MSU LIBRARIES RETURNING MATERIALS: Place in book drop to remove this checkout from your record. FINES will be charged if‘book is returned after the date bggmped below. - "9 H‘O H ’- If: 5 2;? 3,3,7." -h‘.‘1$¥l\ :11. .2: .‘ A I 1" ‘ A' . I . 1‘ ° ‘. I l l I: ... - 1 -1“ Z: "a“; 135I99Q, “—— .1. ES” 9 9 '06“: 1 6’ .g-L: 1 -JS 1 ,1¢£¢; COMMUNITY SERVICE AND OUTRIGHT RELEASE AS ALTERNATIVES TO JUVENILE COURT: AN EXPERIMENTAL EVALUATION By J. Randy Koch A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Psychology 1985 ‘5'” .A I * r3. 1;“; Capyright by JOHN RANDOLPH KOCH 1985 TABLE OF CONTENTS LIST OF TABLES ................ . ............................ vi LIST OF FIGURES . ........................ . .................. viii INTRODUCTION ............................................. 1 Chapter 1. THE HISTORICAL DEVELOPMENT OF DIVERSION ........ . . . . 3 The President's Comission and the Rationale for Diversion . . 3 Ineffectiveness of the Juvenile Court ......... . . . . 4 Juvenile Court Overload and High Costs ........ . . . 5 OverreachoftheSystem.............. 6 Lack of Humanitarian Treatment and Due Process. 7 The Negative Effects of Labeling. . . . . . . . . . . . 9 TheRiseinPwularitVOOOOOOOOOOOOO::::OOOOOOOO0.. 12 Critical Review and Current Status . . . . . ............ . . l4 conCIEJSiOMOOOOOOOOOOOOO ..... OOOOOOOOOOOOOOOOOO 15 2. DIVERSION AND PROGRAM EFFECTIVENESS: A REVIEW OF PRIOR RESEARCH O 0000000000000 O O O ..... O O O O O O O O O O O O O 16 Quasi-Experimental Evaluations ................. . . . . 18 Summary ...... . .................... . . . . . 31 Experimental Evaluations ........................ . . 32 Summary ..................... ..... ....42 3. CONCLUSIONS AND RATIONALE FOR THE CURRENT STUDY . . 44 IsDiversion Effective?........... ................. 44 Implementation of Diversion ................... . 45 Labelmg 52 ProgramCosts................. ...... 54 Recidivism ..... ......... 56 Conclusions.. ...... 66 Rationale for the Current Study ...... . . . . ......... . . 68 OverviewoftheStudy. . ..... .. 69 Overview of the Community Service Program ........... . 70 ProgramRationale........... ......... . ...... 71 ResearchGoals...... ............. ........... .74 Model Predicted Results . .......... . ......... . . 76 iii 4. METHODOFPROCEDURE............................. 79 ResearchSetting....... . ...................... .. 79 TheCountyASite........... .............. .. 80 Subjects 82 Project Participants ............. . . . . . . . . . . . . . 84 Demgn ........ .. ....... ....88 Procedures.. ...... .. ................. ..........90 Referral.................................. 90 Projectlntake.............................. 91 InterventionAlternatives...................... 93 Post-Period Assessment . . . . . . . ............ . . . . 98 TimeFrame................................ 99 Measures ..... .. ................. . ..... 100 Measure Development ..... . .......... . . . . . . . . 101 Implementation of Diversion .................. . 104 InterventionProcess......................... 104 Mediating Processes . ........................ 110 Ultimate Outcomes ......................... . 117 5. RESULTS............... ..... .. ..... ..............121 Intervention Assessment .......... . . . . . .......... . 122 Implementation of Diversion . . . . . . . . . . ..... 122 Description of Alternative Interventions . . . . . . . . . . . 134 EquivalenceofGroups. 147 Intermediate Outcomes ......... . . . . . . . . . . . . . . 149 Analyses of Variance for Labeling and Social Bonding. 149 ExploratoryAnalyses........................ 152 Summary........ ........... ...... .....155 Ultimate Outcomes ................. . . . . . ...... . 156 Project Impact on Recidivism . . . ............... 156 ExploratoryAnalyses........................ 159 Summary . . . .............................. 165 6. DISCUSSION . . . ................................... . 170 Implementation... ......... 170 Implementation of Diversion . . . . . . . . . . ...... 170 Implementation of the Community Service Program . . 175 Intermediate and Ultimate Outcomes . . . . . . . . . . . . . . . . . 179 Conclusions and Recommendations . . . . . . . . . . . ...... . . 184 The Diversion Impact Model . . ............... . . 184 The Community Service Impact Model ........... . 188 SummaryandConclusions..................... 191 FOOTNOTES . . ............................................ . 196 APPENDICES Appendix A. Summary of Diversion Evaluations . . . . ............. 200 Appendix B. The Diversion Project Participation Agreement . . . . . . . 204 iv Appendix C. Community Service Process Interview ............. 206 Appendix D. Court Contact Record ..... . . . ...... . ...... . . 213 Appendix E. Label Encapsulation . . ....................... 214 Appendix F. Perceived Negative Labeling ...... . ........... . 216 Appendix G. Community Service Outcome Measures . . . . . . . . . . . . 217 Appendix H. Placement Supervisor Interview ................ 220 Appendix I. Self-Report of Delinquency .................... 222 LISTOFREFERENCES .......... . ....... 224 Table 1. Table 2. Table 3. Table 4. Table 5. Table 6. Table 7. Table 8. Table 9. Table 10. Table 11. Table 12. Table 13. Table 14. Table 15. Table 16. Table 17. Table 18. Table 19. Table 20. Table 21. Table 22. LIST OF TABLES ProgramOutcomebyLevelofDiversion.................... 61 Program Outcome by Type of Sponsor . .......... . ........ . . 63 Program Outcome by Type of Services ................ . . . . . 64 Reasons for Nonparticipation ...... . . ............ . . . . . . . . 85 Race of Participants and Nonparticipants . . . . . . . . . . ......... 86 Living Arrangement of Participants and Nonparticipants . . . . . . . . .86 Referral Offense of Participants and Nonparticipants ......... . . 87 Experimental Design and Measures ............ . ....... . . . . 89 EvaluationMeasures..................................103 Interrater Agreement for Program Record Data . . . . . . . . . . . . . . 107 Internal Consistency of Community Service Process Scales ...... 109 Community Service Process Interscale Correlations . . . . . . . . . . . 109 Internal Consistency of Label Encapsulation Scales . ......... . . 112 Label Encapsulation Interscale Correlations ..... . . . . ...... . . 112 Internal Consistency of Community Service Outcome Scales . . . . . 115 Community Service Outcome Interscale Correlations ......... . 115 Internal Consistency of Supervisor Rating Scales ............. 118 Test-Retest Reliability of Placement Supervisor Interview . . . . . . 118 Pre-Post Formal-Processing Rates . . . . . . . . . . . . . . . . ........ 125 Reductions in Formal-Processing Rates by Department . . . . . . . . . 126 Correlation of Client Demographic and Court Measures ....... . 136 Types of Placements Recruited and Used ................... 140 vi Table 23. Table 24. Table 25. Table 26. Table 27. Table 28. Table 29. Table 30. Table 31. Table 32. Table 33. Hours of Community Service Completed ................... 142 ANOVA for JJS Contacts Scale ................ . ......... 151 Correlation of Program Process and Labeling Measures . . . . . . . . 153 Correlation of Program Process and Social Bonding Measures . . . 154 Correlation of Pre- and Post—period Measures of Recidivism . . . . . 157 Correlation of Program Process and Recidivism Measures ...... 161 Correlation of Labeling and Recidivism Measures ............. 163 Correlation of Social Bonding and Recidivism Measures . . . . . . . . .164 Prediction of SRD Using Demographic Variables . . . . ..... . . . . . 166 Prediction of SRD Using Demographic _a_n_d Mediating Variables . . . 167 Discriminant Analysis Classifying Official Recidivists .......... 168 vii Figure 1. Figure 2. Figure 3. Figure 4. Figure 5. Figure 6. LIST OF FIGURES DiversionImpactModel........................ ....... . 75 Community Service Impact Model ........................ 75 Conceptual Model for Assessment Instruments . . . . . . . . . . . . . . . 102 Formal-Processing Rates for CPD 1 ...................... 130 Formal-Processing Rates for CPD 2 ............ . .......... 131 Formal-Processing Rates for CSD ........................ .132 viii INTRODUCTION Much of the history of juvenile justice in the United States is a history of society's attempts to find a solution to the problems of juvenile crime and delinquency. Like many "solutions" that have been proposed in the past, diversion of juvenile offenders was proposed with grand expectations (President's Commission on Law Enforcement and the Administration of Justice, 1967) and continued to prosper with little evidence to support the expectation that its promises would be fulfilled (Cressey 6c McDermott, 1973). It has now been 17 years since diversion was first proposed by the President's Crime Commission as a strategy to combat juvenile delinquency. As will be seen from the material presented in subsequent chapters, during the more than 15 years since diversion efforts were first initiated a great deal of time and money has been invested in planning efforts, program operation, and research studies related to juvenile diversion. Unfortunately, despite the substantial attention that has been devoted to diversion many investigators of juvenile diversion believe that few, if any, conclusions can be drawn about its effectiveness (e.g., Binder, 1977; Klein, 1979a; Rutherford 6c McDermott, 1976). As will be demonstrated later, the inability to determine the effectiveness of diversion is mostly due to the inadequacies of previous research and the apparent failure of programs to actually implement diversion. In addition, insufficient emphasis has been placed on the types of services provided by diversion programs. In this paper, an evaluation of a specific diversion program will be 1 2 presented. The study was designed to address the major problems associated with prior evaluations of diversion efforts. In particular, this study included a rigorous evaluation design, preliminary planning steps designed to increase the likelihood of successful implementation of diversion, and a service model with at least some prior evidence of effectiveness and a strong theoretical framework. In the following chapters, an evaluation of three alternatives for handling juvenile offenders will be presented. An experimental design was employed in which youths were randomly assigned to traditional processing (referral to court), diversion without services (i.e., outright release), and diversion to the Community Service Program. The Community Service Program provided a setting for the arbitration of conflicts between the victim and the juvenile offender, the payment of restitution, and the placement of the youth in a position of voluntary community service. The evaluation design included an assessment of diversion and Community Service Program implementation, as well as an assessment of the impact of the alternatives on labeling, social bonding, and recidivism. The report of the evaluation is presented in six major sections. In the first section, the history of diversion is described with a special emphasis on the original rationales for its initiation. In the next section, a comprehensive review of prior research on the effectiveness of diversion is presented. In the third section, the empirical data related to the major goals of diversion is reviewed in light of the methodological adequacy of diversion research. This section concludes with an overall assessment of the state of the art in juvenile diversion and the rationale for the current study. In the fourth section, the methodology employed in the current study is described. In the fifth and sixth sections, the results of the study are presented and the implications of the study for future research and delinquency programming are discussed. CHAPTER 1 THE HISTORICAL DEVELOPMENT OF DIVERSION There appear to be several distinct stages in the history of juvenile diversion even though its history is just over fifteen years in duration. In this section, the history of diversion will be traced from its origin and rise in popularity beginning with the President's Commission on Law Enforcement and the Administration of Justice, through a period of critical review to its current status. Particular emphasis will be placed on diversion's roots in the recommendations of the President's Commission. The President's Commission and the Rationafie forfiversion Juvenile diversion has its roots in the 1967 President's Commission on Law Enforcement and the Administration of Justice (e.g., Lemert, 1981; Vorenberg 6: Vorenberg, 197 3). Statements clearly supportive of the concept of diversion appear throughout the President's Commission's report. The following example was quoted by Carter and Klein (1976): The formal sanctioning system and pronouncement of delinquency should be used only as a last resort. In place of the formal system, dispositional alternatives to adjudication must be developed for dealing with juveniles, including agencies to provide and coordinate services and procedures to achieve necessary control without unnecessary stigma. (p. 10) In addition to advocating the diversion of juvenile offenders, the President's Commission specifically created youth service bureaus (YSB's) to implement diversion efforts in particular communities, and it recommended expanded use of outright release in delinquency cases.1 4 While the President's Commission noted anticipated benefits in having the community assume the responsibility for its youths as a result of diversion efforts, its support for diversion appears to be based largely on dissatisfaction with the operation and impact of the juvenile justice system. Repeatedly the President's Commission noted inadequacies in the system and expressed their judgment that it had failed. to achieve its goals. The following statement is but one example of the President's Commission's evaluation of the juvenile justice system: the great hapes originally held for the juvenile court have not been fulfilled. It has not succeeded significantly in rehabilitating delinquent youth, in reducing or even stemming the tide of delinquency, or in bringing justice and compassion to the child offender. (President's Commission quoted in Carter G: Klein, 1976, p. 7) The President's Commission's criticisms of the juvenile justice system gained widespread support and, then as now, provided the rationale for juvenile diversion. Thus, juvenile diversion grew more out of dissatisfaction with the existing system for handling the juvenile offender than out of a specific theory supporting the presumed benefits of diversion. These dissatisfactions were that the juvenile justice system was ineffective, overloaded, costly, overreaching in its influence, inhumanitarian in its treatment of youths, and adversely affected youths through negative labeling. It was these failures that diversion was suppose to correct. Considerable attention was given to these presumed failures of the juvenile court and they were used as arguments for diversion. Ineffectiveness of the Juvenile Court Statements that the juvenile justice system had been unsuccessful in dealing with juvenile crime were plentiful in the late 19603. One argument for the ineffectiveness of the existing system was the rising juvenile crime rate. For example, in the 7 years leading up to the President's Commission's report "arrests of juveniles for serious crimes increased 59 percent...while the number of persons in the young age group, 10-17, increased 22 percent" (Federal Bureau of 5 Investigation, 1967, p. 1). This startling increase resulted in 1,259,622 juveniles, age 10-17, being arrested in 1967. Juvenile arrests accounted for 22.9% of g arrests and 45.5% of all arrests for index crimes for that year (i.e., criminal homicide, forcible rape, robbery, aggravated assault, breaking and entering, larceny over $50 and auto theft) (FBI, 1967). The President's Commission, in noting the juvenile crime rate, projected that "one in every nine youths—one in every six male youths—will be referred to juvenile court...before his 18th birthday" (quoted in Empey, 1978, pp. 460-461). A second indication of the supposed failure of the juvenile justice system was the high recidivism rate for youths who had been handled by the system. For example, in the Uniform Crime Reports for 1967 it was reported that "of young offenders under 20 released in 1963, 70 percent repeated" (FBI, 1967, p. 1). Another source reported that the rate of recidivism for youths on probation was 50% (Scarpitti 6c Stephenson, 1968). Later studies also provided reason for concern about the relationship between court processing and recidivism. In a well publicized study of male cohorts born in 1945 (_1_1_ = 9,945), Wolfgang, Figlio, and Sellin (1972) found that between the ages of 10-17, 35% of the youths in their study became officially delinquent, and 54% of these youths recidivated. Further, the 1,862 youths who recidivated accounted for 8,601 "recorded offenses." Juvenile Court Overload and Hg‘ h Costs One of the obvious consequences of the increase in juvenile arrests and high recidivism rates was that an ever increasing number of youths were entering the justice system. For example, it has been estimated that in the 4 years just before the President's Commission, 1962-1966, the number of cases received by juvenile court intake departments in the U.S. increased from 550,000 to 745,000, or 35% (Empey, 1978). Such a rapid increase in court caseloads was presumed to have placed a great burden on the nation's juvenile justice system. One purported 6 consequence of the influx of youths into the system was that juvenile courts did not have the time or personpower to provide the individualized justice that was to have been the hallmark of the juvenile justice system (Schur, 1973). Also, since many of the youths entering the system were status offenders and other minor law violators (FBI, 1967), the courts presumably could not devote the time and resources to the youths who were most in need. A second consequence of rising caseloads was that juvenile justice expenditures had to increase dramatically to keep up with the influx of cases, and the cost per case was extremely high. Sarri and Vinter (1976) estimated that in 1974 the "annual per-offender costs for institutions, camps, and ranches average $11,657" (p. 173), and this figure did not include the cost of court hearings, probation and parole services, and so forth. In fact, avoidance of the high costs associated with juvenile justice processing has been one of the major rationales for diversion. In an early planning effort for the Department of Health Education and Welfare, Gemignani (1973) argued that with a systematic attempt to divert juvenile offenders "almost $1.5 billion could be saved in official court costs" (p. 10) over a period of 5 years. Overreach of the System Another criticism of the juvenile justice system which contributed to the support for diversion, was that the system had extended its control over youths in far too many cases. In particular, the justice system was criticized for establishing its jurisdiction in a manner that allowed it to define acceptable morals for young people, and for using vaguely worded laws that allowed courts great discretion in defining youthful misbehavior (e.g., President's Commission, 1967). Examples of such laws are apparent in the Michigan Juvenile Code. The Michigan Code provides for court jurisdiction over any youth... 7 Who repeatedly associates with immoral persons, or who is leading an immoral life; or is found on premises occupied or used for illegal purposes; Who habitually idles away his or her time; Who is repeatedly disobedient to the reasonable and lawful commands of his parents... (Office of the Court Administrator, 1975, p. 1) Much of the criticism about the overreaching influence of the juvenile justice system was focused on the court's jurisdiction over status offenders. Many organizations and commissions (for example, the National Council on Crime and Delinquency (1975) and the President's Commission (1967)) recommended that juvenile court jurisdiction over status offenders be eliminated. Such recom- mendations were based on the belief that these youths were in need of services that could best be provided outside the justice system. Proponents were further motivated by reports of the large number of status offenders who were being housed in juvenile correctional institutions. One survey conducted in 1966 indicated that approximately 30% of the youths in correctional facilities were there for having committed a status offense (Sheridan, 1966). Diversion was seen as one method for restricting the influence of the court and for providing the services these youths needed. Lack of Humanitarian Treatment and Due Process One of the more emotional criticisms of the juvenile justice system was that it had failed to treat the youths in its care in a humanitarian manner. While the "ideal" juvenile court was to be characterized by a judge acting as a "wise parent" (p. 15) dispensing advice, individualized justice, and treatment provided by a professional staff (Empey, 1973), critics charged that the court's actual operation was in sharp contrast to this ideal. Criticisms of the court included the accusation that the judges were incompetent; lacking knowledge of youth and family problems and of juvenile law. 8 The President's Commission noted that half of the judges practicing in juvenile court had no undergraduate college degree (President's Commission, quoted in Carter 6: Kline, 1976, p. 8). Other critics suggested that, far from providing the individualized justice originally promised, juvenile courts instead dispensed what Empey has called "assembly-line justice" (1978, p. 471). Juvenile court hearings, rather than being a time to diagnose a youth's problems and to determine an apprOpriate solution, "often turn out to be little more than attenuated interviews of ten to fifteen minutes duration" (President's Commission, quoted in Carter 6: Kline, 1976, p. 8). Finally, and perhaps the most serious criticism of all, the juvenile justice system was charged with permitting or failing to stop the brutal treatment of youths in their care. The mistreatment of juveniles at the hands of other youths and juvenile justice staff was documented by several sources (e.g., Empey, 1978; James, 1969; Ohlin, Coates, 6: Miller, 1974; Richetti, 1969). Reports of abuses in juvenile correctional facilities were especially dramatic and greatly elevated the level of dissatisfaction with the juvenile justice system. While by some accounts the structure of the juvenile court was originally developed in order to provide the flexibility that was needed to deliver individual- ized justice and rehabilitative treatment, neither of these goals appear to have been attained. Instead, critics charged that the court's informality resulted in the loss of basic constitutional rights for juveniles (Schur, 1973). Until the well known Gault case of 1967, juveniles were not guaranteed the right of notice, the right to counsel, the privilege against self-incrimination, or the right of confrontation and cross-examination (In re Gault, 387 U.S. 1, 14 (1967)). The United States Supreme Court expressed the feelings of many people in the case of Kent vs. United States (384 U.S. 541, 556 (1966)) when it stated: There is evidence, in fact, there may be grounds for concern that the child receives the worst of both worlds: That he gets neither the protection accorded to adults nor the solicitous care and regenerative treatment postulated for children. 9 In summary, in the late 19608 there was a common belief that the juvenile justice system had deprived youths of their due process rights; allowing them to enter a system where they may be abused rather than obtain the treatment they needed. The Negative Effects of Labeling At the same time that the above criticisms were being directed at the juvenile justice system, there was a theory or conception of delinquency that was gaining widespread popularity. This conception of delinquency was called labeling theory. Unlike other theories of delinquency, labeling theory did not focus on the individual as the cause of delinquency; instead, it focused on the justice system's reaction to delinquency and how the system might actually contribute to the youth's delinquent behavior through negative labeling. Support for the labeling perspective of delinquency was based, in part, on the results of self-report studies of delinquent behavior. In particular, it was observed that nearly all youths commit acts that could result in arrest and referral to court (e.g., Williams 81 Gold, 1972). However, it has been estimated that 90% to 95% of all youths' illegal behavior goes "undiscovered or unacted upon by authorities" (Empey, 1973, p. 16). Despite the failure of the juvenile justice system to respond to the delinquent behavior of the majority of youths, most young peeple develop into "law-abiding" adults without ever having entered the justice system. Additional evidence supporting the validity of labeling theory was presented in a well known study conducted by Gold and Williams (1969). These authors found that youths who were caught after committing a delinquent act and subsequently identified as a delinquent, were more likely to recidivate than youths who were not caught for their delinquent behavior. While the evidence presented above did not demonstrate a cause-and-effect relationship between juvenile 10 justice processing, negative labeling and delinquency, it did contribute to the doubts about the value of the juvenile justice system. In general, labeling theory proposes that repeated contact with the juvenile justice system results in a youth being labeled as a delinquent. As a result of this new label and the stigma attached to it, future delinquent behavior is somehow encouraged. The process through which delinquency labeling actually causes delinquent behavior has not generally been stated in operational terms that allow empirical testing. However, the usually vague descriptions relating negative labeling to future delinquency appear to emphasize one of two aspects of the labeling process. One version of labeling is that society's reactions to the youth's behavior (i.e., arrest, court hearings, etc.) sets into motion a self-fulfilling prophecy. According to this perspective, formal processing by the juvenile justice system results in the delinquency label being applied to the youth. This label is reinforced by repeated contact with the justice system and by the reaction of the community to the youth. As a result, the youth comes to. see himself/herself as delinquent and acts accordingly. This perspective proposes that a change in self- perception and self-concept ultimately leads to increased delinquent behavior. Exactly how the youths self-perception is changed and why this change results in delinquent behavior is not made totally clear by labeling theory. The above version of labeling theory which emphasizes a change in self- concept was specifically used by the President's Commission as part of its rationale for limiting the use of formal juvenile justice processing. The Commission presented its interpretation of labeling theory as follows: Official action may actually help to fix and perpetuate delinquency in the child through a process in which the individual begins to think of himself as delinquent and organizes his behavior accordingly. That process itself is further reinforced by the effect of the labeling upon the child's family, neighbors, teachers, and peers, whose reactions 11 communicate to the child in subtle ways a kind of expectation of delinquent conduct. (reprinted in Carter 6: Klein, 1976, p. 9) The second perspective of labeling, rather than focusing on changes taking place in the youth, emphasizes the role of the justice system's and community's response to the youth. Once a youth has been labeled a delinquent a number of reactions from the justice system, parents, neighbors, and teachers may be forthcoming. For example, a new set of rules may be applied to the youth (Lemert, 1971). For the adjudicated youth, probation generally includes rules specifying early curfews, improved academic performance, and so forth. Parents may also expect the youth to abide by additional rules. While ‘these rules are intended to help the youth stay out of trouble, the increase in restrictions may actually make it more likely that the youth will break a rule, thus reinforcing their delinquency label. The delinquency label may have other consequences as well. Once labeled a delinquent, a youth is usually subject to greater surveillance by officials who are then more likely to detect deviant behavior. Other persons in the youth's environment, knowledgeable of the youth's deviant status, may be more likely to scrutinize the youth's behavior and interpret even minor transgressions as confirmation of the child's delinquent tendencies. As a result, behavior that otherwise would have been overlooked as part of the normal behavior of adolescence may be treated as requiring official intervention. In this manner, youths who have been labeled delinquent may be more likely to be reported to the police and referred to court for relatively mild forms of misbehavior. Conclusion. One probably cannot overemphasize the importance of labeling theory to the diversion movement, and vice versa. Labeling theory's focus on the negative effects of the juvenile justice system fits very nicely with the critics' assessment of the system, and this similar perspective may be partly responsible 12 for the widespread support the theory has enjoyed. At the same time, labeling theory provided a scientific justification for diversion. The respectability this provided to the diversion movement added greatly to its acceptability, particularly in the academic community. This was true despite the lack of empirical support for the assumptions of labeling theory. Then, as now, there are few studies which had tested labeling theory, and those that were conducted produced inconsistent results (Klein, 1979). The Rise in Papularity Born out of dissatisfaction with the current system for handling juvenile offenders and given legitimacy by labeling theory, diversion rapidly gained popularity. Beginning with pr0posals for diversion efforts by the President's Commission on Law Enforcement and the Administration of Justice, diversion was eventually endorsed by a wide variety of commissions and criminal justice organizations. Support came from the National Council on Crime and Delinquency (NCCD, 1976) and the International Association of Chief‘s of Police (Kobetz 6r Bosarge, 197 3). An early endorsement was provided by a second commission, the National Advisory Commission on Criminal Justice Standards and Goals. The National Advisory Commission's strong support was indicated in the following statement: Every police agency, where permitted by law, immediately should divert from the criminal and juvenile justice system any individual who comes to the attention of the police, and for whom the purpose of the criminal or juvenile process would be inappropriate, or in whose case other resources would be more effective. (p. 80) More recently, diversion was advocated by the Institute of Judicial Administration-American Bar Association Joint Commission on Juvenile Justice Standards (1977). This commission recommended that court jurisdiction over status offenders should be eliminated, that "mandatory diversion" (p. 78) should be used for misdemeanants with no prior convictions or formal referrals to court, and 13 that "serious consideration should be given to the formal diversion of all other apprehended juveniles" (p. 77). National support for diversion quickly led to the establishment of diversion programs. Youth service bureaus (YSB's), as recommended by the President's Commission, were created to initiate and coordinate diversion efforts in local communities. These programs rapidly proliferated as communities jumped on the diversion bandwagon. Encouraged by federal funding from the Law Enforcement Assistance Administration (LEAA) it was estimated that by 1971 there were over 150 YSB's in operation across the country (Hewlett, 1973). In addition to programs funded by LEAA, diversion programs were initiated by the Department of Health, Education and Welfare (HEW). By 1972, 23 pilot programs with a diversion focus were implemented by the Youth Development and Delinquency Prevention Administration in HEW (Gemignami, 1973). These programs only represented a portion of the growing number of diversion programs. In addition to those programs initiated by the federal agencies, there were a large number of diversion programs initiated by state and local governments as well as diversion efforts added to existing youth service agencies. The proliferation of diversion programs has been noted by other authors. In 1976, Klein, Teilman, Styles, Lincoln, and Laben-Rosensweig referred to the "explosion" in diversion programs and stated that "easily the most striking single dimension of diversion programming is its enormous recent growth" (p. 103). This growth was obviously accompanied by a huge investment of human resources and money. Unfortunately, this investment was not preceded by a prior investment in research and critical review. As will be described below, after years of investing in diversion programs, a number of serious concerns were raised about the Operation and effectiveness of diversion. 14 Critical Review and Current Status After a period of popularity and rapid expansion, diversion suffered a period of criticism. Experience with diversion efforts led to a critical review of program operation and concern about a number of unresolved issues. This concern was highlighted by a special issue of Crime and Delinquency that was devoted to diversion. This October, 1976 issue contained a number of articles that were generally critical of both the concept of diversion and diversion programming. The criticisms of diversion presented were: that the major rationale for diversion, labeling theory, lacked substantial empirical support (Klein, 1976; Lundman, 1976); that diversion programs had great potential for abusing their client's rights (Maron, 1976; Nejelski, 1976); and that diversion programs may increase rather than reduce the number of youths subject to social control (Nejelski, 1976). Finally, the diversion movement was criticized for the prolifera- tion of programs without an adequate evaluation of diversion's effectiveness (Gibbons 6: Blake, 1976; Klein, 1979b; Vorenberg 6c Vorenberg, 1973). Evaluations of diversion programs were considered to be too few and of poor quality. Over the past 8 years, the discussion about various issues and problems in juvenile diversion has continued. After repeated calls for a comprehensive evaluation of diversion, a few such large-scale evaluations have been conducted. The most recent of these was the National Evaluation of Diversion Projects (Dunford, Osgood, 6r Weichselbaum, 1982), which resulted in a disappointingly negative assessment of diversion's effectiveness. The extensive criticisms of diversion combined with the lack of consistent positive findings about diversion's effectiveness may portend the end of the diversion movement. There is some indication that diversion is already losing its popularity. As noted earlier, the decline in publications about diversion, the lack of new funding initiatives, and increasing demand for a "tougher" approach to 15 juvenile crime (Hellum, 1979; Lemert, 1981) may all signal an abandonment of juvenile diversion. Conclusions The history of juvenile diversion can be traced back just over 15 years to the President's Commission on Law Enforcement and the Administration of Justice. Diversion was expected to help youths avoid all of the damaging effects of juvenile justice involvement, and also, to benefit the juvenile justice system. Through diversion efforts, youths were to be spared the stigma of being labeled a delinquent while still receiving the treatment they needed. At the same time, it was expected that the juvenile justice system costs could be reduced, and that more attention could be devoted to the serious offender. These became the major goals and promises of diversion programs; diversion was to reduce delinquency labeling, to provide a more effective alternative to juvenile justice processing, and to reduce juvenile justice system costs. Given the substantial promises of diversion and widespread dissatisfaction with the juvenile justice system, diversion programs grew at an enormous rate. Unfortunately, this growth occurred without adequate attention to possible problems associated with the operation of diversion programs and without empirical support for its effectiveness. Even the major theoretical rationale for diversion, labeling theory, remained largely untested (Binder 6: Geis, 1984). The continued expansion of diversion programs into the mid-1 97 Os was accompanied by growing concern about the operation of diversion programs and increasing demand for evaluation. A number of evaluations of diversion efforts have now been conducted, and these will be reviewed in the next section. CHAPTER 2 DIVERSION AND PROGRAM EFFECTIVENESS: A REVIEW OF PRIOR RESEARCH As has been discussed, the lack of empirical research was a great concern to the critics of juvenile diversion. While diversion was advocated based on the claims that it would provide a less stigmatizing, less costly and more effective alternative to the formal juvenile justice system, there was little evidence that this was true. Indeed, there was even concern that diversion could not be implemented. Beginning in 197 3, a number of evaluation reports of diversion programs began to appear in the professional literature. In this section, the prior research on diversion programs will be reviewed. Particular attention will be devoted to assessing the extent to which diversion programs have succeeded in attaining the major goals of diversion. That is, this section will attempt to assess the success of diversion programs in: (1) diverting youths who otherwise would have been formally processed; (2) reducing the negative labeling of juvenile offenders; (3) providing a less costly alternative to the justice system; and, (4) reducing recidivism of project participants. As will become clear in the following discussion, there is a great variety in the structure and operation of diversion programs. These programs vary in the point in the system where diversion takes place (e.g., police contact or court intake), the sponsor of the service program (e.g., private agency or police department), and the type of services provided to the youths. Since each of these characteristics of diversion programs may influence a program's effectiveness, a 16 17 global assessment of diversion which ignores programmatic differences may mask important trends. If, for example, 20 evaluations of diversion programs had been conducted, 10 with positive results and 10 with negative results, one might conclude that there is little consistent evidence on which to base decisions about diversion programming. However, if one were to look more closely at the 20 studies and find that the 10 positive evaluations were of programs which diverted youths at the point of police contact while the 10 negative evaluations were of programs which diverted youths at court intake, then one would have substantial information for designing future diversion efforts. Therefore, in a later section, an attempt will be made to assess patterns in the effectiveness of diversion programs by considering the relationships between recidivism rates and major programmatic factors, including the point of diversion, the type of program sponsor, and the type of services provided. Finally, the methodological quality of the prior research in diversion will be addressed. In the past, many authors have bemoaned the small sample sizes, the lack of rigorous research designs, and other problems when reviewing the diversion research. Whether the quality of diversion research has been sufficient to provide valid and reliable data is a critical consideration. Without reasonably high quality research there is a limited ability to draw any conclusions about the effectiveness of diversion. The studies included in the review will be evaluations of diversion programs in which recidivism was used as an outcome variable in comparing diversion to a service program with an alternative method for handling juvenile offenders. With the exception of two recently released reports, this review will only include evaluation reports published in the United States before 1983. A total of 15 studies were identified that met the above criteria. Since some of these studies include more than one diversion program, the review actually covers 32 separate diversion programs. 