age Age at time of referralSchool Status Attending School = 1;

Not Attending = 0Referral Separate offense dummy variables; Person offenses

are reference categoryPersonPropertyStatusPublic order

Prior offenses Number of prior delinquent referralsYear Separate year dummy variables; cases processed

during 1999 are reference category199920002001

Community-Level Data (N = 110)Heterogeneity Racial/ethnic Heterogeneity measureUnemployment Rate Rate of all labor force classified as unemployedSpanish-Speaking Households Rate of households with Spanish as the primary

language spoken at homeDelinquency Delinquency referrals

Dependent VariablesCJC CJC = 1;

Other Diversion Program = 0Recidivism Yes = 1; No = 0

FINDINGS

Before a review of the findings from the multivariate analyses, descrip-tive statistics by group (i.e., CJC vs. the other diversion group) in Table 2show that 62% of juveniles in the restorative justice program were boys, ahigher percentage than in the other diversion group (i.e., comparison


114 RODRIGUEZ

group). The average age for juveniles in both groups was 14 years old.The restorative justice group contained a smaller proportion of Hispanic/Latinos than did the other diversion group, but the representation ofblacks was fairly consistent across both groups. Most cases in the restora-tive justice program involved property offenders (63%), whereas statusoffenders comprised the largest proportion of juveniles in the other diver-sion group (65%). Juveniles in both groups had, on average, at least oneprior court referral. More than three fourths of juveniles in both groupswere attending school at the time they were referred to juvenile court.Interestingly, the rate of recidivism did not substantially vary across bothgroups (28.3% versus 31.2%).

TABLE 2. DESCRIPTIVE STATISTICS

CJC Group Other Diversion Group(%) (%)

Individual-Level DataSex*

Boys 62.0 55.0Girls 38.0 45.0

Race/ethnicity*White 65.0 44.2Black 7.5 6.7Hispanic/Latino 27.5 49.2

Age Mean = 14.1; (SD) = 1.5 Mean = 14.3; (SD) = 1.3School Status*

Attending 90.6 86.1Not Attending 9.4 13.9

Referral*Person 13.8 7.5Property 63.1 23.1Status 15.8 64.5Public Order 7.3 4.9

Prior offenses Mean = 1.1; (SD) = .6 Mean = 1.1; (SD) = .6Year of referral*

1999 67.7 7.72000 28.0 85.12001 5.3 7.3

RecidivismYes 28.3 31.2No 71.7 68.8

Community-Level DataHeterogeneity Mean = .41; (SD) = .15 Mean = .46; (SD) = .12Unemployment Rate Mean = .05; (SD) = .03 Mean = .06; (SD) = .03Spanish-Speaking Households Mean = .19; (SD) = .15 Mean = .28; (SD) = .19Delinquency Mean = 523.6 (SD) = 365.8 Mean = 619; (SD) = 361

* p < 0.05; chi-square test for differences between juveniles in the restorative justice programand the comparison group. Comparison of age and prior offenses used a t-test and were notsignificant at the p < 0.05.



Data on Level 2 measures (i.e., community-level data) by group indicatethat the average rate of racial/ethnic heterogeneity across communities ofrestorative justice participants was lower than those communities ofjuveniles in the comparison group. The unemployment measure showsthat an average of 5% of community members in the restorative justicegroup versus 6% in the comparison group were unemployed. Also, onaverage, 19% of households in communities of restorative justice partici-pants reported Spanish as the primary language spoken at home, a lowerpercentage than the comparison group (28%). The crime measure showsthat the average number of delinquency referrals across communities ofrestorative justice participants was lower than those communities ofjuveniles in the other diversion program.

SELECTION PROCESS

To determine whether the mean selection rate of juveniles into therestorative justice program varied across communities, a one-way analysisof variance (ANOVA) model was estimated (see Table 3, Model 1). Thesignificant random effects intercept indicates that the rate of selection var-ies across zip codes (significant at p < 0.05). It is possible that the varia-tion is simply caused by the unequal distribution of important individual-level characteristics across communities. To control for this, a random-coefficient model including only the individual-level measures was esti-mated. Model 2 presents the fixed-effects and variance component esti-mates. Findings show that property offenders were more likely to beselected by juvenile court officials for the restorative justice program. Inparticular, property offenders were 1.9 times (exp [0.638]) more likely tobe selected than were person offenders. On the other hand, status offend-ers were 0.04 times (exp [−3.270]) less likely than were person offenders tobe selected. Black juvenile offenders were 0.73 times (exp [−0.321]) lesslikely than were white offenders to be selected. Variance componentsindicate that the probability of being selected varied across communities(based on the significant intercept) even after controlling for individualcharacteristics. Additionally, the age, status offenses, total priors, andyear coefficients showed statistically significant variance, which indicatesthat their effects on selection process varied across communities (i.e., zipcodes). This result suggests that juvenile court personnel used theseindicators differently across communities in selecting juveniles for restora-tive justice participation.


