Introduction

Minorities are overrepresented in US juvenile justice systems (Bishop 2005; Hsia et al. 2004; Kempf-Leonard 2007). An analysis of federal and state efforts to address disproportionate minority confinement (DMC) found that in 2001, minority youth were overrepresented in every state reviewed and at all decision points (Leiber 2002). “In fact, minorities were on average greater than 2–2.5 times their percentage of the at-risk youth population (i.e., secure detention, 2.63; secure corrections, 2.64; adult jails, 2.01; adult lockups, 2.16; transfers to adult court, 2.55; and probation, 2.03). The exception was arrests, but minority youth were still on average overrepresented (1.38)” (Leiber 2002, p. 10).

Overrepresentation means that although African Americans constituted 16% of the general population between the ages of 10–17 in 2004, they comprised 39.1% of the youth detained, 35.9% of those handled formally in the juvenile courts, 33% of the youth adjudicated delinquent, 38.3% of the juvenile cases resulting in out-of-home placements, and 44% of the youth transferred to adult courts in that year (Stahl et al. 2007). Caucasian youth conversely, constituted 69% of the general population between the ages of 10 to 17 in 2004, and 60.9% of the youth detained, 64.1% of those handled formally within the juvenile courts, 67% of the youth adjudicated delinquent, 61.7% of the juvenile cases resulting in out-of-home placements, and 56% of the youth transferred to the adult courts in that same year (Stahl et al. 2007).

The disparity in these figures cannot be explained by a difference in rates of crime commission. Using both official record and self-report data of “serious offenders” in Phoenix and Philadelphia, Piquero and Brame (2008) found no significant differences in juvenile crime rates by race and ethnicity. Self-report studies of delinquent behavior also challenge the arrest statistics, since the majority of self-report results do not indicate significant racial differentials (Elliot et al. 1983; Piquero and Brame 2008; Weis 1986). And even if African Americans did commit more crime proportionate to their makeup in the general population, the difference is not enough to match their arrest and confinement rates (Blumstein 1993; Joseph 1995; Huizinga and Elliot 1987; Walker et al. 1996).

Previous research suggests that a myriad of factors influence the social phenomenon of minority overrepresentation in the juvenile justice system (e.g., Bridges and Steen 1998; Hill and Atkinson 1988; Pope and Feyerherm 1990; Tollett and Close 1991). Some of these are legal factors, such as prior record, nature of the current offense, etc.; some are extralegal factors such as race, demeanor, etc.; and then there may be a combination of the two types. United States constitutional and criminal law suggests that justice should be based on legal factors, not extralegal factors. Thus, youth should be judged on their behavior, which is presumably under their control not on factors beyond their control. Moreover, the law should apply equally to all persons and not allow differential treatment based on extralegal factors.

A quarter of a century has passed since disproportionate minority confinement/contact (DMC) was identified in the juvenile justice system and concerns were raised about the role extralegal factors such as race/ethnicity play in the system. There have been three amendments to the Justice and Delinquency Prevention Act of 1974 to address DMC, and yet minorities are still overrepresented in US juvenile justice systems. Indeed, disproportionate minority contact has proven to be a very complex issue (Cabaniss et al. 2007).

Though logistic regression has previously been utilized to analyze quantitative data on this topic, no research to date has employed mixed methods and representatives from each stakeholder group to study minority overrepresentation in the juvenile justice system. Existing qualitative research on this topic has looked at only one stakeholder group’s perspective on a single decision point (Aday 1986) or has limited stakeholders to juvenile justice personnel (Frazier and Bishop 1995) or to juveniles and their families (Population and Society Research Center 1993). To develop a better understanding of the role that race plays in the overrepresentation of minorities in Virginia’s juvenile justice system, this study used both quantitative, multivariate techniques as well as in-depth interviews with juvenile judges, Commonwealth’s attorneys, defense attorneys, police officers, juveniles and juveniles’ families to address three research questions in order to develop a better understanding of the role that race plays in the overrepresentation of minorities in Virginia’s juvenile justice system

  1. 1.

    Is there a disparity in Virginia between juvenile justice processing for African American males and Caucasian males?

  2. 2.

    Is there a disparity in Virginia between juvenile justice sanctions for African American males and Caucasian males?