18 The 15 studies will be presented according to the type of research design used. The nine studies employing a quasi-experimental design will be presented first, followed by the six studies employing a true experimental design. This organizational format was used since one might place a differing amount of confidence in the findings of studies which vary in the rigorousness of their research design. Quasi-Experimental Evaluations Juvenile Justice Service Center#Wayne CountyAdichiggn One of the earlier evaluations of a diversion program to be published was conducted by Kelly, Schulman, and Lynch (1976). In this study, youths were diverted at court intake and referred to one of three neighborhood-based programs sponsored by the Wayne County Juvenile Court. A comparison group was formed by youths who were referred back to court intake because they did not live in the service program catchment area. A total of 75 youths were assigned to each condition. The 75 youths in the comparison group were selected from a total of 300 youths who did not live in the catchment area. The method of selecting these 75 youths was not described, but the authors did mention that the comparison youths lived in "selected geographical areas adjacent to the three target areas" (p. 9). Youths in the diverted group were primarily criminal offenders (57%). The average age was 14.8 years, and the group was 73% male and 95% black. Youths in the comparison group were also primarily criminal offenders (64%). Their average age was 14.3 years, and this group was 79% male and 68% black. Services provided to the diverted youths consisted of counseling and service brokerage. These services were provided over a period of 90 days with weekly or l9 biweekly contact with the youths. If problems persisted and the family agreed, services were extended beyond the 90 days. At an average of 6 months after referral from court intake, 17% (n = 13) of the diverted youths had one or more subsequent formal court complaints filed against them while 37% (n = 28) of the comparison youths had subsequent court complaints filed against them (Q < .01). The authors did not report any data reflecting the extent to which project youths were actually diverted from court processing, any assessment of cost- effectiveness, or information regarding the impact of the program on the labeling of its clients. Juvenile Aid Panels in Australia A diversion program in South Australia was studied by Sarri and Bradley (1980). Their evaluation focused on "juvenile aid panels" which were set up to serve as an alternative to court referral for youths charged with less serious offenses. Referrals were made to a local panel by the police. The panel, which was presided over by a police officer and a "community worker," attempted to work out a plan of action that was acceptable to all the parties involved. The panel could counsel and release the youth with no further action, develop a written agreement specifying certain actions to be completed by the youth and/or the youth's parents, refer the youth to counseling services, or even refer the youth to court. The juvenile aid panel program was administered by a public agency, the Department of Community Welfare. The evaluation of the juvenile aid panels consisted of a comparison of the recidivism of youths participating in a panel with the recidivism of youths who were handled by a juvenile court. The comparison used existing groups (youths were not assigned to one group for the purposes of the research). All of the youths in both conditions were first offenders ranging in age from 10 to 18 years 20 old. The diverted group was 71% male and the court processing group was 85% male. In presenting the results of their study, Sarri and Bradley reported that over a 5-year period, 21% of the youths "committed at least one new offense, with the overall figure distributed fairly evenly between the panels and the court" (p. 58). As in the previous study, cost comparisons were not conducted and labeling was not assessed. However, Sarri and Bradley did address the implementation issue, findiig substantial evidence of net-widening. In particular, the investigators found that police use of release with a warning dropped "sharply" after the startup of the panels. They further reported that "processing through the panels increased far more rapidly than did the overall rate of increase in police apprehension of juveniles" (p. 51) and that this increase was not consistently accompanied by a decrease in referrals to court. Based on these findings, Sarri and Bradley concluded that the panels actually increased the number of youths in the system of social control. While this study failed to adequately address the effectiveness of diversion because of its lack of a rigorous research design, it did provide support for the concern about the possible net-widening effect of diversion programs. Police Referral Program in California The remaining quasi-experimental studies used a more rigorous research design in evaluating program effectiveness and, as a general rule, greater confidence can be placed in the results of these studies. The first such study was conducted by Lincoln (1976). Lincoln compared youths referred to a diversion program (5 = 30) with a "typical group" (p. 323) of offenders contacted within the 40 days prior to program startup (n = 250) and with a matched group of youths (n = 30) selected from the "typical group." The youths were matched on several 21 variables including prior record, seriousness of the instant offense, age, race, and sex. Youths in the diverted group were diverted at the point of police contact. A police officer acted as a service broker, referring the youth to 1 of 10 different private agencies. These agencies provided health care, recreational and social opportunities, employment assistance, and counseling and protective services. Youths in the comparison group were handled by the police in their normal manner. The diverted youths were 43% first offenders, 75% male, predominantly nonwhite, and their average age was 16 years. Lincoln did not report any cost or labeling outcome data. In an assessment of recidivism, Lincoln concluded that there were no substantial differences between the diverted group and the "typical group," or between the diverted group and the matched, nondiverted group. For the matched groups comparison, Lincoln found that within 1 year of the instant offense, 53% (g = 16) of the diverted group and 47% (g = 14) of the comparison group had recidivated. In addition, the diverted youths committed an average of 1.7 officially recorded, subsequent offenses as compared to an average of 1.1 offenses for the nondiverted youths. However, Lincoln cautioned that interpretations of this latter data should take into account the small number of youths involved in the comparison. Finally, Lincoln assessed program implementation by examining the actual disposition of youths in the matched comparison group. She found that 12 of the 30 youths were released with no further action and therefore, inferred that nearly half of the "diverted" youths were actually In no danger of formal processing. Lincoln concluded from these data that some officers may have used referral as a form of social control. 22 Southeast Early Diversion Project, Los Angeles County, California The next study also used more than one research design in evaluating program effectiveness. In fact, four different research designs were used by Lipsey, Cordray, and Berger (1981) in their evaluation of a diversion program in Los Angeles County, California. The research designs were: (1) a comparison of youths receiving different levels of services; (2) "tie-breaking randomization;" (3) "regression-discountinuity;" and, (4) matched groups. The diversion program consisted of three separate projects, each receiving referrals from two police agencies. The projects were administered by the city at each site, and a sergeant from the Los Angeles County Sheriff Department served as the director. The projects provided service brokerage for the diverted youths, referring them to agencies that generally provided individual and family coun- seling, as well as some remedial reading, recreational, and substance abuse services. With a few exceptions, the youths in the program were criminal offenders. While Lipsey et al. reported that there was "at least weak evidence that recidivism was lower among some juveniles who received diversion services than among comparable juveniles who did not receive the service" (p. 297), the implications of the study for the effectiveness of diversion are not very clear. First of all, there was ample evidence to indicate that the program did not succeed in diverting many youths. A discriminant function analysis was used to classify referrals from four of the police departments as youths who were similar to those normally referred to court or similar to those normally counseled and released. The discriminant function analysis, which had a classification accuracy of 88%, classified 76% of the "diverted" youths as youths who normally would have been counseled and released. The rate at individual projects ranged from 70% to 84%. Also, across the three sites, there was only a slight decrease in the 23 preportion of arrested juveniles who were referred to court. The decrease ranged between 2.2% and 5.1%, depending on the source of the arrest data. Site specific figures were not reported. Given the small proportion of youths in the study in jeepardy of court referral, the evaluation did not provide an adequate test of diversion. The results of the four research designs used in the study do not indicate support for diversion. In the first design, the recidivism rates of youths receiving different amounts of services were compared. This did not constitute a test of diversion since diverted youths were not compared with youths handled in another manner. While this design provided some indication of the effectiveness of the services, it did not test diversion's effectiveness as an alternative to the juvenile justice system. Results from each of the other three research designs indicate that there were no significant differences in recidivism between the diversion program and the comparison groups. While there were some positive trends in favor of the diverted youths, this was only true when "diverted" youths who resembled counseled and released youths were compared to a similar group of youths. The study conducted by Lipsey et al. may indicate that the program has some potential to be an effective treatment for minor offenders, but it provided no support that it is an effective alternative for youths who are normally referred to court. Alternate Routes, Orange County, California The next study was an evaluation of the Alternate Routes diversion program which Operated in two suburban communities in Orange County, California. While there were two project sites, only one was included in the outcome evaluation. This program was Sponsored by the court and staffed by probation officers. The program provided individual counseling, group counseling, and diagnostic services. 24 The staff reportedly had small caseloads which allowed them to have "intensive contact" (p. 303) with the youths and their families. In judging the success of the program in implementing diversion, it is important to note that the project received referrals from the police, schools, parents, and the youths themselves. While the author, Gilbert (1977), reported that the "majority" (p. 305) of the youths in the project were referred by the police and the schools, exact figures were not provided. The inclusion of youths referred from schools, parents, and neighbors indicates that there is at least a possibility that a large prOportion of the youths in the program were not in jeepardy of formal court processing. 1 To the credit of the program, however, in a survey of the police, Gilbert found that 7 5% of the police referred youths in one site and 80% in the other site would have been formally processed had the program not been available. In addition, Gilbert reported that the two project sites had a greater decrease in juvenile arrests and court referrals than the rest of the county. These data indicate that the program may have diverted some youths, but it is impossible to estimate the magnitude of diversion. In order to evaluate program effectiveness, two groups in one of the sites were compared on recidivism. A random sample of program youths (n = 58) was compared to a random sample of youths with similar arrest records processed by the police in the year prior to the startup of the program (3 = 78). Gilbert reported significantly less recidivism for the diversion program group than for the comparison group. At a 6-month follow-up, the recidivism rates were 29% for program participants and 53% for traditionally processed youths (p < .01), and at the 12-month follow-up the rates were 34% and 65% (p < .01). While no data were provided, it was also reported that the Alternate Routes program was less costly than traditional processing. Again, labeling was not addressed. 25 A Multi—Site Evaluation One of the most ambitious studies employing a quasi-experimental design was conducted by Elliot, Dunford, and Knowles (1978). This study was an evaluation of three separate diversion programs located in Boulder, Colorado, Kansas City, Missouri, and San Antonio, Texas. In all three sites the research design involved a comparison of youths who were either diverted to a service program (DTP), diverted without services (DWS), or given traditional processing (TP) within the juvenile justice system after police contact. These were naturally occurring groups although there was some matching of youths across the three conditions. The sample sizes at each site ranged from 50 to 102 per condition. Youths in the DWS group were released without further legal action or treatment, while youths in the TP group were referred to court. Youths who were in the DTP group were referred to community agencies by the police. The type of services received by the diverted youths varied somewhat across the sites, but were generally traditional. In Boulder the youths primarily received individual counseling and recreational services. In Kansas City the services were predominantly recreational in nature. And in San Antonio most youths received individual counseling. The youths included in the study across all sites ranged in age from 7 to 17 years old, with most being 14 to 16 years old. Across the sites the youths were 71% male. In Kansas City and San Antonio the youths were predominantly white and middle class. Implementation of diversion was not assessed in this study, since the intent of the research was to examine naturally occurring groups rather than to increase (the preportion of youths who were diverted. Also, the relative costs of the different approaches were not assessed. Elliot et a1. did examine labeling and recidivism, however. 26 With respect to labeling, the authors reported that there were no significant differences in self-esteem or perceived labeling between the three conditions in each site. Contrary to labeling theory, there was a trend for DTP youths to report more labeling than youths in the other conditions. With respect to recidivism, there were no significant differences in any site in the percent recidivating or the number of rearrests when controlling for differences in prior record. One positive effect did emerge from the study. That is, a smaller proportion of DTP and DWS youths with prior juvenile justice involvement recidivated, and these youths had a lower total number of rearrests. The California Youth Authority Study The largest study included in this review was conducted by Palmer, Bohnstedt, and Lewis (1978) for the California Youth Authority. A total of 11 programs were included in their study of project outcomes. The 11 programs varied considerably in structure and operation. Diversion took place at the point of police contact, court intake, or both; programs were sponsored by the police, juvenile court, or private agencies; and the services were provided by the diversion program in some sites and by referral to community agencies in other sites. Even the evaluation design varied across the sites. The research design in each site called for a comparison of youths who were diverted to services with those who received traditional juvenile justice processing. The groups in nine sites were formed by matching program youths on age, sex, and ethnicity with pre—project cases identified in juvenile justice records, and the groups in the other two sites were formed by random assignment of project referrals. Since most of the sites were evaluated using a quasi-experimental design and most of the analyses conducted on all sites combined, this study, including the two experi- mental sites, will be discussed in this section. 27 The demographic characteristics of the youths in each condition were quite similar when all sites were combined. Youths in the DTP group were 55.4% male, 66.7% white, and had an average age of 15.0 years. Youths in the TP group were 63.4% male, 61.0% white, and were an average of 15.3 years old. However, on one variable, the average number of prior arrests, there was a substantial pre- difference. The DTP group averaged .5 prior arrests and the TP group averaged .9 prior arrests. The results should be interpreted in light of this difference. Youths served by the 11 programs tended to be referred from sources other than the police and court; however, since Palmer and his associates were interested in evaluating diversion, they included only justice system referrals in their study. Once referred to a program, youths received a variety of different services. Across all sites, 72% received some family counseling, 52% received some individual counseling, and 9% received group counseling. No more than 5% received noncounseling services. On the average, each youth received only 5.5 hours of service over a duration of 6 weeks. Implementation of diversion was assessed in nine of the project sites. Palmer et al. reported that approximately 51% of the youths served by the nine programs had been truly diverted. This estimate was based on across site findings that 28% of the youths in the programs had been referred by nonjustice system sources, an estimated 14% would have been counseled and released and an estimated 7% would have been dismissed at court intake. The estimates were based on comparison group totals. The major outcome variable addressed in this study was simple recidivism. Included in this assessment were 1,345 youths in the DTP group and 1,192 youths in the TP group. When looking at all sites combined, Palmer et al. found that the simple recidivism rate was significantly less for the DTP condition (25.4%) than 28 for the TP condition (30.7%) at a 6-month follow-up (2 < .01).2 However, it is important to note that the positive results were due to the performance of only 3 of the 11 programs. That is, when the results were analyzed on a site by site basis, only three programs were found to perform significantly better than traditional processing. However, for the three successful programs, the difference in recidivism rates (DTP vs. TP) were large, ranging from 25.0% to 34.1%. Two of these three programs were effective despite the fact that the DTP group had a higher average number of prior arrests than the TP group. Finally, Palmer et al. performed one of the most sophisticated cost analyses of those conducted by studies included in this review. The authors reported that the average cost of diversion services was $250 per client, while the average cost of arrest was $511 and the average cost of probation was $477. Overall, in their assessment of cost-effectiveness, Palmer et al. found that diversion to a service program cost $31.95 less than traditional processing. Sacramento County Diversion Project, California The next two diversion programs were awarded "Exemplary Project" status by the Law Enforcement Assistance Administration (LEAA). The first of these was the Sacramento County 601 Diversion Project, evaluated by Baron, Feeney, and Thornton (1973). Unlike the other programs described thus far, the 601 Diversion Project served only status offenders. Youths were diverted at court intake to a program that was Sponsored by the juvenile court and staffed by probation officers. The program provided family crisis intervention and found alternative placements for youths who otherwise would have been placed in a detention facility. The maximum number of sessions of family counseling was five. No demographic information about the youths in the study was provided in the report. 29 The evaluation design included a comparison of youths who were assigned to the diversion program with youths who were handled by the regular court intake unit. Assignment to the two conditions was made according to the day of the week that the youth was referred to court. On 4 days of the week youths referred to court were diverted to the 601 Project and on the other 3 days youths were handled in the normal manner. The days of the week which determined the assignment were rotated monthly. The number of youths assigned to each condition was not reported. The evaluation compared the relative cost of the two approaches and the simple recidivism rate. Neither labeling nor implementation were specifically addressed. While the authors did not address implementation, they did note that only 21.3% of the comparison group was subsequently petitioned. Assuming the groups were comparable, this would suggest that over three-fourths of the experimental group was not actually diverted. Baron et al. reported that the 601 Project was less expensive on a per case basis than traditional processing when the costs of intakes, court hearings, and supervision were included ($29/case vs. $222/case). Finally, concerning recidivism, Baron et al. reported that at 7 months, 35% of the DTP group had recidivated as compared to 45.5% of the TP group. Unfortunately, no inferential statistics were reported. Without knowledge of the number of youths in the study, little meaning can be attributed to the differences found. Community Arbitration Project, Anne Arundel County, Maryland The second Exemplary Project to be reviewed is the Community Arbitration Project (CAP) in Anne Arundel County, Maryland (Blew 6: Rosenblum, 1979). Youths participating in the project were diverted at intake screening which took place prior to actual court referral. The intake screenings were conducted by the 30 County Juvenile Services Administration which also administered the diversion program. Youths who were diverted to the program attended an arbitration meeting along with their parents, and the victim/complainant. These meetings were presided over by a lawyer who determined the disposition. For the first 4,233 youths attending an arbitration meeting, the dispositions were as follows: 38% closed with a warning or dismissed for lack of sufficient evidence; 8% referred to the state's attorney for formal action; 4% continued for additional investigation; 3% referred to court intake or probation; and 47% handled informally. For cases handled informally, the youths were assigned to do voluntary community service work (usually 10 to 25 hours), were asked to make financial restitution if apprOpriate, and were referred for other services, such as counseling, at the discretion of the arbitrator. The program usually lasted about 90 days or the amount of time needed for the youth to complete his/her assignment. Youths served by the program were misdemeanor and minor felony offenders. Nearly half were 16 or 17 years old. Also, the group was 80% male and 75% white. The program was evaluated by comparing a group of youths diverted to the program (3 = 482) with a group that was assigned to traditional court intake (3 = 342). Every fourth youth referred to the diversion program was included in the study and every fifth youth who normally would have been diverted was assigned to traditional intake. At what appeared to be a 12-month follow-up, there was not a significant difference in the percent recidivating; 9.8% of the DTP youths were charged with at least one offense as compared to 14.3% for TP youths (p < .07). However, the DTP group did have a significantly smaller average number of recharges (.415) than did the TP group (.659) (p < .01). Blew and Rosenblum reported that the project was less expensive than traditional processing ($30/case vs. $37/case). The authors also reported the project was 31 successful in maintaining program involvement since 85% Of the youths completed their assignments. Implementation of diversion was partially assessed. The authors reported that there was a 19% increase in police referrals to court, indicating possible net- widening. However, this may have been compensated for by an increase in the use Of dismissals by the program. Before the diversion program was in Operation, traditional intake had dismissed 4.4% of its cases for insufficient evidence. In comparison, the program dismissed 30.6% of its cases. It was also noted that traditional intake had formally processed 75% of its cases prior to the program and the Community Arbitration Project formally processed only 7.2% of its cases. However, the extent to which the total prOportion of formal processing had changed since the inception of the project was not reported. It could be that the youths referred to the CAP were those who were normally dismissed by traditional intake. Sum man In this section nine different evaluations of diversion programs have been reviewed. Each Of these used a quasi-experimental research design. A number of Observations can be made at this point. First of all, despite the importance of labeling and program cost to the rationale Of diversion, most studies failed tO address these issues. In fact, the impact of diversion on labeling was addressed by only one study (i.e., Elliot et al., 1978) and the relative cost of diversion was addressed in four studies (i.e., Baron et al., 1973; Blew dc Rosenblum, 1977; Gilbert, 1977; Palmer et al., 1978). To summarize the results of these studies it should be noted that Elliot et a1. (1977) found no differences with respect to labeling, and all four of the studies listed above found diversion to be less expensive than traditional processing. 32 The second Observation that can be made is that these studies provided very little evidence that diversion was successfully implemented. While all but two of the studies conducted some sort of implementation assessment (only Elliot et al., 1978 and Kelley et al., 1976 did not include an assessment of diversion im- plementation), the results of the other studies provided strong evidence that the diversion programs produced a great deal of net-widening. For those studies that provided some estimate of the proportion Of youths who were actually diverted from further juvenile justice system processing, these estimates ranged form 80% (Gilbert, 1977) to 21% (Baron et al., 1973). Most of the estimates indicate that the preportion of program youths who were actually diverted ranged from 21% to 49% (e.g., Baron et al., 1973; Lincoln, 1976; Lipsey et al., 1981; Palmer et aL, 197 8). Finally, it can be Observed that while these studies provided some support for the effectiveness Of diversion when compared to traditional processing (the studies conducted by Baron et aL, 1973; Blew a Rosenblum, 1977; Gilbert, 1977; Kelley et al., 1976; Palmer et al., 1978 found diversion to a program to be more effective than traditional processing on at least one measure Of recidivism), the methodological problems Of these studies, the lack Of detailed reporting, and the inherent weaknesses Of quasi-experimental designs provide a poor base of in- formation on which tO make conclusive statements about the impact of diversion on recidivism. Experimental Evaluations San Fernando Police Department Crisis Intervention ProgramLCalifornia The first published evaluation of a diversion program to use an experimental design was conducted by Stratton (1974) in San Fernando, California. The diversion program served youths who had been referred to the police or arrested 33 for status and misdemeanor Offenses. The only information provided about the participants was Stratton's statement that the groups were "essentially equal" (p. 11) with respect to race, sex, and age. Shortly after police contact, usually within one hour, youths assigned to the program attended their first session Of family, crisis intervention. The average number of sessions attended by the youths and their families was 2.5. These sessions focused on "problem solving," "expression of feelings," and gaining an "understanding of the problem." The program was sponsored by the police department. Youths were randomly assigned to the diversion program (n = 30) or to traditional police processing (2 = 30). Youths assigned to the TP group were either informally counseled and released, referred to court, or placed in deten- tion. Unfortunately, data was not provided about the final disposition of youths in the TP group. Without this information it is difficult to know what the comparison group was or whether diversion was implemented. Further, since some youths in the TP group were informally counseled and released and youths in the DTP group only attended an average of 2.5 counseling sessions, it is not readily apparent that the two groups were handled very differently. A 6-month follow-up period was used to assess program outcomes. This 6- month period began "following the completion of the counseling portion of the study" (p. 11). Given this description, it is not clear whether the follow-up period was equivalent for all youths or even for the two groups. In a comparison of the recidivism rates, Stratton reported that nearly equal proportions of the two groups were rearrested (DTP = 30%; TP = 33%). However, in a comparison of multiple recidivism, Stratton reported that the DTP group was charged with a total Of 12 Offenses while the TP group was charged with 24 Offenses, a difference that was reportedly significant at p < .05. Stratton did not report any data relevant to implementationior labeling outcomes. However, Stratton did assess the relative cost of the two approaches. 34 He determined that the cost of traditional processing (including probation investigation, probation supervision, and detention) was 2.11 times greater than the cost of the diversion program. These costs appear to be based on the actual handling of the youths in each group, including the costs associated with rearrest during the 6-month follow-up. Juvenile Services Program, Pinellas CountyLFlorida The next study employing random assignment was conducted by Quay and Love (1977), who evaluated a Pinnelas County, Florida diversion program. Youths participating in the study were criminal and status offenders who ranged in age from 12 to 16 years. The average age of the youths was 15.8 years. The youths were 70% white and 72% male. Referrals were made by the police, court, schools, and other community agencies. Services were provided by a private agency and included vocational counseling, job training, job placement, academic education, and individual and group counseling. Diverted youths received these services for an average of 89 days. The evaluation was a comparison of youths randomly assigned to the diversion program (13 = 436) with youths who were randomly assigned to "whatever means were available to the juvenile justice system" (p. 377) (g = 132). While program costs and labeling were not assessed, Quay and Love did report positive program effects for recidivism. In particular, Quay and Love found that the DTP group had a significantly lower simple recidivism rate than the TP group (32% vs. 45%; p < .01), and there was a significantly lower average number Of rearrests for the DTP group (.62 vs. 1.00; p < .01). However, the reported positive effects must be interpreted with a great deal of caution for several reasons. First, the results reported above were based only on the post- program period for DTP youths. When both the post-program and during-program periods are combined, the simple recidivism rate Of the DTP group increases to 35 40% and the average number of rearrests increases to .86. Second, even these latter recidivism rates are based on a shorter time at risk for the DTP group. While the follow-up assessment occurred at an average of 450 days after referral for the TP group, the average duration from referral to assessment for DTP youths was only 400 days. Third, there were differences in the average number of prior Offenses in favor of the diverted youths (1.14 vs. 1.50). Further, it appears that many of the youths participating in the study were not candidates for diversion. Court referred youths had already been adjudicated, and school and agency referred youths had not had any formal contact with the juvenile justice system. Finally, in an assessment of the effectiveness of the program for different types of youths, Quay and Love reported significant differences only when comparing youths with 52 prior juvenile justice involvement. Thus, not only is there reason to question the successful implementation of diversion (no other data relevant to this issue was provided), but it appears that the program was ineffective when dealing with youths who may have been diverted. Community-Based Adolescent Diversion Project, Illinois The next study was unique among those reviewed in that a comparison was made between youths who were diverted to a service program and youths who were diverted without services (DWS) (i.e., released with no further action). Diversion took place at the point of police contact. Youths in the treatment condition were referred to a program sponsored by the University of Illinois at Urbana-Champaign. As a result of the investigators' (Davidson, Seidman, Rappaport, Berk, Rhodes, 6: Herring, 1977) attempts to refine the service program, two different evaluations were conducted in successive years. During the first year youths in the DTP group received a "combination of relationship skills, behavioral contracting and child advocacy" (p. 45) services. The services were provided by student volunteers. Youths met with the volunteers for 6 to 8 36 hours a week for an average Of 4.5 months. A total of 25 youths were randomly assigned to the treatment group and 12 youths were assigned to the no-treatment control group. Across both groups the youths had an average of 2.16 prior police, contacts, and their average age was 14.1 years. The groups were 75.7% male and 75.7% white. While the investigators found no significant differences in self- reported delinquency or in perceived labeling, they did report that the DTP group had a significantly lower recidivism rate (48%) than the DWS group (100%) 27 months after referral to the project. In the second year of the project, two distinct service programs were evaluated; one used behavioral contracting and the other used child advocacy. Also, the duration Of program participation was standardized at 18 weeks for each program. Twelve youths were randomly assigned to each condition. The condi- tions were DTP-contracting, DTP-advocacy, and DWS. As in the first year evaluation, no significant differences were found in perceived labeling or in self- reported delinquency. However, there were significant differences in favor of the treatment groups for simple recidivism (DTP-contracting = 33.3%, DTP-advocacy = 50.0%, DWS = 91.7%) 12 months after project referral. Cost comparisons were not provided. The lack of a comparison group that received traditional processing made an assessment of implementation difficult since the investigators could not look at actual dispositions of a traditionally processed comparison group. Concerning implementation, it was reported that there was a decrease in the number of petitions to the state's attorney (Seidman, 1981); thus, the project may have served youths who would have been formally processed unless the decrease was attributable to another cause. Youth Service Bureau, Mich1g’ an As part of a state-wide evaluation of youth service bureaus in Michigan reported by Davidson, Koch, Lewis, and Wresinski (1981), an experimental design 37 was employed in a comparison of youths who received services from a YSB (g = 59) with youths who were put into a "waiting list" control group (n = 31). Youths were referred to the court sponsored program from a variety Of sources including schools, parents, police, and the juvenile court. The program provided short-term counseling, service brokerage, and parent consultation services. Services were of low intensity. Based on a random sample Of youths served by the - YSB during its history, the median number of program-youth contacts was 1.3, with 76% of the youths having two or fewer program contacts. Again based on a random sample of youths served by the project during its history, program youths were 56.5% white, 59.3% male, and their average age was 13.9 years. The authors found mixed evidence with regards to the implementation Of diversion. Sixty-three percent Of the YSB clients had been referred by either the police or court. This was interpreted by the authors as representing an "upper limit" to the prOportion of program youths who had been diverted. Other evidence, including the young age of the clients and the proportion of court referred youths who were on probation at the time of referral, indicated that the actual prOportion Of diverted youths was actually smaller than 63%. The authors also found no support for diversion implementation in an examination of changes in the prOportion Of the juvenile pOpulation arrested or in the number of petitions received by the court when comparing pre- and post-YSB time periods. However, there was a substantial decrease in police referrals to court. Police referrals to court drOpped from 35.37/1,000 juveniles in the year prior to YSB startup to 31.30/1,000 in the year following YSB startup, and the rate continued to drop to 21.21/1,000 in the fourth year after the YSB was initiated. Program costs and labeling were not assessed, but an assessment Of Official and self-reported delinquency was included in the evaluation. Despite the use of random assignment, large pre—differences were noted in the juvenile justice 38 records of the comparison groups. Youths who received services from the YSB had an average Of .03 prior arrests and .17 prior court petitions, while the comparable figures for the control group were .19 arrests and .03 petitions. Given this finding, the investigators used an analysis Of covariance to control for the prior-record differences. Using this analysis, no significant differences were found in rearrests or court petitions at 3 months after referral to the project. An analysis of self—reported delinquency was hampered by similar "pre" differences, and again the analysis of covariance indicated no significant differences. The authors indicated that there was a substantial amount of subject mortality for interviews (approximately 50%) and that the usefulness of the self-report Of delinquency data was questionable. The Southwest Police Department Study, California One of the more recently reported studies was conducted by Klein (1979b). Klein compared youths who were randomly assigned to four different types of dISpositions: (1) diversion without services (11 = 82); (2) diversion with services (51 = 88); (3) diversion with purchase Of services (11 = 55); and, (4) traditional processing (3 = 81). Youths were referred to the project from nine stations in a southwestern California police department. Police officers acted as service brokers in this project. Referrals were made to over 40 different agencies. These agencies tended to be small and "almost all of the agencies were private" (p. 28). Klein also reported that for the most part these agencies provided "rather traditional" (p. 28) services, with an emphasis on individual, group, and family counseling. Little information was provided about the youths participating in the study although it was noted that they had an average of 1.29 prior offenses and were "deliberately drawn from a more serious Offender pOpulation" (p. 26). The assessment of program outcomes focused on labeling and recidivism. NO data were reported concerning implementation or the relative cost of the four 39 dispositional Options. Regarding recidivism, Klein reported that there was a significant difference in the percentage of youths with one or more subsequent rearrests and that the traditional processing group had the highest recidivism rate. At the time of a 27-month follow-up, the simple recidivism rates for the groups were: (1) DWS = 49%; (2) DTP = 57%; (3) DTP-purchase of services == 62%; and (4) TP = 73%. These differences were not only significant (p < .05), but also quite meaningful, especially, considering the long follow-up period. Despite the differences in Official recidivism, Klein did not find a significant difference in self-reported delinquency. It should be noted that some "cheating" (p. 30) occurred in the random assignment procedures, resulting in a higher incidence of prior records for the TP group than for the diverted group. However, Klein stated that these differences were significant only when comparing the TP group to all other groups combined. However, when controlling for the pre—differences, Klein found the results to be the same as those reported above. With regard to labeling, Klein found some support for diversion based on a labeling theory rationale. Specifically, Klein found that justice label encap- sulation (measured as the number of justice system contacts, the number of peOple who knew about the contact, and the number of people who thought the youth was delinquent) and delinquency label acceptance were positively related to self-reported delinquent behavior (2 < .05), although they accounted for only 8% of the variance. At the same time, there was not a relationship between referral for services and justice label encapsulation. This suggests that diversion to services did not increase delinquency labeling. However, Klein did not find any relationship between label acceptance and disposition. Given this latter finding and the finding of differences in official recidivism but not in self-reported delinquency, Klein concluded that this study provided greater support for the "societal reaction" version Of labeling theory than for the "self-concept" version. 