116 RODRIGUEZ

TABLE 3. HGLM ESTIMATES OF JUVENILESELECTION INTO RESTORATIVE

JUSTICE PROGRAM

Model 1b Model 2c Model 3d

Fixed EffectsIntercepta −0.449** −0.616** −0.540**

(.112) (.221) (.167)Community-Level Indicators (N = 110)

Delinquency 0.001(.000)

Unemployment Rate 13.738**(6.543)

Spanish-Speaking Household −7.748**(1.378)

Heterogeneity −6.081**(1.338)

Individual-Level Indicators (N = 7,264)Sex 0.096 0.064

(.104) (.092)Black −0.321* −0.287*

(.186) (.179)Hispanic/Latino −0.122 −0.222**

(.138) (.119)Age .002 .001

(.031) (.031)School Status 0.157 0.153

(.151) (.134)Property .638** .676**

(.146) (.132)Public Order −.118 −.026

(.198) (.188)Status −3.270** −3.393**

(.238) (.206)Prior Offenses 0.035 0.054

(.110) (.107)2000 −5.133** −5.194**

(.234) (.227)2001 −6.245** −6.400**

(.522) (.490)

Variance Variance VarianceRandom Effectse Components Components ComponentsIntercept 1.149** 4.622** 2.285**Age 0.021** 0.018**Status 1.109** —Total Priors .232** —2000 1.691** 1.283**

a White offenders and person offenses represent the reference category.b Chi-square = 1,028.08; df = 110.c Chi-square = 339.00; df = 40.d Chi-square = 665.258; df = 91.e Only statistically significant random effects are presented here.* p < 0.10; ** p < 0.05; Standard errors in parentheses.



In an effort to explain why the mean selection rate varied across com-munities, community measures were included in the prediction model.Thus, Model 3 incorporates the Level 2 measures as predictors of theLevel 1 intercept. Of the four community-level indicators, three signifi-cantly influenced the selection process. The rate of unemployed residents,Spanish-speaking households, and racial/ethnic heterogeneity significantlyinfluenced the mean rate of selection, while controlling for individualcharacteristics. The unemployment rate of communities has a positiveeffect on the selection process, which indicates that juveniles from commu-nities with a higher percentage of unemployed residents were more likelyto be selected for the program. On the other hand, the effect of the pro-portion of Spanish-speaking households and racial/ethnic heterogeneity isnegative, which reveals that juveniles from communities characterized bySpanish-speaking households and racial/ethnic heterogeneity were lesslikely to be selected for the restorative justice program. Level 1 estimateswere similar to the estimates reported in Model 2. Interestingly, after theinclusion of the community-level data, Hispanic/Latino offenders were sig-nificantly less likely to be selected for program participation than werewhite offenders.

RECIDIVISM

A second set of HGLM models was estimated to assess the effects of therestorative justice program on recidivism and to determine if restorativejustice was more effective in some communities than in others. Table 4contains the results of the recidivism analyses.

The one-way ANOVA model confirmed that the mean rate of recidi-vism varied across zip codes (i.e., random effects intercept significant at p< 0.05) (see Model 1). As with the selection model, to determine if thevariation was simply caused by the unequal distribution of important indi-vidual-level characteristics across communities, a random-coefficientmodel including only the individual-level measures was estimated. Model2 presents the fixed-effects and variance component estimates. Fixed-effects indicate that restorative justice participation decreased recidivism.Specifically, juveniles who took part in the restorative justice programwere 0.81 times (exp [−0.205]) less likely to recidivate than were juvenilesin other diversion programs. Variance components indicate that the meanlevel of recidivism still varied across communities (based on the significantintercept) after controlling for individual-level characteristics. However,the variance of the restorative justice coefficient did not vary across com-munities. Hence, restorative justice participation had the same effect inreducing recidivism across all communities. Had the variance of the


