Social Science Research 44 (2014) 126–140

Contents lists available at ScienceDirect

Social Science Research journal homepage: www.elsevier.com/locate/ssresearch

The color of juvenile justice: Racial disparities in dispositional decisions Jamie J. Fader ⇑, Megan C. Kurlychek, Kirstin A. Morgan School of Criminal Justice, University at Albany, 135 Western Avenue, Albany, NY 12222, United States

a r t i c l e

i n f o

Article history: Received 14 March 2013 Revised 8 October 2013 Accepted 23 November 2013 Available online 9 December 2013 Keywords: Juvenile court Residential facilities Intervention modalities Decision-making Race/ethnicity

a b s t r a c t Existing research on dispositional decisions typically models the outcome as merely placed or not placed. However, this does not accurately reflect the wide variation in residential options available to juvenile court actors. In this research, we combine data from ProDES, which tracks adjudicated youth in Philadelphia, with data from the Program Design Inventory, which describes over 100 intervention programs, to further examine the factors that influence court actors’ decision making in selecting an appropriate program for a juvenile offender. We find that even after controlling for legal and needs-based factors, race continues to exert a significant influence, with decision makers being significantly more likely to commit minority youth to facilities using physical regimen as their primary modality and reserving smaller, therapeutic facilities for their white counterparts. Using focal concerns theory as an explanatory lens, we suggest that court actors in this jurisdiction employ a racialized perceptual shorthand of youthful offenders that attributes both higher levels of blame and lower evaluations of reformability to minority youth. Ó 2013 Elsevier Inc. All rights reserved.

1. Introduction Since the birth of the American reformatory in 1825, the character of residential placements (i.e., their architecture, setting, size, and strategies for addressing delinquency) has been guided by evolving standards of how policy makers and the public define social constructs such as ‘‘children,’’ ‘‘crime,’’ and ‘‘punishment.’’ During eras in which children were conceived of as innately good and malleable and causes of delinquency were viewed as largely outside of their control, reform schools were generally guided by principles of moral education, hard work, and sentimental pastoralism (Bernard and Kurlychek, 2010; Platt, 1969; Schneider, 1993). Many took the form of rural cottages in which Christian husband and wife teams served to replace biological parents who were assumed to be unable or unwilling to properly socialize and supervise their children. Later this image of the youthful offender as a needy child was to be replaced with one of a sick child. As the field of child and adolescent psychology grew, so did the view that the juvenile delinquent could be ‘‘fixed’’ through therapeutic treatments. This new image of the delinquent then led to a reformulation of juvenile institutions from educational and home-like settings to treatment-oriented facilities (Bernard and Kurlychek, 2010). More recently, the image of the juvenile delinquent shifted again, this time attributing to the youthful offender the mens rea (guilty mind) of an adult criminal. Indeed, during the 1990s, the image of the ‘‘superpredator’’ portrayed youthful offenders as cold, calculating, adult-like criminals who should be held responsible for their crimes. Residential placements that previously retained a medicalized therapeutic emphasis to reform delinquents were characterized as coddling vicious predators that needed to do adult, or at the very least, hard time (Bennett et al., 1996). According to Feld, the history of

⇑ Corresponding author. E-mail addresses: [email protected] (J.J. Fader), [email protected] (M.C. Kurlychek), [email protected] (K.A. Morgan). 0049-089X/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.ssresearch.2013.11.006

J.J. Fader et al. / Social Science Research 44 (2014) 126–140

127

shifting conceptions of youthful offending has left behind a modern system of ‘‘schizophrenic formulations of youth – dependent and vulnerable or independent and responsible – [that] enable states to selectively choose between the two constructs to manipulate young people’s legal status, to maximize their social control, and to subordinate their freedom and autonomy’’ (1999, p. 9). The shifting imagery constructed in each of these eras has also left its mark on the landscape of residential placements, resulting in a hodgepodge of facilities and intervention modalities currently available for use by decision makers in the juvenile court. Yet, to date, little attention has been given to the differences in these interventions, specifically as they relate to court decisions regarding which youth are most appropriate for which facilities (e.g. which youth are in need of care, which youth are amenable to treatment, and which youth are simply small criminals). Many scholars have argued convincingly that attributions of delinquent youth as more or less amenable to treatment have been raced, classed, and gendered (Feld, 1999; Grossberg, 2002; Platt, 1969; Schneider, 1993). Much, if not most, of the research on juvenile justice decision making has focused on whether and how these extra-legal factors affect sentencing outcomes at various stages of the process (Beger and Hoffman, 1998; Bridges and Steen, 1998; Leiber and Stairs, 1999; Sealock and Simpson, 1998; Wordes et al., 1994; Wu, 1997). The majority of studies conducted in the last several decades have found that youth of color are more likely than their white counterparts to receive harsher dispositions (Poe-Yamagata and Jones, 2000; Pope and Feyerherm, 1990; Pope et al., 2002), while gender effects have been less consistent (Belknap, 2001; Chesney-Lind and Shelden, 1992; Van Wormer and Bartollas, 2000). Although court actors are certainly less likely than ever to explicitly factor race and ethnicity into their decisions, recent research has identified a number of cognitive mechanisms by which decision makers develop a ‘‘perceptual shorthand’’ (Steffensmeier et al., 1998) that may cast minority youth as more adult like, culpable for their offenses, and less amenable to treatment (Bridges and Steen, 1998) or alternatively, could position class-privileged and/or white youth as exceptional cases which pose an even greater threat than their counterparts (Peterson and Hagan, 1984). However, as previously noted, the body of research on dispositional decisions has not accurately reflected the variety of residential options available to juvenile court decision makers. Studies exploring sentencing in juvenile court either measure severity in terms of sentence length or whether the youth was referred to a residential placement versus a community-based program. This approach masks important differences in the settings, target populations, treatment modalities, and program activities across residential facilities. As previously noted, dispositional options include a variety of placement settings, ranging from traditional cottage-type facilities, to specialized treatment units, to secure institutional facilities. We hypothesize that these distinct differences in placement types are an additional avenue through which disparity may operate. For example, a 90-day boot camp is qualitatively different from a therapeutic substance abuse facility; locked facilities with razor wire surrounding them are meaningfully different from those with open campuses and small cottages. Decision makers undoubtedly recognize these distinctions, as well. However, existing studies treat all placements as monolithic in nature. This paper addresses this shortcoming and expands our understanding of the decision making process by exploring the role of race (if any) in matching juvenile offenders to a type of residential placement facility. Before describing our study in greater detail, we explore two strands of literature that are essential to this undertaking: theoretical and empirical examinations of racial attributions in juvenile justice decision making, and factors predicting dispositional decisions in juvenile courts.

2. Literature review 2.1. Focal concerns and racial attributions A growing literature demonstrates that racial and gendered stereotypes can trickle down into the court’s decision-making processes. This may be particularly true in the juvenile court, which was founded on the principle of individualized justice and thereby encourages greater discretion among decision-makers (Feld, 1995; Horowitz and Wasserman, 1980; Sampson and Laub, 1993). Moreover, decision makers, especially those in overburdened urban courts such as the one in this study, must make a rapid succession of decisions. Research has shown that in these circumstances court actors often apply patterned responses, or rely on ‘‘perceptual shorthand’’ to determine appropriate sentences (Steffensmeier et al., 1998; Steffensmeier and Demuth, 2006). The primary theory that has been applied to understand what guides court actors’ decision making is focal concerns theory, which is based on empirical analyses of adult court decisions but has been extended by other scholars to the juvenile justice system (Bishop et al., 2010). According to this theory, in sentencing decisions, a judge (and/or other court room actors) is trying to balance three key concerns. First is the assessment of the defendant’s culpability and blameworthiness for the crime. In a juvenile court it is evident how this may be intertwined with assessments of maturity and ‘‘adult-like’’ behavior. A second concern is the need for community protection. This assessment necessarily requires the judge to make predictions about whether or not this particular youth will offend again in the future and how serious this offending might be. The inherent ambiguity in such prediction often leads court actors to rely on the aforementioned perceptional shorthand and stereotypes in making such assessment. The final concern is with the practical consequences of sentencing decisions (e.g. cost, existing resources, or the offender’s ability to ‘‘do time.’’) In extending this notion to juvenile court, it might be that those youths seen as least blameworthy and less mature would be the most likely to benefit from an educational or