40 National Evaluation of Diversion Projects The final study to be reviewed in this paper is the National Evaluation of Diversion Projects conducted by Dunford, Osgood, and Weichselbaum (1982). A total of 11 project sites participated in the study, but only four Of these were included in the experimental evaluation of program outcomes. These four sites were located in large cities in different regions of the country. It should also be noted that one Of the sites (Site 1) actually had two separate diversion programs included in the evaluation. The four sites represented a range of Options in the organization Of diversion programs. The points Of diversion used included police contact, court intake, and the state's attorney. Program sponsors included police departments and community agencies. However, the programs did not exhibit much variation on the type Of services provided. ‘ Each site used a service brokerage approach and the second program in Site 1 employed a short-term, crisis-intervention model. Here, the program was the primary provider of services. Across the four sites, services generally included individual, group, and family counseling. Also, programs provided recreation and employment services. The proportion of youths receiving each type of service did vary substantially across sites.3 The evaluation. was very broadly focused; it included an assessment of implementation, program costs, labeling, and recidivism. Implementation was assessed by looking at the proportion of youths who were formally processed at the point of diversion before and after the diversion program was initiated. Dunford et a1. included all 11 sites in this assessment and found confirming evidence of diversion in only three of the sites. In one site, the results were negative and in the other seven sites, the archival data was so unreliable that assessment was impossible. Difficulties were also experienced in the assessment of relative program costs and could be accomplished in only three of the 41 experimental evaluation sites. Using what Dunford et a1. referred to as "a set of highly qualified cost calculations for the juvenile courts" (1982b, p. 16), they reported that the cost per referral to the program was higher than the juvenile court in one site, less costly in one site, and of equal cost in the third site. The research design for the assessment of labeling and recidivism involved random assignment to one of three conditions. The conditions were: (1) diversion to the service program; (2) diversion without services (outright release); and, (3) traditional processing. The number Of cases per condition ranged from 76 to 211 across the four sites. A total Of 19 measures Of labeling were used to determine if levels of labeling differed by dispositional category at 6 and 12 months after referral to the project. Analyses of labeling included a two-way analysis of variance (disposition by time period), an analysis Of variance using the 12-month assessment of labeling with disposition and "selected respondent characteristics" (1982b, p. 7) as independent variables, and an analysis Of variance with the same independent variables as the preceding analyses using the raw change score from the pre-period assessment to the 12-month assessment. In reporting the results of these analyses, Dunford et al. stated that "the few instances in which significant effects were found revealed no patterns or trends, suggesting that disposition had very little effect upon the labeling experiences Of the youths" (1982b, p. 8). The same types of analyses described above were also performed with 10 different measures of self-reported delinquency as the dependent variable. Again, there was no consistent evidence that disposition had an effect. Recidivism was also assessed using Official records. Both simple recidivism and multiple recidivism (Operationalized as 0, 1, or 2 or more subsequent arrests) were measured for three different categories of offenses: felonies, felony- misdemeanor, and felony-misdemeanor-status. Dunford et al. found no significant differences for any recidivism measure at 6- or 12-month follow-up for Site 1 and 42 Site 4. There were a few significant differences at the 6-month follow-up for Site 2 and Site 3; however, these differences were not present at the time Of the 12- month follow-up. In addition, when the number of prior arrests was controlled for, there were no significant differences at any time period for Site 2. In an attempt to more closely examine the potential impact Of the diversion programs, Dunford et a1. looked at the relationship between recidivism and the amounts and types Of services received. Both project records and youth reports were used tO assess the independent variables. The results Of these analyses provided almost no evidence supporting a relationship between amounts and types of services and recidivism. However, Dunford et a1. did find that a large percentage of youths did not receive the services they were referred for. Across the different service programs, between 2% and 24% Of the youths never showed, and between 4% and 42% never completed the service program (Dunford, personal communication). Dunford et al. concluded that "the emphasis on stigma in the rationale for diversion projects has beguiled community based service providers into thinking that any program independent Of the juvenile justice system will be effective" (p. 21). Further, they called for the development of more innovative service programs that are based in theory and subjected to rigorous evaluation. Sum magy In this section, six experimental evaluations of diversion programs were reviewed. Despite the rigorous nature of an experimental design, the picture which emerges from these studies is far from clear. Two Of the studies indicated a positive impact on recidivism when diversion programs were compared to traditional processing (Klein, 1979b; Stratton, 1974), one study found diversion to a service program to be more effective than diversion without services (Davidson et al., 1977) and three studies indicated no difference between diversion to a 43 service program and either diversion without services or traditional processing (Davidson et al., 1981; Dunford et al., 1982; Quay & Love, 1977). While these studies were, in general, more adequately conceived and implemented than the quasi-experimental studies reported earlier, the experimental evaluations also suffered from problems that clouded the results. In these studies, there were too few attempts to adequately assess the implementation of diversion (only Davidson et al. (1977), Davidson et aL (1981), and Dunford et al. (1982) reported an assessment Of implementation). When an implementation assessment was attempted, the results were not generally encouraging. On a more positive note, it is commendable that several studies included an assessment Of labeling (Davidson et al., 1977; Dunford et al., 1982; Klein, 1979b) and several extended their assessment Of recidivism to include self- reported delinquency (Davidson et al., 1977; Davidson et al., 1981; Dunford et al., 1982; Klein, 1979b). Unfortunately, there was only mild support for a relationship between labeling and diversion, and no evidence that diversion programs have a positive effect on self-reported delinquency. CHAPTER 3 CONCLUSIONS AND RATION ALE FOR THE CURRENT STUDY In this section, an attempt will be made to judge whether diversion has been successful in attaining its goals. Of course, in order to make sound judgments about the successes and failures of diversion, the quality of the research must be considered. Therefore, judgments about the performance of diversion will be based on an evaluation of the amount and quality of relevant research. Finally, this section will conclude with an overview of and the rationale for the current study. 13 Diversion Effective? A total of 15 studies addressing the effectiveness of diversion were reviewed. These studies included evaluations Of 32 different projects that represented a variety of program structures and service types. TO some degree, these studies addressed the four major goals Of diversion: apprOpriate implemen- tation, reduction Of labeling, reduction of recidivism, and reduction Of costs. In this section, an attempt will be made to evaluate the 15 studies in terms of how adequately the four goals were assessed and tO what degree they indicate that diversion has been successful in achieving its goals. In addition, an attempt will be made to determine whether these studies presented empirical evidence supportive of the criticisms of diversion prevalent in the literature. These criticisms are that diversion: (1) may actually "widen the net" Of juvenile justice and increase social control; (2) may not prevent delinquency labeling but instead, 44 45 may result in another type of labeling; (3) may not reduce juvenile justice expenditures; and, (4) may not have been adequately evaluated. TO determine whether the studies provided valid information about the goals and criticisms Of diversion, a critique of the various research methods used will also be included. Implementation Of Diversion Implementation success is both difficult to define and difficult to assess. A review Of the 15 studies reveals that two approaches have been used to assess implementation and that these approaches imply two different definitions of implementation success. One approach has been to look for a decrease in more serious handling of juvenile Offenders beyond the point of diversion. This approach implies that the implementation goal is to effect an observable change in police or court handling of juvenile cases. The second approach has been to consider what prOportion Of the youths in the program qualifies as diverted from more serious handling. This approach implies that the implementation goal is to serve only those youths who would have been processed further in the juvenile justice system. The first goal, change in system processing, is reasonable only when the program is designed to be large enough to achieve the desired impact. The second goal, on the other hand, can be achieved by a small program in a large juvenile justice system. However, a goal of serving only apprOpriate youths is also difficult to achieve, and it requires considerable control over the referral process. While both of these goals can be considered reasonable indicators of implementa- tion success, achieving one should not be confused with achieving the other. In the following discussion, the two approaches to assessing implementation success will be considered separately. The discussion will include a consideration of the adequacy Of the methods used and an attempt will be made to determine how much is known about program implementation. 46 Assessng juvenile justice system trends. The first approach to assessing diversion implementation that will be discussed is an analysis of juvenile justice system processing trends. This approach attempts to determine whether or not the initiation of a diversion program is followed by changes that would be predicted if diversion had been successfully implemented. Most Often, investiga- tors try to determine if there was a decrease in the rate Of youths formally processed (e.g., referred to court) beyond the point of diversion (e.g., police contact) ( e.g., Davidson et al., 1981; Dunford et al., 1982; Lipsey et al., 1981). Other investigators look at police rates of. "warning and releasing" in search of evidence Of net-widening (e.g., Blew dc Rosenblum, 1977; Sarri 6: Bradley, 1980). Using this method, they interpret decreases in police use of warning and releasing as an indication that so called "diverted" youths were actually those who would have been released without further formal action. There are a number of methodological problems associated with evaluating diversion implementation through an assessment Of system processing trends. First, processing trends may be influenced by a number of events other than the startup Of a diversion program. While it may be possible to control for some extraneous variables by using equivalent control jurisdictions, alternative hypotheses for Observed differences still cannot be eliminated with a great degree of certainty.4 A second problem with the analysis Of system rates is that this approach implies that a successful diversion effort must be large enough to have a discernable impact on juvenile justice system processing. In many, if not most cases, it is unrealistic to expect that a service program with limited resources is capable Of producing a discernable reduction in formal processing. If this approach is chosen, its use should be preceded by an assessment of the maximum impact the program could have if it served the largest number of youths it was 47 capable of serving. A third and related problem is that the likelihood Of this approach demonstrating implementation success is reduced by including all types of Offenders in the calculations. Most diversion programs restrict eligibility on the basis of age, type of offense, and prior record. If the analysis of system processing is based only on youths in the target population, it would provide a more powerful test Of implementation. Seven of the 15 studies reviewed used an analysis of some type of system processing rate to assess diversion implementation. On balance, the studies provided some evidence that diversion efforts may have reduced formal processing rates. A study reported by Seidman (1981) found that initiation of a diversion program was followed by a reduction in referrals to the state's attorney, and Dunford et al. (1982) reported a decrease in formal processing in three out of four sites. For those studies which compared formal processing rates in a diversion site with rates in other jurisdictions, all reported greater decreases in the diversion site on some measures (Davidson et al., 1981; Gilbert, 1977; Lipsey et al., 1981).5 However, it should be noted that in the Davidson et a1. study, implementation was also assessed for six other sites. In this assessment, only one additional program showed a decrease in formal processing following the startup of the diversion effort. Negative findings were also reported by Sarri and Bradley (1980) and by Blew and Rosenblum (1979). These authors reported that program initiation was followed by a decline in police use of the warn-and-release disposition and not followed by decreases in referrals to court. The results from studies that analyzed system processing rates are Obviously mixed. While the various methodologies employed may cast some doubt on the believability Of results, there is little reason to expect consistent findings. The ability to achieve an impact on juvenile justice processing patterns is situationally determined. Fortunately, whether or not programs achieve this elemental goal, 48 an assessment Of diversion is still possible. In fact, it may be more reasonable to evaluate the performance of diversion in terms of its ability to serve the right pOpulation, to provide a less expensive alternative to formal processing, to reduce labeling, and to reduce recidivism for participants b_e_f_9_r_e attempting to change system processing. Assessiltg client typology. The second approach that has been used to assess implementation is to determine what proportion of referred youths would have been formally processed if the diversion program were not in existence. Several different methods have been used, each with its own methodological problems. One Of the more problematic methods was employed by Gilbert (1977). Gilbert surveyed the referral source to determine which youths normally would have been formally processed. Assuming that a certain type Of referral was requested, the report of the referral source may have been influenced by the strong demand characteristics of the situation. In this instance, the results Obtained may have been an overestimate Of the amount Of diversion taking place. A second method that has been used for estimating the prOportion of referrals actually diverted is tO look at the referral source Of program clients. Here, youths referred from sources other than the juvenile justice system (e.g., parents, schools, and agencies) are not counted as diversion cases because these youths are not considered to be in great jeOpardy Of formal processing. While information about the referral source of program clients may be useful in estimating the prOportion of clients who have been diverted, this information does not necessarily provide an indication of the program's success in implementing diversion. A program may serve nondiverted youths, and still successfully implement a diversion effort. If the program was designed to serve nondiverted youths, a different method must be used to assess implementation. It should be noted that when the referral source is used to assess 49 implementation, the proportion Of youths referred from the justice system represents an upper limit Of the proportion of clients that have been diverted. Within the group of youths referred from the justice system, there are likely to be youths who would not have been formally processed in lieu of referral to the program. Referral source was considered by Davidson et a1. (1981) and Palmer et al. (1978) as Bait. of their assessments of implementation. In both studies, the authors noted that large percentages of program clients were referred by nonjustice system sources.6 If the programs were designed to serve diverted and nondiverted youths, such findings would not necessarily have indicated implementation failure. A third method that has been used to assess implementation is the examination of actual dispositions of youths in a comparison group. The comparison groups used have been of two types. One type has been composed of youths who were randomly assigned to a "traditional processing" condition. The other type of comparison group has been a matched group composed of youths who were in contact with the justice system prior to initiation Of the diversion program. While this method may provide a good estimate of the proportion of program youths actually diverted, there are some potential problems. First, even if random assignment is used, and the comparison group members are all processed further in the justice system, the disposition of these youths may not reflect "traditional" processing. This situation could result from a different type of youth being referred for random assignment than is normally referred for traditional processing. Here, because of net-widening, a new dispositional pattern will emerge. The second possible problem with this method is that a comparison group developed through matching may still not be equivalent to the program group. This is especially true when the variables that are available in the juvenile 50 justice records for use in matching are not strongly related to the dispositional decision. Two studies explicitly used an analysis of comparison group dispositions to assess implementation. In the study conducted by Lincoln (1976), it was reported that 60% of the matched comparison group was formally processed. Using seven matched and two random comparison groups, Palmer et al., (1978) judged 71% of the justice system referred youths to be truly diverted. In a third study, conducted by Baron et al. (1973), youths processed on certain days were assigned to the comparison group. While implementation was not specifically addressed in this study, it was reported that only 21.3% Of the comparison group was formally processed. This indicates that a very small proportion of program youths was actually diverted. The final method Of assessing diversion implementation to be discussed was used in a study conducted by Lipsey et a1. (1981). Lipsey et al. used a discriminate function analysis to classify program participants as typically receiving "formal" or "informal" handling. The classification Of youths according to dispositional type was based on an analysis of youths contacted by the police prior to the startup of the diversion program. Only 24% of the program's clients were classified as "formal" youths using this method. In this study, the discriminant analysis had a classification accuracy of 88%. Use of a discriminant analysis appears tO be a promising method for assessing implementation. Using this method, the researcher can avoid some Of the problems associated with the other methods and still estimate the proportion of program clients that was diverted. One potential problem with this method, however, is that it may not be usable in all situations. For the discriminant analysis to be of value it must have a high level of classification accuracy. However, variables related to the disposition may not always be available in police or court records. 51 Conclusion. Several different methods have been used to assess the implementation of diversion. These methods may be categorized according to two broad types, an assessment Of system processing trends (e.g., a comparison of formal processing rates or warn-and-release rates for the pre- and post-project time periods) and an assessment Of client typology (e.g., classifying diversion program clients as either "formal" or "informal" youths based on the youth's referral source, the breakdown of the disposition Of youths in a comparison group, or the results of a discriminant analysis). However, each method may be inappropriate or impossible to use in a particular evaluation setting, and each method has its own methodological problems that weaken the validity Of the assessment. When diversion implementation is judged in terms of a program's ability to effect changes in system processing trends, mixed results are found. In the studies conducted by Blew and Rosenblum (1977) and Sarri and Bradley (1980), initiation of the diversion effort was found to be followed by a decline in police use Of a warn-and-release disposition and by no change or an increase in the number of referrals to court by the police. In contrast, Dunford et a1. (1982) found that there was a decline in formal processing in three out of four sites following the initiation of a diversion program as did Seidman (1981) for the one site described in this study. To further confuse the results, Davidson et al. (1981) reported that diversion program sites had greater decreases in formal processing (on some measures) than control jurisdictions in two out of seven sites, while Lipsey et al. (1981) reported greater decreases in formal processing for all three diversion sites than their control jurisdictions. Even when a reduction in formal processing was found, the reductions were usually small (e.g., Lipsey et al. (1981) found decreases Of only 2.2% to 5.1% in formal processing depending on the 52 measure used) and, given the nature of the research design, one cannot attribute these decreases to the diversion program. When implementation of diversion is judged by the prOportion of program youths classified as youths who would have been formally processed, the results tend to be negative. The estimates of apprOpriate diversion referrals range from 21.3% (Baron et al., 1973) to 80% (Gilbert, 1977) across the five studies that used this approach. However, the most Optimistic estimate was Obtained using the method that was most likely to result in an overestimation (i.e., the 80% estimate was Obtained by Gilbert (1977) by using the referral sources judgment Of which youths normally would have been given a formal disposition; a method that could have inflated the estimate given the demand characteristics Of the situation). When the results of these studies are combined with the results Obtained in those studies noted earlier (i.e., Blomberg, 1977 ; Klein 6: Teilman, 1976), there seems to be little doubt that there is a tendency for diversion programs to widen the net Of juvenile justice. Labeligg Labeling theory has served as the primary theoretical rationale for diver- sion. As described earlier, it has been proposed that diversion will reduce recidivism because diverted youths will not experience the same level Of negative labeling as those youths who are formally processed by the juvenile justice system. Despite the importance Of labeling to juvenile diversion, only 4 of the 15 studies attempted to assess the impact Of diversion on labeling (i.e., Davidson et al., 1977; Dunford et aL, 1982; Elliot et al., 1978; Klein, 1979b). Even though labeling was addressed in only four studies, it should be noted that these studies included nine separate projects, and that the research methods used in these studies were generally more rigorous than those used in the other studies. For 53 example, these studies used larger sample sizes and longer follow-up periods. In addition, three Of the four studies employed a true experimental design. The results of the four studies were very consistent in demonstrating no support for the hypothesis that diversion will result in less negative labeling than formal processing. This was true regardless of the manner in which labeling was Operationalized. For example, three studies found no evidence that disposition was related to the youth's perception of "negative labeling by parents, teachers, and peers" (i.e., Davidson et al., 1977; Dunford et al., 1982; Elliot et al., 1978).7 Also, three studies found no significant differences between dispositional groups in terms of the youth's gig-perception. Self-perception was defined differently across the studies, and included the youth's "self-image" as "conforming" or "bad" (Dunford et al., 1982), the youth's "self-esteem" (Elliot et al., 1978), and the youths "acceptance" of a delinquency label (Klein, 1979b). While the research cited above provided no support for the effectiveness of diversion in reducing labeling, there was also no evidence provided to support the notion that diversion to a service program may result in label substitution. As was mentioned previously, diversion critics have proposed that while diversion tO a service program may prevent delinquency labeling, it may also result in the youth being labeled as "sick" or "in need or treatment." This possibility was addressed in the studies by Klein and Dunford et al.. Klein found no significant difference between dispositional groups in the youths' acceptance Of an "in need of treatment" label. This finding was supported by Dunford et al.. In this study, there were no significant differences in the youths self-image as "sick," and only one instance of a significant difference in the youths' perceptions that others applied a "sick" label to them.8 Even though the Klein and Dunford et al. studies did not provide support for label substitution, the results may be limited in generalizability. It seems 54 plausible that label substitution is likely to be related to several program characteristics including, the intensity and types of services provided and the orientation of program staff. To the extent that the service programs in the Klein and Dunford et al. studies are not representative Of all diversion programs, one must question the generalizability of the results. This point is not intended as criticism of these two studies. Rather, it is intended to suggest that caution should be exercised in generalizing the findings Of these studies as they relate to label substitution. Conclusions. A few well designed studies have been conducted in which the impact of diversion on labeling has been assessed. These studies consistently found that disposition was not related to delinquency labeling, both in terms of the youths' self-perceptions (Dunford et al., 1982; Elliot et al., 197 8; Klein, 1979b), and in terms of the youths' perceptions of other persons' beliefs about them (Davidson et al., 1977; Dunford et al. 1982; Elliott et al., 1978). In addition, two studies addressed the possibility that diversion to a service program may result in label substitution, but no evidence was found to indicate that label substitution had occurred (Dunford et al., 1982; Klein, 1979b). Prgggam Costs A third goal for diversion is to provide a less costly alternative to traditional juvenile justice system processing. Six studies included some assess- ment of the cost of diversion services. In most cases, the assessment was a simple comparison of the per youth cost of program services with the per youth cost of formal processing. However, in two studies cost comparisons included an assessment of cost-effectiveness (Baron et al., 1973; Palmer et al., 1978). With very few exceptions, investigators have reported that diversion services are less costly than juvenile justice system processing. In general, this 55 finding can be extended to assessments of cost-effectiveness. Baron et al. reported greater cost-effectiveness for the 601 Diversion Project, and Palmer et al. reported greater cost—effectiveness for 7 Of the 11 projects included their study. Despite the positive results reported in these studies, actual cost savings from diversion efforts may be difficult to achieve. In fact, total system costs for serving juveniles may actually increase. The first reason for this is related to the finding that diversion programs have tended to serve many youths who were not actually diverted. These youths never would have been processed by the juvenile justice system and therefore, any costs associated with providing services to them are additional costs to the system. With the added costs Of serving nondiverted youths, diversion programs may not be less costly than the juvenile justice system. In fact, this was found to be true in a study of 11 California Youth Authority diversion programs (Bohnstedt, 1978). Bohnstedt reported that when the costs of serving "nondiverted" youths were considered, the cost of diverting a youth exceeded the cost Of juvenile justice processing. A second reason why diversion may not result in cost savings is that expenditures Of the juvenile justice system must be reduced to produce cost savings.9 If this does not happen, the diversion program will only be an added cost to the juvenile justice system. Unfortunately, none of the studies addressed this issue. However, it may not be reasonable to expect a reduction in juvenile justice system expenditures until after diversion has proved its effectiveness and has become a permanent part of the system. At that time, expenditures by the formal system could be reasonably reduced. Conclusion. To the extent that the diversion programs included in this review are representative of diversion programs in general, it appears that diversion may provide a less costly alternative to formal processing. At least, 56 that potential has been demonstrated. A total of six studies included some assessment of the relative cost of diversion programs and traditional processing (Baron et al., 1973; Blew 6t Rosenblum, 1979; Dunford et aL, 1982; Gilbert, 1977; Palmer et al., 1978). Of these six studies, only Dunford et al. did not find consistent cost savings in favor of the diversion program. However, in the three sites included in their assessment, Dunford et al. did find that diversion to a service program was less costly in one site, more costly in one site and Of equal cost in one site. While some evidence was provided for reducing costs through diversion programming, the ability to produce cost savings will be influenced by four factors: (1) the extent to which diversion programs serve only those youths who would have been formally processed; (2) the impact Of the program on recidivism; (3) the extent to which expenditures of the formal system are reduced; and, (4) the cost of the treatment program. Recidivism It may be obvious at this point that, at best, the research on diversion presents a mixed picture of diversion's efficacy in reducing official recidivism. For the 32 programs evaluated, the results reported suggest that 8 were effective in reducing recidivism, 22 were ineffective, and 2 were effective g1: ineffective depending on the measure of recidivism considered (see Appendix A for a site by site description of each diversion program evaluation). To interpret the mixed results of diversion program evaluations, it is important to ascertain what factors account for the variation. For example, it may be that the overall quality of the evaluation methods that have been used is so poor that there is not yet adequate information for judging the effectiveness of diversion. Or it may be that the differences in program outcomes are attributable to differences in the evaluation methods employed. Perhaps less rigorous 57 research designs or invalid measures are responsible for the reported outcomes. In any case, it is important to consider the methodological quality of the prior research in judging the success of diversion. Another important factor to consider in judging diversion's effectiveness is the wide variety of program types. An evaluation Of diversion is not an evaluation Of a Specific, narrowly defined program. Diversion programs vary on many dimensions including, the type of services provided (if any), the point at which diversion takes place, and program sponsorship. The programs that have been evaluated in the past vary considerably on these dimensions. Given this variation, it may be unreasonable to expect consistency in program outcomes. Instead, it may be more reasonable to explore the relationship between diversion program characteristics and program outcomes. Recidivism and research methods. More than eight years have passed since Gibbons and Blake (1976) first noted the poor state of diversion research. Since that time, many of the criticisms Of the research expressed by Gibbons and Blake have been addressed. Over the past eight years experimental evaluations of eight diversion programs have been published. These evaluations have responded to previous criticisms by generally including larger sample sizes, longer follow-up periods, and multiple measures of program outcomes. While some methodological problems have been avoided in these studies, there are still some major faults that prohibit an accurate assessment Of diversion's effectiveness. The first and most significant problem with prior diversion research is that the methods used to assess implementation have generally been inadequate. Because there has rarely been an accurate assessment of the proportion of program youths who were actually diverted, it is difficult to judge the extent to which diversion has been evaluated. In fact, there is good reason to believe that diversion has never been adequately evaluated. In the material examined for this 58 review, not one report of an experimental evaluation included an estimate of the proportion of program youths who had actually been diverted. While such an assessment was apparently accomplished by Palmer et al. (197 8), the results were not reported separately for the two experimentally evaluated programs. A similar conclusion was reached by Blomberg (1980) who stated that "it remains unclear whether diversion is or is not reducing subsequent behavior difficulties...for intended diversion clients" (p. 579). The second major problem with diversion research is that there has been a lack Of documentation of the handling of youths in the comparison groups. It is not sufficient tO state that youths in the comparison group were handled in the "traditional" fashion or handled as they "normally" would have been. Local juvenile courts vary considerably in the manner in which juveniles are handled. Some courts restrict their activities to legal procedures, while others provide treatment services. It is also important to know how deeply comparison youths penetrated the juvenile justice system. Were these youths referred tO court simply to be dismissed at intake with no actual court contact, or were they adjudicated and thus the recipients of extensive court intervention? Given the presumed role Of labeling, this type of information is critical for judging the meaning of program outcomes. One must know what types of handling are being compared with diversion. The third major methodological problem is that diversion research has tended to ignore the assessment Of process variables and intermediate outcomes. Only rarely have evaluations of diversion programs assessed constructs related to labeling or to other potentially relevant theories. In addition, information about the type and intensity of services received by program youths has been frequently not reported. In fact, studies have rarely stated the theoretical rationales for 59 diversion service programs, let alone subjected them to empirical investigation. Without an attempt to determine if intermediate outcomes are attained (e.g., reduced labeling) or what processes are related to outcomes, (e.g., service intensity and comparison group handling) diversion research will not contribute any generalizable explanations concerning why diversion programs are effective or ineffective. This is the kind of information that is needed to improve future efforts. Another major weakness is that the evaluations have rarely included a comparison of traditional processing, diversion to a service program, an_d diversion without services. Only three studies included all three groups (i.e., Dunford et al., 1982; Elliot et al., 197 8; Klein, 197 9b). Inclusion Of the diversion-without-services group is important for two reasons. First, it allows the investigator to separate the independent effects of diversion from the effects Of the service program. Second, diversion without services may be an appropriate and cost-effective strategy for handling many juvenile Offenders. Its potential effectiveness warrants a thorough evaluation. Finally, most evaluations of diversion 'efforts have failed to use multiple measures of recidivism. Only a few studies assessed self-reported delinquency, and most did not report outcomes of both simple and multiple recidivism. Given the contradictory results of two studies employing measures of simple and multiple recidivism (Blew 6t Rosenblum, 1977; Stratton, 1975), it seems advisable that both measures be used. While no study reported a significant impact on self- reported delinquency, this measure was rarely used despite its Obvious importance to those who are interested in behavioral change. Recidivism results. While the methodological problems associated with the prior research may indicate that few conclusions can be drawn about diversion's effectiveness, it may still be useful to more closely examine the results published 60 to date. As was stated earlier, the findings concerning the impact of diversion on Official recidivism are mixed. While there certainly is no consistent support for the efficacy of diversion, there are several examples of positive findings which should be examined further. In particular, these findings should be examined in terms of their relationship to different program characteristics to determine if there are any trends in diversion outcomes.10 The first program characteristic to be examined is the level of diversion. Diversion has been conducted at several points in the juvenile justice system. Youths have been diverted at the time Of police contact, at court intake, and at the preliminary hearing stage. Given labeling theory as the rationale for diversion, it has been suggested that diversion should take place at earlier points in juvenile justice processing in order to avoid negative labeling more completely (e.g., Dunford, 1977). In order to examine the relationship between the level of diversion and program outcomes, programs were categorized as either police- Or court-level diversion programs. Police-level programs included all diversion efforts which took place 2129}; to a referral to court, while court-level programs included all diversion efforts which took place fie; a referral to court.11 As can be seen in Table 1, 60% of the studies of court-level diversion programs reported positive results as compared to 17% Of the police-level programs.12 However, in an attempt to measure the strength Of the relationship between level of diversion and program effectiveness, a Chi-square analysis (using Yates Correction) indicated that these factors were not significantly related (x2(1) = 1.89, p > .05). Another critical dimension of diversion efforts is the type of organization that sponsors the service program. Sponsorship Of the program reflects who has control over program Operations. This control is evidenced by the ability to design the program, to hire and train staff, and to set program policies. In the 61 Table 1. Program Outcome by Level of Diversion Level of Diversion Prggram Outcome Police Court Effective - 3 (17 96) 3 (60%) Ineffective 15 (83%) 2 (40%) Note. x2(l) = 1.89, p >.os 62 past, program Sponsors have included the formal juvenile justice system (i.e., police departments and juvenile courts), private, community-based agencies, and public agencies. Frequently, the type of sponsor has a bearing on the physical location of the project, the type of staff hired, and program policies related to conditional versus unconditional dismissal. For example, a program sponsored by a juvenile court is more likely to be located at the juvenile court, tO be staffed by probation officers, and to use conditional dismissals. It has been argued that programs that are sponsored by the juvenile justice system are less likely to prevent negative labeling (e.g., Dunford, 1977), more likely to increase social control (e.g., Lemert, 1981), and more likely to use coercion (e.g., Andriessen, 1980). To the extent that such factors are related to program outcomes, one might predict that programs sponsored by the juvenile justice system are less likely to be effective. However, as can be seen in Table 2, there was no support for a relationship between program sponsorship and reported program effectiveness in reducing recidivism (X 2(2) = .29, p > .05).