118 RODRIGUEZ

TABLE 4. HGLM ESTIMATES OFJUVENILE RECIDIVISM

Model 1b Model 2c

Fixed EffectsIntercepta −0.886** −0.947**

(.036) (.038)Individual-Level Indicators (N = 7,264)

CJC −0.205**(.114)

Sex 0.377*(.064)

Black −0.074(.135)

Hispanic/Latino −0.029(.087)

Age −0.084**(.018)

School Status .001(.095)

Property −0.283**(.107)

Public Order −0.009(.147)

Status 0.403**(.118)

Prior Offenses 0.290**(.052)

2000 −0.652**(.107)

2001 −1.932**(.403)

Variance VarianceRandom Effectsd Components ComponentsIntercept 0.041** 0.045**Hispanic/Latino — 0.144**

a White offenders and person offenses represent the reference category.b Chi-square = 163.63; df = 110.c Chi-square = 88.49; df = 40.d Only statistically significant random effects are presented here.* p < 0.10; ** p < 0.05; Standard errors in parentheses.

restorative justice program been significant, the inclusion of Level 2 mea-sures would be warranted in an attempt to explain the variation of pro-gram effect across communities. Given the interest in examiningrestorative justice impact on recidivism and, consequently, to determinewhether recidivism effects varied across communities, findings from the



incorporation of Level 2 effects are excluded from this presentation.9Other significant individual-level indictors of recidivism included sex, age,property offense, status offense, and prior offenses. Boys, status offense,and juveniles with more extensive delinquency records were more likely torecidivate. Yet, older juveniles and property offenders were less likely torecidivate.

DISCUSSION

Community differences have been identified as important aspects ofcommunity justice and offenders’ reintegration (Braithwaite, 1989; Clearand Karp, 1999). However, minimal attention has been given to how com-munity characteristics influence restorative justice and community justiceprograms. Incorporating community characteristics to studies of restora-tive justice expands information of the reintegration process in severalways. First, identifying the individual and community measures that guidejuvenile court officials in their selection of juveniles for program participa-tion reveals the types of juvenile offenders and communities perceived tobe most in need of a reintegrative process versus those more appropriatefor traditional juvenile court supervision programs. Second, by incorpo-rating community characteristics to the examination of program impact(e.g., reducing recidivism), it can be determined whether program impactvaries across communities.

The current findings on selection process reveal that both individual andcommunity characteristics influence the decisions made by juvenile courtofficials. Some researchers have hypothesized that programs tend to tar-get predominantly minority juveniles, whereas others have alluded to thetargeting of predominantly white juveniles. Findings here show that race/ethnicity plays a significant role in the selection decision. Both black andHispanic/Latino juveniles were less likely than were white offenders to beselected for placement in a restorative justice program. This race/ethniceffect found here may not be viewed as evidence of racial disparity giventhat all offenders in this study were selected for diversion processes (vs.formal court processing). However, although all offenders were given anopportunity to not have their cases formally processed in juvenile court,the opportunity to take part in the reintegration process was certainlyinfluenced by race/ethnicity, and it reveals that white juvenile offendersare most likely served and treated by this restorative justice program.

The effect of legal variables shows that property offenders were more

9. Some readers may be interested in the effect of community-level data on recid-ivism. Once the four Level 2 measures were included, only delinquency had a statisti-cally significant effect on mean recidivism; yet its effect was minimal (logit coefficient =0.001).


120 RODRIGUEZ

likely than were person offenders to be selected for program participation.Such a selection process may be a product of two different mechanisms.First, program policy dictates that sex offenders and violent felony offend-ers should be excluded from participating in the restorative justice pro-gram. Not surprisingly, the probability of selecting juveniles with a personoffense is relatively low. Second, offenses in which a victim is directlyinvolved may be perceived by court officials as more fitting for the reinte-gration process. Victim’s conveying the monetary loss and other harmsmay be elements highly sought by officials seeking restorative justicecases. Victim’s presence clearly enables the desired dialogue amongoffender, victim, and community members that is central to the reintegra-tion process.

Although community context plays a central role in the restorative jus-tice process, prior studies have only described the types of communitiesthat may facilitate or challenge the reintegration process. The lack ofinclusion of community measures in prior community justice and restora-tive justice studies made it difficult to ascertain which types of communi-ties were more likely to have a reintegrative process in their communitiesrelative to other communities. Findings from this study expand priorresearch in this area by empirically demonstrating that juveniles fromcommunities characterized by higher levels of unemployment were morelikely to be selected for restorative justice participation; yet, juveniles fromcommunities characterized by higher levels of Spanish-speaking house-holds and racial/ethnic heterogeneity were less likely to be selected.

