یک تحلیل از اثر مداخلات بزهکاری نوجوانان بر نتایج تحصیلی

Abstract This meta-analysis examined the effects of juvenile delinquency interventions on academic outcomes. After retrieving over 250 reports, 15 reports met inclusion criteria and provided 134 effect sizes (92 unadjusted and 42 adjusted) based on 20 separate samples in a variety of settings, including school, community, and juvenile justice settings. Heterogeneity of the samples, generally weak research designs, and the absence of control conditions in many recovered reports was a limitation in the existing research. Overall, there were limited positive effects of juvenile delinquency interventions on academic outcomes. The lack of theory-driven or empirically supported academic interventions was notable. Studies with the weakest designs produced the largest effects on academic achievement, and school attendance outcomes were enhanced only for older adolescents. The implications of findings for future research and policy are discussed.

Introduction The concept of the school to prison pipeline is one of the most pressing concerns related to juvenile delinquency and education. This term refers to the phenomenon in which students gradually become disengaged from school while simultaneously becoming more involved in crime and delinquency ( Christle, Jolivette, & Nelson, 2005). The overall cost to society is a concern. That is, youths (5–17 years) account for 18.9% of the total United States population ( Howden & Meyer, 2011), but are responsible for approximately 16% of all violent crime, and 26% of all property crimes ( Puzzanchera, 2009). According to the Office of Juvenile Justice and Delinquency Prevention (OJJDP), nearly 2 million youths are arrested each year and the overall incidence and costs that result from acts of delinquency are considerable ( Puzzanchera, 2009). Education seems to be a central component of the problem and solution ( Council of State Governments and Public Policy Research Initiative, 2011). Low educational attainment is linked to higher rates of delinquency and recidivism ( Cottle et al., 2001 and Leone et al., 2003), while increased literacy and educational achievement is associated with lower rates of crime ( Cottle et al., 2001 and Keith and McCray, 2002). The educational challenges of populations engaged in juvenile delinquency have been clearly noted in research literature since the 1950s (Wilson, Lipsey, & Soydan, 2003). One often-cited characteristic of juvenile offenders is the high proportion of them who are involved in special education. Special education participation of youths in juvenile justice systems is considerably higher than that of the general population; 35% of juvenile offenders participate in special education (Quinn et al., 2005 and Zabel and Nigro, 1999) versus 8% national rate of the general population of the United States (National Center for Education Statistics, 2010). However, this characteristic alone is not a useful descriptor. The problem is not that youths are enrolled in special education participation per se; it is low academic skill. The average age of adjudicated youth is 15 years, an age at which most youth are entering the 10th grade, but the average reading level of adjudicated youth is aligned with the 4th grade, where children are typically age 9–10 years (Leone et al., 2005 and Vacca, 2008). In light of the close associations among reading level, educational attainment, and crime and recidivism rates, academic skill is a highly relevant consideration for juvenile justice rehabilitation efforts. In spite of the importance of educational attainment, much of the current literature focuses exclusively on behavioral and crime/recidivism outcomes. It is not surprising that there are numerous reports on interventions to reduce crime and improve behavioral outcomes for youth, many of which have examined educational variables and outcomes in some capacity. Yet, despite the clear links between delinquency and education, the literature about juvenile delinquency interventions and educational outcomes is under-developed. Specifically, reports of educational interventions and studies of potential factors that may influence interventions are more or less effective for promoting positive educational outcomes for youth offenders are not clearly summarized in the current literature. A synthesis of the educational effects of juvenile delinquency interventions would provide a meaningful contribution to inform future interventions and research. To address the gaps in the literature, the primary purpose of this investigation is to synthesize existing research about the effects of juvenile delinquency interventions on academic outcomes. We start by briefly summarizing the research about delinquency interventions not specific to education, summarizing some general patterns that guided our quantitative analyses, and then we describe the available literature about educational effects of delinquency programs. 1.1. Prior meta-analyses on the effects of juvenile delinquency interventions Several meta-analyses have summarized effects of juvenile delinquency interventions on behavioral outcomes, primarily recidivism and crime. Although educational outcomes have not generally been the focus of these prior meta-analyses (though some have examined academic outcomes to a limited extent or used education variables as moderators), it is useful to have an overview of the characteristics of successful juvenile delinquency interventions that focus on behavioral outcomes. This framework serves as a background to understand what factors might be important to consider in understanding the effects of juvenile delinquency interventions on academic outcomes. One of the most comprehensive datasets of juvenile delinquency reports has yielded multiple meta-analytic investigations (Lipsey, 1999, Lipsey, 2009, Wilson and Lipsey, 2001 and Wilson and Lipsey, 2007). In the most recent update using that database, Lipsey (2009) included 500 published and unpublished original studies conducted between 1950 