“3 The third major dimension that may be related to program success is the type Of service provided. While it would have been desirable to look at such dimensions as the intensity of services and the treatment approach (e.g., counseling vs. recreation vs. employment) this was not possible. The program descriptions rarely included documentation Of the intensity of services, and most programs used a variety of treatment approaches. However, there was one major dimension on which service programs could be classified. That is, programs either provided services through referrals, or they provided services "in-house." In the former case, the programs Operated as service brokers referring youths to other agencies, or the police or court staff directly referred youths to other agencies. As can be seen in Table 3, 13 programs used a service brokerage approach while 19 programs used a specific service model. While only 15% Of the service 63 Table 2. Program Outcome by Type of Sponsor Tymf Sponsor Policel Private Public Prggram Outcome Court Agency AggncL Effective 5 (31%) 2 (20%) 1 (25%) Ineffective 11 (69%) 8 (80%) 3 (75%) Note. x2(2) = .29, p > .05 64 Table 3. Program Outcome by Type of Services Type of Service Igggram Outcome Service Brokerage Model Effective 2 (15%) 6 (35%) Ineffective 11 (85%) 11 (65%) Mixed 0 2 Note. x2(1) = .65, p > .05 65 brokerage programs fell in the effective category, 35% of the model programs fell in this category. If the two model programs that were found tO have a positive impact on one Of the two recidivism measures were included as successes, then 42% of the model programs would have been effective. Again, however, no significant relationship was found between the program characteristic and reported effectiveness (x2(1) = .65, p > .05). Despite the lack of evidence indicating the relative effectiveness Of a (service brokerage versus service model approach, there is some evidence that youths participating in service brokerage programs may be less likely to receive services. When youths are shuffled about, there is more Opportunity for them to be lost in the system. Even though data related to this issue are rarely reported, there is some evidence that youths do not receive the intended services. For example, Dunford (1977) noted that in one study of a police diversion program using service brokerage, 38% of the youths participating never received the intended services. It is also possible that, in service programs using many service providers, there is an inability to monitor and control service quality. Conclusion. Reports of previous evaluations of diversion programs indicate mixed evidence about the effectiveness of diversion. In the 15 studies reviewed in this report, evaluations of 32 different diversion programs were presented. For those evaluations in which diversion was compared to traditional processing, diversion was found to be more effective in 8 Of the 30 program evaluations. In addition, diversion was never found to be less effective than traditional processing. A meta-analysis of diversion research failed to find any significant relationships between program characteristics and reported effectiveness. However, a greater percentage of service model, as Opposed to service brokerage, programs were reported to be effective, and some evidence was provided to 66 indicate that service brokerage approaches may suffer from extensive client mortality (e.g., Dunford, 1977). Even though there have been many evaluations of diversion, it is virtually impossible to jucge diversion's effectiveness. As was demonstrated by this review, this is due to both methodological problems associated with past research and to the fact that implementation failure was found in some studies and implementation was inadequately documented in others. At this time, the most accurate conclusion may be that diversion has not been tested. Conclusions Diversion efforts were initiated with the ham that they would provide a less costly and more effective alternative to the formal juvenile justice system. After nearly 17 years of experience with diversion programs and numerous evaluations, the review presented here indicates that there is still only the "hOpe" that diversion can achieve its goals. The research that has been conducted to date does not allow one to draw firm conclusions about the effectiveness of diversion. Evaluations of diversion programs have frequently suffered from methodological problems that cast suSpicions on the results. More critically, there is little evidence that evaluations have been conducted on successfully implemented diversion programs. It may well be that diversion has not yet been tested. While it is easy to be pessimistic, there is still reason to be hOpeful. Past experience with diversion efforts has led to the identification of several potential problems associated with diversion including net-widening, label substitution and increased system costs. The intense discussion and debate generated by these problems can provide the foundation for improved program designs. Also on a positive note, it is important to acknowledge that diversion was not found to be _l_es_§ effective than traditional processing in reducing recidivism in any of the evaluations reviewed. While diversion was not consistently more effective than 67 its alternative, it is probably unreasonable to expect highly consistent findings given the wide variety of program types, pOpulations and so forth evidenced in the research. A more reasonable expectation would be to determine what types of diversion work, under what conditions, and with what types of Offenders. The results of prior research indicate that future diversion efforts should be more carefully planned, more thoroughly documented, and more rigorously evaluated. The most important focus for future efforts in diversion is successful implementation. If diversion cannot be successfully implemented, then all other efforts related to diversion are of little value. Despite the importance Of implementation and the frequent criticism Of diversion for implementation failure, there appears to have been little attention devoted to developing strategies that might improve diversion implementation. As discussed earlier, previous evaluations have generally failed to adequately assess diversion implementation. Without such an assessment, it is not possible to determine if an evaluation is a valid test of diversion. The methods used to assess implementation must be improved, and multiple methods should be used in each study in order to compensate for the weaknesses associated with Specific methods. Consistent findings across different methods would provide one with more confidence in the results. Assuming that diversion can be successfully implemented, there will be a need for evaluations of its effectiveness in reducing recidivism. Obviously, this research should attempt to avoid the problems frequently associated with previous evaluations. Specifically, true experimental designs should be used in a comparison Of diversion to services, diversion without services, and traditional processing. Multiple measures of recidivism should be employed, including self- reported delinquency. The actual handling of youths in each comparison group 68 should be thoroughly documented, and outcomes should be examined in relation to process variables. Greater attention must be paid to the types of services provided in diversion efforts. In most cases, diversion programs have provided traditional, counseling- oriented services that have generally been found to have little impact on recidivism (e.g., Lundman dc Scarpitti, 1978; Romig, 1979; Wright 6: Dixon, 1977). Future efforts should include service programs that have prior evidence of effectiveness or that, at least, have strong theoretical support. Given the results Of the studies presented here, it may also be advisable to use a Specific service model rather than service brokerage. Finally, the performance of the diversion service program should always be compared to diversion without services. In the few evaluations that included an outright release group, youths in this condition were never found to have recividated more than youths who were formally processed (e.g., Dunford et al., 1982; Elliot et al., 1978; Klein, 1979b). Given the lower cost Of outright release and the fewer risks associated with its practice (e.g., avoidance of net-widening and some due process problems) it may be the best Option available. Rationale for the Current Study It appears that diversion is a promising, although insufficiently tested, delinquency intervention strategy. A number Of studies have been conducted which indicate that alternative programs and outright release are as effective or more effective than formal juvenile justice processing (e.g., Baron et al., 1973; Dunford et al., 1982a,b; Klein, 197 9b). In addition, in this review no studies were found which indicate that diversion is less effective than juvenile justice system processing. Therefore, there is little reason to doubt that diversion could provide a less costly, less stigmatizing, and more humanitarian alternative to the juvenile justice system. 69 Unfortunately, prior research on diversion does not provide a foundation for drawing firm conclusions about its effectiveness. It appears that there is little evidence to demonstrate that diversion has been successfully implemented (Klein, 1979a), and much of the prior research suffers from serious methodological problems, such as lack of an experimental design (e.g., Kelley et al., 1976; Gilbert, 1977), large, unexplained pre-period differences between comparison groups (e.g., Davidson et al., 1981), and lack of an implementation assessment (e.g., Quay 6: Love, 1975). In general, the prior research provides a very confusing and mixed picture Of the effectiveness of diversion. Overview of the Study Given the relatively untested promises Of diversion, there is a need for more research in this area. This study was designed to address the major problems of past diversion efforts and their evaluations. In particular, this study was designed to reduce problems in three areas: (1) diversion implementation; (2) research methodology; and (3) program services. Diversion implementation. A number of steps were taken in the design Of the diversion effort to increase the likelihood Of successful implementation. First, a site was selected in which a large number and a large prOportion of youths were being formally processed. This strategy was intended to increase the likelihood that there would be a sufficient number of youths that could be diverted. Second, intensive negotiations were conducted with agents of the juvenile justice system to obtain agreements that only those youths who would have been formally processed would be referred. In addition, the agents agreed that all youths fitting Specific eligibility criteria would be referred to the project and that compliance with this agreement could be monitored by the research staff. 70 Research methodology. A rigorous research design was employed in which youths were randomly assigned to traditional processing, diversion with services, and diversion without services. Each Of the intervention alternatives were monitored in order to provide an adequate description of the comparison conditions, and multiple measures Of intervention outcomes were used. Finally, multiple methods were used to assess diversion implementation. These methods were used to assess both system processing trends and the prOportion of project participants that were truly diverted. giggram services. An existing service program was selected and modified for use in this project. Given the findings and issues discussed earlier, the program used a service model rather than a service brokerage approach. The service model was based on the Community Arbitration Project, described earlier. This program model has some prior evidence of effectiveness and, with certain modifications in the program design, it also has a strong theoretical framework. Both social control and labeling theories were used to make modifications in the original program design to strengthen the theoretical foundation. It should be noted that this service model was selected, in part, because Of its close "fit" with the rationale for diversion. In particular, the service model was designed to prevent the negative labeling of program clients and to increase positive labeling. As a result, the program was consistent with the goals of diversion. Overview of the Community Service Prm The design of the Community Service Program was based primarily on the Community Arbitration Project Of Anne Arundel County, Maryland. This project was awarded the LEAA Exemplary Project status in 1978, and like the Community Service Program, it combined arbitration of a formal complaint with community 71 service placements. The arbitration process focused on resolving the conflict between the complainant and the youth to their mutual satisfaction and on facilitating victim restitution when appropriate. The community service placements were used as a means of involving the youth in a constructive activity in a community organization. These placements served the dual function of enabling the youth to provide social restitution and integrating him/her into the prosocial community. In addition, the design of the Community Service Program was strongly influenced by the work of the Westinghouse National Issues Center and the Center for Action Research (Johnson, Bird 61 Little, 1979; Beville 6r Cioffi, 1979). In a document written for the Office Of Juvenile Justice and Delinquency Prevention, these authors presented prOposals for a delinquency prevention strategy that employed job and community service placements. Their proposals were not only complementary with the Community Arbitration Project, but they also strengthened this model by Specifying characteristics of the community service activity that are most likely to reduce future delinquent behavior. 9mm Rationale Arbitration. Several rationales can be presented for each Of the two distinct aspects of arbitration employed in this program; conflict resolution and restitution. At least three rationales can be prOposed for the use of conflict resolution. First, the resolution of a conflict may prevent the immediate crisis from escalating into a situation requiring formal involvement of the juvenile justice system. Second, if arbitration produces a lasting agreement between the parties, then it may prevent future conflict situations that would result in formal juvenile justice system action. Finally, the process of arbitration may set an 72 example for the youth concerning apprOpriate methods for resolving future conflicts. Again, a number of rationales can also be proposed for the use of restitution. These rationales apply to both financial restitution and to social restitution (as in the case Of community service). Schneider and Schneider (1979) have summarized the major rationales for restitution as follows: It is usually believed that the process of "making restitution" should increase the youth's sense of accountability and reSponsibility. This, in turn, is expected to promote the moral development of the juvenile and to encourage law abiding behavior. The exact mechanisms through which the process of "making restitution" Operates to achieve these expected results are, of course, not known. One could speculate, however, that youths who pay restitution are more aware Of the human consequences of their Offenses than are youths who do not pay restitution. Thus, they would be more cognizant of the full costs of delinquent behavior-including the monetary and psychic costs to victims—and for this reason would be less likely to commit delinquent acts in the future. In addition, it has been prOposed that paying restitution gives the youthful Offender a feeling Of accomplishment which could improve his or her self-image. (p. 7) Community service. The design of the community service component was based on two prominent theories Of delinquency: social control theory and labeling theory. From the perspective of social control theory, "delinquent behavior is a direct result Of weak ties to the conventional order" (Elliot, Ageton, 8t Canter, 1978, p. 11). Without ties or "bonds" to conventional society, the youth is free to behave in a delinquent manner. According to Hirschi (1969) there are four components of the youth's bond to society. First, there is the youth's "attachment" to other peOple. Attachment includes such dimensions as respect and affection for significant others in the youth's life. Youths with strong attachments to others may be less likely to engage in delinquent behavior if it would jeopardize their relationships with these peOple. The second element Of the bond is "commitment" to conventional goals 73 such as education and employment. Again, it is proposed that youths with strong commitments to such goals will not engage in behavior that will jeopardize their success in achieving these goals. The third element, "involvement," represents the amount of time the youth spends in prosocial activities. To the extent that the youth is involved in prosocial activities, he/she has less time to perform delinquent acts. Finally, the fourth element of the social bond is "belief" in the ' validity of conventional norms. If the youth accepts the validity of conventional norms, he/she is likely to conform to them. Several studies have been conducted in which empirical support for the relationship between social bonds and delinquency was found (e.g., Hepburn, 197 7; Krohn 6c Massey, 1980). Strong social bonds were found to be related to lower levels Of official and self-reported delinquency (e.g., Linden, 1978), and social control models were found to account for more variance in delinquent behavior than other models Of delinquency (e.g., Aultman, 197 5). The Community Service Program was designed to integrate youths into community organizations in order to strengthen their bonds to conventional society through the following process. First, participation in community service work was intended to increase the youth's involvement in prosocial activities. Second, since the community service setting was structured such that the youth would interact with staff members, the development Of attachments to others and having liligfg that conform to prosocial norms were encouraged. Finally, through positive reinforcement of the youth's behavior, his/her commitment to conventional behavior was to be strengthened. The second theoretical rationale for community service (as designed for this program) is labeling theory. As was described earlier, labeling theorists have prOposed that the negative labeling of juvenile Offenders is a cause of subsequent delinquency. Given the possible consequences of delinquency labeling, the 74 Community Service Program was designed not only to reduce the probability of negative labeling, but actually to increase the positive labeling of program participants. The program was designed to increase the positive labeling of program participants in several ways. First positive labeling was to be accomplished by placing youths in {£12113 positions of community service in which the youths could be seen as constructive and contributing members of the community. Second, the program was to be publicized as a youth volunteer agency and not as a delinquency program. Third, the program was to serve youths recruited from the high schools and other nonjustice sources to further emphasize the program's positive community function. In addition, it should be noted that staff at the community service placements were not to be told which youths were juvenile justice referrals. Finally, positive feedback about the youth's contributions were to be provided to the youth's family and to other relevant persons. Each of these program elements was intended to increase the positive labeling of all participants. Research Goals In general, the goals of the study were to test the models of diversion and community service that were described earlier. The specific models that were tested are presented in Figure 1 and Figure 2. More specifically, the study was designed to answer the following questions.14 Intervention 1. TO what extent was diversion successfully implemented? (Police Records: change in rate Of police referrals to court) 2. What was the intervention process for each Of the three conditions? (Court Contact Record, Community Service Program Records, Community Service Process Interview) 75 EEEEOP. 00.60..) MEOOEDO flak—25.45 Emngooa 00.60._| MSOOEDO mE5 .H 9:63 .9509: 0>30M0= 0260...: mmmOOmm 025.4:sz 553E090 5:58:50 0 5 x83 53:33 no 00.85.5209! 305:8 5531530; 3060.: 0:0 553500.. :50; 035.5) mmmOOMm zo_Ezm>mw.—.z_ 3230.. 00:3 u:0>0.a| mmmOOMm ZO—Ezm>mm92_ 76 3. To what extent was the Community Service Program successfully implemented? (same as #2) IntervenirgProcesses 1. Did the intervention alternatives produce different levels of labeling? (Label Encapsulation, Perceived Negative Labeling) 2. Did the intervention alternatives produce different levels Of social bonding? (Community Service Outcome Measures) 3. Was there a relationship between the nature of the intervention received by youths in the Community Service Program and the level Of labeling and social bonding? (same as #1 and #2) Ultimate Outcomes 1. Did the intervention alternatives produce different levels Of Official and self-reported delinquency? (Police and Court Records, Self- Report of Delinquency) 2. Was there a relationship between the intervening process of labeling and social bonding-and the ultimate outcomes of official and self- reported delinquency? 3. Was there a relationship between the nature of the intervention received by youths ill the Community Service Program and the ultimate outcomes of official and self-reported delinquency? Model Predicted Results Given the impact models described for diversion and the Community Service Program, certain patterns of results were predicted for the mediating processes of labeling and social bonding and the ultimate outcome of recidivism. For labeling, the diversion impact model led to the prediction that both diverted groups Of youths would experience less labeling than the group of youths who were referred to court. In addition, since the Community Service Program was designed to increase the positive labeling of program participants, it was predicted that youths diverted to the Community Service Program would experience less delinquency labeling than those youths who were diverted without services. 77 With respect to social bonding, the impact model for the Community Service Program led to the prediction that youths who performed community service work would deveIOp stronger social bonds than youths who were diverted without services and youths who were referred to court. The impact models for diversion and the Community Service Program do not Specify a differential effect on social bonding for diversion without services and referral to court, and therefore, it was predicted that there would be no difference between these alternatives in their impact on social bonding. Finally, the impact model of diversion led to the prediction that youths diverted without services and youths diverted to the Community Service Program would recidivate less than youths who were referred to court. In addition, the impact model for the Community Service Program led to the prediction that youths participating in this program would recidivate less than youths diverted without services. At this point it should be pointed out that recidivism is not necessarily a unidimensional construct. As noted earlier, recidivism has been measured using Official records and using the youth's self-report. While Official records and self— report may be alternative measures of recidivism, there is good reason to believe that these measures actually reflect different dimensions of recidivism. Specifically, while self-reported delinquency may be a measure of actual delinquent behavior, Official delinquency may be a measure of the "social reaction" (Klein, 1979b) of the juvenile justice system to the youth and his/her label. Thus, official records may be more of an indication of the system's response to the youth than an indication of the youth's behavior. Still, the impact models for diversion and the Community Service Program led to the same prediction about recidivism regardless of the measure used. That is, diversion and the Community Service Program were expected to reduce both actual delinquent 78 behavior and the Official delinquency that may result from the response to delinquency labels. CHAPTER 4 METHOD OF PROCEDURE Research Setting This study was part of a much larger study of diversion that was conducted in the State of Michigan. This larger study (known as the State Plan for Diversion) developed in reSponse to many of the same concerns about diversion that were discussed earlier. In particular, there was concern that a large investment had been made in diversion programs without any knowledge about the effectiveness of these programs. Before proceeding further in establishing programs, the state planners responsible for diversion services decided to conduct a rigorous evaluation of diversion. Since diversion programs can have many different structures and methods Of Operation, and since the prior research provided few indications of what types of programs would be most effective, it was decided that different types of diversion efforts should be evaluated. The final research plan called for the initiation Of diversion programs in four sites around the state. These programs were tO vary on such dimensions as the point of diversion, program Sponsorship, and the type of services provided. However, in each site a rigorous experimental evaluation was to be conducted. The first major task to be accomplished was the selection of program sites. A number Of criteria were outlined to guide this process. First, the four sites were to be representative of the State in terms of their size and geographical location. Second, the local juvenile justice system had to be one which formally processed a higher than average proportion of youths who entered the system. 79 80 This was considered to be an indication that there was a "need" for diversion, or at least an indication that there was "room" for diversion to be implemented. Finally, local juvenile justice Officials had to be willing to accept an experimental diversion effort that Operated according to certain prespecified guidelines. One of the sites that met these criteria and subsequently agreed to participate in this project is the focus of the study described in this report. This county will be referred to as County A. The County A Site County A ranks in the upper quartile of Michigan counties in terms of pOpulation size. County A's two largest cities border each other and they are centrally located in the county. County A's participation in the project was initiated by the city police department of one of these cities (subsequently referred to as CPD 1). Several early meetings were held to explain the project and to begin the negotiation of the program's design. These meetings were attended by represen- tatives of the juvenile court, the county criminal justice commission, and the four major police departments located in central County A (i.e., CPD 1, CPD 2, township police department (TPD), and the county sheriff department (CSD)). The first step in the negotiations involved obtaining an agreement that the program would Operate in certain prespecified ways. In fact, these program design features were not negotiable. First, and most important, the program was to be experimentally evaluated with youths being randomly assigned to traditional processing, diversion to services, or diversion without services. Second, only those youths who would have been formally processed were to be referred to the project. Third, participation by the youths was to be entirely voluntary and could be discontinued at any time without penalty. Also, youths assigned to one of the two diversion conditions were to be diverted unconditionally; thus, once diverted, 81 these youths could not be subject to further juvenile justice processing for the referral offense. Finally, the research staff was to have complete access to the police and court records that would be necessary to monitor compliance with the administrative agreements. There were also a number of program design decisions that had to be made. The first of these was the type of services to be provided to youths assigned to the diversion to services condition. Decision makers in each site could choose one of three service models that had been develOped. The models were: the Community Service Program (described earlier); the Youth Skills Training Program, which was to teach youths communication, job seeking, study, and use Of community resource skills; and the Family Support and Education Program, which was to teach parents child management and family advocacy skills. Since County A was the first site to enter this stage of the negotiations, representatives from County A were given their choice of the three service models. There was nearly unanimous support for selecting the Community Service Program. A second issue that was addressed was the point at which diversion was to take place. Diversion could have taken place after police contact but before court referral, at the time of court intake screening, or at the preliminary hearing stage. Largely because Of the lead role taken by the city police department, CPD 1, and their interest to the project, it was decided that diversion would take place at the police diSposition stage. For these same reasons, it was agreed that CPD 1 would serve as the sponsor (i.e., administrator) of the program. The final issue to be decided was the type Of youths who were to be diverted. In order tO increase the probability of successful diversion implementa- tion, it was decided that specific eligibility criteria would be developed and that 111 youths who fit these criteria would be referred. Using this procedure, police discretion would be reduced and compliance could be monitored. Before the 82 eligibility criteria were negotiated, 1 year of police contact data was collected from CPD 1. This data was used during the negotiations to ensure that the eligibility criteria were not so restrictive that there would be too few eligible youths to permit an adequate evaluation of diversion. It was also hoped that this data would allow a prediction of the types of youths (i.e., age, offense, number of prior offenses, etc.) who were normally "warned and released." This would have provided some guidelines as to the types of youths who could have been excluded from the project because they were unlikely candidates for formal processing. Unfortunately, a sufficiently accurate prediction of police disposition could not be made using the available descriptive data from police records. The eligibility criteria were negotiated subsequent to the data collection. In general, it was agreed that all first and second offenders from 12 to 16 years of age who were normally referred to court would be referred to the project. The exceptions to these criteria were to be youths charged with serious offenses against persons, property offenses with loss Of property over $500, and serious drug Offenses. In order to provide the police with the flexibility needed to handle "unusual" cases, the agreement also allowed the police to use complete discretion in the dispositional decision for up to 10% of the "eligible" cases. After a lengthy series of meetings, the administrative agreement was completed. Each of the four police departments mentioned earlier agreed to participate by referring youths, and CPD 1 was chosen to serve as the program Sponsor. The project was unanimously endorsed by the county criminal justice commission and approved by the city commission (i.e., the city corresponding to CPD 1). Subjects Subjects for this study were male and female youths referred by the two city police departments, the township police department, and the county sheriff 83 department; all located in County A. In order to be eligible for project participation, a youth was to be 12 to 16 years of age, a resident of County A, and could not be a resident of an institution or have any current court involvement (i.e., the youth could not be on probation, have a petition pending, or be receiving services from a court program). While the project was to accept all youths who met the above referral criteria, the participating police departments were not Obllg' ated to refer youths with more than one prior Offense, or youths whose instant offense was one of the following: arson, homicide, criminal sexual conduct, armed robbery, felonious assault, Offenses involving a weapon, breaking and entering an occupied building, theft or damage exceeding $500, and sale/delivery Of a controlled substance. Finally, participation in the project was entirely voluntary. If either the youth or the youth's parents did not want to participate in the project, the referral was returned to the police for continued processing. A total of 310 referrals Of eligible youths were received by the project. Out of these 310 referrals, 60 youths and/or parents (19.4%) chose not to participate in the project. The decision not to participate was made by the parents only in 31 cases (51.7%), by the youth only in 11 cases (18.3%), and by both the parent and the youth in 18 cases (30.0%). There were another 7 cases (2.6%) in which the family could not be located so that an explanation Of the project could be given to them. These cases were included under the general category of nonparticipants. It should be noted that the relatively large percentage Of cases in which the decision not to participate was made by parents was primarily due to the fact that parents were contacted first to arrange a meeting at which time the project would be explained. However, during this initial contact parents occasionally refused to attend the meeting; thus, the youth never had an Opportunity to make a decision about project participation. 84 In Table 4 the reasons given by the youths and parents for not participating in the project are presented. More than one reason could be given. The most frequent reason given by both parents and youths for not participating was the belief that the youth was innocent Of the charges and that the youth should go to court to prove his/her innocence. This was given as the reason for not participating by 34.7% of the parents and 44.8% of the youths. In contrast, relatively few of the parents (20.4%) and youths (10.3%) decided not to participate as a result of some aspect Of the research requirements of the project (e.g., the use of random assignment or participation in interviews). For eligible youths referred to the project who did not participate, the average age was 14.5 years, 55.2% were male, 71.2% were white, and 39.1% lived with both natural parents. The most frequent type of referral Offense for the nonparticipants were status Offenses (35.8%), though this was only slightly more frequent than a referral for some type of larceny (34.3%). A breakdown of the nonparticipants by race, living arrangement, and type of referral offense is presented in Tables 5 through 7, respectively. Project Participants A total Of 243 youths participated in the study and, therefore, were randomly assigned to one Of the three alternatives. Seventy-eight youths were assigned to traditional processing (TP), 79 youths were assigned to diversion to the Community Service Program (DCSP), and 86 were assigned to diversion without services (DWS). Project participants were very similar to nonparticipants. Participants had an average age of 14.7 years, 73.7% were white, 57.2% were male, and 34.6% lived with both natural parents. In contrast to the nonparticipants, the most common type Of referral Offense was some type Of larceny. Over 62% of the participants were referred for this type of Offense, while 11.9% of the participants were referred for a status offense. Again, a 85 Table 4. Reasons for Nonparticipation Reason Parent Youth Feels youth should be punished 3(6.1%) 0 Dislikes outright release alternative 3(6.1%) 0 Dislikes service alternative 9(18.4%) 7(24.1%) Dislikes use of random assignment 8(16.3%) 2(6.9%) Dislikes other research requirements 2(4.1)% 1(3.4%) Believes youth is innocent (i.e., wants to go to court) 17(34.7%) 13(44.8%) The service program requires too much time 1(2.0%) 5(17.2%) Youth/family is already receiving services elsewhere 4(8.2%) 2(6.9%) Believes court will not do anything 5(10.2%) 3(10.3%) Does not want outside interference 6(12.2%) 3(10.3%) Othera 16(32.796) 2(6.9%) aFor example, no transportation available to community service agency, offense is not serious enough to warrant 36 hours of community service, and the problem is not apprOpriate for the program. 86 Table 5. Race of Participants and Nonparticipants Race Participants Nonparticipants White 17 9(73.7%) 47(71.2%) Black 55(22.6%) 18(27 .396) Hispanic 7(2.9%) 1(1.5%) American Indian 1(0.4%) 0(0.0%) Asian 1(0.4%) 0(0.0%) Unknown 0 1 Table 6. Living Arrangement of Participants and Nonparticipants LivigArranggment Particflants Nonparticipants Both Natural Parents 84(34.6%) 21(32.8%) Natural Parent and Step- Parent 34(14.0%) 13(20.3%) One Parent 103(42.4%) 25(39.1%) Othera 22(9.1%) 5(7.8%) Unknown 0 3 aFor example, natural parent and boyfriend/girlfriend, grand- parent(s), and friends. 87 Table 7. Referral Offense of Participants and Nonparticipants fipe of Offense Participants NonParticipants Status 2901.996) 24(35.8%) Larceny/ShOplifting 152(62.6%) 23(34.3%) Malicious Destruction Of Property 18(7.3%) 5(7.5%) Minor in Possession/ Controlled Substance 6(2.4%) 2(3.0%) Breaking and Entering 9(3.7%) 4(6.0%) Assault 14(5.7%) 4(6.0%) Miscellaneous 18(6.2%) 5(7.5%) 88 breakdown of the participants by race, living arrangement, and type Of referral offense is presented in Tables 5 through 7, respectively. In addition to those variables that were assessed for both participants and nonparticipants, descriptive information was Obtained for participants on their employment status, school status, and prior Offense record. At the time of project intake, 90.2% of the youths were unemployed, 98.4% were enrolled in school and currently attending, and in the 1 year prior to project intake they had an average of 1.22 recorded offenses as indicated in the records of the four referring police departments. Finally, it should be noted that there was a small amount of subject mortality. Post-period interviews could not be completed on eight youths (3.2%) who had participated in the project. For the most part, subject mortality was due to an inability. to locate the youth, although in one case permission for the interview was refused. There were four cases of subject mortality in the diversion without services group, and two cases in each Of the other groups. Official recidivism data was Obtained for all cases. M An experimental design was used to assess the impact of the alternative interventions on recidivism and the relationships between intervening processes and dispositional alternatives. In this design, subjects were nested in conditions. A total of 243 subjects were randomly assigned to one of the three conditions. The conditions were diversion to the Community Service Program, diversion without services, and traditional court processing. Pre-period measures Of official and self-reported delinquency were used to assess the equivalence Of the groups. Two types of dependent variables were used to assess the effectiveness of the alternative interventions. Measures Of recidivism (1-3 in Table 8) were used as indicators of ultimate outcome, and Table 8. Experimental Design and Measures 89 Condition fie Post Diversion tO Community Service Program 1 - 3 1 - 8 Diversion Without Services 1 - 3 1-6, 8 Traditional Processing (Referral to Court) 1 - 3 1-6, 8 Measures: 1. Police Record (frequency and average seriousness of police contacts) 2. Court Record (frequency and average seriousness Of court complaints) 3. Self-Report of Delinquency 4. Label Encapsulation 5. Perceived Negative Labeling 6. Court Contact Record 7. Community Service Process Interview 8. Community Service Outcome Measures (social bonding) 90 measures of labeling and social bonding were used as indicators of the intervening processes (4,5,8 in Table 8). Finally, each of the three intervention conditions was assessed using condition-specific measures of intervention processes. Procedures15 Referral Youths were referred to the project by each of the following police departments: city police department 1, city police department 2, the township police department, and the county sheriff department. In general, juvenile contacts in these police departments were normally screened by one officer who made the dispositional decision. For this project, this same officer made the decision concerning referral of the youth to the research project. In most cases, the officer making the referral decision had not had contact with the juvenile. Therefore, the decision was based solely on whether or not the youth fit the eligibility criteria and the recommendation of the arresting Officer. Since the referral decision was made after the initial police contact, the youth in question was usually not aware that the referral had been made until he/she was contacted by the project intake worker.16 In making a referral to the project, the referring officer completed a "Referral Form." On this form the Officer recorded general identifying informa- tion, as well as the type Of Offense with which the youth was being charged and the referral date. The referral was given to the project intake worker shortly after this form was completed. Upon receiving the referral, the intake worker confirmed that the youth was eligible for project participation (i.e., the youth's age and residence were apprOpriate and he/she had not participated in the project in the past). This confirmation process included checking with the court to make sure that the youth was not on probation, was not receiving any court services, and did not have a complaint pending against him/her. 91 Once the youth's eligibility had been confirmed, the intake worker contacted the youth's parent(s) in order to provide them with a very brief explanation Of the project and tO ask them to bring the youth to an intake meeting at which time a more detailed explanation Of the project would be given. This initial contact with the parents was to take place within one working day of receiving the referral. Project Intake The intake meeting was to take place within 7 days Of the referral date. The youth and at least one parent/guardian was always present at the intake meeting. This meeting usually took place either at the intake worker's Office or at the family's home, depending on what was most convenient for the family. The intake meeting was tape recorded (unless the youth or his/her parents objected) to facilitate the supervision of the intake worker by the principal investigator. During the intake meeting, the intake worker explained the purpose of the project, emphasizing that the youth's participation was voluntary. A detailed explanation of the project was also given including: (1) the nature of the three possible dispositions; (2) the use of random assignment to make the dispositional decision; (3) the type of archival and interview-based data that was to be collected; and (4) the consequences Of nonparticipation. On this latter point, it was explained that should the youth not participate the referral would be returned to the referring police department which would then refer the youth to court as would normally have been the case. In presenting the above information to the family, the intake worker followed an outline provided by the "Diversion Project Participation Agreement" (see Appendix B). After describing the project, the intake worker asked the family if they would like to participate. Both the youth and his/her parent(s) had tO agree tO participate in order for the youth to be included in the project. If one or both Of the parties decided not to participate, the referral was returned to the police 92 department with a note to this effect. For those families who did not participate, some basic demographic information was collected and their reason for not participating was Obtained. If the youth and his/her parents decided to participate in the project, they were asked to sign the participation agreement mentioned above. This agreement described the Operation of the project, affirmed that the youth's participation was voluntary, and gave the project permission to check the youth's police and court records. Following the signing of the participation agreement, extensive demographic information was Obtained from the youth and his/her parents and this information was recorded on the "Intake Record." The demographic information collected included the youth's sex, race, date of birth, employment status, school status, and living arrangement. The youth's parent(s) were also asked to provide information about their educational background and current employment status. This information was used as an indicator of the family's socioeconomic status. Once the Intake Record had been completed, the intake worker met with the youth alone in order to conduct the "Self-Report of Delinquency" (this instrument will be described in detail later). This part of the interview was also tape recorded unless objected to by the youth. Following the completion of the Self-Report Of Delinquency, the youth and his/her parents were brought back together. At this time, the youth was randomly assigned to one of the three conditions (i.e., diversion to the Community Service Program, diversion without services, and traditional processing). The assignment was stratified by sex. The condition to which the youth was assigned was written on a slip of paper and enclosed in a sealed envelope. The envelOpes were prepared by the research staff, and the intake worker did not know the results of the assignment until the envelOpe was Opened in the presence of the youth and his/her parent(s). The slips 93 of paper with the assignment decision on them were placed in a random order into the envelopes. The envelOpes were numbered and the intake worker was instructed to open the envelOpes in sequence. In order to stratify the assignment by sex, two sets Of envelOpes were used (one set for males and one set for females). In cases when two or more siblings were referred to the project for the same Offense, they were assigned to the same condition. Intervention Alternatives Once the random assignment was made, the subsequent procedures varied with the condition to which the youth was assigned. Presented below are the intervention activities associated with each Of the dispositional conditions. Traditional processing. For youths assigned to the traditional processing condition, the referral was returned to the referring police department for continued processing. Each participating police department had agreed to refer the youth to court using their normal procedures. The juvenile court was not to be informed that the youth was participating in the research project. Once referred to the juvenile court, youths were to be handled using the court's normal procedures. The project was not to intervene in the court's handling of these youths. Diversion without services. Youths assigned to the diversion without services condition were turned over to their parents for the parents to handle the situation, and the charges for the instant offense were dismissed. However, these youths were informed that any future Offenses that they might commit would be handled in the manner judged best by the contacting police department. While youths were not to be referred to the Community Service Program or any other human service agency, they were free to pursue any services that they desired to receive. However, neither the intake worker nor program staff were to make a 94 referral for them. The only future involvement these youths were to have in the project was to participate in the post-period interview. Following the intake meeting, the intake worker notified the referring police department that no further action was to be taken with a youth who had been assigned to diversion without services. Diversion to the Community Service Prog£a_n_1. All charges for the instant Offense were dismissed for youths assigned to the diversion to the Community Service Program condition. However, any future Offenses were to be handled by the contacting police department in the manner they judged as apprOpriate. The referring police department was notified Of the assignment of youths to this condition so that the department and the court would not take further action on the instant Offense. However, youths in this condition were to have extensive involvement with the Community Service Program over the next 16 weeks. Following the assignment of a youth to this condition, an appointment was made for the youth to begin the first stage of the program, participation in an arbitration meeting. As described earlier, the purpose Of the arbitration meeting was to negotiate a mutually satisfactory agreement between the youth and the victim/complainant, and to arrange for financial restitution when a monetary loss was suffered. In order to accomplish these goals, an arbitration meeting was held at which the youth, his/her parents, and the victim were tO be in attendance. The arbitration meeting was to begin with an explanation by the arbitrator of the purpose of the meeting,17 the voluntary nature of participation, and the confidentiality Of the proceedings. The arbitrator was then to proceed to determine the facts of the situation (i.e., the offense) and to attempt to reach a settlement agreeable to all parties. When restitution was called for by the agreement, the youth was informed that the necessary money could be earned 95 through the completion of a community service activity. Funds were available for restitution in the budget of the Community Service Program (CSP). In all cases, the community service component Of the program was explained to the youth and to the others in attendance. Included in this explanation was to be a description Of the types of community service placements that were available and the amount Of time that the service activities would require Of the youth. A Special emphasis was to be placed on explaining the rationale for community service, including the intended benefits for the youth and the philOSOphy of social restitution. In particular, the youth was to be told that all illegal acts have some cost to society. These costs may be direct costs as in cases of prOperty damage, or the cost may be indirect, such as the costs Of time spent by law enforcement and juvenile court Officials in the handling of a specific Offense. Further, the youth was to be told that through his/her service activities the community would be repaid for the losses it had suffered. This same general explanation was to be made for financial restitution to the victim when it was relevant. The youth was also to be told that the community service placement was intended to be far more than a means for paying a debt to the community. That is, the intent was for the youth to participate in an organization that would allow him/her to explore his/her interests and to learn new skills. Immediately after completing the arbitration meeting, the youth and his/her parent(s) were to meet with one of the two community service workers in order to discuss the selection of a community service placement. The intent of these placements was for the youths to engage in constructive activities that would be Of benefit to themselves, the host organization, and the community at large. It was intended that the youths would build skills, become integrated into the prosocial community, and be recognized for their positive contributions through their involvement in the service activity. 