Interestingly, the decision to select restorative justice participants maybe related to a preceding decision-making process. That is, before juvenilecourt officials decide on the appropriateness of a juvenile for restorativejustice participation, they may first attribute the delinquent act to some-thing and/or someone. One possibility is that court officials seek to reinte-grate juveniles from socially disorganized communities, for example, thosecommunities characterized by high unemployment.10 Juveniles from thesecommunities may be perceived most in need of the reintegration process.Other community characteristics such as racial/ethnic heterogeneity maybe viewed by court officials as indicators of areas with high crime rates andperceive juveniles from such communities as more responsible for theirdelinquent act and less appropriate (i.e., less deserving) for reintegration,which supports the assumption that the more racially/ethnically mixed acommunity, the more challenging the reintegration process becomes.Most studies examining the effects of race/ethnic composition in decision-

10. Median family income and poverty level are other measures of possible socialdisorganization. Neither measure was a significant predictor of selection; thus, theywere eliminated from the analyses.



making processes have focused on white and black populations. Accord-ing to Census 2000 data, Hispanic/Latinos comprise 25% of the populationin Maricopa County, whereas blacks make up only 4% of the population.Findings on the role of racial/ethnic composition found here stress theimportance of examining groups other than blacks in studies of racialthreat and disparate treatment.

Another explanation for the effect of community characteristics in theselection process of juveniles may be more practical in nature. Although itis unlikely that court officials are aware of specific community measures(e.g., racial/ethnic heterogeneity and unemployment rate at the zip codelevel), they may certainly be able to identify those communities that canfacilitate a reintegrative process. For example, communities characterizedby high unemployment may have a higher concentration of communitymembers that can participate in the restorative justice program. Courtofficials’ decisions to select offenders may be indirectly influenced by theability to have community members (i.e., volunteers) and parents fromthose communities take part in the committees.

Communities characterized by a high percentage of Spanish-speakinghouseholds may present a challenge to juvenile court officials. Court offi-cials may be less likely to select juveniles from those communities giventhe difficulty in obtaining community members and family members whospeak English and wish to participate in the program. The language bar-rier, which can be overcome with the presence of a translator, requiresadditional resources from the court that may not be readily available oraffordable.

Prior studies of restorative justice programs have produced mixed find-ings on the impact such programs have on recidivism. Findings from thisstudy indicate that juveniles who participated in the restorative justice pro-gram were less likely to recidivate than were offenders in standard super-vision programs. The lack of variation in the effect of restorative justiceparticipation across communities indicates that the program reduces thelikelihood of recidivism for all offenders regardless of their communitycharacteristics. Although community variation occurred in the selectionprocess and community characteristics were significant predictors of whowas selected, these aspects did not play a significant role in the effective-ness of the program. In the end, regardless of whether the selection deci-sion made by court officials is attributed to their perception of juveniles,communities, or mere artifacts of program capacity (e.g., availability ofvolunteers), lower recidivism by restorative justice participants was exhib-ited across all communities.

Findings form this study have theoretical and practical importance.


122 RODRIGUEZ

First, the inclusion of community measures in this study enabled an empir-ical test of how particular community characteristics influence the reinte-gration process. By identifying the influence that community measuressuch as unemployment, racial/ethnic heterogeneity, and Spanish-speakinghouseholds have in the selection of juveniles for reintegration, a directempirical relationship can be substantiated between restorative justice andcommunities. Second, the focus on the decision-making process of juve-nile court officials vested in implementing and overseeing the reintegra-tion process has been neglected in prior studies. It is important torecognize that juvenile court officials, before the implementation of anyevaluation design (e.g., randomization, matched design), identify thoseappropriate for reintegration. These decisions have a direct influence onwhere community justice is likely to be “nested.” Third, identifying thatcertain communities are less likely to take part in reintegrative processesprovides valuable information regarding the perceived appropriateness ofcorrectional supervision. Findings here demonstrate that certain commu-nities are less likely to be granted access to reintegrative processes andmore likely to be exposed to traditional correctional supervision mecha-nisms. The difference in access to restorative justice programs providesinsight on how court officials allocate court resources and where restora-tive justice programs may be expanded. Lastly, the reduction in recidivismexperienced by restorative justice participants provides support for theexpansion of community-based programs that seek to incorporate commu-nity members and victims in the justice process.