  3. 3.

    If a disparity exists in juvenile justice processing and sanctions for African American and Caucasian males, what role does race play in this disparity?

Methods

Quantitative Data

The quantitative data set (n = 2,920) is a disproportionate, stratified, random sample of juvenile cases from all 35 Virginia Court Service Units (CSU) where each CSU was treated as a separate stratum. These data were collected by the Joint Legislative Audit and Review Commission (JLARC) in an examination of court processing and outcomes of delinquents and status offenders in Virginia.

JLARC collected data on the juveniles’ previous felonies; previous misdemeanors; previous violations of probation/parole; previous status offenses; recent criminal charges, intake action on those charges, pre-disposition(s) of those charges, court disposition(s) of those charges; and demographics such as sex, race, data of birth, CSU, and geotype (urban, suburban, rural). For a subset of these cases, data included information from the youth’s social history, which required judicial request.

The author was granted permission to use the JLARC data and filtered it for race (Caucasian and African American only) and sex (male only) for a final sample 2,233. The small representation of females in this sample from Virginia posed the potential for a confounding effect, and most of the research nationwide has found equally small groups of youth in racial categories other than Caucasian and African American (though the group of Latinos/as is growing). This study uses Caucasian and African American labels because that is what most of the data collection instruments in Virginia use and because African American has been suggested as a preferred term by many in the African American community (Walker et al. 1996). For classification purposes, in instances where cases included multiple arrests and multiple petitions, the most recent arrest and the most severe petition (as determined by the Crime Severity Index) and sanction were used. Finally, this study’s protocol was approved by a university institutional review board.

Logistic regression was used because it presumes a nonlinear connection between the dependent variable and the independent variables based on its logarithm (Dattalo 1994, p. 124). Two logit models were therefore constructed to examine the role of various factors at two points in the juvenile justice system—one early, diversion, and one late, incarceration. Diversion is a measure of juvenile justice processing, whereas incarceration is a measure of juvenile justice sanctions. Logistic regression is well-matched to the research questions because the dependent variables were both dichotomous (1 = diverted, 0 = not diverted; 1 = incarcerated; 0 = not incarcerated) and this technique allows the researcher to describe simultaneous effects of several independent variables as well as estimate the probability of being diverted/not diverted or incarcerated/not incarcerated.

Based on previous research, the two dependent variables, diverted and incarcerated, were regressed with the following independent variables: family income, family structure, geotype (urban, suburban, rural locale), grade repetition, most recent crime committed, and number of prior misdemeanors (Bishop 2005; Frazier and Bishop 1995; Pope and Feyerherm 1990; Tollett and Close 1991).

Qualitative Data

A purposive sample (n = 36) of juvenile judges, Commonwealth’s attorneys, defense attorneys, police officers, and youth and their families was interviewed by the author in 1997. The juvenile justice personnel were from six Court Service Units across the state, including two urban, two suburban, two rural, two from Region I, two from Region II, and two from Region III. Given access and confidentiality issues and to maximize variation of the data, the youth and their families were all selected from one Court Service Unit (CSU) located in an urban geotype with a population of approximately 250,000. Participants from each CSU were chosen to provide maximum diversity in perspectives and experiences, and thus varied by race, sex, and age; and the justice personnel also varied in length of employment, educational discipline and educational attainment.

Survey instruments were developed with open- and closed-ended questions to interview the 36 stakeholders. One non-participating ‘expert’ from each stakeholder group was selected to review the face validity of the instruments prior to data collection and the instrument had two versions—one for service stakeholders and one for juveniles and their families. The interview schedules were based on those used by Frazier and Bishop (1995) in their study of the Florida Supreme Court’s Commission on Race and Ethnic Bias, and the Population and Society Research Center’s (1993) study of race and juvenile justice in Ohio. The surveys were administered in face-to-face interviews by the author.

The closed-ended responses were tabulated and the findings and conclusions reported. The open-ended responses were unitized (Lincoln and Guba 1985) and clustered into categories (Miles and Huberman 1984) in order to better understand how participants perceived the role of race in Virginia’s juvenile justice system.