96 In general, the interests and preferences of the youth were to guide the selection Of a community service placement. The youth was to be presented with a list Of potential placements along with an explanation of the types of activities associated with each placement. In making the initial selection, the youth was to be encouraged to consider his/her abilities and schedule, the location of the placement, and the transportation he/she had available. Unless the youth indicated that he/she knew which placement he/she wanted, the youth was to be given 2 or 3 days to think over the Options that were available before making a selection. A written description Of the placements was to be provided to help in this task. A second meeting was to be held with the youth within 2 or 3 days of the first meeting. At this time, the youth was to make a selection of one or two placements that he/she wanted to. pursue further. After the selection was made, the staff person was to arrange for an interview with the project liaison at the community placement agency.18 The youth was to be prepared for this interview by discussing such issues as apprOpriate dress, promptness, arranging for transportation, and by role playing the interview. The interview was to provide the youth with an Opportunity to learn more about the organization and the role he or she could serve. The interview was included to provide the youth with an Opportunity to assess the extent to which the placement would fit with his/her interests and abilities. Following the interview, the youth was to meet with the community service worker to discuss the prospective placement and the outcome of the interview. If the youth expressed strong negative feelings about the placement or wanted to explore another placement that interested him/her, an interview with a second placement Option was to be arranged. Following the second interview, the youth was to choose a community service placement. 97 Once the youth had been placed into a community service organization, the majority of staff time was to be spent monitoring the quantity and quality Of the youth's work. In general, this was to be accomplished through regularly scheduled contacts with the youth and the project liaison, and through examining records kept by the project liaison. Face-to-face meetings were to be held with the youth after the youth had completed two sessions Of community service and during the 6th and 10th weeks of community service. In addition, the project staff person was to contact the youth by phone during those weeks‘when a meeting was not held. These telephone calls were also to be used as a means of keeping the youth's parents informed Of their child's performance. The project liaison was also to be contacted on a regular basis. Telephone contacts were to be made at least biweekly with the project liaison and face-to- face meetings were to be held on a monthly basis. These contacts were to be used by project staff to learn what service activities the youth was performing, and to assist the project liaison in designing activities that met the expectations of the program. The youth was to perform the community service work for 3 to 4 hours per week, for a duration of 12 weeks. Upon the completion of this stage Of the program, a meeting was to be held with the youth and his/her parent(s). This meeting was to be used to give recognition to the youth for his/her positive contributions to the community. Part Of this recognition was to involve the presentation of a plaque acknowledging the youth's performance. At this meeting, the youth was also to be encouraged to continue his/her positive involvement in the community. This could have been a continuation of the current community 3 service position, or Obtaining a new community service or employment position. If one of the latter two Options was chosen, the project was to act as the youth's consultant in this task. 98 As described earlier, one of the major goals of the program was to increase the positive labeling of the youths. There were certain aspects of the program that were designed to increase the potential of the program to accomplish this goal. First, the program was to strive to become known as a youth volunteer agency, not as a program for delinquency prevention. In presenting the program to the community, the staff were to emphasize that it was designed to provide positive and legitimate Opportunities to youths and to provide a service to the community. Second, in order to further foster this positive image, the program was to serve nondelinquent youths, also. These youths were to be recruited from the schools and other sources. In addition, staff at the community service placements were not to be told which youths were referred by the police and which were recruited from the community at large. Finally, every attempt was to be made to provide positive feedback to significant others about the youth's contributions in order to foster a positive image about the individual youths participating in the project. Post-Period Assessment Sixteen weeks after the intake meeting, a post-period interview was conducted with each youth assigned to one of the three conditions. The interviewers were paid staff, selected and trained by the principal investigator. Each youth was given $5 for participating in the interview in order to decrease subject mortality. For youths who were assigned to the Community Service Program, this interview always tOOk place after the program termination meeting. The post-period interview was tape recorded, as was the intake meeting, and the interview generally took place either in the youth's home or at the interviewer's office.19 In general, this interview lasted between 45 minutes and 1.5 hours. 99 Time Frame Presented below is the time frame for all research activities, from police contact to the post-period interview. Week One 1. Police contact 2. Police referral to project intake worker 3. Project intake meeting with youth and youth's parents 4. Research interview with youth to collect pre-period data 5. Random assignment to one of the three alternatives The following steps apply only to those youths assigned to the CSP: 6. Referral of the youth to the Community Service Program 7. Begin arrangement for arbitration . meeting (i.e., notify victim/complainant, set time for meeting to occur, etc.) Week Two 1. Arbitration meeting conducted 2. Community service worker holds first meeting with youth and his/her parents to explore apprOpriate community service placements 3. Community service worker holds second meeting with the youth to make the initial selection Of a community service placement 4. Interview is arranged for the youth with the community service organization Week Three 1. Youth has interview with prospective community service placement 2. Youth meets with project staff to discuss the interview and the apprOpriateness of the particular placement Note: If the youth and project staff person decide that another community service placement should be explored, then this process was repeated. 100 Week Four to Week Fifteen 1. Youth spends 3 tO 4 hours per week participating in a community service activity Week Sixteen 1. The project staff person holds a meeting with the youth and his/her parents in order tO provide the youth with positive feedback for the youth's community service activities and to present a plaque of recognition for the youth's accomplishments 2. Post-period interview The actual time frame maintained by the project closely fit this ideal schedule. For example, the mean number of days from referral to the intake meeting and the random assignment was 5.8 days (s_q = 4.6 days), while the mean number of weeks from the intake meeting/pre-period interview to the post-period interview was 16.4 weeks (g = 12.1 days)., However, there was one major deviation from the prOposed time-frame; a deviation that was beyond the project's control. That is, the mean duration between police contact and referral to the project was 11.8 days (g = 11.0 days). Measures The study was designed to test both diversion and a specific service model, the Community Service Program. As described earlier, it has been proposed that diversion will decrease delinquency by preventing the negative labeling that results from formal juvenile justice processing. The Community Service Program was based primarily on social control theory. It was prOposed that constructive participation in a community organization would increase the youth's bonds to society, thereby decreasing the likelihood of further antisocial behavior. The measures that were used in this study were specifically designed or selected tO test both diversion and the Community Service Program. More specifically, measures were used to assess the intervention process (program 101 services, diversion without services, and traditional processing), the hypothesized intervening processes (labeling and social bonding), and the ultimate outcomes (Official and self-reported delinquency). In addition, measures were included to assess the implementation of diversion. The conceptual framework for the measures used in this study is presented in Figure 3, and a list of the measures organized by condition and time period is presented in Table 9. Measure Development Scales were developed from the interview items in order to increase measure reliability and in order to reduce the large number of variables, thus simplifying subsequent analyses and the interpretation Of results. Since the same procedures were used to develOp all interview-based measures, the general approach will be described here. The overall strategy of scale development used in this study may be described as a rational-empirical approach (Jackson, 1971). As will be presented later, each of the measures administered in this study contained a priori defined rational scales. These rational scales were analyzed in order to provide empirical guidelines for scale modification. The goal of these modifications was to create scales with maximal internal consistency and independence. The scale development task began with an assessment Of the internal consistency of the rational scales, as well as an assessment of the item-scale and interscale correlations (internal consistency was assessed using the "Reliability" program of the Statistical Package for the Social Sciences (Nie, Hull, Jenkins, Steinbrenner, 6t Bent, 1975) which computes Cronbach's Alpha as the indicator of internal consistency). The original rational scales were then modified using the following procedures and decision rules. First, items were deleted from scales when the removal of the item would increase the alpha level for the scale or when the item-total correlation was not significant. Second, when items were removed 102 3:05:50:— u:0:_mm0mm< ..0: :0002 53:00:00 .m 0.55.: 13:0 05:0» 5050.:— 00_>..0m 5555500 3 :32039 30:23:: ..0£>..0a:m 50:002.: «0.5502 0E0050 00_>:0m 355.500 .H 05080 38m 3:025:00 :0 5000:...«8m .m 5:000: 0>30w02 00300.0: .N 000000: 2:00 0:: 00:0: 4 53053000: 000..— .: 50:30:00: 9:000..— 0>30w0z mm—ZOOEDO MERE—PAD mmmOOmm 02322092 00000: “03:00 «.500 .: 5600005 5:03:00; 0:000: “03:00 2:00 A 00_>._0m 505E :0_m._0>_o 0:000: «03:00 “.500 .m 305 L35 $000.5 00_>._0m 3:58:80 .m 0:000: 53:00 .0 .m 0:000: 00_>._0m 3:55:80 .m E05023: 53053.3. .N 5:0: 535:0: 53953.2 A 80.30:: 00_>._0m .< 53:55:00 3 56:03: mmmOOmm zogzm>mm.—.ZH 103 Table 9. Evaluation Measures Diversion to the Community Service Program Pre/Intake Mgasures 1. Intake Record 2. Self-Report of Delinquency 3. Police and Court Records Program Records 1. Arbitration- Reporting Form 2. Arbitration Agreement 3. Community Service Record 4. Contact Record 5. Termination Record Post Measures 1. Label Encapsulation 2. Perceived Negative Labeling 3. Self-Report of Delinquency 4. Court Contact Record 5. Community Service Process Interview 6. Community Service Out- come Measures (bonding) 7. Placement Supervisor Interview (bonding) Diversion Without Services Traditional Processing Intake Record Intake Record Self-Report of Self-Report of Delinquency Delinquency Police and Court Police and Court Records Records Label Label Encapsulation Encapsulation Perceived Negative Labeling Self-Report of Delinquency Court Contact Record Community Service Process Interview Perceived Negative Labeling Self-Report of Delinquency Court Contact Record Community Service Process Interview 104 from a scale, the item's correlation with other scales was examined to determine if the item could be moved to a different scale where it was highly correlated and rationally "fit" with the content of that scale. Third, items were moved from one scale to another if the item correlated more highly with the second scale, and if the item rationally fit in the second scale. Implementation of Diversion The assessment of the implementation of diversion relied primarily on archival data collected from police and court records. Two different types of measures were used. First, court records were used to determine the actual disposition of youths assigned to the "traditional" processing condition. As described earlier, the prOportion of youths in this type of comparison group who were formally processed is one indication of diversion implementation. One would expect that, if diversion was successfully implemented, all or a very high prOportion of the youths assigned to this condition would have been referred to court. The second measure used to assess implementation was the prOportion of police contacts given a formal diSposition 1 year before project initiation and 1 year after project initiation. The rate of formal processing was assessed for all juvenile cases and for those cases that fit the diversion eligibility criteria (based on age, type of offense, county of residence, and the number of prior offenses). Intervention Process Several measures were designed to describe the three types of interventions that participating youths may have received. Those measures designed to assess the Operation of the Community Service Program were also used to examine the extent to which the program was implemented successfully. Intervention process measures included program records kept by the staff of the Community Service 105 Program, the Community Service Process Interview administered to youths at the conclusion of program participation, and the Court Contact Record administered to youths in all three conditions. Mam records. Several different instruments were used by program staff to record information about the quantity of program services and the outcome of specific program activities. The first two instruments of this type were the Arbitration Reporting' Form and the Arbitration Agreement. These two instru- ments were completed by the arbitrator at the conclusion of each arbitration meeting. The Arbitration Reporting Form was used to provide a description of the time and duration of the meeting and a record of those who attended. Also recorded on this form was the amount of the loss suffered by the victim and the amount of restitution provided. The Arbitration Agreement was used to record any agreements between the complainant and the youth or the youth's parent(s) that resulted from the arbitration meeting. Two additional types of program records were used to describe the community service component of the program. The first of these was the Community Service Record which was completed by the project liaison at each placement. On this form, the project liaison recorded the date, the amount of time worked, and the type of work performed for each session of community service. This information was used by the principal investigator to assess program Operation, and it was used by the program staff to monitor the youth's performance. Finally, at the conclusion of program services for each youth, the Termination Record was completed. On this form the community service worker recorded the beginning and ending dates of program participation and community service work, the name of the community placement, the reason for termination 106 of program participation, and a summary of the information recorded on the Community Service Record. The reliability of the coding Of program record data was assessed as the percent agreement between raters. The interrater agreement ranged from 86.796 to 10096 for these variables (see Table 10). Community Service Process Interview. The Community Service Process Interview (see Appendix C) was part of the post-period interview that was conducted 16 weeks after the intake meeting. Of course, the Community Service Process Interview was only conducted with youths who were assigned to the community service condition. This interview used a structured format and had 48 items. Both fixed-alternative and open-end items were included; however, the vast majority of the items were Of the first type. This instrument was developed by the principal investigator in order to assess several dimensions of program Operation that were specific to the Community Service Program. These dimensions and the corresponding items are outlined below. 1. Arbitration satisfaction: Two items were included to assess the youth's satisfaction with the arbitration component of the program (#1,2). These items addressed the extent to which the youth thought that the arbitration meeting was "helpful" and was a "good idea." 2. Prggram satisfactio : The youth's general satisfaction with the Community Service Program was also assessed in terms of the youth's perception that the program was "helpful" or "harmful" (#38) and "a bad thing to happen to you" or "a good thing to happen to you" (#39). 3. Staff satisfaction: The youth's satisfaction with the community service workers was assessed by the extent to which the youth thought that the staff were "friendly" (#35), "treated" the youth "fairly" (#36), and "did a good job" (#37). 4. Type and amount of community service work: Four Open-end items addressed where the youth performed his/her community service work (#3), what type of work he/she performed (#4), what new things he/she learned how to do (#11), and the number of hours the youth worked (#5). 107 Table 10. Interrater Agreement for Program Record Data Table Percent Agreementa Arbitration Record 1. Complainant present at arbitration (Y/N) 100% 2. Amount of loss by victim 100% 3. Restitution provided (Y/N) 100% 4. Amount of restitution to be paid 100% 5. Victim-youth agreement reached (Y/N) 93.3% Termination Record 1. Number of weeks in program 100% 2. Termination reason 100% 3. Number of weeks of community service 100% 4. Total hours of community service 100% 5. Sessions of community service 93.3% 6. Type of placement 93.3% Contact Record 1. Number of contacts with youth 100% 2. Number of contacts with parents 86.7% 3. Number of contacts with placement 100% 82:15 108 5. Work ggoups: Items #12-15 were included to assess the extent to which the youth worked alone or with other peOple. If the youth was to develop a social bond to the organization, the youth must have been given an Opportunity to interact with the organization's staff. 6. Supervision: The amount and style of supervision provided by the placement was assessed by items #7, 17.-25. 7. Interpersonal attachment: The extent to which the youth became "attached" to members of the placement organization was assessed by items #26-30. This attachment was Operationalized as the number of friends the youth made (as perceived by the youth) (#26), the frequency with which the youth talked to other staff in general (#27) and about personal problems (#28), and the likelihood that the youth would talk to another staff person if the youth had a problem he/she wanted to talk about (#29). 8. Work attitudes: The youth's attitudes toward the community service work were assessed in terms of the youth's perception that the work was "important" (#6), helpful (#8), and "interesting" (#9). 9. Prggram supervision: The youth's perception Of the intensity of supervision of the community service work was assessed by items #42 and #43. 10. Perceived staff attitudes: The youth's perceptions of the community service worker's attitudes towards him/her were assessed by a series Of semantic differential items (#33,34), on the following scales: troublesome-cOOperative; law abiding- delinquent; good-bad; and disliked you a lot-liked you a lot. 11. Parental support: The level of parental support for the youth's community service work was assessed by the youth's report of how "important" his/her parent(s) thought the work was (#46), how much "encouragement" his/her parent(s) provided (#47), and how frequently his/her parent(s) asked the youth about what he/she was doing (#48). The scale develOpment procedures described earlier in this section were used to refine the Community Service Process Interview. A total of seven scales reflecting different dimensions of the intervention process were develOped using this procedure. These scales, their item content and their alpha levels are presented in Table 11. The interscale correlations are presented in Table 12. As can be seen from these tables, the final Community Service Process scales had 109 Table 11. Internal Consistency of Community Service Process Scales Scales Items Alppha Project Satisfaction 1,2,38,39 .79 Staff Satisfaction 35,36,37 .68 Work Groups 12,13,15 .80 Attachment to Placement Members 23,25,27,28,29 .58 Perceived Negative Labeling by Program Staff 33a, 33b,33c,34 .85 Parental Support 47 ,48,49 .65 Attachment to Community Service Work 6,8,9 .65 Table 12. Community Service Process Interscale Correlations Project Staff Attach. Attach. to Satis- Satis- Work to Place- Staff Parental Community faction faction Groups ment PNL Support Service Project 1.00 Satisfaction Staff .38* 1.00 Satisfaction Work .29* -.01 1.00 Groups Attachment .53“ .57* -.00 1.00 to Placement Staff -.62* -.58* -.16 -.37* 1.00 PNL Parental .19 .24* .11 .33* -.18 1.00 Support Attachment .94* .27 "‘ .28* .57* -.66* .15 1.00 to Community Service Note. The correlations were computed with an 3 ranging from 74 to 75. The interscale correlations were corrected for attenuation. *p < .05 110 alpha levels ranging from .58 to .85, and interscale correlations ranging from -.01 to .94, with a mean Of .35.20 Court Contact Record. The Court Contact Record (see Appendix D) was administered at the post-period interview to youths in all three conditions. This instrument was used to assess the number of contacts with juvenile court staff (judges/referees and probation Officers) and the "typical" duration of these contacts. This instrument was particularly important for the youths assigned to the traditional processing condition since it provided some basic descriptive information about the intervention process, and since it provided some information that could be used to assess the extent to which the traditional processing condition was implemented. The Court Contact Record provided relevant data for youths assigned to the two diversion conditions. Even though these youths did not experience court intervention for the referral Offense, it was possible that these youths committed subsequent offenses that resulted in a referral to court. Since such an event would have changed the intended intervention for a given youth, it was important to monitor the extent to which this occurred. Mediating Processes Three different instruments were designed to assess the mediating processes that provided the hypothesized link between the type of intervention and subsequent delinquency. The first two instruments were designed to measure different aSpects of delinquency labeling, while the third instrument was designed to measure social bonding. Label Encapsulation. The Label Encapsulation measure (see Appendix E) was a modified version Of a measure used by Klein (1979b). This measure was originally designed to "distinguish between those with more or less exposure to 111 labeling" (p. 15). Label encapsulation was conceptualized by Klein as having three separate dimensions. The first dimension was simply the number of justice contacts. In this study, this dimension was Operationalized as the number of police and court contacts the youth had since the first contact for the referral Offense. The second dimension was "label spread." This dimension refers to the number of peOple who knew about the police contact that resulted in the referral to the project. The youth was asked to report the number of relatives, friends, school personnel, neighbors, and peOple living in his/her house who knew about the police contact. The third dimension was "label applications." This dimension was Operationalized as the number of people that the juvenile perceived as thinking of him/her as someone who will ”get into trouble in the future." This question was asked in reference to the youth's friends, school personnel, and people living in the youth's household. In addition, for those youths assigned to the Community Service Program, the number of people (friends, school personnel, and members of the youth's household) that knew about their participation in the program was assessed. Knowledge on the part of others that the youth participated in a "program for delinquents" may have been just as labeling as knowledge about the police contact. This last dimension, "program label spread," was included so that this possibility could be explored. The results of the scale-development analyses are presented in Tables 13 and 14. The Program Label Spread Scale was dropped because of the extremely low level of internal consistency (alpha = .22), while three items from the Label Spread Scale were deleted in order to increase the scale's internal consistency to the modest level of .44. The Label Applications and Juvenile Justice Contacts Scales were retained with their original item content. The alpha levels for these scales were .45 and .51, respectively. The interscale correlations ranged from .27 to .61, and had a mean of .44. 112 Table 13. Internal Consistency of Label Encapsulation Scales "Scale Items Alpha Label Spread 5,7,8 .44 Label Applications 9,10,11 .45 Juvenile Justice Contacts 1,2 .51 Table 14. Label Encapsulation Interscale Correlations Label Label J JS Spread Applications Contacts Label 1.00 Spread Label .61 * 1.00 Applications JJS Contacts .45 * .27 * 1.00 Note. {1 = 235. The interscale correlations were corrected for attenuation. *p < .05 113 Perceived Negative Labeling. The Perceived Negative Labeling measure (see Appendix F) was included in order to assess the extent to which the youth perceived that significant others (parents, teachers, and peers) applied a delinquency label to him/her, and the extent to which the youth adapted the same self-perception. This measure was a modification Of a measure originally develOped by Elliot, Ageton, Hunter, and Knowles (197 5). This measure used a 7- point, semantic differential format in which the youth rated the level of perceived delinquency labeling on three scales. The adjective pairs for these scales were: (1) troublesome-cooperative: (2) law abiding-delinquent; and, (3) good-bad. As with the other measures of intervening processes, this measure was administered at the post-period interview. As a result of the high intercorrelation Of the parent, teacher, and peer perceived negative labeling scales, these scales were combined to create a total Perceived Negative Labeling Scale (PNL). The alpha level for this scale was .85. The alpha level of the Label Acceptance Scale was .79, and thus it was retained with its original item content. It should also be noted that these two scales were highly correlated (5 = .72), but they were retained as separate scales due to the rational difference in the dimensions represented by their item content. Community Service Outcome Measure. The Community Service Outcome Measure (see Appendix G) was developed to assess each of the four dimensions of social bonding. Many of the items were selected from those used by Hirschi (1969). Specifically, those items were selected that were related to Official and/or self-reported delinquency and that Hirschi believed adequately reflected the four dimensions of a social bond. The four dimensions of bonding and the corresponding items are outlined below. 114 1. Attachment: Several items were selected from Hirschi (with a few modifications) to assess the youth's attachment to parents (#7-10) and peers (#11-13). 2. Commitment: Two items were selected from Hirschi to assess commitment to employment (#1,2) and four items were selected to assess commitment to education (#3-6). 3. Involvement: Five new items were developed to assess the youth's involvement in conventional/prosocial activities. Involvement was Operationalized as the number of hours the outh "usually" spends doing each of five types of activities working, performing volunteer work, participating in clubs, playing sports and doing "things" with his/her parents) (#14-18). 4. Beliefs: The youth's beliefs in conventional norms were assessed using three items developed by Hirschi (#19-22) and one item developed by Simons, Miller, and Aigner (1980). These items were statements expressing conventional or unconventional values to which the youth expressed his/her agreement on a 5- point, Likert-type scale. The results of the scale development analyses are presented in Tables 15 and 16. A total of three scales were developed from the original pool of items. These scales are Commitment to School, Attachment to Parents and Prosocial Beliefs. The item content of the scales and the alpha levels (which ranged from .52 to .68) are presented in Table 15. The interscale correlations ranged from .27 to .38, with a mean of .34, and are presented in Table 16. Placement Supervisor Interview. In addition to assessing social bonding through youth interviews, social bonding was also assessed from the perspective of the placement supervisors for a subset of 43 youths assigned to the Community Service Program.21 The interviews with placement supervisors were conducted an average of 4 weeks after the youth had completed his/her volunteer work. These interviews were conducted regardless of the number of hours of work completed by the youth, as long as the youth actually performed some work. The agency staff member who most frequently supervised the youth was selected as the respondent. The interview followed a very structured format and was conducted 115 Table 15. Internal Consistency of Community Service Outcome Scales Scale Items Alpha Commitment to School 3,4,5 .68 Attachment to Parents 7,8,9,10 .52 Prosocial Beliefs 20,21,22 .52 Table 16. Community Service Outcome Interscale Correlations Commitment Attachment fiosocial to School to Parents Beliefs Commitment 1.00 to School Attachment .27 * 1.00 to Parents (3 = 158) Prosocial .36* .38* 1.00 Beliefs (g = 233) (Q = 159) Note. 3 = 235. The interscale correlations were corrected for attenuation. *p < .05 116 by one of three interviewers. The original, rationally constructed scales are presented below. 1. Commitment: Three items were included to assess the supervisor's perception of the youth's commitment to employment (#4—6) and two items were included to assess the supervisor's perception of the youth's commitment to education (#7,8). 2. Prosocial Beliefs: Three items were included to assess the supervisor's perception of the extent to which the youth holds prosocial beliefs (#9—11). In addition to the above items reflecting social bonding, other items were included in the interview to assess dimensions of the intervention process that were also assessed in the youth interview. That is, the supervisors were asked to rate the importance of the work performed by the youth (#1), the extent to which the youth worked with other peOple (#2), and the number of friends the youth made at the placement (#3). In addition, the supervisors were asked to provide Open-ended feedback on what they liked best about the program (#13) and what changes they would like to see in the way that the program Operated (#14). Finally, an item was included as a check on the implementation Of one of the design components of the program. As described earlier, one of the intents of the program was to prevent the negative labeling of program participants. One way in which this was to be accomplished was by including community recruited youths in the program and not informing placement agencies which youths were police referred and which were recruited from the community. The effectiveness of the project in concealing the referral source of a participating youth was assessed by an item which asked for the supervisor's rating of the likelihood that a given youth was referred by the police (#12)(see Appendix H for a copy of the interview schedule). Again, the scale development procedures described earlier were used for the social bonding items of the Placement Supervisor Interview. The original scales were retained, although their item content was slightly modified. The final 117 scales, their item content, and their alpha levels are presented in Table 17. The intercorrelation of these scales was .42. It should also be noted that the test-retest reliability of the Placement Supervisor Interview was also assessed. Seven placement supervisors were reinterviewed an average of 4 weeks after the initial interview. The percent agreement for the 12 closed-ended items are presented in Table 18. As can be seen in this table, the test-retest agreement ranged from 28.6% to 100%. The mean percent agreement for the 12 items was 57.1%. Finally, it should be noted that the independent assessments of commitment and prosocial beliefs across source (i.e., community service youths and their supervisors) were correlated -.06 for prosocial beliefs and .44 (p < .05) for commitment. Ultimate Outcomes The most important measures for evaluating the impact of juvenile diversion and the Community Service Program are the recidivism measures. In this study two different types Of recidivism measures were used; Official and self-reported. It is particularly important that both types of measures be used in an evaluation of diversion. As described earlier, from a labeling perspective all youths commit acts that could result in police contact and formal juvenile justice processing. However, labeling theorists argue that only some youths are "caught" and formally processed for their misconduct, and it is this event and subsequent social reactions that are the true causes of delinquency. One such social reaction is the increased probability of formal juvenile justice intervention. Given this perSpective, one might predict that there would be a difference in outcomes when assessing "official" versus "actual" (self-reported) delinquency. In fact, many studies have found such differences (e.g., Davidson et al., 1977 ; Klein, 197 9b). 118 Table 17 . Internal Consistency of Supervisor Rating Scales Scale Items Alpha Commitment to School and Job 6,7,8 .76 Prosocial Beliefs 4,9,10,11 .82 Table 18. Test-Retest Reliability of Placement Supervisor Interview Item P—ercent AEeementa 1. 85.7% 2. 100.0% 3. 28.6% 4. 71.4% 5. 42.9% 6. 71.4% 7. 28.6% 8. 28.6% 9. 71.4% 10. 57.1% 11. 42.9% 12. 57.1% 119 Official records. Both police and court records were used to assess the Official recidivism of project participants. Data from these sources were collected for the 1 year period prior to project intake and for the 16 week period after project intake. Court record data were collected from the county juvenile court. Police record data were collected from each of the four referring police departments (i.e., CPD 1, CPD 2, TPD, and CSD). Four measures of juvenile justice involvement were assessed for the 1 year prior to project intake and for the 16 weeks after project intake. These measures were: (1) the number of police contacts/charges; (2) the average seriousness of police contacts/charges; (3) the number of complaints filed with the juvenile court; and, (4) the average seriousness of the complaint. The seriousness of the offense resulting in police contact or court complaint was assessed on a 3-point scale in which status offenses were scored as 1, misdemeanors were scored as 2, and felonies were scored as 3. While more discriminating systems exist for assessing Offense seriousness, there was not sufficiently detailed information available to permit the use of these scoring systems. The police and court record data were collected by the principal investigator and an assistant who was "blind" to the condition youths had been assigned to. The inter-rater reliability of the police data was assessed for both data collection and data coding, but the inter-rater reliability of the court data was assessed only for data coding (the procedures required by the court for data collection made an assessment of inter-rater reliability extremely difficult). The percent agreement between raters relative to the collection Of police data was 100% for each measure. The percent agreement between raters for the coding of police and court data was also 100% for each measure except the average seriousness of prior court complaints. For this variable the percent agreement was 96.0%. 