This study has empirically substantiated the important impact individualand community characteristics play in restorative justice. However, sev-eral limitations of this study should be noted. Data on parents’ maritalstatus and/or living arrangements were not available for the juvenile popu-lation under study. Given the important role families play in the restora-tive justice process, the omission of such measures necessarily ignoresparents’ role in restorative justice. Second, findings from this study areonly representative of white, black, and Hispanic/Latino juveniles.Although other racial/ethnic groups are processed through the program,their numbers were relatively small and therefore excluded from theseanalyses. Thus, findings of this study cannot represent the experiences ofNative-American, Asian, or other racial/ethnic offenders. Third, becausedrug offenders were excluded from this study, it is unknown whetherrestorative justice can be effective in dealing with substance abuse as pro-posed by some (Braithwaite, 2001). Fourth, this study included only caseswhere juvenile offenders successfully completed the terms of diversion. Itis possible that terminated cases (e.g., dropouts) experienced differentrecidivism outcomes than those presented here. Lastly, many juvenile riskfactors (e.g., psychological, social, educational) were not available for



examination in these analyses. It is certainly possible that these excludeddata have a direct or indirect influence on the selection decision and recid-ivism outcome. In sum, although these limitations do not enable full gen-eralization of these findings, this study has significantly advanced existingknowledge of restorative justice by addressing the role community charac-teristics play in reintegration programs.

CONCLUSION

This research set out to incorporate community measures to the study ofrestorative justice processes. By using official juvenile data and Censusdata, this study found that individual- (e.g., race/ethnicity) and commu-nity-level (e.g., unemployment rate, racial/ethnic heterogeneity, propor-tion of Spanish-speaking households) characteristics significantlyinfluenced the selection of offenders to a restorative justice program.When examining restorative justice impact, findings of recidivism showthat offenders who took part in the restorative justice program were signif-icantly less likely to recidivate than were offenders in the comparisongroup. Moreover, the effect of restorative justice on recidivism did notvary across communities. Hence, all offenders in the restorative justiceprogram, regardless of their community’s characteristics, were less likelyto recidivate.

Findings from this study present several directions for future research.First, studies of restorative justice must continue to examine the processesthat guide the placement of juveniles into programs. Examining how courtofficials perceive offenders and communities when reintegration policiesare involved is important because the selection decision is the first formalstep in the reintegration process. As such, future research should examinefront-end processes of restorative justice programs and address whethercourt officials (e.g., probation officers) consider individual juvenile attrib-utes, which may be indirectly correlated with community characteristics,or whether they consider their experiences and knowledge of those com-munities when making decisions. Few studies have relied on interviewdata of court personnel to assess decision-making processes. Studiesbased on interviews with court officials can provide insight to how bothindividual and community characteristics influence juvenile courtprocesses and outcomes. Second, studies should incorporate multiple com-munity measures to explore how restorative justice and, in general, com-munity justice is shaped by community characteristics. In particular,racial/ethnic composition should be considered in future studies regardlessof whether community justice programs are nested in homogeneous or


124 RODRIGUEZ

heterogeneous communities. The “dynamic” nature of communities, espe-cially the role that the growing Hispanic/Latino population plays in com-munities, should be recognized and addressed in future studies. Third,studies must continue to explore the important role of individual charac-teristics in examining restorative justice programs. Although a fewresearchers have identified the types of offenders most likely to succeed inthe reintegration process, future research should examine whether particu-lar offenders are most likely to succeed in restorative justice programs.Further, studies must examine the possible interactive effects betweenindividual- and community-level data to expand our understanding of therestorative justice process.

REFERENCES

Bazemore, Gordon1992 On mission statements and reform in juvenile justice: The case of the

“Balanced Approach.” Federal Probation 56:64–70.1999 The fork in the road to juvenile court reform. Annals of the American

Academy of Political and Social Science 564:81–108.

Bazemore, Gordon and Curt T. Griffiths1997 Conferences, circles, boards, and mediations: The “new wave” of

community justice decisionmaking. Federal Probation 61:25–37.

Bazemore, Gordon and Dennis Maloney1994 Rehabilitating community service: Toward restorative service sanctions in

a balanced justice system. Federal Probation 58:24–35.