128

J.J. Fader et al. / Social Science Research 44 (2014) 126–140

therapeutic setting and least likely to be able to handle the more ‘‘hard time’’ or a physically based or disciplinary based program such as a boot camp. Prior analysis of adult sentencing in Pennsylvania finds that race, gender, and age interact such that young black males receive harsher penalties than any other group (Steffensmeier et al., 1998). Interviews with judges revealed that they assessed the criminal records of young black men as more serious and presenting greater future risk of offending and viewed this group as more blameworthy for their offenses. Moreover, they viewed young black men as less reformable, less tied to the community, and their criminality as less mitigated by prior victimization (Steffensmeier et al., 1993; Ulmer, 1997; Ulmer and Kramer, 1996). Focal concerns theory may be extended by incorporating causal attribution theory, which suggests that sentencing decisions are an attempt by court actors to link assumed causes of criminality and an appropriate punitive response (Carroll and Payne, 1976; Hawkins, 1981). Attributions are differentially adhered based, among other factors, whether a cause is internal or external to a person (Graham and Lowery, 2004). According to this theory, criminal justice actors view the cause of crime as either stemming from internal factors, such as antisocial personality or lack of remorse, or external (i.e., environmental) factors, like delinquent peers, poverty, or lack of parental supervision. Characteristics attributed to internal sources may be considered the least amenable to change or be viewed as the product of choice and thereby be assigned a greater degree of blame by decision makers. In short, these attributions may lead to an emphasis on different focal concerns. For example, internal attributions of crime may lead court actors to affix greater blame to the juvenile or perceive a greater risk of recidivism (Bridges and Steen, 1998). External attributions may lead them to weigh practical concerns in such a way as to extend scarce or costly resources to those viewed as neediest or most reformable. In probably the best-known study examining the locus element of causal attribution, Bridges and Steen (1998) analyzed narrative reports written by juvenile probation officers. They found that, even when controlling for offense severity and prior offenses, officers were more likely to attribute crimes committed by black youth to negative internal characteristics and those by whites to negative external characteristics. Moreover, negative internal attributions were more influential in assessments of future risk than were external negative attributions. Despite the gains made by racial and ethnic minorities in recent decades, it seems clear that cultural stereotypes still operate, often in ways that are masked by other institutional processes (Spohn and DeLone, 2000). This intricate, contextual nature of racial discrimination in decision making is reflected in the often conflicting literature on the role of race in both adult and juvenile courts. In the next section, we briefly review prior research in this area, highlighting some of the other factors that have been shown to affect dispositional decisions. 2.2. Prior research: predictors of dispositional decision making in juvenile court As in the literature on adult sentencing, the extant body of research on juvenile justice decision-making has largely revolved around the role of extra-legal variables such as race (and to a smaller degree gender) in dispositional outcomes. An early meta-analysis performed by Pope and Feyerherm (1990) concluded that two-thirds of the studies on race effects on juvenile court decision making were indicative of racial bias. Subsequent analyses spanning 1989–1999 have found similar results (Leiber, 2002; Pope et al., 2002). Since many – but not all – jurisdictions studied were found to exhibit either direct or indirect race effects, 21st century scholarship has employed a contextual perspective, asking not whether, but when race matters (Spohn and DeLone, 2000). This perspective also acknowledges that race effects in decision making can result in either more severe or more lenient outcomes dependent upon other offense and offender characteristics (Leiber and Mack, 2003; Mears and Field, 2000). Moreover, dispositions for residential placements of juveniles cannot be understood solely within the framework of traditional legal and extra-legal factors. As the juvenile justice system is based on philosophies of individualized justice and rehabilitation, many needs-based factors of the youth and his/her family setting may enter into the equation. For example a youth may receive a commitment when there is evidence of physical or sexual abuse or severe neglect within the youth’s family, if there is a high degree of family stability, or if the youth presents substance abuse or mental health needs that cannot be treated in the community. With some exceptions (see Bishop et al., 2010), this critical point has been largely overlooked in the criminological literature, probably because most studies draw from official records that typically do not contain information on treatment needs. This may be a key reason why research on sentencing predictors often explains so little of the variation in dispositional decisions (Feld, 1995). Studies that have been able to incorporate needs-related predictors have found that substance abuse, family problems, or school problems are considered along with offense-related factors (self-identifying reference; Campbell and Schmidt, 2000; Sanborn, 1996) or are even better predictors than legal factors (Horowitz and Wasserman, 1980). Some find that the effects that are conditioned by race and gender (Leiber and Mack, 2003), suggesting that images of presenting needs and correctional solutions may be racialized and/or gendered as well (see also Frazier and Bishop, 1995; Pope and Feyerherm, 1990). For example, decision-makers may use race as a proxy for ‘‘broken’’ families or single-parent households, attributing greater future risk to black youth because of a perceived lack of supervision and proper socialization by black families (Bishop and Frazier, 1996; Wordes et al., 1994). In addition, as previously noted, most studies of court decision making have relied on the simple dependent variables of incarcerated or not, and length of sentence when incarcerated. Based on the history of the juvenile court, however, we have demonstrated that decision-makers (judges and probation officers, who recommend placements) must not decide

J.J. Fader et al. / Social Science Research 44 (2014) 126–140

129

only whether or not to place a juvenile out of home, but must select the specific program to which the youth will be committed.1 Based upon our extension of focal concerns theory to juvenile courts, and the existing empirical evidence on the role of race in sentencing decisions we hypothesize that: 1. Minority youths, particularly black youths, will be more likely to receive placement in facilities that emphasize physical activity while white youths will be more likely to receive commitments to therapeutic facilities. We base this hypothesis on the findings that court actors are more likely to attribute criminal acts to internal causes for minority youth and that such attributions are further associated with a person’s inability to change. Thus, the more costly, less widely-available therapeutic services will be reserved for white youths who appear more ‘‘amenable’’ to treatment. 2. Race will interact with other legal and extra-legal factors such that minority youth accused of violent crimes will be even more likely to receive physical rather than therapeutic programming. On the other hand, white youth with presenting needs will be even more likely to receive therapeutic programming. If black youth are, as we noted, more likely to be viewed as internally responsible or blameworthy for their crimes, we believe that when such youth commit a violent crime (typically seen as more adult like behavior) this will further impact court actors assessment of these youth as young criminals, evoking a court response more in tune with the get tough movement’s philosophy. On the other hand, for white youth, already seen as less responsible for their crimes, when there are other presenting needs-based factors, this will further emphasize the appropriateness of therapeutic treatment and the youth’s ability to change.

3. Data and methods Data for this study are drawn from ProDES (Program Development and Evaluation System), which tracked all Philadelphia youth who were adjudicated delinquent and court-committed to intervention services during a ten-year period. ProDES data were gathered at four points: (1) from the juvenile’s court file, or ‘‘J-file,’’ (2) at the time of intake into the program, (3) at the time of discharge from the program, and (4) six months after discharge from the program. In addition to official records, which were culled at the point of disposition and again at follow-up, the dataset includes risk and needs measures recorded by program staff at intake and discharge, self-report scales completed by youth at intake and discharge, and a phone survey conducted with young people and guardians at follow-up. Between 1994 and 2004, ProDES collected data on over 40,000 juveniles with 17,695 being committed to residential facilities thereby constituting our original sample of interest. In this paper we examine juvenile court file (‘‘J-file) data for youth referred by the court to private residential facilities contracted through the city’s Department of Human Services. These rich data represent all information available to decision-makers at the point of disposition. ‘‘J-file’’ data includes information gathered by probation officers during interaction with the juvenile and his or her family, mental health and/or psychological evaluations, prior offense history records and details of the referring ‘‘instant’’ offense, school records, and progress reports from programs to which the child may have been committed previously. ProDES data are supplemented by the Program Design Inventory (PDI), a detailed program description for each of the over 100 delinquency intervention programs available to Philadelphia’s decision-makers. The PDI was constructed by visiting each program in person and interviewing administrators and staff about the programs’ mission, treatment modalities, program activities (including weekly dosage), short- and long-term treatment goals, program location, licensing, and characteristics of the staff.2 Because we wanted to consider cases for which a variety of dispositional options were available to decision-makers, we excluded cases for whom correctional options were likely constrained by youths’ special needs or long histories of failure in prior placements. This includes commitments to RTF (Residential Treatment Facility), which are certified facilities for juveniles with severe mental health or co-occurring disorders, institutions targeted toward sex offenders, fire starters, or developmentally delayed youth (2,280 juveniles were excluded on this basis). When young people present these needs or offenses, decision making is constrained to a small number of facilities that accept them. Another potential bias may also be encountered by including youths who are sent to state programs operated by the Pennsylvania Department of Public Welfare. These facilities are often considered the ‘‘last stop’’ in the juvenile justice system and are typically reserved for youth with long histories of serious offending and/or demonstrated inability to adjust to less secure facilities. Thus a ‘‘decision’’ to send a youth to these facilities may reflect a long history of system failures which limits choice rather than true decision1 In Philadelphia’s juvenile justice system, adjudicatory and dispositional hearings are held separately for serious offenders or those with significant needs related to substance abuse, mental health, or family functioning. Probation officers, who are responsible for conducting a pre-sentence investigation on the juvenile, make recommendations to judges about the most appropriate facility to address the youth’s offense and needs. Often, the probation officer has contacted the facility to ensure that the youth would be accepted and that the bed space is available before making such a recommendation. Studies conducted in other jurisdictions (Kingsnorth and Rizzo, 1979; Spohn, 2002) and anecdotal evidence gathered by the first author in six years of conversations with juvenile probation officers in Philadelphia suggests a high level of concurrence between probation officer recommendations and judges’ official dispositions. 2 The first author was involved in PDI data collection in 2001–02 and personally visited the programs in our final sample (most on several occasions).