120 Self-Report of Delinquency. A slightly modified version of the Self-Report of Delinquency measure develOped by Blakely, Kushler, Parisian, and Davidson (1980) was used in this study (see Appendix I). This measure consisted of 29 items reflecting status Offenses, prOperty Offenses, person offenses, drug offenses, and misbehavior in school. The measure was developed to include a wide variety of delinquent acts that are frequently committed by juveniles. Therefore, the items do not include extremely serious offenses such as homicide and criminal sexual conduct. This measure was administered at intake and at the time of the post- period interview. At each time period the youth was asked to report the number Of times he/she had committed each delinquent act "in the last 16 weeks." Each of the 29 items was read to the youth and the interviewer recorded the response. Given the high intercorrelations of the Self-Report of Delinquency (SRD) items, a Total SRD scale was created. This scale had an alpha level Of .72. CHAPTER 5 RESULTS The analyses for this study were conducted in order to answer each of the research questions that were presented earlier. In particular, the following research questions related to describing and assessing the implementation of the three intervention conditions were posed. 1. TO what extent was diversion implemented? 2. What was the intervention process for each of the three conditions? 3. To what extent was the Community Service Program successfully implemented? Three other research questions were posed to address the intervening processes that were hypothesized to link the intervention tO the ultimate outcomes. 1. Did the intervention alternatives produce different levels of labeling? 2. Did the intervention alternatives produce different levels of social bonding? 3. Was there a relationship between the nature of the intervention received by youths in the Community Service Program and the level of labeling and social bonding? Finally, three questions were posed to address the ultimate outcomes of the alternative intervention strategies. 1. Did the intervention alternatives produce different levels of official and self-reported delinquency? 121 122 2. Was there a relationship between the intervening processes of labeling and social bonding and the ultimate outcomes of official and self- reported delinquency? 3. Was there a relationship between the nature of the intervention received by youths in the Community Service Program and the ultimate outcomes of Official and self-reported delinquency? Intervention Assessment Implementation of Diversion According to the impact model tested by this study, the diversion Of juvenile Offenders reduces recidivism by decreasing the negative labeling that results from formal processing. In order to adequately test this model one must first demonstrate that diversion was, in fact, implemented. In this section, results will be presented for the two methods that were used to assess the extent to which diversion was successfully implemented: an assessment of the final police disposition of youths assigned to the traditional processing group and an assessment of the police formal-processing rates before and after project initiation. Disposition of traditional processing youths. The percentage of youths assigned to the traditional processing condition that were actually referred to juvenile court was used as one indication Of the implementation of diversion. Successful implementation would have been indicated by a large proportion of these youths actually having been referred to court. The basic assumption underlying this method is that youths assigned to one of the two diversion conditions would have been handled in the same manner as those who were referred back to the police for traditional processing. 123 In order to carry out this assessment, court records were checked to determine the police disposition of the referral offense for youths assigned to the TP group. These records were checked a minimum Of 6 months following a youth's referral to the project, thus allowing sufficient time for the offense and its disposition to be recorded. Out Of a total of 78 youths assigned to the TP group, 75 (96.2%) were found to have been referred to court for the referral offense. If one assumes that youths assigned to the two diversion groups would have normally been handled in the same manner as those assigned to traditional processing (given the equivalence Of the groups due to random assignment), then the results obtained using this method indicate a high level Of diversion implementation. That is, the results indicate that approximately 96% of the youths in the "diversion" groups had, in fact, been diverted. Pre-post baseline of formal processmg° rates. The second method used to assess implementation was an analysis of formal-processing rates. That is, the rate of formal processing (referral to court) by the participating police departments was compared for the year prior to project initiation (March 15, 1981 to March 14, 1982) with the year after project initiation (March 15, 1982 to March 14, 1983). Successful implementation of diversion would have been indicated by a decrease in the rate of formal processing from the pre-period to the post-period. In order to conduct this assessment, the police disposition of all juvenile contacts for the two, 1 year time periods was collected from three of the four participating police departments.22 While one would expect successful implementation to have been reflected by a decrease in the overall formal- processing rate, it must be recognized that the police departments only agreed to divert certain "eligible" youths. These eligible youths were the target population for the project, and one would expect that the greatest decrease in formal 124 processing would have occurred for this group Of youths. Given this, the assessment of formal processing was conducted separately for a_ll_ juvenile contacts and for those contacts with youths who met the project's eligibility criteria.23 The results of the assessment of formal-processing rates are presented in Tables 19 and 20. As can be seen in Table 19, the results are presented by police department for all youths contacted and for the subset of youths who were eligible for project referral. In the first column, the formal processing data is presented for the year prior to project initiation. Using CPD 1 as an example, one can see that in the 12 months prior to project initiation, CPD I referred 1038 cases to court out of a total of 1702 juvenile contacts: Thus, CPD 1 had a formal- processing rate of 61.0% for this time period. An examination of the pre-period, formal-processing rates reveals that there were substantial differences in the number Of cases handled by the different police departments, and that there was also a large range in the rate of formal processing. That is, the total number of juvenile contacts ranged from 263 for CSD to 1702 for CPD 1, while the formal processing rates ranged from 61.0% to 89.6% for all youth contacts and from 59.6% tO 86.9% for eligible youth contacts. In the next three columns of Table 19, three different estimates of the post- period, formal-processing rates are presented. These three estimates may be considered to represent a continuum of more liberal to more conservative estimates of formal-processing rates. In the first of these columns ("Post: Initial Disposition"), only the police departments' initial disposition was considered in the determination of an "informal" versus "formal" disposition. That is, if the youth was referred to the research project he/she was considered to have received an informal disposition, regardless of the ultimate disposition of the case. Thus, the youths who were determined to be ineligible, the youths who chose not to 125 ...30350 5:0» 350:0: :0 »:000:00 05 5 000::05 «0: 0:03 «00:05 05 »0 0:50:05 00 0« 0055:300 05:0 .0 ...:05:0:5: 00 00550300 0:03 05:50:00 :0:0:0>:0 025 05 :0 0:0 0« 00:0:000 0:0 «00.:05 05 0« 00::0:0: 0:03 0:3 05:0» 0005 »::00 :.:05:0:5: 00 00550300 0:03 5:050 «500 05 0« 00:0:000 0:03 0:3 «:0 «00:05 05 5 0305:0555 0:3 05:0» .05 530350 5:0» 350:0: :0 »:000:00 05 55:3 000::05 «0: 0:03 «00:05 05 »0 0:50:05 00 0« 0055:0«00 05:0» 05:0» ::05:0:: 00 00550300 0:03 0«:0::0:«:0::020 .:0:0:0:05 00 0« 0055:300 0:03 »05 :0 35:05:05 0: «0: 000:000 »05 00:0000 «500 0: 00::0:0: 0:03 :0 :500 0« 00:0:000 »:0«05:«:: 0:03 05:0» 0500 :0: 005000: »05 55:00:50 0«05:«:: 05 :0 000:0:000: .055» ::05:0:5: 00 00550300 0:03 50:05 05 0: 00::0:0: 05:0» ::<0 0:00:05 «0: 003 .30 60:00:00: :0 »«::00 .000 0.55» 05 :0 5:05:85 0050 .00:0::0 :0 0:»« 05 :0 0:000 05 :0 0055:0«00 00 »::0 0::00 : 0:0 :0: :30350 5:0» 0:50:90 300033300 300.030330 $00.00:0:\00 3006000330: 0:03:00 :«:0> 350:: :000.:.0v00:\00: 300030030: 300.0300Qv0 30003000000 300500 500% :2: 0:00 3004.300000: 300.0300030: 300030000: 300.032030: 0:03:00 5:0» 350:: 30043003000 30063000200 30008000000 30003003000 300500 5:0» :0. 0 0:0 300.032.2300 300.035.0300: 300.032.3300 30003002300 0303000 500% 350:: 300.032.3300 30053003300 300.033.3000 300. 80005000: 300500 550% :0. : 0:0 05:00:05 000: 05:00:05 0:: 5050:0500 00:0: :05: :8: 355 :8: 030:: 050000051555: :00:u0:: .0: 0:00.:. 126 Table 20. Reductions in Formal-Processing Rates by Department Initial Final Police Department Disposition Post Disposition CPD 1 All Youth Contacts 13.7% 9.7% 8.7% Eligible Youth Contacts 15.3% 12.4% 11 .2% CPD 2 All Youth Contacts 20.7% 14.0% 8.3% Eligible Youth Contacts 38.9% 29.6% 19.0% CSD All Youth Contacts 41.0% 29.8% 21.1% Eligible Youth Contacts 60.4% 49.4% 33.5% TOTAL All Youth Contacts 18.0% 12.8% 10.3% Eligible Youth Contacts 21.5% 17.1% 13.8% 127 participate and were returned to the police for court referral, and the youths who were assigned to court were all considered to have received an informal disposition. These cases were considered "informal" because of the apparent willingness of the police departments to divert these youths. However, including nonparticipants and ineligible youths as part of the informal group may have resulted in an overestimate of the amount of actual diversion. For example, the police departments could have purposely referred youths who they knew to be ineligible or who they believed were unlikely to participate in order to give the appearance of fulfilling the referral agreement. To control for this possibility, a second more conservative method for estimating formal-processing rates, excluding nonparticipants and ineligible youths, was used. The formal-processing rates presented in the last column of Table 19 ("Post: Final Disposition") were estimated by using the final police disposition. That is, only youths who were warned and released without referral to the project and the youths who were referred to the project and subsequently assigned to one of the two diversion groups were considered to have received informal dispositions. This estimate, the most conservative of the three, results in the highest rate of formal processing. There are two major problems with using this method. First, since random assignment was not stratified by police department, differential rates in the assignment of youths to diversion versus court conditions could have resulted in differential formal-processing rates between the departments that were due to random assignment and not due to the actual level of diversion.24 That is, one police department may appear to have implemented diversion to a lesser extent than other departments as a result of a greater proportion of their referrals having been assigned to the formal processing condition. The second major problem with this method is that it may not accurately represent the amount of diversion that would have taken place had there not been an experimental 128 evaluation of the project. Assuming the same willingness to refer youths to the project if there had not been random assignment to court, then all youths referred to the project would have been diverted. Therefore, all youths would have been handled informally, thus reducing the formal-processing rate. The third column of Table 19 ("Post") represents a "middle of the road" estimate for the post-period, formal-processing rates. This estimate was calculated by including as formal those who chose not to participate and those who were determined to be ineligible for referral, since both groups were to be returned to the police department for referral to court. In addition, this estimate includes as informal those participating youths assigned to court because these youths were eligible candidates for diversion who were handled formally only due to the requirements of the evaluation design. Because they represent a middle, and perhaps more reasonable estimate, the post-period, formal-processing rates presented in column three of Table 19 will be used in the presentation of additional results from the assessment of diversion implementation. A comparison of column three (the post-period, formal-processing rates) and column one (the pre-period, formal-processing rates) reveals that there was a decrease in the formal-processing rates after the initiation of the diversion project. This decrease was found for the total group and the eligible group within each of the three police departments included in the assessment. Further, the decrease was greater for youths who were eligible for project referral, as predicted. In Table 20, the percentage decrease in the formal-processing rates is presented for each category of youth contact within each police department and across the three police departments. As can be seen in this table, the decreases in the formal-processing rates ranged from 9.796 to 29.896 for all youth contacts and from 12.696 to 49.4% for eligible youth contacts. Across all police 129 departments combined, there was a 12.8% reduction in formal processing for all youth contacts and a 17.1% reduction for eligible youth contacts. As might be expected, the police department with the smallest decrease in formal processing was the one with the lowest rate of formal processing prior to the startup of the project (i.e., CPD 1). Figures 4, 5, and 6 present a graphic representation of the formal processing trends for the three police departments during the 12 months preceding and the 12 months following project initiation. These graphs help illustrate the decline in formal-processing rates. The formal-processing trends for CPD 1 are presented in Figure 4. As can be seen in this figure, the monthly formal-processing rate for CPD 1 ranged from 43% to 76% during the 12 months prior to project startup. Immediately after project initiation there was a sharp decline in formal processing. In fact, during the 12 months after project initiation there were only 2 months (October and February) in which the formal-processing rate exceeded the formal-processing rate during corresponding months of the pre—period.25 The same presentation of formal-processing rates can be seen in Figure 5 for CPD 2. In this case, one will notice that the pre-period, formal-processing rate ranged from 67% to 100%. Following the initiation of the project, there was a small but steady decrease in the rate of formal processing. Again, the post- period, formal-processing rates were consistently lower than the pre-period rates. In every case, the post—period, formal-processing rate for a given month was lower than pre—period, formal-processing rate for the corresponding month. It is also noteworthy that after a steady decline in the rate of formal processing during the post-period, the rate remained at a lower level (with the exception of 1 month) than at any time during the pre-period. 130 s 98 .3. 8:: 2.8o8aéesm .e 2.6; "3 ,2 .< ,2 h ...o 2 :83 :83 :83" :83. .139.“ .890 .83. .85» rfica 1&2: a 96 ..8 8.3 gaseooaéesm .m 95m: anon - o...— —z 131 —i- o <_._. ___ <2 __ —o duo -—< a —a —S -< #2 10 42 m 2 4 fl :3: :83 v88 :83. nicer. nice -83. :83 Lisa 1&2: 132 El 98 .8 8%: 2.§8&.§§§ a 23E ..< Ohmuf qLZ -.fl’u -—H dj—Q ...2 ..PO ..-4! duh-H -—H _1_E 4-2 ..p-a -—D 4.2 +4» .._< _—H ..lu—F: ..-2 _L< -F: .83 .8:« .85» $3.. 18...». r899 1.82. 18.5 .8ca #82: 133 The formal-processing trends for CSD are presented in Figure 6. While the formal-processing rates for CSD exhibit the least stability of the participating police departments (probably due to the extremely small number of cases handled by that department), the change in the formal-processing rate following project initiation was especially dramatic. In the 5 months preceding project startup, the formal-processing rate for CSD was 100%. This was followed by an immediate decline to a formal-processing rate of 0% after project startup. While the rate of formal processing continued to be quite variable during the post-period, it remained substantially lower than the pre-period. In fact, there was only 1 month during the post-period when the formal-processing rate was higher than the same month during the pre-period (May). Summary. Two different methods were employed to assess diversion implementation. First, the actual police disposition of youths assigned to the traditional processing condition was assessed to determine the proportion of these youths that were referred to court. Given the presumed equivalence of the groups due to random assignment, it was assumed that youths assigned to one of the two diversion conditions would have been formally processed at the same rate as youths assigned to the TP group, if the diversion alternatives had not existed. Since 96.2% of the traditional processing group was referred to court, support was provided for successful diversion implementation. The second method used to assess diversion implementation involved an assessment of formal-processing rates for the participating police departments. One indication of successful implementation would have been a decrease in the rate of formal processing after the initiation of the project. In fact, such a decrease was found. Across all three police departments there was a 12.8% reduction in the formal-processing rate for all youth contacts, and a 17.1% reduction in the formal-processing rate for those cases specifically targeted by 134 the project (i.e., eligible youth contacts). Thus, the type of "systems impact" that would be predicted by successful implementation of diversion was found. Description of Alternative Interventions Youths who participated in this study received one of three types of treatment; traditional processing (i.e., referral to court), diversion to the Community Service Program, or diversion without services. In this section, information will be provided to more specifically describe the actual treatment provided to youths assigned to each of the three conditions. This description will focus on the intensity of any court intervention received by the youths in each of the three conditions and a description of program services received by youths assigned to the Community Service Program. Traditional processing. The major source of information about the intensity of court intervention was the Court Contact Record. This measure was administered to participating youths during the post-period interview, and it was used to record the youth's report of the intensity of court intervention during the 16 weeks following project intake. This measure was administered to 76 of the 78 youths assigned to the traditional processing condition. As reported by these youths, 34 (44.7%) had one or more contacts with someone from the county juvenile court. Only 10 youths reported that their contact was with a judge or referee (13.2%), while 29 youths reported one or more contacts with some other type of court staff (38.2%). For those youths who reported some contact with a judge, the mean number of contacts was 1.6 and the average duration was 59 minutes. For those youths who reported some contact with other court staff, the mean number of contacts reported was 3.3, with an average duration of 46 minutes. As noted earlier, there were three youths assigned to the traditional processing condition who were never referred to court. According to court 135 dispositional records, very few youths assigned to the traditional processing condition received formal processing beyond the court intake stage for the referral offense. In fact, 53 of the 75 youths actually referred to court (70.7%) were not processed beyond this point. In 7 of these 53 cases the complaint was denied, in 1 case the youth was placed on informal probation, and in 45 cases the youth was referred to one of the court's own service programs.26 An additional 20 youths (26.7%) attended a preliminary inquiry/hearing as the last step in court processing and only 2 youths reached a formal (adjudicatory) hearing. Across all levels of court processing, 27 (36.0%) of the cases were eventually denied or dismissed without further court intervention, 45 (60.0%) of the cases were referred to a court-sponsored service program, and 3 (4.0%) were placed on either formal or informal probation.27:28 Given the relatively nonintensive nature of court intervention received by youths assigned. to traditional processing, a set of exploratory correlations were computed in order to examine the relationship between subject characteristics and the intensity of court intervention. Intensity of court intervention was defined as the seriousness of the final court disposition (i.e., how far into the court system the youth penetrated, with court intake ranked as minimum penetration and a dispositional hearing ranked as maximal penetration) and the number of contacts with a judge and/or other court staff. Client characteristics included several demographic variables and prior, off icial—delinquency variables. The client characteristics included in these analyses were generally available to juvenile court staff, and this information may have influenced their handling of particular youths. The results of the correlational analyses are presented in Table 21. As can be seen in this table, there was a tendency for the youths' prior record to be related to the intensity of court intervention. It appears that while. a greater 136 Table 21. Correlation of Client Demographic and Court Measures Demographic Court Disposition No. of Characteristic Seriousness Court Contacts Age -.04 -.24* Sex -. 10 .05 Family Status .02 . 17 Race .23"I . 10 School Status - . 06 -. 04 Employment Status .04 -. 10 No. of Prior Offenses .43* .22* Seriousness of Prior Offenses -.03 -.25* Note. The n for the correlations with the seriousness of the court disposition was .75 and the n for the correlations with the number of court contacts was .76. a"E<.05 137 frequency of prior offenses was related to more intensive court intervention, there was a tendency for less serious prior offenses to be related to more intensive court intervention. In terms of the demographic characteristics, the youth's age was negatively correlated with the number of court contacts and the youth's race was positively correlated with the seriousness of the court disposition, indicating that more serious court dispositions were related to being nonwhite. No other correlations were found to be significant. Diversion without services. While youths assigned to the diversion without services condition were to be released without any further contact with the juvenile justice system as a result of the referral offense, it was still possible that these youths would have court involvement resulting from a new offense. The Court Contact Record was administered to youths in this condition in order to assess any such involvement during the 16 weeks following project intake. As would be predicted given the nature of this alternative, very few youths assigned to the diversion without services condition reported any court contact during the post-period. Of the 82 diversion without services youths completing this measure, only 5 (6.1%) reported any court contact. Two youths reported one or more contacts with a judge and four youths reported one or more contacts with other court staff. While youths assigned to the diversion without services condition were not to be referred to court for the project referral offense, court records were examined to check on this aspect of the experimental manipulation. Through this assessment it was discovered that seven of the youths assigned to this condition had actually been referred to court by the police for the project referral offense. However, the court records indicated that these youths had received very nonintensive court intervention. The complaint was denied/dismissed without further court action for five of the youths (one at court intake and four at a 138 preliminary hearing/inquiry), while two were referred to a court service program (one from court intake and one as a result of a preliminary hearing/inquiry). Diversion to the Community Service Progm. The Community Service Pro- gram intervention was assessed both in terms of the nature of court contact experienced by program clients and in terms of the services provided to these youths. This latter information was collected in order to describe the program process and in order to assess the extent to which the Community Service Program was implemented successfully. Like youths in the diversion without services condition, youths in the diversion to community service condition reported very little court involvement during the 16 weeks after project intake. Only 5 of the 77 community service youths (6.7%) reported any court contact. One youth reported contact with a judge while four youths reported contact with some other court official. Again, in this condition there were a number of youths who were supposed to be diverted who were instead referred to court by the police. Fourteen of the 79 youths (17.7%) assigned to diversion to community service were referred to court for the project referral offense. Nine of these youths had the complaint against them denied or dismissed without further court action (three by court intake and six as the result of a preliminary hearing/inquiry). The other five youths were referred to court service programs by the court intake unit. Additional information concerning youths assigned to the diversion to community service condition was obtained from records kept by program staff and from the Community Service Process Interview conducted as part of the post- period interview. These two sources were used to provide a detailed description of the Community Service Program and the services provided to program participants. 139 Program records were used to describe several aspects of the intervention process. These included the arbitration meeting, the types of community service placements used by the program, the amount and type of community service work performed by the youths, and the intensity of staff supervision/monitoring provided by the program. An arbitration meeting was held for 77 of the 79 youths assigned to the program.29 There was a victim in 63 of these cases (81.8%), but he/she attended the arbitration meeting in only 12 cases (19.0%). In the vast majority of cases the victim was a retail store. It is also notable that it was very rare for the referral offense of program participants to have resulted in the loss or destruction of property. As a result of this, restitution was provided in only two cases. A total of $138 in restitution was paid to these two victims; one received $51 and the other $87. In addition to these two cases in which restitution was provided, there were three cases in which an agreement was negotiated between the victim and youth in order to resolve a dispute between these parties. After completion of the arbitration component of the program, youths were to select a community service placement where they would perform 36 hours of voluntary service. In the nearly 18 months of program Operation covered by this study, a total of 41 different organizations were recruited to serve as placements for program youths. A list of the different types of organizations recruited and the number of each type is presented in Table 22. As can be seen in this table, a wide range of human service organizations participated. The most frequent types of organizations were day care centers, senior centers/nursing homes, and libraries; with seven of each serving as placements for program youths. These same types of organizations were also the most frequently used by the program. As can be seen in Table 22, day care centers were selected as the site for 140 Table 22. Types of Placements Recruited and Used Type of Placement :Recruited Used Day Care Center 7 14(18.7%) Senior Center/Nursing Home 7 10(13.3%) Library 7 17(22.7%) Hospital 2 0 Community Center 5 6(8.096) Church 3 5(6.7%) Human Service Agency 2 0 Other (e.g., parks and recreational department, symphony and nature center) 8 23(30 .7%) 141 volunteer work by 14 youths (18.7%), senior centers/nursing homes were selected by 10 youths (13.3%), and libraries were selected by 17 youths (22.7%). Youths assigned to the Community Service Program were to perform 3 to 4 hours of volunteer work per week, for 12 weeks. The actual number of hours of community service performed by program participants ranged from 0 to 197 hours, with the median number of hours being 36. The number of individual sessions of community service work ranged from 0 to 45, and the median number of sessions was 12.8. A total of 48 youths (60.8%) completed the expected minimum of 36 hours of community service. A more detailed breakdown. of the number of community service hours completed by program youths is presented in Table 23.30 Of the 79 youths assigned to the program, 71 (90.0%) maintained at least some level of involvement for the full duration of program services. Three youths (3.8%) did not complete the program because they moved out of the county, two youths (2.5%) were withdrawn from the program by their parents, and three youths did not complete the program for other reasons.31 The types of community service activities performed by the youths reflected the variety represented by the different types of placement agencies available to the program. The specific types of volunteer activities performed ranged from assisting in direct patient care at a medical facility to assisting in maintenance work at a church. Listed below are examples of job titles and the responsibilities associated with them. 1. Activity Director Aid: Assisting in the design, organization, and running of activity groups for residents of a nursing home. 2. Tutor: Helping students with school work at a community center's tutorial program. 3. Barnyard Assistant: Assisting in the care and feeding of animals at a nature center. 142 Table 23. Hours of Community Service Completed Hours Nowm 0-10 6 11-20 4 21-30 13 31-40 39 41-50 12 over 50 4 143 4. Recreation Assistant: Assisting handicapped children in organized recreational activities and Sporting events. 5. Medical Clerk Office Assistant: Assisting staff in the organization and maintenance of medical records and equipment. 6. Graphics Production Assista4nt_: Selecting, evaluating, and preparing pictures for circulation at a library. Finally, program records were used to document the amount of supervision and monitoring provided by program staff. Staff were to maintain at least weekly contact with program participants and at least biweekly contact with the youths' supervisors during the 12 weeks of community service work. The Contact Records kept by program staff indicate that the mean number of youth contacts was 15.5 and the mean number of placement agency contacts was 5.9. Additional information about the program intervention process was gathered through the Community Service Process Interview conducted with the youth after program completion (scheduled for 16 weeks after project intake for all youths). Items included in this measure were intended to address specific dimensions of the program that were prescribed by the program model. These included the amount of time the youths spent working with others, the youths' attachment to the community service placement, the provision of services not intended to be part of the program model, and the amount and the nature of interaction with placement staff. In addition, the youths' satisfaction with the program, the program staff, and the community service placement was assessed. One of the goals of the program was to facilitate the development of a social bond (attachment) between the youth and the members of the placement agency. Given this goal, an attempt was made to structure the community service work in such ways that this would be more likely to occur. The primary method by which social bonding was to be encouraged was by creating a situation in which the youth would work alongside other members of the organization. Therefore, items were included in the interview to assess this dimension of 144 program Operation. The Work Group Scale described previously addressed this dimension. On a 5-point scale in which a high rating indicated more participation in work with other people, the mean score was 2.22. Specifically, 15 youths (20%) indicated that they worked alone "more than half of the time," and 18 youths (24%) indicated that they worked alone "all of the time." The ability of the community service experience to foster social bonding may also have been influenced by the amount and nature of the youth's interaction with staff at the placement agency, especially his/her supervisor. Several items were included to address this dimension. While the youths indicated that there was usually someone available to help them if they needed assistance (mean score of 4.53 on a 5-point scale), they also indicated that they were not closely supervised (mean item scores ranged from 1.38 to 1.74 on a 4-point scale for items assessing the frequency with which supervisors checked the youth's work, showed the youth how to do the work, and suggested ways of doing the work better). Program participants also reported that they did not frequently talk to agency staff members about personal problems (1.28 on a 4-point scale) or other nonwork related topics (1.43 on a 4-point scale). At the same time, the youths indicated that their supervisors gave them compliments frequently (4.01 on a 5- point scale) and were very friendly towards them (4.69 on a 5-point scale). Overall, program participants' responses also indicated at least moderate attachment to the community service agency and the work performed. The youths' responses indicated that the youths felt that the work they performed was relatively important (2.42 on a 3-point scale), helpful to the placement agency (3.69 on a 4-point scale), and interesting (3.03 on a 4-point scale). In general, the youths' responses also indicated substantial satisfaction with the program and program staff. Specifically, the youths indicated that the program was helpful to them (4.51 on a 5-point scale) and was a good thing to 145 happen to them (4.11 on a 5-point scale). Their responses also indicated a great deal of satisfaction with program staff. The mean score for the Staff Satisfaction Scale was 4.72 (5-point scale). Two items were also included to monitor program staff's compliance with two restrictions on service delivery. That is, staff were prohibited from referring clients to other service programs and from providing "counseling" services to program clients. Of the 77 youths included in this assessment, only 1 youth reported that he had been referred for additional services, and the mean item score for an item which assessed the frequency with which staff talked to the youth about personal problems was 1.88 on a 5-point scale, with a 5 indicating high frequency. Finally, an item was included in the interview with placement supervisors to assess their perception of the likelihood that individual participants were referred to the project by the police. As described earlier, the program was to include youths referred by the police 9M. youths recruited from the general community in order to reduce the probability of delinquency labeling of program participants. While the program was relatively unsuccessful in recruiting community youths (four such youths were recruited), it appears that the youths' supervisors did not generally identify program participants as having been referred by the police. In fact, of the 43 program participants included in this assessment, supervisors identified 15 (37.2%) as either "definitely" or ”probably" referred by the police, and they identified 20 (46.5%) as "definitely not" or "probably not" referred by the police. For the other 7 youths (16.3%), the supervisors reported that they were unable to make such a judgment. Summa_ry. In this section, information was presented to describe the nature of the intervention given to youths in each of the three conditions. First, the intensity of court involvement subsequent to project intake was described for 146 each condition. It was reported that several youths in the two diversion conditions were inappropriately referred to court, and that their involvement in court was very nonintensive. The majority of cases were denied/dismissed prior to a formal hearing without any further court intervention, and very little court contact was reported by the youths. In fact, only 6.1% of the diversion without services youths and 6.5% of the diversion to community service youths reported any contact with the juvenile court. For the youths assigned to traditional processing, 96.2% were actually referred to court. The subsequent court processing of these youths tended to be nonintensive and informal. Only 44.7% of these youths reported actual contact with the juvenile court, and very few (2.6%) were processed beyond the preliminary hearing/inquiry stage. Finally, for youths assigned to traditional processing, there was a 60.0% referral rate to court- sponsored programs. Also presented in this section were program process results for the diversion to community service condition. According to these results, there were very few cases in which the arbitration meeting was used to arbitrate conflicts or to negotiate restitution. This appears to be due to three factors. First, there were many cases in which there was not a victim. A fairly large proportion of the cases involved victimless crimes such as, truancy, possession of marijuana, running away, and so forth. These victimless crimes accounted for 18.2% of the program participants. Second, even when there was a victim the victim did not usually attend the arbitration meeting. In fact, only 12 (19.0%) of the victims attended the arbitration meeting. Finally, there were very few cases in which there was an actual loss of property by the victim. There were only a few cases of malicious destruction of property, and in almost all larceny cases the property was recovered undamaged. 147 Results were also presented which described the community service work. There were many placement options available to the youths, and the placements provided a great variety of work activities. It was also determined that once placed in a position of community service the youths usually completed the program, with very few withdrawing as a result of some dissatisfaction with the program. A majority completed the expected 36 hours of community service, and many youths far exceeded this amount. In addition, program participants reported that they were very satisfied with the program, program staff, and the community service work they performed. In general, it was found that most of the program guidelines were followed. The program staff maintained the prescribed amount of contact with the youths and placement agencies, and they apparently refrained from providing "counseling" services and referrals to other service programs. However, program youths reported less interaction with the staff at the placement agencies than was intended. Program clients reported that they tended to work alone much of the time and to talk to staff about nonwork related topics infrequently. Still, the youths indicated that the staff were generally friendly and that they found the work to be interesting. Equivalence of Groups Before presenting the results of the intermediate and ultimate outcomes analyses, information will be presented that is related to the equivalence of the three groups at the time of project intake. The groups were compared on several demographic and delinquency variables. The demographic variables included the youths' race and sex, as well as their age, school status, employment status, and living arrangement at the time of project intake. The delinquency variables included in this analysis were the type of referral offense, the number and average seriousness of offenses and court complaints during the year prior to 148 project intake, and the total number of self-reported delinquent acts during the 16 weeks prior to project intake. The results of the comparisons on the demographic variables indicated that the groups were nearly identical on these variables. Specifically, across the three groups the average age ranged from 14.6 to 14.8 years ($2,240) = .32, p > .05), the percentage of white youths ranged from 73.1% to 74.7% (x2(2) = .06, p > .05), and the percentage of males ranged from 54.7% to 59.5% (X 2(2) = .41, p > .05). There were no significant differences between the groups on any of the demographic variables. The results of the comparisons on the delinquency variables were somewhat less consistent. While there were no significant differences between the groups on the type of referral offense ( X 2(54) = 49.39, p > .05), the total number of self- reported delinquent acts (£2,240) = .38, p > .05), and the average seriousness of prior offenses (F(2,240) = 1.36, p > .05) and court complaints (_F_‘(2,102) = 6.00, p > .05), there were significant differences between groups for the number of prior offenses (132,240) = 3.23, p < .05) and for the number of prior court complaints (§(2,240) = 3.21, p < .05). For both the number of offenses and the number of court complaints, the formal processing group had a more serious prior record than the other two groups.32 However, upon closer examination it appeared that the pre—period differences were largely due to a relatively small number of youths assigned to the formal processing condition who had extensive delinquency records. For example, the number of prior court complaints ranged from zero to two for the two diversion groups, while the formal processing group had one youth with seven court complaints and one youth with six court complaints. ' In order to control for these outlying cases, the number of offenses and court complaints were recoded to 0, 1, and 2'or more, and an analysis of variance 149 was conducted again. This time, there were no significant differences on either measure of prior delinquency (for offenses, fi(2,240) = 2.74, p > .05 and for complaints {(2,240) = 2.30, p > .05). Given the above results, the groups were considered to be equivalent at the time of project intake for the purposes of subsequent analyses of post-period measures. Intermediate Outcomes In this section, the results of the labeling and social bonding intermediate outcome analyses will be presented. This presentation will include analyses of variance (AN OVA) that were used to test for significant differences between the groups on several dimensions of labeling and social bonding, and exploratory analyses that were used to examine the relationship between these intermediate outcomes and several process variables. Analyses of Variance for Labeling and Social Bonding A one-way analysis of variance was used to test for significant differences between the groups on each dimension of delinquency labeling and social bonding. These analyses were conducted using two different sets of project participants. As the reader will remember, there were 24 youths who were not handled by the police in the manner prescribed by the results of the random assignment. That is, there were youths assigned to one of the diversion conditions who were referred to court and there were youths assigned to traditional processing who were not referred to court. Because of this situation, two sets of analyses were conducted. The first Set included all project participants (5 = 243) and the second set included only those project participants who actually received the police disposition indicated by the random assignment (5 = 219). However, since the results of the analyses were the same regardless of the subject population used, 150 only those analyses based on the entire group of project participants will be presented. Delinquency labeling. An ANOVA was conducted for each of five measures of delinquency labeling. These measures were: the number of juvenile justice contacts within 16 weeks after project intake; Label Spread (the number of people who knew about the youth's juvenile justice contact); Label Applications (the number of people the youth thought perceived him/her as delinquent); Perceived Negative Labeling (the extent to which the youth perceived others as believing he/she was delinquent); and Label Acceptance (the extent to which the youth perceived himself/herself to be delinquent). The results of these analyses indicate that there was a significant between groups difference on only one measure of labeling, the number of juvenile justice contacts (F(2,232) = 3.82, p < .05) (see Table 24). Post hoc analyses (Scheffe tests) indicate a significant difference between the traditional processing group and the diversion without services group (20,232) = 6.50, p < .05). There were no significant between group differences for the other measures of delinquency labeling. These included Label Spread (£2,232 = .31, p > .05), Label Applications (20,232 = .37, p > .05), Perceived Negative Labeling (32,229) = .87, p > .05), and Label Acceptance (29,232) = 2.19, p > .05).33 Social bonding. The results of the ANOVA for each of the three measures of social bonding do not indicate that the groups were significantly different on any of the following dimensions: commitment to school (F(2,231) = l.27,p > .05); attachment to parents (F(2,156) = .65, p > .05); and, prosocial beliefs (32,231) = .73, p > .05).34 151 Table 24. ANOVA for JJS Contacts Scale Source (if~ SS; W143— F *1"? m7 Between Groups 2 31.35 15.68 3.82 .02 .023 Within Groups 232 951.01 4.10 Total 234 982 . 