Bazemore, Gordon and Mark S. Umbreit1995 Rethinking the sanctioning function in juvenile court: Retributive or

restorative responses to youth crime. Crime & Delinquency 41:296–316.2001 A Comparison of Four Restorative Conferencing Models. Washington,

D.C.: Office of Juvenile Justice and Delinquency Prevention, U.S.Department of Justice.

Blalock, Hubert M.1957 Toward a Theory of Minority Group Relations. New York: Wiley.

Blau, Peter M.1977 Inequality and Heterogeneity: A Primitive Theory of Social Structure.

New York: Free Press.

Bortner, M. A., Marjorie Zatz, and Darnell F. Hawkins2000 Race and transfer: Empirical research and social context. In Jeffery Fagan

and Franklin Zimring (eds.), The Changing Borders of Juvenile Justice.Chicago, Ill.: University of Chicago Press.

Braithwaite, John1989 Crime, Shame, and Reintegration. Cambridge, U.K.: Cambridge Univer-

sity Press.2001 Restorative justice and a new criminal law of substance abuse. Youth and

Society 33:227–248.



Bridges, George S. and Robert D. Crutchfield1988 Law, social standing, and racial disparities in imprisonment. Social Forces

66:601–616.

Bridges, George S., Darlene Conley, Gina R. Beretta, Rodney L. Engen, TownsandPrice-Spratlen, Margo Rankin, et al.1993 Racial Disproportionality in the Juvenile Justice System: Final Report.

Olympia, Wash.: Washington Department of Social and Health Services.

Britt, Chester2000 Social context and racial disparities in punishment decisions. Justice

Quarterly 17:707–732.

Bursik, Robert J.1988 Social disorganization and theories of crime and delinquency: Problems

and prospects. Criminology 26:519–551.

Bursik, Robert J. and Harold G. Grasmick1993 Neighborhoods and Crime: The Dimensions of Effective Community

Control. Lexington Books.

Clear, Todd R. and David R. Karp1999 The Community Justice Ideal. Boulder, Colo.: Westview.2000 Toward the ideal of community justice. National Institute of Justice

Journal 21–27.

Crutchfield, Robert D., George S. Bridges, and Susan R. Pitchford1994 Analytical and aggregation biases in analyses of imprisonment: reconciling

discrepancies in studies of racial disparity. Journal of Research in Crimeand Delinquency 31:166–182.

Elliott, Delbert S., William J. Wilson, David Huizinga, Robert J. Sampson, AmandaElliott, and Bruce Rankin1996 The effects of neighborhood disadvantage on adolescent development.

Journal of Research in Crime and Delinquency 33:389–426.

Feld, Barry C.1999 Rehabilitation, retribution, and restorative justice: Alternative conceptions

of juvenile justice. In Gordon Bazemore and Lode Walgrave (eds.),Restorative Juvenile Justice: Repairing the Harm of Youth Crime.Monsey, N.Y.: Criminal Justice Press.

Frazier, Charles E, and Soon R. Lee1992 Reducing juvenile detention rates or expanding the official control nets:

An evaluation of a legislative reform effort. Crime & Delinquency38:204–218.

Frisbie, W. Parker and Lisa Neidert1977 Inequality and the relative size of minority populations: A Comparative

Analysis. American Journal of Sociology 82:1007–1030.

Gottfredson, Denise C., Richard J. McNeil, and Gary D. Gottfredson1991 Social area influences on delinquency: A multilevel analysis. Journal of

Research in Crime and Delinquency 28:197–226.

Hagan, John1992 The poverty of a classless criminology. The American Society of

Criminology: 1991 Presidential Address. Criminology 30:1–19.


126 RODRIGUEZ

1994 The new sociology of crime and inequality in America. Studies on Crimeand Crime Prevention 3:7–23.

Hawkins, Darnell F.1993 Crime and ethnicity. In Brian Forst (ed.), The Socio-Economics of Crime

and Justice. Armonk, NY: M.E. Sharpe.1999 What can we learn from data disaggregation? The case of homicide and

African Americans. In M. Dwayne Smith and Margaret Zahn (eds.),Homicide: A Sourcebook of Social Research. Thousand Oaks, Calif.:Sage.

Hayes, Hennessey and Kathleen Daly2003 Youth Justice Conferencing and Re-offending. Justice Quarterly

20:725–764.

Hoffmann, John P.2003 A contextual analysis of differential association, social control, and strain

theories of delinquency. Social Forces 81:753–785.