130

J.J. Fader et al. / Social Science Research 44 (2014) 126–140

Table 1 Characteristics of traditional, therapeutic, and physical regimen facilities.

# of Facilities Utilization Dosage Cost Security level Mission Counseling activitiesc

Licenses a b c d

PDI available Commitments per year Average length of stay Average per diem Opena Education MH &/or D&A Individual Group Substance abused OMH, BDAP, Dept. of Health, D&A, MH/MR

Traditional

Therapeutic

Physical regimen

6 466 9 months $131.60 3 5 2 3 2 2 4

12 302 8 months $148.31 13b 4 7 12 12 8 11

4 588 7 months $115.83 0 0 0 1 3 0 0

Security level data were available for 1 additional facility in each category. 2 Programs had secure wards for use as needed. Counseling activities are counted when they are regularly scheduled, not simply available as needed. AA or NA meetings are not counted.

maker discretion. For this reason, we examined the data both with these youth in the sample and without them. As we believe the estimates excluding the 2508 state-committed youth from the sample provide a more conservative estimate we present those models herein.3 The final sample utilized for the tables presented below included 12,906 youths sentenced to 28 residential programs.

3.1. Dependent variable To create the dependent variable, program type, we analyzed data from the PDI, paying special attention to the security level (e.g., open or staff secure); the problems, needs, and offense behaviors of the program’s target population; mission and objectives; licensure; and activities/dosage. Table 1 presents the number of facilities with these characteristics, as well as the annual utilization and average cost per juvenile per day (per diem). Each author separately developed an inductive classification strategy using these factors and then compared both the categories and the programs contained within those groups for consistency across authors. There was remarkable consistency across classification categories with initial agreement on all but one facility.4 Our final typology of residential placements, which mirror the historical development of the juvenile justice system, included three categories: traditional reform schools (7 facilities), therapeutic programs (14 facilities), and programs emphasizing physical regimen (7 facilities). Traditional reform schools, modeled upon the nation’s first reformatories, tended to emphasize educational objectives most heavily and mentioned education in 5 of 6 of the mission statements available in the PDI. These training schools were often structured like college preparatory schools, requiring students to wear blazers and ties. They frequently referred to themselves in their promotional literature and in the PDI as ‘‘schools’’ or ‘‘academies’’ and to their clients as ‘‘students.’’ Photos on their websites show young people studying, learning trade skills, playing sports, and holding diplomas. The average length of stay at these facilities was nine months and successful discharge was often tied to completing the academic year. These facilities were the most likely of the three program types to be run by non-profit organizations. Therapeutic programs, consistent with the medical model of juvenile justice, were those whose objectives and activities (e.g., counseling) were most clinical in nature and where counseling activities were most likely to be delivered by licensed social workers. This category is largely comprised of substance abuse treatment and mental health programs (as long as they were not RTF-certified, noted in our exclusion criteria). These programs are more likely to refer to their residents as ‘‘patients’’ or ‘‘clients.’’ Architecturally, all of these facilities were open settings, although two of them had secure wards for use as needed. They were generally the most modern and physically decentralized into dorms or cottages. Their average length of stay was eight months, although the most heavily utilized programs were 10–14 months. Successful discharge at these facilities was most often tied to a treatment plan and progress through a therapeutic level system. For this reason, release could be especially challenging to earn because residents were required to demonstrate both cognitive and behavioral change (self-identifying reference). These programs are also distinguished from traditional and physical regimen 3

Alternate models on larger sample provided consistent results on the role of race in dispositional decisions and are available from the authors upon request. This particular facility’s activities were largely education-focused as were most traditional reform schools, but we classified it as physical regimen modality because it is both known for its athletics program and for using physical confrontation between ‘‘peers’’ as its primary modality. After review of the supplemental information on this facility all authors were in agreement on the final classification. However, as a precaution, findings were replicated with the alternative specification of this facility. These subsequent analyses showed similar magnitude and direction of overall findings although the absolute values of coefficients necessarily changed slightly. In the 5 cases where the PDI data were not available, classification was determined by the first author, who had conducted site visits and worked with the facilities as part of the ProDES project. Mission statements, information on target population and services, and photographs of the facilities were often available on facility webpages. 4

J.J. Fader et al. / Social Science Research 44 (2014) 126–140

131

programs by their high levels of licensure from Department of Health, Bureau of Drug & Alcohol Programs (BDAP), Office of Mental Health and were often certified as inpatient/outpatient substance abuse and/or mental health facilities. Finally, programs emphasizing physical regimen included boot camp, wagon train, wilderness programs, and one program that encouraged peers to physically confront one another to create a ‘‘positive peer culture.’’ These programs were more likely to target their rehabilitative efforts upon the body than the brain. All were staff secure. These were most often short-term ‘‘shock’’ incarceration programs that involved military drills or other physical activities such as wilderness survival. Young people went through the program with a ‘‘platoon’’ or squad of others and were released after a fixed period of time (the most heavily-utilized were three or four months) as long as they had not engaged in ‘‘complete non-compliance’’ with the programs’ rules. None of these programs referred to education, mental health, or substance abuse in their mission statements, nor had licenses beyond standard residential and educational licenses. Physical regimen programs had the highest utilization, the lowest cost per diem, and were the most likely of the three categories to be run by for-profit organizations. It should be noted that these categories are sorted by emphasis and designed to represent ‘‘ideal types’’ of residential facilities. All such facilities were required by law, for example, to provide access to education for all school-aged youth. Many traditional reform schools also offered counseling programs, although they were much less likely to identify substance abuse or mental health problems in their mission statements, target populations, or primary objectives. We contend that physical regimen and traditional programs are distinct from therapeutic programs in ways that are particularly meaningful to decision makers. First, beds within therapeutic programs were fewer in number than in facilities that used other modalities. This scarcity made them especially valuable and consequently, reserved for those who were viewed as most deserving, most needy, or most amenable to treatment. Open beds at the largest physical regimen programs, by contrast, were especially plentiful because new platoons started every few weeks and ran only 90–120 days. Traditional reform schools, typically located in large ‘‘halls’’ or ‘‘academies,’’ were also able to accept large numbers of youth referred by the Philadelphia court. Second, substance abuse and mental health treatment programs (comprising 12 of 14 of those we classify as therapeutic programs) were particularly valuable because they were not available on an outpatient basis during the study period. If education was the primary concern of the decision-maker, by contrast, educational needs could be met by either communitybased (day treatment) or residential settings. Therefore, youth who presented severe treatment needs were more likely to be sent by judges to residential programs than to remain in the community (self-identifying reference) because no communitybased alternative was available. It should be noted that, although the per diem costs of these three types of facilities varies, with therapeutically-oriented facilities costing more per day and per stay than the other two types of facilities, this was unlikely to affect decision-making. Since the funding for all delinquent commitments was provided by the city’s Department of Human Services, decision-makers at Family Court did not have to consider youths’ families’ ability to pay or the accountability to City Council for overspending. 3.2. Independent variables The independent variables for use in the study were selected to represent notions of legal considerations, traditional extra-legal considerations, and needs-based characteristics. The independent variables used here largely mirror those used in [self-identifying reference] and Bishop and Frazier (1996), the most widely-cited study on race and decision making. As our primary research question addresses the relationship between extra-legal variables such as race and gender on disposition, we begin with a consideration of race/ethnicity coded as white (reference), black or Latino.5 We also include gender (female reference) and age (continuous) in the model. Next we moved to the consideration of legal factors that are known to influence decision-making. To capture the nature and severity of the presenting offense, we classify offenses in two ways. The first is a simple typology of crime into personal, weapon, drug, property (reference), and other categories. In cases in which multiple charges were generated, the most serious referring charge was used. Second, we adopted from the Pennsylvania Commission on Sentencing an offense gravity scoring (OGS) technique used by the adult courts in Pennsylvania to determine offense severity. This system classifies offenses on a range from 1 to 14 with 14 being the most severe. This coding scheme utilizes the offense (such as assault, theft, robbery, etc.) as well as subcategories of the offense which are designed to capture further nuances in the actual impact of the crime such as amount and type of drug in a drug offense and injury to the victim in a personal offense. For many crimes, offense gravity may vary drastically based on such factors. For example, the offense gravity score for retail theft ranges from 2 (under $50 dollars) to 7, (over $100,000 of merchandise); in cases in which it was not possible to decipher such distinctions from the data, a median offense gravity score calculated from the possible range for the primary offense was assigned (e.g., in the above case a median score of 4.5 would be assigned). Another important legal consideration is a youth’s prior involvement with the justice system. This is important as previously noted because the juvenile court oft employs a graduated approach in which more lenient strategies are tried first, with offenders later ‘‘graduating’’ to more severe outcomes. Our measures of prior court involvement include whether the youth was under court supervision at the time of the current offense (yes/no), whether the youth had another arrest between 5 We use the term ‘‘Latino’’ instead of ‘‘Hispanic’’ because the vast majority of Hispanic youth in Philadelphia are Puerto Rican. The sample did not contain sufficient numbers of other minorities groups such as Native Americans or Asian–Americans for inclusion of these categories in the current study. Instead, we created a fourth category combining other groups with a small number of cases with no race/ethnicity data.