37 Note. The mean number of contacts for the TP, DCSP, and DWS groups was 1.07, .34, and .24 respectively. 152 Exploratory Analyses A set of exploratory analyses were conducted to examine the relationship between certain process dimensions of the Community Service Program and the intermediate outcomes of labeling and social bonding. Because there were no differences between the three groups on the extent of delinquency labeling and social bonding, an examination of the correlations between program operation measures and these intermediate outcome measures was conducted to provide information that may help explain the results and that may be useful in future program develOpment. Pr_ogram process and delinquency labeling. The correlations between eight measures of Community Service Program process and five measures of delinquency labeling are presented in Table 25. Out of a total of 40 correlations, 18 were significant (p< .05). Two measures of program process were consistently correlated with the measures of delinquency labeling. First, the extent to which the youths believed that program staff thought them to be delinquent was positively correlated with four of the five labeling measures. Secondly, the youth's attachment to the community service work (i.e., the extent to which he/she thought the work was interesting, important, and helpful) was negatively correlated with each dimension of labeling. However, even though there were a large number of significant correlations, the relationships were not very strong. The significant correlations ranged from .20 to .56, with 12 of the 18 significant correlations falling between .20 and .29. Prggram process and social bonding. In contrast to the delinquency labeling results, there were very few significant correlations between the program process measures and the five measures of social bonding. These correlations are presented in Table 26. Three of the measures of social bonding were obtained for 153 Table 25. Correlation of Program Process and Labeling Measures JJS JJS Label Label Label Contacts Spread Application PNL Accejtance Project Satisfaction -.24* -.23* -.31* -. 17 -.19 Staff Satisfaction -.22* -.04 -.24* -. 15 -.09 Work Groups .07 -.06 .01 -.05 -.02 Attachment- Placement -.02 -.19 -.22* -.05 -.06 Staff PNL .34* .11 .28* .56* .41* Parent Support -. 17 .02 -.26* -.27* -. 18 Community Service Attachment -.20* -.23* -.31* -.25"‘ -.27* Hours of Community Service .14 .04 . 14 .22* .00 Note. The correlations were computed with an E ranging from 74 to 75. *p < .05 Table 26. Correlation of Program Process and Labeling Measures 154 Com mitment Attachment Prosocial Commitment to Prosocial to School-Y to Parents Beliefs-Y Job/School-S Beliefs-S Project Satisfaction .06 .07 .35* .13 .19 Staff Satisfaction -.07 .07 .05 .13 .21 Work Groups .12 .07 .12 .16 -.05 Attachment- Placement -.14 .20 .10 .14 .21 Staff PNL -.32"‘ .04 -.28* -.42* -.32"‘ Parent Support .00 .47“ .32* . 16 . 17 Community Service Attachment .10 .00 .38* .04 -.07 Hours of Community Service .05 -.03 .15 -.08 -.07 Note. The correlations of program process with the youth measures of social bonding were computed with an E ranging from 49 to 75, while the correlations of program process with the supervisor measures of social bonding were computed with an 3 ranging from 39 to 43. It should also be noted that measures obtained from youth interviews are indicated by a "Y," while measures of the same construct obtained from supervisor interviews are indicated by an "S." *p < .05 155 all project participants during the youth interviews, and two of the measures were obtained only for Community Service Program participants during interviews with community service supervisors. As can be seen in Table 26, only 8 of the 40 correlations were significant and, with the exception of the Staff PNL process measure, there were no consistent relationships between program process and social bonding. Sum mm Analyses of variance were conducted to test for between-group differences on social bonding and delinquency labeling. Results of these analyses indicate that there were no between-group differences for any of the three measures of social bonding and only one for the five measures of labeling (the number of juvenile justice contacts). A post hoc analysis on the number of juvenile justice contacts (Scheffe) found that the traditional processing group had a higher frequency of juvenile justice contacts during the 16 weeks following project intake than did the diversion without service group. In terms of the exploratory analyses, there were few consistent patterns of significant correlations between program process variables and intermediate outcome variables. Perceived negative labeling by program staff was consistently related to both labeling and social bonding (positively related to labeling and negatively related to social bonding), and the youths' attachment to the community service work was consistently related (negatively) to labeling. Higher levels of perceived negative labeling by program staff was associated with higher levels of other types of labeling and lower levels of social bonding, and higher levels of attachment to community service work was associated with lower levels of labeling. However, in almost all cases these relationships were relatively weak, accounting for only a small amount of variance (_r_2 ranged from .04 to .10 for Community Service Attachment and the various labeling measures; from .12 156 to .31 for Staff PNL and labeling; and from .08 to .18 for Staff PNL and social bonding) even when the correlations were significant. Ultimate Outcomes In this section, the results of the analyses related to official and self- reported recidivism will be presented. Included in this presentation will be a Chi- square analysis of simple recidivism and an analysis of variance for multiple recidivism and self-reported delinquency. Each of these measures of recidivism was for the time-period covering the 16 weeks after project intake. In addition to the above analyses, exploratory analyses will be presented that examine the relationships between recidivism and the intermediate outcomes, and between recidivism and community service processes. Finally, the ability to predict recidivism from participant demographic, prior record, and intermediate outcome data will be examined. Project Impact on Recidivism In the analyses of recidivism, the three groups of youths were compared on both official and self-reported measures. While four different measures of official recidivism were made for this study (i.e., number of offenses, average seriousness of offenses, number of court complaints, and average seriousness of court complaints), these measures were found to be extremely intercorrelated. The intercorrelations of the four measures ranged from .59 to .80 (see Table 27). Given this finding, all subsequent analyses of official recidivism used only one measure of official recidivism, the number of offenses. This measure was selected because it had the highest average correlation with the remaining three measures. It should also be noted that, as with the analyses of intermediate outcomes, the analyses of recidivism were conducted for all project participants (11 = 243) 157 Table 27 . Correlation of Pre- and Post-Period Measures of Recidivism Measure Correlation of Recidivism Coefficient Number of Offenses .04 Average Seriousness of Offenses -.06 Number of Court Complaints .09 Average Seriousness of Court Complaints .00 Note. 3 = 243 158 and for those youths who received the police disposition indicated by the random assignment (N = 219). Again, the results were consistent regardless of the subject population used. Therefore, only those analyses including all project participants will be presented. Simple recidivism. A Chi-square analysis was conducted for the crosstabulation of condition by simple recidivism. Simple recidivism was defined as one or more offenses during the 16 weeks after project intake. Overall, there was very little recidivism. The total amount of offense recidivism was 13.6% across the three groups, and the Chi-square analysis indicates that there was no significant relationship between condition and simple recidivism (x 2(2) = 2.51, p > .05).35 Multiple recidivism. A one-way analysis of variance was conducted in order to examine the ultimate outcome of multiple recidivism for the 16 weeks after project intake. The number of offenses was coded as 0, 1, and 2 or more. Again, there were no differences between conditions on this measure (£12,240) = 1.80, p > .05).36 An analysis of covariance (ANCOVA) was also conducted in order to test for group differences on official multiple recidivism. In this analysis, the number of pre—period offenses served as the covariate. Again, no between-group differences were found (_F_(2,239) = 1.71, p > .05). Self-reported delixguiency. Finally, an analysis of variance was conducted for the total number of self-reported delinquent acts during the 16 weeks after project intake. As would be expected, the number of self-reported delinquent acts indicated a much greater frequency of recidivism than did official records of delinquency. However, no between-group differences were evidenced (29,232) = .62, p > .05). It should be noted that an ANCOVA was also conducted 159 using the total number of self-reported delinquent acts during the 16 weeks prior to project intake as the covariate. The AN COVA was conducted in order to increase precision, but again, no between-group differences were found (30,219) = 1.04, p > .05).37 Exploratory Analyses In an attempt to develop additional insights into the operation of the diversion effort and to help explain the results of the outcome analyses, several sets of exploratory analyses were conducted. These analyses included: (1) an assessment of project impact on recidivism for participants who received the "complete treatment" indicated by their random assignment; (2) an assessment of the relationship of recidivism with the intermediate outcomes (labeling and social bonding) and with community service process; and (3) an assessment of the ability to predict recidivism using various participant characteristics (i.e., demographic, delinquency history, labeling, and social bonding variables) as predictors. Recidivism for the "complete treatment" mg. As described earlier, there were many cases in which youths assigned to a particular treatment condition did not receive the treatment intended for those youths. In some cases, the police did not give the youth the disposition indicated by the random assignment, while in other cases the youth received the correct disposition but did not receive the intensity of treatment originally planned. In an effort to more closely examine the impact of the three intended alternatives, an attempt was made to eliminate youths from each condition who did not receive the complete treatment regimen. The purpose of using this strategy was to create groups that more clearly represented the three alternative treatments. Theoretically, this strategy should have maximized the likelihood that the groups would have exhibited differences in recidivism.38 160 The new groups were formed in the following manner. First, all youths who did not receive the disposition indicated by the random assignment were excluded (this was the same procedure described earlier). Second, specific criteria were applied to each treatment condition in order to eliminate cases that did not receive an acceptable level of the planned treatment. The remaining youths were considered to have received the "full treatment." For the TP condition, only those youths who were processed mg court intake (as indicated by court records) and who reported personal contact with a court official (2 = 15) were included. For the diversion to community service condition, only those youths who completed the program with 36 hours or more of community service work were included (3 = 42). Finally, for the DWS condition, only youths who reported that they did 92]; receive any human/social services during the 16‘ week post- intake period were included (3 = 62). Once again, an ANOVA indicated no difference between the groups on either self-reported delinquency (£32,116) = 1.54, p > .05) or official delinquency (212,116) = 2.40, p > .05). Community service process and recidivism. The two measures of recidivism were correlated with eight measures of community service process (see Table 28). Overall, little relationship was observed between the nature of the services received (including the amount of services, the organization of the service activity, and the youth's satisfaction with the program) and the amount of official recidivism. However, self-reported recidivism was significantly correlated with three measures of community service process. Specifically, self-reported delinquency was negatively correlated with Project Satisfaction and Community Service Attachment, and positively correlated with Staff PNL. 161 Table 28. Correlation of Program Process and Recidivism Measures No. of Total Offenses SRD Project Satisfaction -.06 -. 28* Staff Satisfaction -.10 -.13 Attachment to Placement .08 -.01 Community Service Attachment . 02 - . 26 * Hours of Community Service . 05 - . 07 Work Groups .09 - .04 Staff PNL .10 .25* Parent Support -.01 -.03 Note. The p for these correlations ranged from 74 to 7 5. *2 < .05 162 Delinquency labelinLand recidivism. The same two measures of recidivism were correlated with the four measures of delinquency labeling. The results of these analyses are presented in Table 29. As can be seen in this table, a consistent relationship between labeling and recidivism was found. Out of the 10 correlations, 8 were significant in a positive direction (p< .05). In addition, all measures of labeling were significantly correlated (positively) with self-reported delinquency. Social bondiand recidivism. The results of the correlational analyses between social bonding and recidivism are presented in Table 30. Out of a total of 10 correlations, 3 were significant (all in a negative direction). Again, all three significant correlations were with self-reported delinquency. Also, it should be noted that only the three measures of social bonding assessed through the interviews with the youth were correlated with self-reported delinquency. In all cases, the significant correlations were negative, indicating that lower levels of social bonding were related to higher levels of self-reported delinquency. Prediction of recidivism. To further explore the relationship between the intervention process and recidivism, multivariate analyses were conducted. Two multiple regression analyses were conducted to predict recidivism as measured by self-reported delinquency: one included only pre-period measures of participant demographic and prior, official-delinquency variables as predictors; the second included these same variables plus post-period measures of labeling and social bonding. Stepwise regression analyses were conducted in which variables selected for inclusion in the equation met the minimum criteria of an 2 value greater than 3.89 (this correSponds to probability level of .05) and a tolerance level greater than .01. For the first regression analysis, four demographic/prior official record 163 Table 29. Correlation of Labeling and Recidivism Measures No. of Total Offenses SRD JJS Contacts .25"I .32" JJS Label Spread . 12 * . 33 "' Label Applications . 03 . 28" PN L . 17 * . 40" Label Acceptance . 09 . 20* Note. The p for these correlations ranged from 232 to 235. *9 < .05 164 Table 30. Correlation of Social Bonding and Recidivism Measures No. of Total Offenses SRD Commitment to School -.01 -.34* (p = 234) (p = 234) Attachment to Parents .03 -.29" (p = 159) (p = 159) Prosocial Beliefs-Youth - . 08 - . 35 "' (p = 234) (p = 234) Commitment to Job/School—Supervisor -.08 . 18 (p = 39) (p = 39) Prosocial Belief s-Supervisor - . 1 1 - . 05 (p = 43) (p = 43) *p<.05 165 variables entered the equation, yielding an R2 of .12. Thus, approximately 12% of the variance in self-reported delinquency was explained by the pre-period measures. A list of the predictors and a summary table for the regression analysis is presented in Table 31. For the second regression analysis, the demographic/prior official record variables were entered first, and these were followed by the labeling/social bonding variables. Within each group of variables, the stepwise procedure described above was used. For this analysis, in addition to the four demographic/prior official record variables, five measures of labeling and social bonding entered the equation resulting in an I}? of .35. A list of the predictors and a summary table for this regression analysis is presented in Table 32. In addition to the above analyses, two discriminate function analyses were conducted to examine the association between the same set of predictor variables and a dichotomized measure of official recidivism. For the dependent measure, youths were classified as recidivists or nonrecidivists based on whether or not they had an officially recorded offense during the 16 weeks following project intake. For these analyses, no variables entered the equation. The discriminant analyses performed were stepwise discriminant analyses using Wilk's Lambda (equivalent to the largest overall multivariate E) for selection of variables entering the equation at each step. Additional criteria used included, an E to enter of 3.89 (p < .05) and a tolerance level of .01. None of the predictor variables met the criteria used, thus preventing the classification of participants into recidivist and nonrecidivist groups (see Table 33). Sum may The results of the analyses of ultimate outcomes clearly and consistently indicate that there were no differences between the three groups on subsequent official and self-reported delinquency. In addition, while a consistent relationship 166 .momcoto hora no mmocmsorom cacao; new .82 £8 .03 p.83 cos—26o 2: 35 ooaoaco go: 333...; .802 «a. am 3. 5.. Scfimcoov 2. «H. mm. «a. 56 3.2 vmém £53m doocom .w no. ma. 5. no. :4. SJ 3.3 moms—etc nor—m mo .02 .m 2.. S. mm. «o. and mod 3.2 man—3m acofizanm .m 3. me. 3. «a. SJ. 36 3.2 £53m 323m .H magma «M fl M aoucm m no mm m wfiaoucm 95m um .58 8 N 33st; monta> oEmwamoEoQ mEmD Qmm mo 530695 .3 035. 167 00530000 822 0:0 6:230:30 302 600.50 323 $020.30 nova «o 0005:0200 0m0~0>0 .000.— .x0m .03 0003 550.60 05 35 c0830 «0: m032§> .0uoz 2.2 3.3.. 9:32.09 as. mm. mm. ma. mm.” pm. «p.- goocom OH HCQEu—mEEOO om S. 2. E. 8. 3.... 2.. 2%- $5.80 8 202532 .m 8. S. 3. S. S; 3; $4.. 38:30 a... 4. 3. Z. S. 8. $5 2... 2.3- 3230 388.5 a S. 2. 3. 8. 3.2 $5. :6 :20 38. .m 2. 2. S: 2. 3... SJ 3.0 8288 03.5 .«o 308:2 .v 2. S. a. 2. $4 3.3 24;” 2:80. 82% .m 8. S. E. S. :3 $5 3.3 2.35 208335 .N 3. 3. 2. S. 35 $5 3; 25¢. 35.; A w l l I I I o cwzo «m m m .525 m .8 mm 0 228:0 85 mm .cma 3 m 832...; 833.5 95282 28 023.853 was: 95 go 5:285 .3 033. 168 mmfi n ma 883689. 983 a >20 .330 cmH 2: «O 6032...; 95.33286 98:. .8 one :0 $33 9:82: 3 26 @8338 0.53 momma co>omnbcwfi .oHoz 8 . 8 . H 8 . H 8 . H 823 H3885 8H 8. 8H 8H 388m 3 83832 8. 8H .8. H 8. H 388. 3 255880 8 . H 8 . 8 . H 8 . H 8:388on H83 8. 88 8H 8H 42¢ 38. 8 . 8. 8 . H 8 . H 828233 H83 8. 8 . N 8. H 8. H 828 H83 8 . 8 . H 8. H 8. H 88:8on notoaomufiom norm 8. 8H 8H 8H 8888 .55 mo mmocmsotmm mmgm>< 8. 8H 8. H 8. H 8283 SE 8 .882 8H 8. 8H 8H 83H 8 . 8 . 8 . H 8 . H 888 Hcoszanm --- -1... 8 . a 8 . a 888 388 8. 8.8 8H 8H 8:8 8:58 88. Hm.H cc.H ¢¢.H xom 8. 8. 8H 8H 8< 25qu $53 .885 3 m 8:838. 8:238. 338...; 53552 «fiazsoom H2080 85586 £885 888285 .8 mam: 169 was not found between community service process and social bonding and the ultimate outcome of official delinquency, these constructs, as well as labeling, were consistently related to self-reported delinquency. While these relationships were not strong, they were in the direction predicted by the theories underlying the diversion and community service impact models. Finally, the results of the exploratory correlational analyses were supported by the results of the predictive analyses. That is, while the variables used in this study did not permit the classification of participants into recidivist and nonrecidivist groups for official delinquency, these variables were able to account for a moderate amount of the variance in self-reported delinquency. CHAPTER 6 DISCUSSION Earlier in this report, several questions that were to be addressed by this study were presented. These questions were related to three general areas, including successful implementation for both diversion and the Community Service Program, the relative impact of the three alternatives on the intermediate outcomes of labeling and social bonding, and the relative impact of the three alternatives on recidivism. In this section, the conclusions that can be drawn from the results of the current study as they relate to the research questions will be discussed. In addition, recommendations for future diversion programming and research will be made. Implementation Implementation of Diversion Successful implementation is certainly one of the most critical issues in diversion programming and research. As noted earlier, documentation of successful implementation is a requirement for a valid test of diversion, and diversion programs have been widely criticized for their failure to appropriately implement diversion (e.g., Blomberg, 1977; Klein 6: Teilman, 1976). In this study, several steps were taken to improve diversion implementation and its assessment. In particular, a site was selected that had a relatively high rate of formal processing and a sufficiently large number of formally processed 170 171 youths to satisfy program needs. In addition, the participating police departments signed a written agreement that specified the types of youths to be referred to the project, and they provided research staff with the ability to monitor the departments' compliance with the agreement. Finally, two different methods were used to assess implementation. These were an assessment of the actual disposition of youths assigned to the traditional processing condition and an assessment of the change in formal-processing rates from the year prior to project initiation to the year after project initiation. The assessment of formal- processing rates was conducted separately for youths who met the diversion referral criteria in order to provide a more sensitive measure of diversion implementation. The assessment of diversion implementation provided consistent support for successful implementation. Averaged across the three departments included in this assessment, there was a 96.296 court-referral rate for youths assigned to the traditional processing condition, a 12.896 reduction in the formal-processing rate for all youth contacts, and a 17.1% reduction for cases which fit the referral criteria for the project. Critique. While the methods used to assess diversion implementation provided support for successful implementation, caution must be exercised in interpreting the results. Each of the methods used to assess implementation had certain weaknesses. The final disposition of youths assigned to the traditional processing condition was assessed in an attempt to determine the prOportion of youths in the diversion conditions that would have been formally processed. Use of this method was based on the assumption that, given the equivalence of the three groups, youths assigned to the diversion conditions would have received the same disposition as youths assigned to traditional processing. However, this method 172 may have provided misleading or incomplete information if the police departments referred youths to court who normally would have been warned and released because they knew that the court would simply deny the complaint with no further action. Thus, the police could have referred youths to the project who normally would have been released, in the hope that the youths would receive services, but with the confidence that they would not receive formal court processing. In essence, this method could not have provided evidence that youths referred to the project were the same types of youths who normally would have been referred to court. While a finding that a small proportion of youths in the traditional processing group were actually referred to court would have provided strong evidence of implementation failure, the results presented here do not provide conclusive evidence of implementation success. The assessment of formal-processing rates, as used in this study, also had a major weakness. That is, the observed decreases in the formal-processing rates after the initiation of the diversion project cannot be attributed to the project. In fact, other factors not associated with the project may have caused the decreases. For example, the decreases in the formal-processing rates may have been the result of increases in the recording of cases in which youths received an informal (i.e., warn and release) disposition. That is, the rate of formal processing could have appeared to be lower when, actually, the police may have been recording cases in which the youth was warned and released that they previously had not been recording. This change in record keeping practices would have artificially inflated the number of informally-handled cases (when compared to the pre-period), thus causing an apparent reduction in formal processing. Evidence that such a phenomenon had occurred might have been indicated by an increase in the number of juvenile contacts recorded from the pre-period to post- period. In the current study, however, each police department exhibited a 173 decrease in the total number of juvenile contacts from the pre-period to post- period (see Table 19). While one cannot rule out changes in record keeping of the type described above, no evidence was found to indicate that this happened. A second alternative explanation for the reduction in formal processing is that some factor, other than the project, caused the reduction. For example, if there had been a change in the juvenile code which decriminalized status offenses, this might have caused a reduction in formal processing. Unfortunately, the research design used to assess implementation does not allow one to rule out extraneous factors as the cause of observed decreases in formal processing. Additional evidence to address this threat to internal validity could have been developed through the inclusion of a matched control jurisdiction (see Campbell 6: Stanley, 1963). Diversion from what? While this study provided consistent support for successful implementation of diversion, the results of the process assessment raise some legitimate concerns about how well this study tested the diversion concept. As described earlier, the actual handling of youths assigned to the traditional processing condition was quite informal and, some would say, quite benign. Court processing in this study was apparently very different from the malevolent handling one envisions given the writings of the early advocates of juvenile justice reform (e.g., Empey, 1978; James, 1969; President's Commission on Law Enforcement and the Administration of Justice, 1967). For example, only 18 of the 78 youths (23.1%) assigned to traditional processing went beyond the court intake stage, and only one of these youths went so far as an adjudicatory hearing for the referral offense. In addition, a majority of the youths (58.7%) assigned to traditional processing became the recipient of counseling or other services from a court—sponsored program as a result of the intake disposition. Also, these services were very nonintensive. Youths assigned to one program 174 received only one visit from a court volunteer, and youths who reported any contact with court staff reported an average of 3.3 contacts, each lasting an average of 46 minutes. In general, traditional processing in this community appears to have been either nonintervention or nonintensive casework services, at least for the population of youths studied. Given the nature of the intervention accorded to youths in the traditional processing condition, it may not be surprising that these youths did not experience greater levels of delinquency labeling and, therefore, greater levels of recidivism. While the project was apparently successful in diverting youths from court intervention, court intervention may not have been intensive enough or formal enough to produce the negative effects predicted by labeling theory. Summary. While there were weaknesses in the methods used to assess diversion implementation, the available evidence indicates that the project was successful on this dimension. The prOportion of the group assigned to traditional processing that was referred to court was extremely high, there was a substantial decrease in the rate of formal processing by the participating police department, and no evidence was found to support the identified alternative explanations for the observed results. It should be noted that the results reported here indicate greater implementation success than was achieved by other diversion efforts that were evaluated using similar methods for assessing implementation. For example, a few studies found that there was no change or an increase in court referrals following initiation of a diversion project (e.g., Blew 6c Rosenblum, 1979; Sarri 6: Bradley, 1980), while the most Optimistic finding reported was that the formal- processing rate was reduced by 5.196 (Lipsey et al., (1981)). These results stand in contrast to the results reported for this study in which there was a 12.8% reduction in formal processing for all youth contacts and a 17.196 reduction in formal processing for youths meeting the referral criteria. However, despite 175 apparent successful implementation, it was observed that youths assigned to traditional processing were not subjected to intensive, formal court intervention. This finding must be kept in mind when interpreting the results related to the intermediate and ultimate outcomes. Implementation of the Community Service Prom An examination of the implementation of the service program was a critical component of this study. Like the assessment of diversion implementation, the assessment of program implementation was necessary to document that the study actually tested the community service impact model. That is, if one wants to determine if this program model is a viable alternative to court referral, one must first demonstrate that the program was operated as designed. Without such information, conclusions about the effectiveness of the program model cannot be reached. The assessment of program implementation was also important in order to document that youths received services of sufficient intensity that one might expect the services to be able to produce a positive impact. That this issue is a concern in diversion research can be demonstrated by the low intensity of services frequently provided to program participants. For example, in the evaluation conducted by Baron et al. (1973) clients usually received a maximum of five counseling sessions, and in the study reported by Davidson et al. (1981) clients had a median of 1.3 contacts with program staff and 76.2% had no more than two contacts with program staff. Even more enlightening was a report by Dunford (1977) that 3896 of the youths in one police diversion program using a service brokerage approach never received the intended services. In general, the results of the program implementation assessment indicate that the program was operated according to the prescribed guidelines, that most 176 youths received the intended level of services, and that these services were relatively intensive when compared to the norm for diversion programs. As you will recall, the Community Service Program consisted of two program components. The arbitration component was to provide a setting in which victim restitution could be arranged and conflicts between the youth and the victim/complainant could be resolved. The community service component was designed to place youths in voluntary positions in community agencies in order to increase their involvement in prosocial activities, facilitate an attachment to prosocial norms and people, help the youth build new skills, and increase positive labeling. With a few exceptions, the program did operate as designed. Youths were successfully placed in community organizations where they performed, on average, the expected number of hours of voluntary service. Further, the service activities performed were generally of a constructive rather than "make-work" nature, and the youths expressed strong satisfaction with the type of work performed. The program staff also performed the required amount of youth monitoring and placement monitoring, and they complied with the prohibitions against engaging in counseling and making referrals to outside agencies. Implementation problems, however, did occur in four areas of program operation. As described earlier, there were very few cases in which restitution was provided. However, this does not appear to have been the result of noncompliance by program staff or the result of the arbitrators' inability to negotiate restitution. Rather, it appears that restitution was not appropriate for the overwhelming majority of cases. In fact, there were only four cases in which a financial loss was suffered by the victim. In one of these cases the victim refused to accept restitution, and in another case the youth's parents insisted on paying for the restitution themselves rather than using the program's funds. The second implementation problem experienced by the program was the low rate of 177 attendance at the arbitration meeting by the victims/complainants. The victim/complainant attended the arbitration meeting in only 12 out of 63 cases. In many cases the victim/complainant was a large retail store that sent a representative to the arbitration meeting during the early months of program operation but discontinued their participation as time progressed. One possible explanation for this phenomenon is that the store representatives were satisfied with the handling of the cases by the program during the early months of operation and later felt no need to continue attending the arbitration meetings. Again, the lack of attendance by victims/complainants at the arbitration. meeting appears to have been largely a result of the type of offenders referred to the program (i.e., shoplifters from large retail stores). Had there been a greater rate of youths participating in the program who were involved in an ongoing dispute with the complainant or who had damaged the victim's property, then there may have been greater attendance by victims/complainants at the arbitration meeting. The Community Service Program also experienced some implementation problems in the recruitment of community volunteers. The intent of this aspect of the program was to increase the positive image of the program as a volunteer agency and to prevent the identification of specific youths in the program as police-referred (i.e., delinquent) youths. Accomplishment of these two goals was intended to prevent the negative labeling of program participants and to increase positive labeling. Unfortunately, only four community youths were recruited to participate in the program. Despite this failure (the original expectation was that approximately 5096 of the youths participating in the program were to be community-recruited youths), the program appears to have been rather successful in preventing the identification of specific youths as police referred and in fostering a positive image of the program. In the process of recruiting placements, program staff continued to explain that both police-referred and 178 community-recruited youths would be participating in the program, and when placing individual youths they were not identified as being police referred or community recruited. In a check on the successfullness of this aspect of the program, placement supervisors identified 37.296 of the police-referred youths as either "definitely" or "probably" referred by the police, and 46.596 of these youths as "definitely no_t" or "probably 3%" referred by the police (in 16.396 of the cases the supervisors reported that they could not make such a judgment). In a few cases it was known that the youth volunteered the fact that he/she was police referred; thus, perhaps, inflating the percentage of youths identified as police referred. Overall, despite the failure to recruit community youths, the program appears to have been relatively successful in achieving the intended goal of this aspect of the program design. Finally, successful implementation of the Community Service Program was hindered by the degree to which program participants performed their volunteer work along with other organizational members. It was intended that participation in work groups would increase the opportunity for the youths to interact with other placement staff, thereby increasing the probability of prosocial bonding. Prosocial bonding was identified by Hirschi (1969) as being negatively related to delinquency. However, interviews with program participants revealed that they did not frequently talk to placement staff about nonwork related topics and that they usually performed their volunteer work alone (44% of the youths reported that they worked alone either "all of the time" or "more than half of the time"). In addition, the results of the youth interviews were confirmed by interviews with the placement supervisors. While the results of the program implementation assessment indicate that the program was not successful on this dimension, the importance of this outcome is related to the program's ability to increase social 179 bonding and the relationship of social bonding to delinquency. This will be examined in the next section. Intermediate and Ultimate Outcomes This study attemtpted to test two impact models. According to the diversion impact model, diversion from the juvenile justice system decreases recidivism by decreasing delinquency labeling. According to the community service impact model, participation in community service decreases recidivism by both decreasing negative labeling and by increasing social bonding. In order to test these impact models, the three intervention alternatives (traditional processing, diversion without services, and diversion to the Community Service Program) were compared on the intermediate outcomes of labeling and social bonding and on the ultimate outcome of recidivism. As was presented in the results section of this report, there was a significant difference between the three alternatives on only one outcome measure, the number of juvenile justice contacts. These results do not add confirming evidence in support of the diversion and community service impact models. While the results of this study do not support the diversion and community service impact models, one possible explanation for such a finding is that the interventions were not successfully implemented. In the preceding section, mixed evidence was presented regarding the successful implementation of diversion and the Community Service Program. While this study provided consistent evidence that diversion was successfully implemented, one could legitimately question whether the traditional processing condition was sufficiently formal and intensive to produce delinquency labeling, especially at such a level that increased recidivism would result. While this is certainly an empirical question, the level of formal court intervention experienced by youths in the traditional processing condition did not allow a very 180 strong test of the diversion impact model. With only 23.1% of the youths assigned to traditional processing having been processed beyond the point of court intake, it does not appear that, as a group, youths assigned to this condition would have experienced much delinquency labeling. In fact, the results of this study, given the