Karp, David R.2001 Harm and repair: Observing restorative justice in Vermont. Justice

Quarterly 18:727–757

Karp, David R. and Todd R. Clear2002 The community justice frontier: An introduction. In David R. Karp and

Todd R. Clear (eds.), What is Community justice? Case Studies ofRestorative Justice and Community Supervision. Thousand Oaks, Calif.:Sage.

Karp, David R., Jodi Lane, and Susan Turner2002 Ventura county and the theory of community justice. In David R. Karp

and Todd R. Clear (eds.), What is Community justice? Case Studies ofRestorative Justice and Community Supervision. Thousand Oaks, Calif.:Sage.

Kurki, Leena2000 Restorative and community justice in the United States. In Michael

Tonry (ed.), Crime & Justice: A Review of Research. Vol. 27. Chicago,Ill.: University of Chicago Press.

Latimer, Jeff, Craig Dowden, and Danielle Muise2001 The Effectiveness of Restorative Justice Practices: A Meta-analysis.

Ottawa, Canada: Canada Department of Justice.

Luke, Garth and Bronwyn Lind2002 Reducing Juvenile Crime: Conferencing versus Court. Crime and Justice

Bulletin. NSW Bureau of Crime Statistics and Research.

Maxwell, Gabrielle and Allison Morris1993 Family, Victims, and Culture: Youth Justice in New Zealand. Wellington,

New Zealand: Victoria University of Wellington, Social Policy Agencyand Institute of Criminology.

1997 Family group conferences and restorative justice. Journal on CommunityCorrections 8:37–40.

2001 Family group conferences and re-offending. In Allison Morris andGabrielle Maxwell (eds.), Restorative Justice for Juveniles: Conferencing,Mediation and Circles. Oxford, U.K.: Hart.



McCold, Paul and Benjamin Wachtel1998 Restorative Policing Experiment: The Bethlehem Pennsylvania Police

Family Group Conferencing Project. Pipersville, Penn.: CommunityService Foundation.

McElrea, Frederick W.1996 New Zealand youth court: A model for use with adults. In Burt Galaway

and Joe Hudson (eds.), Restorative Justice: International Perspectives.Monsey, N.Y.: Willow Tree Press.

McGarrell, Edward F.2001 Restorative Justice Conferences as an Early Response to Young Offend-

ers. Washington, D.C.: Office of Juvenile Justice and DelinquencyPrevention, U.S. Department of Justice.

Melton, A.1995 Indigenous justice systems and tribal society. Judicature 70:126–133.

Morris, Allison2002 Critiquing the critics: A brief response to critics of restorative justice.

British Journal of Criminology 42:596–615.

Morris, Allison, Gabrielle Maxwell, and Jeremy Robertson1993 Giving victims a voice: A New Zealand experiment. Howard Journal of

Criminal Justice 32:304–321.

Myers, Martha and Susette M. Talarico1987 The Social Contexts of Criminal Sentencing. New York: Springer-Verlag.

Niemeyer, Mike and David Shichor1996 A preliminary study of a large victim/offender reconciliation program.

Federal Probation:30–34.

Nugent, William and Jeffrey Paddock1995 The effect of victim-offender mediation on severity of reoffense. Media-

tion Quarterly 12:353–367.

Nugent, William, Mark Umbreit, Lizabeth Wiinamaki, and Jeffrey Paddock2001 Participation in victim-offender mediation and reoffense: Successful

replication? Research on Social Work Practice 11:5–23.

Osgood, D. Wayne and Jeff M. Chambers2000 Social disorganization outside the metropolis: An analysis of rural youth

violence. Criminology 38:81–115.

Peterson, Ruth D. and John Hagan1984 Changing conceptions of race: Towards an account of anomalous findings

of sentencing research. American Sociological Review 49:56–70.

Presser, Lois and Patricia Van Voorhis2002 Values and evaluations: Assessing processes and outcomes of restorative

justice programs. Crime & Delinquency 48:162–188.

Raudenbush, Stephen W. and Anthony S. Bryk2002 Hierarchical Linear Models: Applications and Data Analysis Methods.

(2d ed.). Thousand Oaks, Calif.: Sage.


128 RODRIGUEZ

Rountree, Pamela W., Kenneth C. Land, and Terance Miethe1994 Macro-micro integration in the study of victimization: A hierarchical

logistic model analysis across Seattle neighborhoods. Criminology32:387–414.