132

J.J. Fader et al. / Social Science Research 44 (2014) 126–140

Table 2 Description of sample (n = 12,906). Variable

Value

Percent

Program type (DV)

Traditional Therapeutic Physical regimen

32.2 26.3 41.5

Black Latino Race missing/other

71.9 13.7 2.1 15.84 (SD = 1.400) 93.5

Demographic factors Race/ethnicity

Age at disposition (mean) Gender Legal factors Offense gravity score (mean) Prior arrests New arrest since instant offense Under supervision at time of disposition Offense type

Needs-based factors Any dependency referral Parental substance abuse history Parental criminal history Sibling arrested Prior out of home placement Chronic or active drug abuse Occasional or past drug abuse Chronic or active alcohol abuse Occasional or past alcohol abuse

Male 1–14 1–3 priors 3 + priors 1 or more Person Drug Weapons Other

5.629 (SD = 2.900) 52 8.3 28.8 36.3 31.3 27.9 5.4 2.4 28.6 27.4 17.7 31.8 16.8 37.3 28 7.9 16.8

the current offense and its disposition (yes/no), whether or not the youth had a prior out of home placement, and number of prior arrests. This last measure was coded as none (reference), one to three, and more than three. While the original measure is continuous, it is highly skewed, with the majority of youth having no prior or 1 or 2 priors. This categorical measure therefore better captures what might be seen as types of youth that would influence attributions (e.g., the first time offender, the minor offender, and the chronic offender). Our final category of variables taps needs-based concerns. As the historic mission of juvenile court historically maintained the promise of ‘‘treatment,’’ we also needed to account for a juvenile’s presenting problems that might affect selection of a therapeutic option over either traditional reform schools or physical regimen programs. Here we consider a youth’s history of drug and alcohol abuse, which we hypothesize would lead to a preference for therapeutic programs. For both behaviors, no indication is used as the reference category with dummy variables added for any history of abuse, and chronic/current abuse. As the juvenile court also must consider the youth’s home environment, similar indicators of parental drug and alcohol abuse, parental criminality, sibling criminality, and prior dependency referral are also included in the models. These measures are also included as dichotomies with no indication of problem or arrest serving as the reference category. Table 2 provides basic descriptive statistics on all variables.

3.3. Statistical methods We selected multinomial logistic regression for our analyses as the categories of our dependent variable are distinct in nature rather than representing an ordinal pattern of severity. This method allows us to compare the impact of each independent variable as an increase or decrease in the odds of receiving a given program type (traditional, physical regimen) as compared to our reference category (therapeutic). The reference category was rotated in a series of analyses to allow for a greater understanding of the direct relationships between the individual categories. Models with ‘‘therapeutic’’ as the reference category were selected for presentation simply because these models provide positive coefficients on key independent variables of interest which makes discussion of the findings more straightforward for the reader. To explore the proposed interaction effects between race and offense type and needs-based factors, separate models were fitted for White, Black and Latino youth. A Z-statistic was used to explore the whether the differences observed in coefficients across models were significant (Paternoster et al., 1998). Use of separate models also allows for the exploration of model fit to determine whether our traditional juvenile court variables are perhaps better at predicting outcomes for a given racial/ ethnic category.

133

J.J. Fader et al. / Social Science Research 44 (2014) 126–140 Table 3 Distributions of dispositions across program modalities (%) (n = 12,906).a Category

Variable

Traditional

Therapeutic

Physical regimen

Race/ethnicity

White Black Latino Race missing/other Age at disposition (mean) Male Female Offense gravity score (mean) No prior arrests 1 to 3 prior arrests More than 3 prior arrests New arrest since instant offense Under supervision at time of disposition Not under supervision Person Property Drug Weapons Other Any dependency referral No dependency referral Prior out of home placement No prior out of home placement Parental substance abuse history No parental substance abuse Parental criminal history No parental criminal history Sibling arrested No sibling arrested Chronic or active drug abuse Occasional or past drug abuse No record of drug abuse Chronic or active alcohol abuse Occasional or past alcohol abuse No record of alcohol abuse

20.0 34.7 28.7 41.7 15.2 32.0 35.5 5.8 32.2 31.0 20.4 31.5 27.9 34.6 35.7 34.2 26.3 34.2 23.8 32.3 32.2 16.4 35.4 28.5 33.5 32.9 32.0 32.5 32.1 15.1 27.6 43.3 21.0 33.3 35.3 32.2

57.4 19.9 33.3 19.5 16.1 24.9 47.4 5.5 26.3 25.5 26.2 25.3 25.5 26.8 24.0 25.5 30.0 24.3 30.9 27.1 26.0 26.0 26.3 30.2 24.8 24.2 26.7 23.6 27.6 52.3 34.0 14.0 41.0 22.1 21.3 26.3

22.7 45.4 38.0 38.7 16.2 43.2 17.1 5.6 41.5 43.6 53.5 43.2 46.5 38.6 40.4 40.3 43.8 41.5 45.3 40.6 41.8 57.6 38.3 41.3 41.6 42.8 41.2 43.9 40.3 32.6 38.3 42.7 38.0 44.6 43.4 41.5

Age Sex Offense score Arrest history

Supervision Offense type

Dependency referral Placement history Family history

Substance abuse history

Total a

Percentages may not add to 100 because of rounding.

4. Results Table 3 provides a summary of distribution for all independent variables across the three types of residential programs.6 Most notable is the stark variation in modal facility type across racial/ethnic categories. Specifically, the modal disposition for white youth was a therapeutic facility (57%), while the modal disposition for black and Latino youth was a physical regimen facility (45% and 38% respectively), although Latino youths were much more evenly distributed than their black counterparts across program categories. Interpreted another way, black youth were nearly three times less likely than white youth to be committed to therapeutic facilities and twice as likely to be committed to programs relying on physical regimen as their primary modality. Black youth also were significantly more likely than white youth to be committed to traditional reform schools (35% and 20% respectively), although the difference is not as stark as in the other two placement types. Several other legal and extra-legal factors were also correlated with program type at the bivariate level. Gender differences were not apparent in traditional reform schools, but girls were almost twice as likely as boys (47% vs. 25%) to be placed in therapeutic facilities. Regarding legal factors, dispositions to physical regimen and traditional programs appeared to vary by arrest history; as youth accumulated arrests, they were more likely to be committed to physical regimen programs and less likely to be committed to traditional reform schools. Similarly, youth already under supervision at the time of their disposition, meaning that they had engaged in new criminal behavior or had violated the conditions of probation, were somewhat more likely to be placed at physical regimen facilities (47%) than the average placement rate (42%) for these programs. Youth with a prior out-of-home placement (suggesting repeated failure) were substantially more likely than average to be committed to physical regimen programs (58% vs. 42%). Youth with drug offenses (30%) and other offenses (31%) – most of which were public order offenses – were more likely than average (26%) to be committed therapeutic programs.

6 Bivariate correlations of independent variables were also explored, although not presented herein, with the highest correlation indicated being .639 for the categories of black and Latino. The next highest correlation was between offense types (.420 for drug and personal offenses) with no other correlations approaching a magnitude of concern. Overall, it was determined that multicollinearity of measures did not pose a problem for the current analyses.

134

J.J. Fader et al. / Social Science Research 44 (2014) 126–140

Table 4 Multinomial model with sociodemographic variables (n = 12,906). Variables

Intercept Black Latino Race missing/other Age at disposition Male

Traditional

Physical regimen

B

Std. error

5.142*** 1.486*** .843*** 1.612*** .426*** .633***

.303 .074 .092 .185 .018 .087

Exp(B)

B

Std. error

Exp(B)

4.420 2.324 5.015 .653 1.883

3.477*** 1.742*** 1.029*** 1.570*** .071*** 1.514***

.311 .069 .085 .182 .018 .103

5.708 2.798 4.804 1.073 4.544



p < .05. p < .01. Pseudo R2 (Nagelkerke) .164. Model chi-square 2319.17.  p < .001. 

Moving to the needs-related factors, neither dependency referral nor parental or sibling criminal history appears to be associated with commitments to facilities using different modalities. However, juveniles presenting histories of parental substance abuse were somewhat more likely (30%) than average (26%) to be sent to therapeutic facilities. Next to race/ethnicity, substance abuse history shows the greatest correlation with program type. Young people with chronic or active substance abuse problems were twice as likely (52% vs. 26%) as the average to be committed to therapeutic programs; this pattern is similar, but less strong for chronic or active alcohol abusers (41%). Table 4 presents the results from the first multinomial model which includes our key independent variable of interest, race, and controls for other demographic factors (gender, and age at disposition). Here it is easy to see that each of the variables has a strong independent effect on residential placement type. The odds ratio for a black youth being committed to a physical regimen program as compared to a white youth was 5.71, while the odds ratio for a Latino youth as compared to a white youth was 2.80.7 Another way to think about this is that for every white youth receiving a commitment to a program emphasizing physical regimen instead of a therapeutic program, there were 5.7 black youth and 2.8 Latino youth committed. The impact for males was also quite large with an odds ratio of 4.54. Age had the smallest effect, although still statistically significant, with an odds ratio of 1.07 indicating that one year increase in age increases the likelihood of commitment to a physical regimen program by 7%. Findings are similar in direction and significance for traditional reform schools although slightly smaller in magnitude, with black youths having an odds ratio of 4.42 and Latino youth 2.32 as compared to white youths to be committed to this modality. The odds ratio for males is 1.88, indicating that males are approximately 88% more likely than females to be placed in traditional reform schools over therapeutic programs. However, older offenders were less likely to be committed to traditional programs than to therapeutic facilities, with a one-year increase in age decreasing the likelihood of commitment to this type of facility by 35%. Table 5 presents the findings from a more complete model that introduces what we deemed as legally relevant factors to see whether these may explain some of the large differences originally found between demographic factors and program modality. All legal factors operate in the anticipated direction; however, only offense severity, prior record, and type of offense achieve statistical significance. In specific, when comparing physical regimen to therapeutic programs we find a one point increase in offense severity to increase the likelihood by about 4.5%. Having a 1–3 prior arrests (compared to 0), increased the likelihood by about 14%, and having over 3 prior arrests increased the likelihood by 32%. Interestingly, of the different types of crime examined (drug, property, personal, weapon, other), the only to achieve significance in this model was drug offense with this type of offender more likely than other types to be committed to therapeutic programs. For example, an offender adjudicated on a drug offense was about 26%% less likely than other types of offenders to be committed to a physical regimen program over a therapeutic program. When comparing traditional reform schools to therapeutic programs, the results show that an increase in offense gravity also increased the likelihood of receiving this type of commitment, although only slightly (about 3%). Having a prior arrest record worked in the opposite direction by decreasing the likelihood of receiving a traditional reform placement rather than a therapeutic program and having committed a drug offense again led to a decreased likelihood of receiving a commitment to a traditional reform school over a therapeutic program. Most importantly, however, we find that for neither the likelihood of commitment to a physical regimen program or a traditional reform school rather than a therapeutic program does the inclusion of legal factors significantly reduce the primary effects of race, gender, and age. Thus, the preference for traditional reform schools and even more so for physical regimen programs for black and Latino offenders is not explained by the type or severity of the crime committed by the youth, the youth’s prior record, or his/her current supervision level with the court.

7 To compute this percentage, we use the formula 100(eb  1) (Allison, 1991, p. 29). All models were reassessed with just black and just Latino in at one time to assess the impact of possible multicollinearity. As might be expected the impact of black and Latino were even larger in these alternate models.

135

J.J. Fader et al. / Social Science Research 44 (2014) 126–140 Table 5 Multinomial model adding legal factors (n = 12,769). Variables

Intercept Black Latino Race missing/other Age at current disposition Male Offense gravity score 1–3 Prior arrests More than 3 prior arrests New arrest since instant offense Under supervision Personal offense Drug offense Weapons offense Other offense

Traditional

Physical regimen

B

Std. error

4.601*** 1.553*** .963*** 1.613*** .397*** .689*** .024* .024 .261* .037 .049 .104 .336*** .098 .726***

.321 .075 .095 .186 .019 .090 .010 .061 .114 .055 .064 .070 .065 .116 .211

Exp(B)

B

Std. error

Exp(B)

4.725 2.620 5.020 .672 1.992 1.024 1.025 .770 .964 .952 .902 .715 1.103 .484

4.093*** 1.812*** 1.148*** 1.580*** .088*** 1.505*** .044*** .128* .280** .070 .097 .053 .307*** .097 .133

.328 .071 .088 .184 .019 .106 .010 .058 .097 .052 .059 .068 .060 .111 .165

6.124 3.151 4.853 1.092 4.506 1.045 1.137 1.324 1.072 1.102 .949 .736 1.102 .875

Pseudo R2 (Nagelkerke) .197. Model Chi-Square 2452.08***. * p < .05. ** p < .01. *** p < .001.

Table 6 Full multinomial model including needs-based variables (n = 12,618). Variables

Intercept Black Latino Race missing/other Age at current disposition Male Offense gravity score 1–3 Prior arrests More than 3 prior arrests New arrest since instant offense Under supervision Personal offense Drug offense Weapons offense Other offense Any dependency referral Parental substance abuse Parental criminal history Sibling arrested Prior out of home placement Occasional or past drug abuse Chronic or active drug abuse Occasional or past alcohol abuse Chronic or active alcohol abuse

Traditional

Physical regimen

B

Std. error

3.97*** 1.26*** .850*** 1.39*** .294*** .673*** .017 .061 .061 .013 .058 .119 .204** .110 .444* .041 .110 .120 .129* .588*** .434*** 1.196*** .189** .594***

.336 .079 .098 .197 .020 .094 .011 .064 .122 .057 .067 .073 .068 .119 .221 .057 .059 .069 .056 .087 .071 .069 .066 .108

Exp(B)

B

Std. error

Exp(B)

3.53 2.34 4.03 0.75 1.96 1.02 1.06 0.94 0.99 1.06 0.89 0.82 1.12 0.64 0.96 0.9 1.13 1.14 0.56 .648 .302 .828 .552

4.44*** 1.53*** 1.02*** 1.36*** .166*** 1.49*** .040*** .077 .107 .094 .058 .055 .214*** .074 .110 .091 .035 .145* .168*** .342*** .362*** 1.004*** .303*** .472***

.341 .074 .091 .193 .020 .109 .010 .061 .104 .053 .062 .070 .062 .114 .170 .054 .055 .064 .052 .071 .069 .066 .061 .087

4.62 2.79 3.9 1.18 4.46 1.04 1.08 1.11 1.1 1.06 0.95 0.81 1.08 0.9 0.91 0.97 1.16 1.18 1.41 .696 .366 .739 .624

Pseudo R2 (Nagelkerke) 0.253. Model Chi-Square 3196.12. * p < .05. ** p < .01. *** p < .001.