Confidentiality

As the protection of human subjects is especially important with youthful offenders, the author discussed confidentiality verbally with each interview participant in addition to using a written consent form. In all cases, participants were assured that their participation was strictly voluntary, was not a waiver of their rights, was not an admission of guilt, would not affect their treatment by the courts, and that not only would their responses be kept confidential, but so would their decision to/not to participate.

Results

Univariate and bivariate quantitative analyses were run on the following variables: diversion, incarceration, family income level, grade repeated, family structure, geotype, severity of last crime committed, and number of prior misdemeanors, in order to examine the data and determine whether there were disparities between juvenile justice processes and sanctions for Caucasian and African American youth.

Eighty-six percent (n = 652) of the African American subsample (missing data excluded), compared to 63% (n = 348) of the Caucasian subsample, reported an annual family income less than $25,000. African American males were also more likely to have repeated a grade: 39.3% (n = 383) of the African American subsample had done so, compared to 28.1% (n = 353) of the Caucasian subsample. The most frequently reported family structure for Caucasian males was the two-parent family (44.2%, n = 556) while African American males mostly (42.8%, n = 417) reported mother-only families. And, whereas Caucasian families seemed to be almost evenly distributed across urban (32.5%, n = 409), suburban (33.4%, n = 420), and rural (33.6%, n = 423) locales, more African American families lived in urban locales (51.9%, n = 506), followed by rural locales (28.0%, n = 273), and then suburban locales (19.6%, n = 191). No real pattern emerged in crime commission. Finally, regarding prior record as measured by prior misdemeanors, more Caucasian young men had no prior misdemeanors (73.8%, n = 928) than African American young men (59.8%, n = 583).

Juvenile justice processing was measured by an early step in the juvenile justice system, official entry into the justice system or official diversion from the justice system. Of the Caucasians (n = 1258), 22.5% (n = 283) were diverted as compared to 15.3% (n = 149) of the African American subsample (n = 975); the majority of youth from both groups were petitioned to court: Caucasians 76.9% (967) and African Americans 84.4% (n = 823). Thus, there was only a slight disparity in processing as measured by diversion; Caucasian males were a little more likely to be diverted than were African American males (Table 1).

Table 1 Distribution of diversion by race
Full size table

Juvenile justice sanctions were measured by the final step and most severe sanction in the juvenile justice system, which is incarceration in a juvenile correctional center. Of the Caucasian subsample (n = 1258), 8.9% were incarcerated (n = 112). Of the African American subsample (n = 975), 19.4% (n = 189) were incarcerated. Again, the majority were not incarcerated. Sixty-eight percent of Caucasians (n = 855) and 64.5% of African Americans (n = 629) received other sanctions. Data were missing for 23.1% (n = 291) of the Caucasian sample and 16.1% (n = 157) of the African American sample. Although the high rates of missing data complicate interpretation, a disparity was clearly present in that African American males were more than twice as likely to be incarcerated as were Caucasian males (Table 2).

Table 2 Distribution of incarceration by race
Full size table

Multivariate Quantitative Analysis

To test for the role of race while acknowledging the influence of six other major independent variables identified in the literature, two logit models were constructed and run on the two dependent variables, diverted (diverted = 1, not diverted = 0) and incarcerated (incarcerated = 1, not incarcerated = 0). The independent variables used were: race, family income level, grade repeated, family structure, geotype, severity of the crime committed, and number of prior misdemeanors.

Severity of the crime was the only significant predictor variable (p < .001) in the diversion regression and it produced an exponentiated (β) of .9107. Crimes were coded from most severe to least severe and the logistic regression suggested that the less severe the crime committed by the youth, the greater his chance of being diverted. Overall, the model using these seven independent variables demonstrated a 93.33% accuracy rate in measuring a youth’s likelihood of being diverted (Table 3).

Table 3 Logistic regression with diversion as the dependent variable
Full size table

Four of the seven independent variables significantly increased the chances of incarceration: race, grade repeated, severity of the crime committed, and number of prior misdemeanors. With an exponentiated (β) of 1.6276 (p < .01) for race, this finding suggests that African American young men were 1.62 times as likely to be incarcerated as Caucasian young men. The other extra-legal factor which increased the likelihood of incarceration was grade repeated. Juveniles who had repeated a grade were 1.6 times as likely (exponentiated (β) = 1.6030, p < .01) to be incarcerated as those who had not repeated a grade. Both legal factors were also predictive: young men with prior misdemeanors had a 1.42 greater chance (exponentiated (β) = 1.4203, p < .001) of being incarcerated than those without prior misdemeanors and those who committed more severe crimes had a 1.04 greater chance (exponentiated (β) = 1.0363, p < .001) of being incarcerated than those who committed less severe crimes. Overall, this model demonstrated a 79.71% accuracy rate in measuring a youth’s likelihood of being incarcerated (Table 4).

Table 4 Logistic regression with incarceration as the dependent variable
Full size table

The multivariate analyses revealed that whereas the race of a young man did not predict his likelihood of diversion, African American young men were 1.62 times as likely to be incarcerated as were Caucasian males, holding the other six variables constant. Thus, being African American increased the likelihood of incarceration.

Qualitative Analysis

The juvenile justice professional sample (judges, Commonwealth’s attorneys, defense attorneys, and police officers) is comprised of 24 individuals of whom (n = 17) were male and (n = 7) were female; (n = 15) were Caucasian (n = 8) were African American, and (n = 1) was Hispanic. The average age was 39, with a range from 26 to 58; the average number of years spent in Virginia’s justice system was 10, with a range from six months to 33 years; and finally, the average percentage of time spent on delinquency case processing was 50%, with a range from 10 to 100%.

The sample of youth and their family members was comprised of all male juveniles, five mothers and one father. Four of the six families were African American and two were Caucasian. The average age of the youth was 16, with a range from 15 to 18; the average grade in school was ninth, with a range from seventh to twelfth; and the self- reported times questioned by the police ranged from 0 to 100. The average age of the parent was 41, with a range from 35 to 50; the average household size was five members with a range from two to eight; the highest grade completed by the parent ranged from 9th to high school graduate on (one parent).

Respondents were read the quantitative findings from this study and then asked whether or not their experiences and/or perceptions of the juvenile justice system were congruent with the findings. They were also asked how commonly they believed instances of racial or ethnic bias occurred in Virginia.

Responses to the open-ended questions were unitized, or reduced into units of information which will serve as the basis for defining categories (Glaser and Strauss 1967; Lincoln and Guba 1985). The units were then clustered and categorized based on the content of the units (Miles and Huberman 1984). Using the constant comparison method (Bulmer 1979), informative categories provided the basis for the qualitative findings (Miles and Huberman 1984). Because this design utilizes data units and the data were then categorized, responses were unattributable and were presented in aggregate form.

Professional respondents were much more likely to respond that race was indeed a factor on juvenile justice processing and sanctions when it was mentioned alone than when it was mentioned with other extralegal factors. When mentioned with other factors, 71% (n = 17) of professionals said that familial factors influenced a youth’s treatment, and the average rank of familial factors for all respondents was 4.47 on a scale from 1 to 5, with 5 having the greatest impact on a youth’s treatment within the system. Three respondents specifically mentioned racial/ethnic factors and the average ranking of racial/ethnic factors for all respondents was 3.66.

Interviews with stakeholders, thus, supported the quantitative finding that race played a role in a youth’s treatment in Virginia’s juvenile justice system. Juvenile justice professionals as well as youth and their families cited racial bias by individual decision-makers and by the overall system, and noted that this bias was most likely to occur by the police during the Alleged Act or Informal Handling stages. However, although race was considered a factor, when compared to other factors, professionals did not think race played a dominant role in affecting a youth’s treatment within the juvenile justice system.

To triangulate the quantitative findings with the qualitative findings (Jick 1983), the juvenile justice professionals were informed of the study’s quantitative findings, and were asked if they felt that a disparity existed in processing and sanctions for African American males and Caucasian males. Eighteen of the juvenile justice professionals stated that they felt a disparity existed, four did not feel that a disparity existed, and two indicated that they did not know.

The youth and their families were also asked to comment on the quantitative findings, and were asked to comment on their experiences and/or perceptions of the fairness of the juvenile justice system, as well. Six said that they had been treated fairly, four responded that they had not been treated fairly, one answered “Don’t Know”, and one indicated that he had been treated both fairly and unfairly. Those who noted unfair treatment, were then asked if they thought this treatment had anything to do with their income, education, race, family, person, or politics. Three indicated income, two responded education, three cited race, no one responded family, one said person, and three answered politics/political reasons. One person stated that he did not know.