Roy, Sudipto1993 Two types of juvenile restitution programs in two Midwestern counties: A

comparative study. Federal Probation 57:48–53.

Sampson, Robert J. and W. Byron Groves1989 Community structure and crime: Testing social disorganization theory.

American Journal of Sociology 94:774–802.

Sampson, Robert J. and John H. Laub1993 Structural variations in juvenile court processing: Inequality, the under-

class, and social control. Law & Society Review 27:285–311.

Sanborn, Joseph B.1994 Remnants of parens patriae in the adjudicatory hearing: Is a fair trial

possible in juvenile court? Crime & Delinquency 40:599–615.

Secret, Philip E. and James B. Johnson1997 The effect of race on juvenile, justice decision making in Nebraska:

Detention, adjudication, and disposition, 1998-1993. Justice Quarterly14:445–478.

Sherman, Lawrence, Heather Strang, and Daniel Woods2000 Recidivism Patterns in the Canberra Reintegrative Shaming Experiments

(RISE). Centre for Restorative Justice, Research School of SocialSciences, Australian National University.

Umbreit, Mark S.1994 Victim Meets Offender: The Impact of Restorative Justice and Mediation.

Monsey, N.Y.: Willow Tree Press.1995 Holding juvenile offenders accountable: A restoratives justice perspective.

Juvenile and Family Court Journal 46:31–42.2000 The Restorative Justice and Mediation Collection: Executive Summary.

Washington, D.C.: Office for Victims of Crime.

Umbreit, Mark S. and Robert Coates1993 Cross-site analysis of victim-offender mediation in four states. Crime &

Delinquency 39:565–585.

Umbreit, Mark S. and Susan Stacey1996 Family group conferencing comes to the U.S.: A comparison with victim

offender mediation. Juvenile and Family Court Journal 47:29–38.

Umbreit, Mark S. and Howard Zehr1996 Restorative family group conferences: Differing models and guidelines for

practice. Federal Probation 60:24–29.

Umbreit, Mark S., Robert Coates, and Betty Vos2001 The impart of victim-offender mediation: Two decades of research.

Federal Probation 65:29–33.2002 The Impact of Restorative Justice Conferencing: A Review of 63

Empirical Studies in 5 Countries. Center for Restorative Justice &Peacemaking. Available at: http://ssw.che.umn.edu/rjp.



Wilson, William J.1987 The Truly Disadvantaged: The Inner City, the Underclass, and Public

Policy. Chicago: University of Chicago Press.1996 When Work Disappears: The World of the New Urban Poor. Chicago:

University of Chicago Press.

Nancy Rodriguez is an Associate Professor in the Criminal Justice and CriminologyDepartment at Arizona State University at the West Campus. Her research interestsinclude sentencing policies, juvenile court processes, and substance abuse. Her recentpublications have appeared in Justice Quarterly and Crime & Delinquency.


130 RODRIGUEZ

APPENDIX ALeve1 1 and 2 models for the examination of both dependent variables arepresented below:

Level 1: (individual)hij = b0j + b1j (X1ij – X̄1j) + . . . . +_bpj (Xpij – X̄pj) + eij (1)

Level 2: (community)b0j = g00 + g01w1j + . . . . + g0qwqj + u0j (2)

Equation 1 includes variables at the individual level, whereas Equation 2includes zip code level measures. In Equation 1, hij represents the log-odds of the dependent variables (i.e., log-odds of selection in the selectionmodel and the log-odds of recidivism in the recidivism model) for juvenileoffender i in zip code j. In the Level 1 model, (Xpij – X̄pj) refers to theindependent variables measured at the individual level. The independentvariables are centered around the Level 2 means (group mean centering)given the interest in examining potential differences across zip codes. Theb coefficients in this model estimate the magnitude of the independentvariables’ impact on the dependent variable. Subscript p represents thenumber of individual-level variables (error term (eij) represents the varia-tion in error among juveniles). Subscript q represents the number of com-munity-level variables. In Equation 2, b0j represents the intercept inEquation 1. This model attempts to predict mean differences in the Level1 outcome variable across zip codes. In the Level 2 model, w refers to theindependent variables measured at the zip code level, whereas the g coeffi-cients represent their effects. Error term in the Level 2 model (upj) repre-sents the error across zip codes.

restorative justice, communities, and delinquency: whom do we reintegrate?

Documents