If these effects are not explained away by legal factors, it may still be that race and gender serve as a proxy for some other underlying familial or needs-based consideration. Table 6 presents the model including our measures of needs-based factors often considered by the juvenile court at the dispositional stage. Importantly, when controlling for these additional factors in the model, the race and gender effects remain large in magnitude and statistically significant. Although not all extra-legal factors considered achieved statistical significance, those predictive of a commitment to a physical regimen program were having a parent with a criminal past (15% increase) having a sibling with a prior arrest

136

J.J. Fader et al. / Social Science Research 44 (2014) 126–140

(18% increase), and having a prior out-of-home placement (41% increase). Those factors that reduced the likelihood of receiving commitment to a physical regimen program were histories of occasional or past drug abuse (31%), chronic or active drug abuse (63%), occasional or past alcohol abuse (26%) and chronic or active alcohol abuse (38%). Sibling (but not parental) criminal history was also influential in the decision between therapeutic and traditional reform school settings, increasing the likelihood by 14%. Similarly, histories of occasional or past drug abuse (35%), chronic or active drug abuse (70%), occasional or past alcohol abuse (17%), chronic or active alcohol abuse (45%), and prior out-of-home placement (44%) all served to further decrease the odds of receiving a commitment to this modality over a therapeutic program. As noted, prior research suggests the possibility of interactive effects between race, gender and age (Steffensmeier et al., 1993) and several theories of racial imagery and threat suggest that race may also interact with type of crime (particularly drug offenses) to indicate a stereotypically ‘‘dangerous’’ offender (Steen et al., 2005). In specific, we anticipated a possible interaction effect between type of offense as well as presenting needs-based factors. To explore these possibilities, separate models were fitted for white, black and Latino youth. For ease of presentation, the results are presented in Tables 7 and 8 with Table 7 presenting the results comparing therapeutic programs to physical regimen programs and Table 8 providing the comparison of therapeutic programs to traditional programs. As shown in Table 7 that compares therapeutic settings to physical regimen programs, we predicted that commission of a type of crime such that minority offenders convicted of a violent offense would be even more likely to receive commitment to a physical program. While we do not find support for this hypothesis, rather we find that interaction exists for weapons related offenses. For white youth having been adjudicated of a weapons offense actually decreases the odds of being placed in a physical regimen program, while it increases the chances for both black (odds ratio 1.305) and Latino (odds ratio 1.416) youth. Another interesting interaction that was not hypothesized arises among Latino youth with a prior arrest record. For white and black youth the prior record measures appear to work in the anticipated direction increasing the chances of receiving a traditional placement over a therapeutic program. However, for Latino youth, oddly, prior record appears to work in the opposite direction, significantly decreasing the chances of a traditional placement. As shown in Table 8, the most significant difference in the models for selecting a traditional rather than a therapeutic program appears to be in the impact of having committed a drug related offense. For all racial categories, having been adjudicated for a drug offense reduces the odds of being placed in a traditional setting rather than a therapeutic setting. However, this reduction is much greater for white youth (odds ratio .52 indicating a reduction of 48%) than for either black youth (odds ratio of .86 indicating a 14% reduction) or Latino youth (odds ratio of .84 indicating a 16% reduction). This difference is statistically significant when comparing black to white youth and approaches significance for Latino youth as compared to white youth. The same scenario arises when looking at a youth history of drug abuse. Again, for all racial categories such a presenting history reduces the chances of being placed in a traditional setting thus indicating that these youth are more

Table 7 Models predicting commitments to physical discipline programs for white, black, and latino youth. Variables

White N = 1538 B

Intercept Age Male Offense gravity score 1 to 3 prior arrests More than 3 prior arrests Arrest between current/ disposition Under supervision Personal offense Drug offense Weapon offense Other offense Referral Mother/father drug or alcohol abuse history Mother/father criminal history Sibling arrested Prior out of home placement History of drug abuse Chronic drug abuse History alcohol abuse Chronic alcohol abuse

Black N = 9080

Z-Score

Z-Score

Exp(B)

B

Std. error

Exp(B)

White/ Black

White/ Latino

.398 .024 .113 .014 .078 .149 .069

194.610 0.745 1.667 1.010 1.081 1.050 0.980

3.202 0.226 1.150 0.035 0.150 0.828 0.229

0.845 0.051 0.271 0.030 0.167 0.318 0.143

24.582 0.798 3.158 1.036 0.861 0.437 0.795

0.846 0.469 1.267 0.357 1.236 0.822 1.135

0.963 1.281 0.697 0.437 1.919 2.509 1.885

0.109 0.036 0.136 0.266 0.098 0.069 0.142

.082 .094 .082 .151 .271 .069 .071

1.115 0.965 0.873 1.305 0.907 0.933 0.868

0.015 0.156 0.178 0.348 1.757 0.033 0.010

0.169 0.210 0.164 0.324 0.779 0.151 0.164

0.985 0.856 0.837 1.416 0.173 1.034 1.010

1.059 1.742 1.340 2.941 0.952 1.184 0.158

0.382 0.827 1.008 2.469 0.977 0.497 0.781

1.285

0.084

.081

1.088

0.141

0.172

1.151

0.736

0.403

1.412 0.483 0.485 0.238 0.675 0.448

0.086 0.606 0.405 1.164 0.149 0.554

.068 0.105 .084 .082 .083 .150

1.090 0.546 0.667 0.312 0.862 0.575

0.196 0.613 0.187 0.970 0.243 0.714

0.138 0.221 0.198 0.190 0.158 0.232

1.217 0.542 0.829 0.379 0.784 0.490

1.422 0.392 1.410 1.249 1.272 0.876

0.683 0.313 1.861 1.682 0.640 0.264

Std. error

Exp(B)

B

4.414 .324 .883 .019 .317 .364 .184

.932 .057 .271 .021 .177 .353 .166

82.599 0.723 2.418 1.019 1.373 1.439 1.202

5.271 0.295 0.511 0.010 0.078 0.049 0.020

.113 .382 .462 .780 .767 .145 .170

.193 .175 .229 .322 .648 .167 .162

0.893 0.682 0.630 0.458 0.464 1.156 0.844

.212 .169 .290 .210 .200 .173 .240

.251 

.345 .727 .724 1.434 .393 .802

2

2

Z-score is calculated using Z  (b1  b2)/sqrt(SEb1 þ SEb2 ).

Std. error

Latino N = 1744

137

J.J. Fader et al. / Social Science Research 44 (2014) 126–140 Table 8 Models predicting commitments to traditional programs for white, black, and latino youth. Variables

Intercept Age Male Offense gravity score 1 to 3 prior arrests More than 3 prior arrests Arrest between current/ disposition Under supervision Personal offense Drug offense Weapon offense Other offense Referral Mother/father drug Mother/father criminal history Sibling arrested Prior out of home placement History of drug abuse Chronic drug abuse History alcohol abuse Chronic alcohol abuse

White N = 1538

Black N = 9080

B

Std. error

Exp(B)

B

2.974** 0.079 1.678*** 0.025 0.282 0.614* 0.065

.972 0.056 0.361 0.020 0.167 0.273 0.157

0.051 1.082 5.355 1.025 1.326 1.848 1.067

3.242*** .182*** 1.398*** .040** .138 .233 .059

0.134 0.188 0.649** 0.388 0.543 0.164 0.039 0.231

0.172 0.264 0.211 0.285 0.441 0.162 0.148 0.195

0.875 0.829 0.523 0.678 0.581 0.849 1.040 1.260

.095 .012 .149* .184 .093 .099 .073 .099

0.425** 0.650***

0.154 0.296

1.530 1.916

.166 .250**

0.801*** 1.232*** 0.369* 0.635***

0.212 0.193 0.161 0.200

0.449 0.292 0.691 0.530

.279*** .959*** .270*** .428***

2

Std. error

Latino N = 1744

Z-score

Z-score

White/ Black

White/ Latino

0.085 1.122 4.918 1.035 0.705 0.446 1.091

0.11 0.73 0.61 0.79 1.26 0.04

0.04 0.18 0.26 2.74*** 3.83*** 0.11

0.153 0.374 0.202 0.152 0.308 0.138 0.147 0.155

1.090 0.834 0.839 1.231 0.733 0.991 1.003 1.307

1.22 0.72 2.23** 1.79 1.28 0.37 0.69 0.63

0.96 0.01 1.62 1.85 0.43 0.73 0.17 0.15

0.196 0.482**

0.125 0.168

1.217 1.619

1.55 1.30

0.465* 0.981*** 0.332* 0.523**

0.192 0.179 0.145 0.187

0.628 0.375 0.717 0.593

2.30** 1.31 0.55 0.89

Exp(B)

B

Std. error

Exp(B)