and 1996. These earlier syntheses of delinquency interventions contribute several helpful points to understand to what extent, when, and how interventions reduce delinquency. There are four consistent findings related to juvenile delinquency interventions and behavioral outcomes. First, even among delinquency interventions that offer moderate success, recidivism is only reduced by about 12% overall (Lipsey, 1999), with generally small effect sizes (d = .10 to .15) ( Lipsey & Cullen, 2007). Second, the most effective delinquency interventions are those based on theory and a clear approach for change, often with a solid research base. Third, interventions conducted as standard field practice are somewhat helpful, but less so than more intensive implementations of interventions that are closely supervised by researchers. Fourth, positive, as opposed to punitive, behavioral interventions to address aggression or disruptive behaviors typical in juvenile offender populations appear to be helpful, even when delivered as routine programs apart from a rigorous research-driven approach in schools ( Lipsey and Cullen, 2007 and Wilson, Lipsey and Derzon, 2003). These four overall findings are clear in terms of behavioral outcomes, and they provide useful guides for additional synthesis for the same population. In terms of the effects of juvenile delinquency interventions on academic outcomes, it is prudent to consider whether intervention characteristics found to be associated with positive behavioral outcomes may also be associated with positive academic outcomes. That is, would academic outcomes mirror the effect sizes of the behavioral outcomes? Further, we use these prior meta-analyses as a guide in the current meta-analysis because examining similar questions will allow findings from this meta-analysis to be integrated and compared with those of prior syntheses of juvenile delinquency intervention effects. 1.2. Educational programs that address behavior During the past two decades, there have been a number of studies of school-based interventions geared at improving positive behaviors while also reducing “acting out” and delinquency-type behaviors among the general student population. Many of these studies are showing positive effects on both academic and behavioral outcomes. The effects of those programs have been summarized in two separate, comprehensive synthesis investigations (Durlak et al., 2011 and Wilson et al., 2001), indicating that behavior-focused programs can be successful at addressing both behavioral and academic concerns. One group of programs, the Social and Emotional Learning (SEL) programs, was the target of a recent meta-analysis that included academic and behavioral outcomes in school settings. SEL programs produced moderate positive effects on conduct problems and academic performance (d = .22 and .27), even when academic intervention was not the primary program component ( Durlak et al., 2011). Another approach, Response to Intervention (RTI), a data-driven approach in education to address academic concerns ( Edmonds et al., 2009 and Vaughn et al., 2010) has emerging support for behavioral concerns as well ( Gresham, 2007). Positive Behavior Support programs (PBS; Sugai & Horner 1999) have also been designed to promote academic and behavioral success, often for the same youths that are served by juvenile justice centers. Overall, research on these school-based universal prevention and targeted early intervention programs indicates that they are helpful in promoting positive academic outcomes, even when the focus is on reducing disruptive and aggressive behavior (Wilson et al., 2001). The body of education research generally shows that school based prevention and early intervention programs, even when the focus is behavioral and not academic, help improve behavior and academic achievement. In light of these findings, it would seem important to clarify the effects of interventions with populations engaged in delinquency on academic outcomes because there may be academic benefits even of juvenile delinquency programs that have no academic component. 1.3. Academic intervention research in juvenile corrections While it would seem that even behavioral and conduct focused interventions can have an impact on academic outcomes, it also seems like common sense to suspect that those juvenile delinquency programs that include an academic component might be most successful at improving academic outcomes. There have been a handful of studies examining the effects of using academic interventions with a reading focus within juvenile correctional detention facilities (e.g. Campbell et al., 1993, Drakeford, 2002, Leone et al., 2005 and Malmgren and Leone, 2000). In fact, a research synthesis on the topic was conducted for studies carried out in correctional facility settings (Leone et al., 2005). The scholars could not draw firm conclusions about overall effects of such programs because there were only six studies available which met their inclusion criteria (Krezmien and Mulcahy, 2008 and Leone et al., 2005). The authors noted the difficulty of carrying out reading research in detention and corrections settings as one of their main findings. However, community services and public schools also address that academic needs of juvenile offenders and programs in these settings are relevant to include in a synthesis for this population. Youths involved in juvenile delinquency typically spend less than 30 days in detention facilities, and only about 14% of all youths who are involved in juvenile justice at any given time are in detention. So the great majority of these youth spend the most of their time in community, school, and other service agencies, where they receive intervention (OJJDP, 2008). To date, a number of studies that have examined the effects of academic and reading-specific intervention programs with juvenile offenders outside the detention setting have not been included in a synthesis of research. In addition, academic intervention research for youths in secondary education, the age range typical for youths engaged in juvenile delinquency, has increased within the most recent 15 years (Edmonds et al., 2009). In sum, given the expansion of educational intervention research, as well as the wide range of settings within which intervention for juvenile delinquency is known to occur, there seems to be a pressing need to synthesize and compare the effects of the most recent juvenile delinquency intervention programs on academic outcomes using a range of approaches and conducted in a range of settings. Finally, and as previously alluded to, while some past syntheses examining the effects of juvenile delinquency interventions (many of which have already been mentioned) have examined their effects on academic outcomes (Leone et al., 2005, Wilson, Lipsey and Derzon, 2003 and Wilson, Lipsey and Soydan, 2003), many of the most comprehensive synthesis studies (e.g., Lipsey, 1999, Lipsey, 2009 and Lipsey and Cullen, 2007) may not accurately reflect the current state of academic interventions in practice. This would seem to be problematic given that a number of innovations in education research interventions have come about in recent years. To explain more, in the largest set of studies used in meta-analysis of juvenile delinquency interventions (n = 305) that includes academic outcomes, the most recent study was conducted in 1996 ( Wilson, Lipsey and Derzon, 2003 and Wilson, Lipsey and Soydan, 2003). Meanwhile, the most effective reading interventions in education settings, especially with upper elementary and secondary education levels (grades 6–12) are reported later (1990–2010) (e.g. Edmonds et al., 2009 and Wanzek et al., 2010). Notably, studies reporting outcomes of targeted reading interventions in education research have increased dramatically since the mid-1990s ( Edmonds et al., 2009 and Wanzek et al., 2010). In sum, recent advances in educational interventions would not yet be captured in syntheses of delinquency intervention literature given the dates of most studies included in previously published meta-analytic studies. As such, it seems important to conduct and up-to-date meta-analysis on the effects of juvenile delinquency interventions on academic outcomes. 1.4. Moderators of interest Finally, there are several aspects of delinquency interventions that are important to consider as potential moderators of effects. Moderators examined in past syntheses related to juvenile offending or education interventions typically include: research design, demographic information (e.g., age, sex, grade level), the duration of the intervention, the training or type of professional delivering the intervention, and the type of setting (Edmonds et al., 2009, Lipsey, 2009, McCarty and Weisz, 2007, Weisz et al., 2006 and Wilson and Lipsey, 2007). All of these factors have been found to explain some of the variability in outcomes for delinquency interventions and/or within the psychosocial and educational intervention literature. As such, we consider these factors in this synthesis of research on juvenile delinquency intervention effects on academic outcomes. 1.4.1. Research design It is necessary to consider whether the findings are due to the difference between groups or the effects of the intervention, rather than research design characteristics (Weisz et al., 2006 and Wilson and Lipsey, 2001). Randomized controlled trials are rare within juvenile delinquency literature, but quasi-experimental designs with appropriate controls are acceptable and are often available in this literature (Lipsey, 2009). Typically, higher quality research designs produce smaller effect sizes and more trustworthy results (Edmonds et al., 2009, Lipsey, 2009 and Weisz et al., 2006). We expect that this will be the case in the current synthesis as well. However, the extent to which randomized experiments differ from quasi-experimental designs with or without equating groups remains an important question to address in this synthesis. 1.4.2. Age of participants In studies about juvenile delinquency, a younger age of arrest is typically predictive of a more problematic level of criminal involvement (Cottle et al., 2001). Older youths tend to respond better to delinquency interventions (Lipsey, 1999). At the same time, however, younger youths respond more rapidly to educational interventions (Kroesbergen & Van Luit, 2003), while older youths may require more intense and targeted educational interventions when they reach grade 6 or higher (Edmonds et al., 2009). This moderator is of interest, but is considered exploratory in this study. 1.4.3. Gender and ethnicity/race Gender may also be important, as girls tend to have higher educational achievement and may respond differently to intervention compared to boys in juvenile delinquency target samples (Zabel & Nigro, 2007). Boys are more likely to engage in crime in general (Cottle et al., 2001). In a meta-analytic study conducted by Wilson, Lipsey, and Soydan (2003), delinquency interventions had small differences in effects, but overlapping confidence intervals, for White compared to minority cultural groups. The question of whether there are differences in the effects of intervention between gender and cultural or ethnic groups remains important to consider. 1.4.4. Setting The type of setting is a moderator of interest that has been included in some prior meta-analyses (Wilson, Lipsey and Derzon, 2003 and Wilson, Lipsey and Soydan, 2003), but is not always available depending on the focus of the synthesis. In their synthesis of meta-analyses, Lipsey and Cullen (2007) reported that community-based programs have more favorable effects than detention settings, but highlight that the reasons for that could be due to the differences in samples, not settings or programs. This moderator is important in order to understand the effects of various types of services available for rehabilitation and supporting education. In line with prior synthesis in this area of research, we also examine setting as a moderator. 1.4.5. Training and staff characteristics The type of training, overall qualifications of the staff providing the intervention, supervision and support of staff during implementation, and type of professional delivering the intervention have all been noted moderators (Wilson & Lipsey, 2007) in prior studies that would be relevant to the current study. If an intervention is designed and carried out by a researcher, as opposed to staff who are not directly affiliated with the intervention per se, the effects usually positive and more pronounced (Wilson & Lipsey, 2007). In addition, interventions delivered by staff with higher levels of training or advanced degrees also yield larger effects (Weisz et al., 2006). It is expected that this investigation would identify similar patterns. 1.4.6. Intervention type and dosage In the intervention literature, including educational, psycho-social or behavioral interventions, the type and duration of intervention also moderate outcomes (McCarty and Weisz, 2007 and Wilson and Lipsey, 2007). Specific interventions that target an identified skill deficit or symptom, rather than a more general approach, tend to produce larger effect sizes (Edmonds et al., 2009). For example, a focused reading comprehension intervention rather than a general broad tutoring approach for academic skills, or using cognitive–behavioral therapy as opposed to more general counseling for mental health is recommended. In addition, interventions that are delivered over period of about 6 to 12 months tend to also produce larger effect sizes than briefer interventions (McCarty and Weisz, 2007 and Wilson and Lipsey, 2007). The intervention type will be noted with as much specificity as possible, particularly for any academic intervention aspects. The duration will also be noted. 1.5. Summary of current gaps in the literature The literature examining the effects of juvenile delinquency interventions is currently limited in several ways. Prior meta-analyses have been limited in terms of the setting of studies included in the analyses, and included studies are not reflective of recent advances in educational interventions. Further, even those meta-analyses that have examined academic outcomes, little attention has been paid to characteristics of the research design, academic intervention, setting, or sample that may influence the effects of juvenile delinquency programs on academic outcomes. Future research would benefit from a systematic analysis that examines the effect of juvenile delinquency interventions on academic outcomes. In addition, the current study offers more depth in terms of the level of specificity for the academic and behavioral/psychosocial intervention components that will be examined. The findings from this study will address which components are most effective for improving academic outcomes of youths engaged in juvenile delinquency. Academic intervention may not be the only program that improves educational attainment, and improvements may occur in school, juvenile justice, or community settings. That is, even program components geared toward behavioral outcomes may improve academic outcomes and may do so across a wide variety of settings. With these considerations in mind, the goals of this research synthesis were twofold. First, we examined the overall effects of intervention programs designed to reduce juvenile delinquency across school, community, and juvenile justice settings on academic achievement, school attendance, and school attitudes. Prior syntheses of research that examined academics as a specific outcome of juvenile delinquency intervention effects were either including studies conducted in juvenile detention facilities only, or in schools only, but not across the multiple settings where juvenile delinquency programs are offered. Second, in this meta-analysis, we investigated several moderators that could potentially influence the effectiveness of delinquency interventions on academic outcomes. We predicted that juvenile delinquency programs that include more rigorous designs, more highly trained staff delivering the intervention, and an academic component will show the greatest improvement in academic achievement, though we expected that even interventions without an academic component could have some positive effect on academic achievement.

. Conclusions This synthesis provides a marker of the state of the literature, and there is considerable room for increased knowledge to inform juvenile delinquency research in some meaningful ways. Given that educational variables are malleable and dynamic (Edmonds et al., 2009), and are closely associated with criminal behavior (Cottle et al., 2001 and Zabel and Nigro, 2007), incorporating educational aspects into rehabilitation programs seems intuitive in terms of a needs-response or responsivity principle (Andrews, 1995). It is clear that rehabilitation programs, particularly those based on specific needs of the offender, are helpful in reducing criminal activity (Lipsey & Cullen, 2007) Moreover, it seems likely that rehabilitation programs might be most advantageous when educational outcomes are also improved. Unfortunately, the effectiveness of those programs that have been tested for improving academic outcomes seems to be lackluster, or at least the evidence is lacking. A possible avenue for future research is to examine the effects of juvenile delinquency programs that make use of the recently expanded variety of programs and resources from the educational intervention literature that have been found to be most effective. The relationship between low academic achievement and delinquency is well established (Cottle et al., 2001), but the field needs to move on to exploring ‘how’ and ‘why’ this link is present in order to improve and inform future delinquency interventions and interrupt the school to prison pipeline. It would be helpful for juvenile justice and educational researchers to draw on interdisciplinary approaches and collaborate in order to address the ongoing and related problems of juvenile delinquency and school failure.