For the most part, the professionals cited legal factors as the primary factors influencing a youth’s processing and sanctions in the Virginia juvenile justice system. When asked about extra-legal factors, several professionals cited family structure as a factor, though no parents or youth cited family structure and neither logistic regression suggested an effect of family structure on diversion or incarceration. No one in either group identified grade repeated as a predictor of incarceration, though there was much discussion of education in both groups and neither group felt that most juvenile offenders had an adequate education. Judges and juveniles alike agreed that education was the key to avoiding contact with the juvenile justice system. This sentiment was captured in one interviewee’s response: “The majority of court-involved youth cannot read or write sufficiently to succeed in today’s society.”

Discussion

There are limitations of this study to note. Some of the JLARC variables were weakened by missing data. Also, for the quantitative analyses, psychological variables would have added another dimension, but they had such small sample sizes and incomplete cells that even any real effects would not have withstood the regressions. For the qualitative analyses, the study was designed to include two additional stakeholder’s voices: intake officers and probation officers. The study was limited without these participants because juvenile probation officers often provide a middle ground between the state’s case—Commonwealth’s attorneys and police officers, and the youth’s case—the adolescent and his/her family, because the probation officers know the system’s expectations of the youth and are more familiar with the day to day lives of their probationers and their families. Addition of these “middle ground” voices would have been helpful. Finally, mid-way through the qualitative interviews, the respondents were specifically asked to comment on the possible effects of race in juvenile justice which could have biased their responses by drawing race to their attention. The decision to conduct race-focused research should never be made lightly.

Univariate and bivariate findings suggest that in Virginia, there are disparities between juvenile justice processing and sanctions for African American and Caucasian males. Multivariate findings suggest that a legal factor: the severity of the crime was the only variable that predicted diversion in the logit model: the less severe the crime, the more likely a youth was to be diverted. Four variables predicted incarceration, two legal and two extralegal: crime severity, prior record, race, and grade repeated: youth who had committed more severe crimes, had more prior misdemeanors, were African American, and had repeated a grade were more likely to be incarcerated than other youth.

This study was unique in that the qualitative findings triangulated and added depth to the quantitative findings. They also confirmed the disparity in processing and sanctions for African American males and Caucasian males. Three-fourths of the juvenile justice professionals said that disparity existed in processing and sanctions for African American and Caucasian males. Further, whereas most of the professional respondents (diversion n = 19; incarceration n = 24) cited the legal factors of crime severity and prior record as most predictive of diversion and incarceration, half of the youth and their parents thought the extralegal factor of race affected both diversion and incarceration. Several professionals cited family structure as a contributing factor yet no parents or youth cited family structure and neither logistic regression suggested an effect of family structure on diversion or incarceration. Although no one mentioned grade repeated as a predictor of incarceration, all respondents felt that the lack of education for most youth in the juvenile justice system impeded their success and may have contributed to their delinquency.

Implications for Social Work Practice

Social work practice intersects with juvenile justice through multiple avenues including: pre-sentence reports, intake assessments, mental health and substance abuse counseling, various intervention and therapy programs, probation and parole services, corrections, and law-making/policy-setting for offenders, victims, and their families. This study adds to our understanding of the factors affecting minority overrepresentation in the juvenile justice system.

Social workers are trained in cultural competence, diversity, and social justice issues and are well-suited to mediate the extralegal role of race in justice environs. Many also implement community-based efforts, which could take the form of supplementary, education programs. Moreover, the social work profession often advocates for evidence-based, structural change which could include processing and sanction changes that focus more on legal factors than extralegal factors.

One stakeholder in this study commented,

Washington [D.C.] talks out of both sides of their mouth, on the one hand they want us to study and fix DMC [disproportionate minority contact], on the other hand they want us to get tough on crime. Don’t they know that if we get tough on crime without making some changes first, we’ll only lock up more and more black kids?

Finally, Cabaniss et al. (2007) suggest that “communities can transcend the emotionally charged atmosphere that often envelops discussions of social injustice by reviewing data and decision point maps that clearly outline the extent of the problem” (p. 399). This research adds scientific data to those discussions.