.400 .024 .129 .014 .074 .130 .065

0.04 1.20 4.05 1.04 1.15 1.26 1.06

2.47*** 0.115* 1.593*** 0.034 0.349* 0.807*** 0.087

0.833 0.049 0.298 0.029 0.158 0.251 0.128

.076 .090 .076 .145 .227 .065 .066 .077

1.10 1.01 0.86 1.20 1.10 0.91 0.93 1.10

0.086 0.182 0.176 0.208 0.311 0.009 0.003 0.268

.064 0.088

1.18 1.28

.082 .078 .077 .121

0.76 0.38 0.76 0.65

1.15 0.49 1.17 0.95 0.17 0.41

2

Z-score is calculated using Z  (b1  b2)/sqrt(SEb1 þ SEb2 ).

likely to be evaluated as in need of therapeutic services. However, again, this impact is significantly greater for white youth (odds ratio of .45 indicating a reduction of 55%) than for black youth (odds ratio of .76 indicating a reduction of 24%). Combined, these findings seem to indicate that white youth with a presenting drug problem are more likely to be deemed appropriate for the therapeutic treatment programs than either their black or Latino counterparts. Also similar to the previous model, there is an interesting interaction regarding prior record, the presence of which increases the odds of physical regimen programming for white and black youth and decreases it significantly for Latino youth. These models reveal one other interesting finding. That is, the traditional legal, extralegal factors, and needs-based factors do a better job at explaining the outcomes of white youth than of either black or Latino youth. For example, the model involving only white youth, has an pseudo R2 value of 24.6%. This decreases to 16.7% for black youth and to only 12.3% for Latino youth. While none of these R2 is particularly impressive in our ability to explain the variance in dispositions, it is interesting that these traditional factors work better when assessing the outcomes of white (majority) youth than for minority populations.

5. Discussion The present study contributes to the burgeoning literature on the contextual effects of race upon sentencing decisions by moving beyond a simple dichotomy of in-community versus residential placement decisions for adjudicated youths. Other research has found that race effects are more difficult to find in the later phases of the court process because disparities are often cumulative and begin with earlier decisions (Bishop and Frazier, 1996; Rodriguez, 2010; Wu et al., 1997). Our use of two unique datasets that distinguish between commitment decisions to traditional reform schools, therapeutic facilities, and programs emphasizing physical regimen allows us to discover whether disparities may also be buried within the process of matching youth to residential programs. Our findings suggest that race is a strong predictor of commitment to facilities using different modalities, with black youth most likely to be sent to programs emphasizing physical regimen, such as boot camps or wilderness programs, and white youth most likely to be committed by the court to therapeutic programs, such as drug and alcohol or mental health treatment facilities. Decision outcomes for Latino youth are most similar to those for black youth, but they are substantially less likely to be sent to physical regimen programs and more likely to be committed to therapeutic programs than their black counterparts. The predictive power of race and ethnicity, as well as other demographic characteristics such as gender and age, remain significant and strong after controlling for both legal factors and social or familial (i.e., needs-based) characteristics. This finding provides support for Hypothesis 1 and the notion that court actors in this jurisdiction do indeed view minority youths differently from white youths in key ways that impact their placement decisions.

138

J.J. Fader et al. / Social Science Research 44 (2014) 126–140

Primarily, we suggested that, therapeutic programs and physical regimen oriented programs represent two poles of juvenile justice philosophy. Therapeutic treatment programs and programs mixing dependent and delinquent populations in a cottage setting epitomize the traditional parens patriae stance of the court, whereby young offenders are seen as presenting needs that are the root cause of their delinquent behavior. They are thereby subject to diminished criminal responsibility for their offenses. By contrast, boot camp and wilderness programs, which assume that delinquency can be ameliorated through a demanding physical regimen, embody the more modern rhetoric the accompanied the ‘‘get tough’’ movement of the 1990s. In these programs, the blame for delinquency is shifted from factors external to the young person’s control and placed firmly upon the individual’s poor decision making. Our findings that white youth are more likely to receive dispositions to therapeutic modalities and black youth to physical regimen programs thus suggests that white youth are seen as youth as less blameworthy for their offenses and responsive to treatments while black, and to a less extent Latino, youths are more culpable for their behaviors and deserving of placements more similar to viewing them as small adult criminals. While this effect did not interact with offense type as predicted (with minority youth convicted of violent offenses having an elevated chance of commitment to a physical program) it did interact with offense type in two interesting ways. First, black youth convicted of an offense involving a weapon were indeed more likely to receive commitment to the physical regimen programs, which suggests that perhaps this type of offense, rather than just a typical offense against a person, signifies blameworthiness or heightened risk. There was also an interesting effect for drug offenses. This type of offense led to a greater likelihood of therapeutic programming across all races; however, this effect was muted for black and Latino offenders. That is, it mattered less. The other major finding in this study is the importance of needs-based factors in the determination of whether to place youth in different types of residential programs. When we added factors such as prior out-of-home placement, history of sibling or parent arrest, and history of substance abuse, many of the previously-significant legal predictors of commitment decisions disappeared. This would suggest that the general philosophy of the court at this stage of the decision making process is consistent with parens patriae, although these factors are considered simultaneously with demographic characteristics such as race, gender, and age. Particularly encouraging perhaps to those who retain a rehabilitative philosophy is that a juvenile’s history of drug and alcohol abuse increases the likelihood of a receiving a commitment to a therapeutic facility even when controlling for legally relevant factors. Thus, other factors such as offense severity or prior record do not negate the historical focus of dispositional decisions in the ‘‘best interests of the child.’’ Again, however, this effect was strongest for white youth and less significant for black and Latino youth. Thus, while there is still some consideration of these needs-based factors for minority youth, it is not given equal weight as for majority youth. On the other hand, the predictive role of parent or sibling arrest history in increasing the probability of a commitment to a program emphasizing physical regimen may speak to the question of reformability. When young people have these family characteristics, all else being equal, decision makers may again be more likely to assign blame for actions to the individual or to label the youth and family as perhaps beyond reform. Moreover, presenting a history of out-of-home placement (which may be a proxy of adjustment to prior residential programs) increases the odds of being committed to a program emphasizing physical regimen over programs that are therapeutic in nature, but decreases the odds of being sent to traditional reform school instead of a therapeutic program. The first condition may speak to the decision-maker’s assessment of the malleability of the youth (e.g. he/she did not change last time in treatment) while the second may influence the actor’s assessment of the potential role of the family in a rehabilitative program. While these considerations of family variables may be considered unwarranted disparity by some, they were and continue to be a unique feature of the juvenile court. Most simply, unlike adult offenders who have choices regarding residence after release from an institution, juveniles must either return home or become a ward of the state and placed through the foster care system. Therefore, those working in the juvenile system continue to take seriously the role of the family in instigating and promulgating delinquency. A few limitations of this study must be considered. First, with the small proportion of females in the study (less than 7% of the sample), our generalizations about the role of gender in decision making are necessarily less strong than our conclusions about that of race and ethnicity. Residential facilities are segregated by gender, with a limited number of options available to decision makers because of the significantly smaller number of programs that accept girls. Second, we are forced to follow the court’s classification of youths into black, white and Latino categories, conflating race and ethnicity. Ideally, we would be able, as the Census now does, to classify young people as black Latino or non-black Latino. Third, while there was high researcher concordance on the classification of programs into categories, as noted above, these categories suggest but do not equal punitiveness. That is, while a therapeutic facility would traditionally be seen as helping the youth and a facility emphasizing physical regimen as perhaps providing more punishment, there are multiple characteristics of these programs that might impact dispositional decisions. For example, one large institution that we classify in the physical regimen category has a strong athletics program that draws college recruiters, providing an opportunity to disadvantaged youth (as most in our sample are) that is unlikely available otherwise. Moreover, the boot camps in this study are short-term programs lasting 30–90 days, whereas the average length of stay at most substance abuse treatment programs is substantially longer since clients must demonstrate therapeutic progress in order to earn release. Another practical sentencing consideration might be distance from home and ability of parents to travel to visit children. Since family preservation is a mission of the Pennsylvania juvenile justice system, if race is in some way related to ability of parents to visit children in various program locations, this would confound the observed relationship between race and program type. Or perhaps, put more simply, ‘‘race’’ would merely be a proxy for such underlying but unobserved variables.

J.J. Fader et al. / Social Science Research 44 (2014) 126–140

139

Finally, we note that our sample is limited to one jurisdiction in a single state. Thus, those characteristics that are unique to this jurisdiction may limit the generalizability of our findings to other locations, particularly those that may vary significantly in racial composition and political ideologies. Despite these limitations, we believe the strength of our findings suggest that race continues to serve a prominent role in dispositional decisions. Our findings reveal that beyond a mere ‘‘in/out’’ or ‘‘length of treatment’’ variable it is important to also consider the type of setting to which a youthful offender is sentenced. Finally, despite the nation’s recent get-tough movement which transformed the nature of juvenile justice, at least on the books, our findings suggest that the more historical notions of needs-based, or ‘‘child-saving,’’ considerations still exert influence in determining dispositions.

References Allison, Paul D., 1991. Logistic Regression Using the SAS System: Theory and Application. SAS Institute, Cary, NC. Beger, Randall R., Hoffman, Harry, 1998. The role of gender in detention dispositioning of juvenile probation violators. J. Crime Just. 21, 173–188. Belknap, Joanne, 2001. The Invisible Woman: Gender, Crime, and Justice. Wadsworth/Thomson Learning, Belmont, CA. Bennett, William J., DiIulio, John J., Walters, John P., 1996. Body Count: Moral Poverty and How to Win America’s War Against Crime and Drugs. Simon & Schuster, New York. Bernard, Thomas J., Kurlychek, Megan C., 2010. The Cycle of Juvenile Justice. Oxford University Press, New York. Bishop, Donna M., Frazier, Charles E., 1996. Race effects in juvenile justice decision-making: findings of a statewide analysis. J. Crim. Law Criminol. 86, 392– 414. Bishop, Donna M., Leiber, Michael, Johnson, Joseph, 2010. Contexts of decision making in the juvenile justice system: an organizational approach to understanding minority overrepresentation. Youth Violence Juvenile Just. 8, 213–233. Bridges, George S., Steen, Sara, 1998. Racial disparities in official assessments of juvenile offenders: attributional stereotypes as mediating mechanisms. Am. Sociol. Rev. 63, 554–570. Campbell, Mary Ann, Schmidt, Fred, 2000. Comparison of mental health and legal factors in the disposition outcome of young offenders. Crim. Just. Behav. 27, 688–715. Carroll, John S., Payne, John W., 1976. The psychology of the parole decision process: a joint application of attribution theory and information processing psychology. In: Carroll, J.S., Payne, J.W. (Eds.), Cognition and Social Behavior. Erlbaum, Hillsdale, NJ, pp. 13–32. Chesney-Lind, Meda, Shelden, Randall G., 1992. Girls, Delinquency, and Juvenile Justice. Wast/Wadsworth, Belmont, CA. Feld, Barry, 1995. The social context of juvenile justice administration: racial disparities in an urban juvenile court. In: Leonard, K.K., Pope, C.E., Feyerherm, W.H. (Eds.), Minorities in Juvenile Justice. Sage, Thousand Oaks, CA, pp. 66–97. Feld, Barry, 1999. Bad Kids: Race and the Transformation of the Juvenile Court. Oxford University Press, New York, NY. Frazier, Charles, Bishop, Donna, 1995. Reflections on race effects in juvenile justice. In: Kempf-Leonard, K., Pope, C., Feyerherm, W. (Eds.), Minorities in Juvenile Justice. Sage, Thousand Oaks, CA, pp. 16–26. Graham, Sarah, Lowery, Brian S., 2004. Priming unconscious racial stereotypes about adolescent offenders. Law Hum. Behav. 28, 483–504. Grossberg, Michael, 2002. Changing conceptions of child welfare in the United States, 1820–1935. In: Rosenheim, M.K., Zimring, F.E., Tanenhaus, D.S., Dohrn, B. (Eds.), A Century of Juvenile Justice. The University of Chicago, Chicago, IL, pp. 3–41. Hawkins, Darnell F., 1981. Causal attribution and punishment for crime. Deviant Behav. 2, 207–230. Horowitz, Allan, Wasserman, Michael, 1980. Formal rationality, substantive justice, and discrimination: a study of a juvenile court. Law Hum. Behav. 4, 103– 115. Kingsnorth, Rodney, Rizzo, Louis, 1979. Decision-making in the criminal courts: continuities and discontinuities. Criminology 17, 3–14. Leiber, Michael J., 2002. Disproportionate Minority Confinement (DMC) of youth: an analysis of state and federal efforts to address the issue. Crime Delinq. 48, 3–45. Leiber, Michael J., Mack, Kristin Y., 2003. The Individual and joint effects of race, gender, and family status on juvenile justice decision-making. J. Res. Crime Delinq. 40, 34–70. Leiber, Michael J., Stairs, Jayne M., 1999. Race, contexts, and the use of intake diversion. J. Res. Crime Delinq. 36, 56–86. Mears, Daniel P., Field, Samuel H., 2000. Theorizing sanctioning in a criminalized juvenile court. Criminology 38, 983–1019. Paternoster, Raymond, Brame, Robert, Piquero, Alexd, 1998. Using the correct statistical test for the equality of regression coefficients. Criminology 36 (4), 859–866. Peterson, Ruth D., Hagan, John, 1984. Changing conceptions of race. Toward an account of anomalous findings of sentencing research. Am. Sociol. Rev. 49, 56–70. Platt, Anthony M., 1969. The Child Savers: The Invention of Delinquency. University of Chicago Press, Chicago. Poe-Yamagata, Eileen, Jones, Michael A., 2000. And Justice for Some. Youth Law Center, Washington, DC. Pope, Carl E., Feyerherm, William, 1990. Minorities and the Juvenile Justice System. Office of Juvenile Justice and Delinquency Prevention. U.S. Department of Justice, Washington, DC. Pope, Carl E., Lovell, Rick, Hsia, Heidi M., 2002. Disproportionate Minority Confinement: A Review of The Research Literature from 1989 Through 2001. Juvenile Justice Bulletin, NCJ 198428. U.S. Department of Justice, Office of Juvenile Justice and Delinquency Prevention, Washington, DC. Rodriguez, Nancy, 2010. The cumulative effect of race and ethnicity in juvenile court outcomes and why preadjudication detention matters. J. Res. Crime Delinq., 1–23. Sampson, Robert J., Laub, John H., 1993. Structural variations in juvenile court processing: inequality, the underclass, and social control. Law Soc. Rev. 27, 285–312. Sanborn Jr., Joseph B., 1996. Factors perceived to affect delinquent dispositions in juvenile court: putting the sentencing decision into context. Crime Delinq. 42, 99–113. Schneider, Eric C., 1993. In the Web of Class: Delinquents and Reformers in Boston, 1810s–1930s. New York University Press, New York. Sealock, Miriam D., Simpson, Sally S., 1998. Unraveling bias in arrest decisions: the role of juvenile offender type-scripts. Just. Quart. 15, 427–457. Spohn, Cassia C., 2002. How Do Judges Decide? The Search for Fairness and Justice in Punishment. Sage Publications, Thousand Oaks, CA. Spohn, Cassia, DeLone, Miriam, 2000. When does race matter? An analysis of the conditions under which race affects sentence severity. Sociol. Crime, Law Deviance 2, 3–37. Steen, Sara, Engen, Rodney L., Gainey, Randy R., 2005. Images of danger and culpability: racial stereotyping, case processing, and criminal sentencing. Criminology 43, 435–468. Steffensmeier, Darrell, Demuth, Stephen, 2006. Does gender modify the effects of race-ethnicity on criminal sanctioning? Sentences for male and female white, black, and hispanic defendants. J. Quant. Criminol. 22, 241–261. Steffensmeier, Darrell, Kramer, John, Streifel, Cathy, 1993. Gender and imprisonment decisions. Criminology 31, 411–446. Steffensmeier, Darrell, Ulmer, Jeffery, Kramer, John, 1998. The interaction of race, gender, and age in criminal sentencing: the punishment cost of being young, black, and male. Criminology 36, 763–798. Ulmer, Jeffery T., 1997. Social Worlds of Sentencing: Court Communities under Sentencing Guidelines. State University of New York Press, Albany.

140

J.J. Fader et al. / Social Science Research 44 (2014) 126–140

Ulmer, Jeffery T., Kramer, John H., 1996. Court communities under sentencing guidelines: dilemmas of formal rationality and sentencing disparity. Criminology 34, 383–408. Van Wormer, Katherine Stuart, Bartollas, Clemons, 2000. Women and the Criminal Justice System. Allyn & Bacon Inc., Needham Heights, MA. Wordes, Madeline, Bynum, Timothy S., Corley, Charles J., 1994. Locking up youth: the impact of race on detention decisions. J. Res. Crime Delinq. 31, 149– 165. Wu, Bohsiu, 1997. The effect of race on juvenile justice processing. Juvenile Family Court J. 48, 43–51. Wu, Bohsiu, Stephen, Cernkovich, Christopher, S. Dunn, 1997. Assessing the effects of race and class on juvenile justice processing. J. Crim. Just. 25, 265–277.