measures used, support this observation. Of course, one cannot determine whether the failure to find greater levels of delinquency labeling in the traditional processing group was due to the implementation problem or to the incorrectness of the theory that formal court intervention produces negative labeling. It might be argued that while just being referred to court should have been sufficient to (produce delinquency labeling, court assigned youths experienced no greater levels of labeling than youths diverted to the Community Service Program because of the nature of the intervention provided to program participants. In fact, the Community Service Program was relatively intensive and intrusive. Participation in this program certainly involved more contact for youths with authority figures and required more of their time than did referral to court. Further, participation in an arbitration meeting and an emphasis on restitution may be viewed as very similar to some aspects of traditional court processing. Given this, one might propose that the failure to find greater levels of labeling for the TP group was due to the fact that the label producing aspects of the Community Service Program were sufficient to "erase" the label prevention aspects of diversion. However, it must be kept in mind that the TP and DCSP groups did not experience significantly greater levels of labeling than the diversion without services group. This group, with very few exceptions, did not experience any formal intervention beyond the point of initial police contact for the referral offense. In fact, according to the youths' self-report, only 15 of these youths (18.396) received any type of human/social service intervention during the 16 weeks after project intake; and for those who did receive such services, they 181 were usually very nonintensive. If one were to argue that the Community Service Program intervention was very similar to some forms of court intervention (e.g., in terms of intensity, the emphasis on restitution, and the similarity between the arbitration meeting and a court hearing), the results of this study certainly call into question the hypothesis that court intervention produces delinquency labeling. Additional evidence for implementation failure as an explanation of the results can be found in the assessment of program implementation. Again, no significant differences were found between the three groups in the level of social bonding. Further, it was found that one of the major program elements that was designed to increase social bonding, high levels of youth-staf f interaction as a result of the youth performing the volunteer service in work groups, was not completely implemented. Whether increases in the use of work groups and increases in youth-staff interaction produce increases in social bonding, thereby decreasing recidivism, remains an unanswered question. The most important outcome assessed in this study was the impact of the alternatives on recidivism. As noted earlier, no significant differences were found on any of the measures of recidivism: simple, official recidivism; multiple, official recidivism; and self-reported delinquency. While certain implementation problems were experienced that may have resulted in the lack of a differential impact on recidivism, additional exploratory analyses failed to provide support for implementation failure as an explanation for these results. That is, analyses which included only those youths who received a relatively complete and "pure" form of the prescribed treatment still did not indicate any between-group differences on recidivism. DeSpite the above results, some support for the impact models was provided by the exploratory analyses. A consistent, although moderate, positive 182 relationship was found between labeling and recidivism, and a consistent, negative relationship was found between the youths' report of social bonding and one measure of recidivism. That is, all five measures of labeling were significantly correlated with the number of self-reported delinquent acts, and three measures of labeling were significantly correlated with the measure of official recidivism. For social bonding, all three measures from youth interviews were negatively correlated with the number of self—reported delinquent acts. Thus, even though the relationships between labeling and social bonding and the ultimate outcome of recidivism were not strong, the study did provide some support for one aspect of the impact models. These results indicate that one possible explanation for the failure of the diversion alternatives to reduce recidivism was their failure to decrease labeling and to increase social bonding. This explanation is consistent with the diversion and community service impact models. The above discussion suggests that the current study's failure to confirm the proposed impact models may have been due either to the incorrectness of the impact models or to implementation failure. Still a third possibility exists. That is, the measures employed in this study may not have provided a valid assessment of the specified constructs. For example, if the measures used to assess labeling and social bonding were not valid measures of these constructs, one would not expect to find a difference between groups on these measures. In essence, the constructs specified by the impact models would not have been assessed. Unfortunately, the design of this study allows only a limited assessment of the validity of the labeling and social bonding measures. However, given this limited ability, moderate support for construct validity is indicated. Specifically, support for construct validity of the labeling and social bonding measures is indicated by the significant correlation of these measures with recidivism in the direction predicted by labeling and social bonding theories. However, it is 183 important to note that these relationships were consistently found only when the labeling and social bonding measures were correlated with self-reported delinquency, not official delinquency. Given that these measures (labeling, social bonding, and self-reported delinquency) were administered at the same time, using the same method (i.e., personal interview), and under the same conditions, the observed relationship may be attributable to method variance. The validity of the measures could have been more adequately assessed if they had been administered at the pre-period assessment so that the predictive validity of the measures could have been determined. While self-reported delinquency, in contrast to official delinquency, was consistently related to social bonding and labeling, this may not be totally attributable to method variance. As noted earlier, official and self-reported delinquency may not be independent measures of the same construct; instead, they may represent two different constructs. For example, while self-reported delinquency may represent actual delinquent behavior, official delinquency may represent society's "reaction" (Klein, 1979b) to the youth's behavior as well as a response to his/her race, sex, reputation, and so forth. If official and self—reported delinquency do represent different constructs, one might not expect a similar pattern of correlations between these measures and measures of labeling and social bonding. Still another possible explanation for the weaker relationship of official delinquency to labeling and social bonding is that official delinquency is largely a random event. That is, the act of being caught for a delinquent action is frequently a result of being in the wrong place at the wrong time. Therefore, given the supposed random nature of official delinquency, one would not expect it to be consistently and strongly related to other constructs. This explanation may be particularly true in the early stages of delinquency (as represented by 184 participants in this project) before the delinquent reputations are established that allow societal reaction to play a more important role. Conclusions and Recommendations This report has presented the results of an experimental test of two delinquency intervention impact models; the diversion and community service impact models. On the basis of this study, conclusions can be drawn about the adequacy of these impact models and recommendations can be made for future programming and research. The Diversion Impact Model Diversion of juvenile offenders from formal juvenile justice processing has been proposed as one intervention strategy for reducing delinquency. According to supporters of this strategy, diversion should lead to a reduction in recidivism by reducing the negative labeling that has been hypothesized to result from involvement in the juvenile justice system (e.g., President's Commission, 1967). In order to confirm the validity of this impact model it must be demonstrated that diversion had been implemented, that diverted youths experienced less labeling than youths processed through the formal system, and that diverted youths recidivated at a lesser rate than youths processed through the formal system. The successful implementation of diversion and its assessment were major emphases of this study. As discussed previously, many previous studies of diversion efforts found little evidence that diversion was successfully implemented (e.g., Lincoln, 1976; Sarri 6: Bradley, 1980), or the methods used to assess implementation had serious weaknesses (e.g., Gilbert, 1977), or implementation was not assessed at all (e.g., Kelly et al., 1976). While'the negative results of past research may lead one to conclude that diversion programs widen the net of juvenile justice, the real issue is: Can diversion 185 programs be designed and operated such that diversion is successfully implemented? This study, despite the methodological weaknesses already mentioned, provided consistent evidence that diversion can be implemented. Even though the exact ingredients leading to successful implementation cannot be identified, it appears that some combination of the methods used in this project were effective. These methods included the use of a written referral agreement indicating the types of youths that were to be diverted, the ability to monitor compliance with the agreement, and the selection of project sites on the basis of high rates of formal processing. This last criterion may be particularly important. Even within the three police departments included in the assessment of diversion implementation, the department with the lowest pre-project rate of formal processing (CPD 1) exhibited the smallest decrease in formal processing (see Table 20). Given the results of this study, the methods used here should be used in future diversion efforts, and other methods should be explored to improve diversion implementation. The results of this study also indicate that while it is important to demonstrate that youths were actually diverted from the juvenile justice system by the project, it is also important to examine the nature of juvenile justice system processing. In this study it was found that youths referred to juvenile court did not experience intensive, formal court intervention. In fact, traditional processing for the participants appeared to be either nonintervention or nonintensive casework services. Given this finding, this study may not have been a very strong test of the diversion impact model. While this study did evaluate diversion from traditional processing as it was practiced in a particular community for a particular type of youth, the results may not be generalizable to the more punitive courts described by diversion advocates (e.g., President's Commission, 1967). In the future, evaluations of diversion programs should 186 include a traditional processing condition in which youths receive more formal and intensive intervention by the juvenile justice system. This could be accomplished by one of three methods: (1) a site could be selected in which the juvenile justice system normally provides a more formal and intensive type of intervention; (2) the "traditional processing" condition could be designed and standardized by the researchers such that all youths assigned to this condition would receive formal, intensive court intervention; and (3) only more serious offenders could be included in the study since these youths would be more likely to receive formal, intensive court intervention. While it is important to develop new methods to improve diversion implementation, it is also important to conduct adequate assessments of implementation. In this study, two methods were used. These were an assessment of the actual disposition of youths assigned to traditional processing and an assessment of the change in formal-processing rates after the initiation of the project. Future research on diversion should continue to use these and other methods to assess implementation. Consistent results from the use of multiple methods will provide stronger evidence than the use of any one method with its idiosyncratic weaknesses. The assessment of system processing rates, as used in this study, could be improved in future research by the inclusion of matched-comparison jurisdictions (as was accomplished by Davidson et al. (1981) and by Lipsey et al. (1981)) and by using multi-year, pre- and post-period baselines.’ The latter recommendation would help the researcher to determine if changes in the formal-processing rates were part of normal, seasonal trends. For example, in this study there were sharp decreases in the formal-processing rates around March 15, 1982. While these decreases may have been due to the startup of the diversion project, the decreases may also have been a normal occurrence for the police departments at 187 that time of year. This possibility could not be adequately examined using only a 1-year baseline. Labeling and recidivism. Despite the successful implementation of diver- sion, the assessment of labeling and recidivism outcomes did not provide supportive evidence for the diversion impact model. There were no between- condition differences on any of the recidivism measures and only one between- condition difference on the five labeling measures. The one significant difference on labeling was for the number of juvenile justice contacts. However, this difference was an expected result of the random assignment, since youths assigned to traditional processing would naturally have more justice-system contacts. In addition, it was found that labeling was consistently related to recidivism, especially self-reported delinquency, although the relationship was not strong. Two interpretations of these results, as they relate to the diversion impact model, are possible. First, it may be that, contrary to the diversion impact model, traditional processing (court intervention) does not result in increased delinquency labeling. This interpretation seems reasonable given the lack of differences among the three groups on four of the five labeling measures and the consistency of these findings with past research (Davidson et al., 1977; Elliot et al., 1978; Klein, 1979b). However, an alternative interpretation of the results is that the traditional processing condition was not sufficiently formal or intensive to produce increased delinquency labeling. While this is an empirical question, the nature of the intervention received by youths in the traditional processing condition certainly makes this a reasonable alternative. Given the possible validity of this interpretation, the recommendation made earlier to conduct future research in which the traditional processing condition consists of a more formal and more intensive type of intervention, is repeated here. 188 The Community Service Impact Model In the impact model constructed for the Community Service Program, it was prOposed that, given successful program implementation, recidivism would be reduced as a result of decreased labeling and increased social bonding. Again, this study provides moderate support for successful implementation but fails to provide support for an impact on the intermediate and ultimate outcomes. Implementation. Overall, the Community Service Program was successfully implemented. As described earlier, participating youths generally completed the program with the expected number of hours of community service, and they expressed substantial satisfaction with the program. However, four areas were identified in which there was less than optimal implementation. First, the program did not provide a substantial amount of restitution to the victims of the program's participants. However, this result does not appear to have been due to a failure on the part of the program or its staff; instead, it appears to have been due to the type of offender referred to the program. That is, there were very few cases in which a victim suffered a permanent loss of property through theft or vandalism. There were very few cases of vandalism, and in most cases of theft, the property was returned. In general, the arbitration/restitution component of the program was inapprOpriate for the vast majority of program participants. In the future, if implementation of this program is considered, it should be preceded by a careful examination of the types of offenders that would participate. If it is determined that the arbitration/restitution component of the program would be inappropriate for the great majority of potential clients, serious consideration should be given to not including this component of the program. The second implementation problem experienced by the program was the low rate of attendance at the arbitration meeting by victims/complainants. There was a victim in 81.896 of the cases, and he/she attended the arbitration meeting in 189 only 19.0% of the cases. Again, much of this problem may have been due to the type of cases handled by the project. When there was no conflict to be arbitrated and no restitution to be paid, there may have been little incentive for the victim to attend the meeting. Also, when the victim was a store owner/representative who was repeatedly involved in the cases, attending the arbitration meeting on one occasion may have been sufficient to satisfy the victim's needs; thus, he/she may not have wanted to attend again. In order to address this possible problem, future implementation of the program should be preceded by an assessment of the types of offenders to be referred in order to determine the appropriateness of the arbitration component of the program. If it appears that a sufficiently large proportion of the cases will include victims who may have an interest in attending the meeting (as determined by the number of victims who would be eligible for restitution and the number of cases that would involve interpersonal conflicts), then methods shduld be explored to increase victim attendance. The third implementation problem experienced by the program was the failure to obtain community-recruited participants. While one-half of the participating youths were to have been recruited from the community, only four such youths participated in the program. The program failed to recruit community youths despite repeated attempts, including presentations to school classes and assemblies, articles in school newspapers, and presentations to various youth organizations. One possible explanation for the program's lack of success in this area is that the local community already had an organization that recruited and placed youths in voluntary service positions. This organization had been active for several years and had been successfully recruiting large numbers of youths. The Community Service Program, therefore, was competing with an established organization that had ongoing agreements with the schools and other agencies to assist in the recruitment of volunteers. Recruitment attempts by the 190 Community Service Program were sometimes resisted by organizations because of their prior ties with the established youth volunteer agency. Also, staff reported that many youths who were likely to do volunteer work had volunteered through the established agency in the past and wanted to continue with this agency. Finally, it should be noted that a substantial amount of anecdotal evidence was gathered to indicate that the established volunteer agency viewed the Community Service Program as a threat to their success in the community. As a result, this agency refused to participate in any joint recruitment efforts or to provide any other types of assistance (e.g., by referring some of their youths to the Community Service Program). In any future attempts to implement the Community Service Program, the community should be examined to determine if the program would be in competition with an existing agency. If so, negotiations should be conducted prior to program initiation in order to work out a cooperative relationship with the established agency. In addition, other methods for recruiting community volunteers should be developed and tested. The last area in which implementation problems were experienced was in the lack of placement staff-youth interaction. Youths reported infrequent interaction with staff about nonwork related matters, and both youths and their supervisors reported that the youths tended to work alone most of the time. In any future replications of the Community Service Program, attempts should be made to increase youth-staff interaction. Perhaps this could be accomplished by conducting training/orientation sessions with the placement supervisors prior to program initiation in order to encourage them to interact with the youths and to help them to develop the interpersonal skills required to do this. Obviously, having placement supervisors attend training sessions and interact more with the youths would require a bigger investment of their time. In this project, there were no means available to encourage and reward such an investment. Therefore, 191 in any future replications of this program, methods should be developed to reward placement supervisors and their organizations for participating in the program. Intermediate and ultimate outcomes. Again, this study provided little support for the community service impact model. While moderate support was found for a negative relationship between social bonding and recidivism (i.e., the three measures of social bonding from youth interviews were negatively correlated with self-reported delinquency), there were no between-group differences for social bonding, labeling, or recidivism. In this case, there is little, if any, evidence that the results can be explained by implementation failure. While there were implementation problems, there is no indication that these problems were responsible for the lack of program effectiveness. That is, while the program was unsuccessful in recruiting substantial numbers of community youths, interviews with placement supervisors revealed that the majority of the program participants were not identified as delinquent youths. Yet despite this, program participants reported as much delinquency labeling as youths in the other two conditions. There is also reason to believe that the low levels of youth-staff interaction may not have influenced program outcomes. While the intent of this aspect of the program was to facilitate the develOpment of strong social bonds on the part of the youths, the results of the correlational analyses indicate that the extent to which the youths worked with others (Work Group Scale) and the degree to which the youths interacted with placement members (the Attachment to Placement Members Scale) were not related to any measure of social bonding used in this study (see Table 33). Summary and Conclusions Two delinquency intervention impact models were tested in this study. Both of these models, diversion and community service, represent alternatives to 192 formal justice system processing for juvenile offenders. Consistent evidence was found to indicate that diversion had been successfully implemented. At the same time, no evidence was found that the two diversion alternatives were effective in reducing either delinquency labeling or recidivism. However, these results must be interpreted in light of the relatively mild handling received by youths assigned to traditional processing. While one may conclude that there were no differences in effectiveness for the intervention strategies tested, these results may be specific to the pOpulation studied and the type of court intervention (as well as the type of diversion program) included in this study. Given the type of intervention provided to youths in the traditional processing condition, this study may not have been an adequate test of diversion. Despite the problem caused by the relatively mild intervention associated with the traditional processing condition, the contributions of this study must be considered in terms of previous diversion research. Perhaps the greatest contribution of this study was in the assessment of diversion implementation. Despite a substantial body of prior research which indicates that diversion has often widened the net of juvenile justice rather than prevented formal juvenile justice intervention (Klein, 197 9a), this study found consistent evidence that diversion was successfully implemented. In fact, both methods used to assess diversion implementation indicated greater levels of successful diversion than was found in previous studies that used similar assessment methods. While this study found that 96.2% of the youths assigned to traditional processing were actually referred to court, studies by Baron et al. (1973), Lincoln (1976), and Palmer et al. (1978) found that the percentage of youths in the traditional processing condition who actually received a formal diSposition was 21.3%, 6096, and 7196, respectively. Also, while this study found a decrease in the formal processing rate of 12.896 for all youth contacts and 17.1% for eligible youth contacts, the 193 most positive finding reported in previous research reports was that initiation of a diversion project was followed by a 5.196 decrease in formal processing (Lipsey et al., 1981). Thus, this study has provided evidence that, while it may be very difficult, it is possible to successfully implement diversion. This study also made a unique contribution to diversion research through the assessment of program implementation. As discussed earlier, reports of diversion research have rarely provided a detailed description of the operation of the service program. Such descriptions are necessary if one is to determine the implications of evaluation outcomes. That is, descriptions of program Operation allow one to more specifically define what "worked" (or did not work), to determine if the program was successfully implemented, and to judge whether or not the program was intensive enough that one could reasonably expect it to have had an impact. In this study, both the proposed and actual operation of the program were described in detail. Further, documentation was provided that the program did operate as designed in terms of the type and amount of services received by program participants. This finding stands in contrast to the findings of other research in which large numbers of youths were reported to have never received the intended services (e.g., Dunford, 1977). Finally, this study provided documentation that the services received by program participants were relatively intensive when compared to other diversion programs. In terms of program intensity, the Community Service Program should have been more capable of having an impact on delinquent behavior than those programs described in other reports of diversion research (e.g., Baron et al., 197 3; Davidson et al., 1981). In terms of intermediate and ultimate outcomes, this study provided results that are generally consistent with those of prior diversion research. For example, like four previous studies which examined the impact of diversion on labeling (i.e., Davidson et al., 1977; Dunford et al., 1982; Elliot et al., 1978; Klein, 1979b), this 194 study found no evidence that diversion reduced delinquency labeling, regardless of the manner in which labeling was defined. The results of this study are also quite consistent with prior research in terms of the impact of diversion on subsequent delinquency. Like all four previous studies which reported the impact of diversion on self-reported delinquency (i.e., Davidson et al., 1977; Davidson et al., 1981; Dunford et al., 1982) no between-group differences were found on this measure. In terms of official delinquency, the results of this study are also relatively consistent with the findings of previous research. For those three studies which compared diversion to a service program, diversion without services, and traditional processing, two of the studies (representing seven out of eight project sites) found no between-group differences on any measure of official delinquency (Dunford et al., 1982; Elliot et al., 1978). The third study (Klein, 1979b) did find significant between-group differences, with the diversion without services group exhibiting the least amount of recidivism and the traditional processing group exhibiting the most amount of recidivism. Finally, the results of the attempts to predict recidivism can be compared to the findings of similar attempts in previous research. While in this study attempts to classify youths as recidivists-nonrecidivists failed, others have reported success in predicting recidivism. For example, Quay and Love (1975) reported that by using a combination of demographic, prior offense, academic performance, behavioral, and personality measures they could accurately classify 7196 of the nonrecidivists and 6596 of the recidivists who participated in their study. Similarly, using a multiple regression approach, Elliot et al. (1978) found that sex, ethnicity, self-reported delinquency, and "offender type" were predictive of official delinquency. However, the results reported for the current study are similar to those reported by Palmer et al. (1978). In this study, Palmer and his 195 associates used a stepwise multiple regression analysis to predict recidivism. Using a combination of demographic and prior delinquency variables, Palmer et al. found that they could only account for 6.496 of the variance in recidivism. Recommendations for future research. , Given the findings of this study, future studies should be conducted in which diversion is compared to court processing that is more formal and more intensive than was provided in the current study. The methods used in this study to improve diversion implementation should be replicated and other methods should be explored as well. In addition, given the lack of support for the community service model, future attempts at designing diversion efforts should continue to investigate other types of programs. The programs should have prior evidence of effectiveness, when possible, and a strong theoretical rationale. Finally, given the finding that diversion without services was as effective as traditional processing and diversion to the service program, future studies should continue to examine the effectiveness of a "no-treatment" alternative. Given cost-effectiveness as a criterion for judging the relative value of the various alternatives for responding to juvenile offenders, diversion without services warrants serious consideration. FOOTNOTES l. FOOTNOTES The apprOpriate definition of diversion has often been debated. For the purposes of this paper, diversion will be defined as the act of stopping the normal, formal handling of an accused juvenile offender once he/she has entered the juvenile justice system (JJS) (i.e., after police contact or after referral to juvenile court) but before adjudication. Diversion may include referral to a human service agency, but for the purposes of this report, it will include outright release with a dismissal of the complaint (i.e., diversion without services). In any case, the youth does not proceed to the next stage of formal JJS processing. For a discussion of the definitional debate surrounding diversion the reader is referred to Klein, 197 8, pp. 10-13. It is interesting to note that one of the "unsuccessful" programs included in this study was also evaluated by another group of investigators (Binder, Monahan, 6: Newkirk, 1976; Binder 8: Newkirk, 1977) with different results. Binder et al. reported an evaluation of the Youth Service Program in Orange County, California. This was one of the random assignment sites identified by Palmer et al. However, in the Binder et al. study DTP youths were compared to youths who were released without further action or service provision. Using this comparison, Binder et al. found that the DTP group did significantly better than the DWS group in terms of the average number of rearrests 6 months after project referral (.14 vs. .37). A site-by-site description of these programs is presented in Appendix A. The threats to internal validity using this type of design have been described by Campbell and Stanley (1963). Davidson, et al. (1981) reported a decrease in police referrals to court, but not in arrests, or petitions received by the court. In the Davidson et al. study this method was used in four sites with the finding that nonjustice referrals ranged from 27% to 88%. In the Palmer et al. study nonjustice referrals averaged 28% across nine sites. It should be noted that in the Davidson, et al. study there was no comparison of diverted youths to youths who were formally processed. This study does not, therefore, represent an adequate test of the labeling reduction goal of diversion. In one of the four sites there was a significant difference in the youths' perception that teachers thought of them as "sick." This would be. true unless the diversion program were so effective in 196 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 197 reducing recidivism that the additional costs of the program were compensated for by cost savings resulting from lower arrest rates, reduced probation services, and so forth. Quite obviously, this is not an adequate test of the relative effectiveness of different types of diversion programs. However, the intent is only to determine if there are any trends which might indicate the most fruitful directions for future research. It should also be noted that the type of analysis presented here is part of a general strategy known as "meta- analysis" that has been popularized by recent publications such as, Hunter, Schmidt, and Jackson (1982) and Glass, McGaugh, and Smith (1981). Many diversion efforts operated at both the police and court levels. A program was categorized as a police- or court-level diversion program when 70% or more of its referrals came from one source. Seven programs could not be categorized using this guideline, and they were excluded from this analysis. Those studies which reported mixed results when using two measures of official recidivism were not included in the calculations. One of these studies operated at the police level (Stratton, 1975) and one operated at the court level (Blew 6c Rosenblum, 1977). Perhaps a more useful approach would have been to use the classification scheme developed by Rutherford and McDermott (1976) in which programs were characterized as "legal," "paralegal," and "nonlegal." Unfortunately, evaluation reports rarely included the information that is needed to use this classification scheme. The measures used to address each question are in parentheses. The procedures described in this report were reviewed and approved by the Human Subjects Committee at Michigan State University. Youths referred by CPD 1 were notified by that department that they had been referred to the research project. The arbitrator was the Community Relations Officer for the sponsoring city. This person was trained in the arbitration procedures by the principal investigator. A description of the types of placements used in the project will be included in the Results section of this report. In a few cases when the youth had moved, the post-period interview was conducted over the telephone (n = 2). The interscale correlations for this and all other measures to be presented later were corrected for attenuation. The Placement Supervisor Interview was implemented after a large number of youths had already completed the program. Therefore, these ratings could not be obtained for some youths. 22. 23. 24. 25. 26. 27. 28. 29. 198 One police department (TPD) was not included in this assessment due to the extremely small number of youths processed by that department and the difficulty in accessing their records. As described earlier, the eligible youths were those who were: 12 to 16 years old; first or second offenders; residents of the county; not living in an institution; not charged with arson, homicide, criminal sexual conduct, armed robbery, felonious assault, breaking and entering an occupied building, delivery of a controlled substance, or an offense with property loss/damage in excess of $500; and were not currently involved in juvenile court. With respect to the last criterion, it should be noted that the police did not normally know the court status of a given youth. Therefore, no youths were considered ineligible on that basis for the purposes of this aspect of the study. For CPD 1, information was not available for most of the variables used to determine eligibility. In fact, the only criterion that could be used was the type of offense. This is eSpecially likely to happen when the total number of referrals for a given department is very small. Such was the case for CSD (n = 48). In Figure 4, one might observe that there was a sharp decrease in formal processing beginning in October and continuing through January. By way of explanation, this increase corresponded to the time when the referring police officer at CPD 1 was absent and the responsibility for the disposition of juvenile cases was turned over to another officer who was unfamiliar with the project. Upon the original officer's return, the formal-processing rate again declined. While detailed information concerning the nature and intensity of services provided to these youths was not available, conversations with court staff indicated that these services frequently involved either one contact with a court volunteer or rather nonintensive casework services provided by court staff. It should be noted that the small number of youths who were observed to have received formal court dispositions at the time of the post-period data collection was not due to a failure to allow sufficient time for a final court disposition to have occurred. In fact, a final court disposition was available for all but two youths when the post-period court data was collected. While there appears to be a discrepancy between the number of youths referred to court and the number reporting court contact, it should be noted that a complaint could have been denied at court intake without there ever having been any contact between the youth and the court. (e.g., when a case is denied for lack of evidence). In such cases, while the official records would have indicated the youth was referred to court, the youth would have accurately reported that he/she did not have any court contact. Arbitration meetings were not conducted in two cases. In one case the youth withdrew from the program before the meeting was conducted and in the second case the youth was institutionalized before the meeting could be conducted. 30. 31. 32. 33. 34. 35. 36. 37. 38. 199 All data on the number of hours and sessions of community service work performed was recorded on the Community Service Record by supervisors at the community service placements. The two youths withdrawn by their parents were a result of some dissatisfaction on the part of the parents, while the "other" reasons for the withdrawal for three youths included one case in which the youth was dissatisfied with the program, one case in which the youth was institutionalized, and one case in which the placement agency asked the youth to discontinue his volunteer work at their agency. The mean number of offenses for the three groups were as follows: TP=1.37; DCSP=1.15; and DWS=1.15. The mean number of court complaints was as follows: TP = .35; DCSP = .10; and DWS = .10. The group means for each of the measures of delinquency labeling were as follows: Label Spread—TP = 14.26, DCSP = 13.60, and DWS = 14.96; Label Applications—TP = 2.82, DCSP = 1.97, and DWS = 1.80; Perceived Negative Labeling—TP = 2.80, DCSP = 2.60, and DWS = 2.67; Label Acceptance- TP = 2.83, DCSP = 2.48, and DWS = 2.56. The group means for each of the social bonding measures were as follows: Commitment to School—TP = 48.96, DCSP = 50.93, and DWS = 50.04 (standardized scores); Attachment to Parents—TP = 50.60, DCSP =49.15, and DWS = 49.72 (standardized scores); and, Prosocial Beliefs-TP = 4.18, DCSP = 4.19, and DWS = 4.29. The percentage of participants recidivating in the TP, DCSP, and DWS groups were 14.1%, 17.7%, and 9.3%, respectively. The mean number of offenses for the TP, DCSP, and DWS groups were .14, .24, and .12, respectively. The pre-measures and post-measures of self-reported delinquency were correlated .70 (p < .05). Obviously, the results of this analysis strategy must be interpreted with great caution since the random assignment of youths to condition has been undermined, and the groups can no longer be considered equivalent. APPENDICES 200 on: on: :2 :2 9:1 H88 2:8 H28... :8: .3 8 88.8: 9.: 3:8 :2 + an: :2 I. :z 8:68 8:6: H82: 38: 8:28 :2 new: :8: :z y- mz 88mm, 2:883 H88. :8: 83 8 8mm 3:38:30 00:0: :z .5. + :z 8% H38 3.8. .88 .82.. N 38 :2 :2 on: :2 8w 8:84 $88 88% :88. H 38 88: .He 8 85:: 525E506 cam: i- m... + 8:... :2 88mm 8:8 M8883 38...: 52: 5.55.58 3.28 8: an: 8% an: 88 .888» .93... H38 88.9.2: q 28 3.55.506 Sea. >058? 0223 . 9338 am: on: am: an: «we 88:9 2:8 8288.... n 88 3:55:80 0223 am: on: am: am: I 8»! 88% 8:6m .8893... a 35 3.23 am: am: am: :mz 8» 3.6 8:8 .82.. H 88 38: .3 8 288: 30:8. 8 02:0on: i lieutullviuunuuii $03.5» I on: + + :mz 8» 88% 8:6: H88. :8: .He 8 8833 3:55:30 an: 8.85: 89%». III 2:68.. ............ .18.: $1- 8:83 81.8.5 89.88 83: Ho .62 .28: 88.8: 8 Ho»... ugh i anon—825 32!.— acufion .35.—.29me 02,—. sz_Em zoaflm—ZD m0 %m<223m < “Sn—2934‘ 201 :2 :2 9:2 :2 8: 888 8:18 8:232: a 8:: 3.3:: :2 ..... .1 :2 on: :2 8:1 8:8: 83a 8:883 a 8:: 3.20: :2 :2 an: :2 8a 8:a 8:8 18:28an g 8:: 28$ .8 8883 00.20: :2--I--Iaw2 H I.-- :2 :2 8:8 E8 88... :33 .8 8:98: :2 - :2 -I+Il!-.mm.- 8: 8:8 “89.88 .82: 33.856 30on 83.. - mm: 82 an: :2 88.8% 8:8 8.883 a 8.: 4.23 2:52:58 0323 ..IWom.-l.. 2.-...- I... 28 an: am: am: :2 8.28:... 8:8 «$88.... a 8:: $23 2:52:53 928:: 8.. I... - 83.. 5.2 on: 32 :2 8.88m. 8:8 8:932: _ 3.: 381a .8 8 85: :23: 3.55.58 3.20: 8m. - $3-2..- - 3-.. - -..:--I.. :2 + on: :2 88 88:81 :58 88... :8: 53:88: a 22: 8:92. Imm-....:-3-.:.m-:- I . :2 88 .88 8:8 -- 88... as: «8 8 88: Immwl::m.8: