J Child Fam Stud (2015) 24:427–433 DOI 10.1007/s10826-013-9853-8 ORIGINAL PAPER Females in the Juvenile Justice System: Influences on Delinquency and Recidivism David E. Barrett • Song Ju • Antonis Katsiyannis Dalun Zhang • Published online: 19 October 2013 Ó Springer Science+Business Media New York 2013 Abstract The role of mental health history and family dysfunction or disruption on female juvenile delinquency was examined. Using large sample archival data from a state juvenile justice agency, we examined the behavioral and demographic predictors of repeat offending for a sample of approximately 34,000 females who had been referred for criminal offenses. Then, after merging these data with those from multiple state agencies, we compared the family and mental health histories of the delinquent females with those of females from a matched control group of the same number, constructed from the records of the state department of education. Drug use, family delinquency, severity of first offenses, and age of first offending were predictors of repeat offending for the females in the delinquent sample. Compared with non-delinquent females, delinquent females were more likely to be eligible for free or reduced lunch, and were more likely to have been in foster care or child protective services. The strongest predictor of membership in the delinquent sample was a DSM-IV diagnosis of a mental health disorder related to aggression or impulse control. All variables associated with delinquency remained significant when other predictors were statistically controlled. Implications for prevention of female juvenile delinquency were addressed. D. E. Barrett (&)
A. Katsiyannis Department of Teacher Education, Clemson University, Clemson, SC 29634, USA e-mail: bdavid@clemson.edu S. Ju University of Cincinnati, Cincinnati, OH, USA D. Zhang Texas A&M University, College Station, TX, USA Keywords Female delinquency
Child maltreatment
Mental health and delinquency
Child protective services
Juvenile justice Introduction In 2009, 1,906,600 juveniles were arrested in the United States for violent crimes (85,890 murder and non-negligent manslaughter, forcible rape, robbery, and aggravated assault), property crimes (417,700 burglary, larceny-theft, motor vehicle theft, and arson), and non-index crimes (1,403,010 crimes such as other assaults, drug abuse, disorderly conduct, violation of liquor laws, status offenses). Regarding violent crimes in 2009, 47 % involved white youth, 51 % black youth, 1 % Asian youth, and 1 % American Indian youth. For property crime arrests, 64 % involved white youth, 33 % black youth, 2 % Asian youth, and 1 % American Indian youth (Puzzanchera and Adams 2011). In 2009, 578,500 females were arrested. These arrests accounted for 18 % of juvenile Violent Crime Index arrests and 38 % of juvenile Property Crime Index arrests. Of nonindex crimes, arrest rates for females were 78 % for prostitution, 55 % runaways, 42 % embezzlement, and 34 % other assaults. From 2000 through 2009, arrests of juvenile females decreased less than male arrests in offense categories such as aggravated assault, vandalism, and drug abuse violations. Female arrests increased while male arrests decreased for crimes such as simple assault, larceny-theft, and disorderly conduct. Further, whereas in 1980, the juvenile male violent crime arrest rate was 8 times greater than the female rate, in 2009 the male rate was just 4 times greater (Puzzanchera and Adams 2011). Overall females are more likely than males to be detained for status offenses (truancy, running away, underage 123 428 drinking) while males are more likely to commit more serious offenses (Boesky 2002; Puzzanchera and Adams 2011). Female delinquency follows specific patterns relating to age and context. Early puberty when paired with family conflict and neighborhoods characterized by poverty, unemployment, and single parent families is a unique risk factor for females (Zahn et al. 2010). Disorganized communities tend to exacerbate the frequency of violent acts (Burman 2003), with gang membership associated with more violent behavior among females (Zahn et al. 2008a). There is some evidence that aggressive behavior among female youth is associated with girls’ ambivalence regarding obedience to parental authority. For example, girls are more likely to fight with family members (Franke et al. 2002) than with non-family members. Data on arrests for intra-family aggressive behavior must be interpreted cautiously, however; domestic behaviors which under certain conditions might be considered ‘‘ungovernable’’ (and result in a referral for a status offense) might, in a domestic situation, result instead in an arrest for simple assault (Chesney-Lind and Sheldon 2004; Gaarder et al. 2004; Zahn et al. 2008a). Despite the increasing rates of female juvenile delinquency, particularly for crimes traditionally associated with males, addressing the needs of females in the juvenile system has been a persistent challenge (Boesky 2002; Quinn et al. 2005; Teplin et al. 2002). The American Bar Association and National Bar Association Report (2001) have concluded that there is a critical lack of prevention, diversion, and treatment alternatives for girls in the juvenile justice system. For example, not only do females tend to commit less serious crimes than males, they often receive differential treatment for similar crimes (Barrett et al. 2010; Miller et al. 1995). Consequently, many professionals argue that the majority of female delinquency cases should be diverted from court proceedings (Bishop and Frazier 1992; Quinn et al. 2005). Females are also prone to emotional, physical, or sexual abuse (Acoca and Dedel 1998; Berlinger and Elliot 2002; Teplin et al. 2002). In fact, physical assault by a parent or caregiver, sexual assault or neglect by a parent or caregiver, and neighborhood disadvantage are key risk factors for delinquency (Hawkins et al. 2009) along with family criminality, drug use, and deviance (Zahn et al. 2010). In contrast, the presence of a caring adult, school success, and religiosity have been shown to serve as protective factors (Hawkins et al. 2009). In the present study we were able to obtain detailed background information on the early experiences of over 34,000 females with records of juvenile delinquency. Using information from a state department of juvenile justice, we examined the role of selected family and demographic 123 J Child Fam Stud (2015) 24:427–433 variables in predicting female recidivism. Using information from the state’s budget and control board, we then constructed a control group of 34,000 female youth without histories of delinquency and matched on birth year and race. By linking records on delinquency with records from other state agencies, we were able to examine the influences of important child personal and experiential variables including mental health history, maltreatment and foster care on juvenile delinquency. Our study addressed two major research questions. First, among delinquent females, what are the personal and family background variables which are useful in predicting female recidivism? Second, to what extent can we predict membership in the delinquent group versus the non-delinquent control group on the basis of females’ emotional/behavioral problems and early adverse family experiences? Method Sample Data from eight cohorts of female juvenile offenders were drawn from the South Carolina Department of Juvenile Justice (SCDJJ) Management Information System. The sample included 34,414 female juvenile offenders born between 1981 and 1988, each of whom had been referred to SCDJJ on at least one occasion. The SCDJJ is a state cabinet agency which covers 43 of 46 counties in South Carolina. When a juvenile is arrested or referred by a Circuit Solicitor or a school, a SCDJJ county office will perform the family court intake and make a recommendation to the Solicitor’s Office with advisory recommendations (e.g., diversion or prosecution). The family court intake involves collecting data from parents or guardians on the child’s gender, ethnicity, and date of birth; documenting the nature of the referral offense; and performing risk and needs assessments. Data collection may also involve other social-demographic variables, including family income, family history of delinquency, child substance use, educational history, and family structure (see Barrett et al. 2010). The sample for this analysis included 34,614 female juveniles whose age at first referral ranged from 5 to 19 (M = 14.67, SD = 1.80). A small percentage (1.4 %) of juveniles who were not African Americans or Caucasian (n = 486) were excluded from this study. Demographic information is presented in Table 1. As shown in Table 1, the racial composition was 18,007 (51.3 %) African American and 16,607 (47.3 %) Caucasian. With respect to numbers of referrals, 13,300 (38.4 %) had one referral only, 13,915 (40.2 %) had two referrals, and 7,399 (21.8 %) had three or more referrals. At the statewide level, J Child Fam Stud (2015) 24:427–433 429 Table 1 Demographic characteristics of delinquent females Characteristics N % Black 18,007 52 White 16,607 48 Total N 34,614 Race Family delinquency Yes 6,001 53.4 No 5,232 46.6 Total N 11,233 Family income \$15,000 7,751 48.9 C15,000 Total N 8,096 15,847 51.1 Yes 4,396 25.1 No 13,102 74.9 Total N 17,498 Drug use history diagnosed as having disorder of aggression and/or impulse control based on DSM-IV classification); see Barrett et al. 2013b for information regarding criteria for evaluating a child as having an aggression-related disorder. Variables created from the different agency files were merged to create a new master dataset for the delinquent sample. Finally, a comparison, non-delinquent sample was randomly selected from a previously created control sample of 99,602 youth using data made available from SCDE; the non-delinquent sample was constructed to have the same proportions of birth years, gender, and ethnicities as the DJJ cohort. Additional information about the matching procedure has been reported (Barrett et al. 2013b). The same data collected from the ORS for the delinquent sample were extracted for the control group. The delinquent and non-delinquent files were merged for the purpose of statistical analyses. Data Analysis 1st referral severity Misdemeanor 8,348 86.3 Felony 1,324 13.7 Total N 9,672 SCDJJ assigns all offenses a severity rating on a scale of 1–25, with ratings less than 2 representing status offenses (e.g., truancy, running away), 2–3 representing misdemeanor offenses (e.g., simple assault and battery, criminal domestic violence), 5–8 representing nonviolent felonies (e.g., grand larceny, carrying a weapon on school grounds), and 8.5–25 representing violent felonies (e.g., assault and battery of a high and aggravated nature, sexual assault, armed robbery). For analysis purposes, we recoded severity of offenses into two levels, status offense or misdemeanor (SCDJJ ratings 1 through 3) and felony (SCDJJ ratings 5 through 25). Next, data from the DJJ were merged with data from the SC State Budget and Control Board’s Office of Research and Statistics (ORS). ORS data were collected from three different state agencies: the Department of Social Services (SCDSS), the Department of Mental Health (SCDMH), and the Department of Education (SCDE). Each child in the DJJ file has been assigned a unique ID generated through a linkage algorithm and matched ORS data were then linked to each child (see Barrett et al. 2013b for details). In this study, variables extracted from ORS files included: (1) foster care placements (i.e., whether or not a child had ever been placed in foster care); (2) maltreatment [i.e., whether or not a child had ever been placed in the custody of child protective services (CPS)]; (3) free lunch (i.e., whether a child was eligible for free and/or reduced lunch); and (4) aggressive behaviors (i.e., whether a child had been The data analysis plan included two steps. First, ‘‘proportional hazards regression’’ (also termed Cox regression) was used to examine the risk of recidivism for female offenders. This analysis technique is a type of survival/ failure analysis (Singer and Willett 2003) and has been used previously to examine the likelihood of and timing of recidivism (Zhang et al. 2011). This type of survival analysis typically examines how the risk of an adverse outcome changes over time in relation to possible covariate effects. In this study, we used this technique to predict recidivism (repeated offense) by time and by associated juvenile characteristics which have been identified from previous literature, including race, presence/absence of family history of delinquency, family income, presence/ absence of personal drug use history, offense severity at 1st referral, and age at first referral (Zhang et al. 2011; Barrett et al. 2010). Data were obtained from the original SCDJJ files (see Barrett et al. 2010). The hazards analysis was performed using the augmented Cox regression model of Lunn and McNeil (1995). The variable ‘‘number of referrals’’ was dummy coded with ‘‘one referral’’ coded as ‘‘0’’ and ‘‘multiple referrals’’ coded as ‘‘1’’; covariates were included in the model. Censoring is used to determine the termination of the period of time. In this study, censoring was created for age 21, as this ended time to juvenile offense. Second, using the merged delinquent and control files and data obtained from the ORS files, we examined differences between the behavioral, academic, and mental health characteristics of females in the delinquent sample and those in the control sample. First, we used Chi square and phi coefficient analyses to compare proportions of each group (delinquent and non-delinquent) showing 123 430 J Child Fam Stud (2015) 24:427–433 presence or absence of each indicator. Next multivariate logistic regression (Hosmer and Lemeshow 2000) was chosen to examine the variables that predicted whether a female would be in the delinquent or the non-delinquent sample. This analysis allowed us to examine the unique as well as the collective contributions of each risk variable. Examined covariates included free lunch, placement in CPS, placement in foster care and disorders of aggression. Race was also included in the model. Adjusted odds ratios and effect sizes are reported for the logistic model. Results Influences on Recidivism The results from the Cox regression analysis are presented in Table 2. Controlling the risk factor of race, all covariates with the exception of family income were significant individual predictors of recidivism. Specifically, offenders who were younger at the first referral had a higher risk of recidivism (b = -.194, p \ .001) with the chance of recidivism expected to decrease by 18 % for each additional year if all other variables are held constant. Offenders who had a drug use history were almost twice as likely to have a second offense as those not having drug use history (b = .666, p \ .001). Compared with offenders who committed more severe offenses, offenders who committed less severe offenses were 30 % more likely to have a second offense (b = -.357, p \ .001). In addition, the effect of family criminal history was also significant at p \ .001. Offenders with family criminal history background were at a high risk for recidivism. On average, it took about 1.06 (SE = .01, 95 % CI 1.05–1.09) years for female offenders to be referred again. Delinquents Compared with Non-delinquents As shown in Table 3, the two groups of juveniles (delinquent versus non-delinquent) were compared on four demographic variables, including free lunch eligibility, CPS services, foster care and disorder of aggression. Descriptive statistics and results from the Chi square test are summarized in the table. The two groups differed significantly on all four demographic variables. More than 50 % of juveniles were eligible for free lunch in both groups; however, the percentage in the delinquent group was larger than in the non-delinquent group. Overall, there were more females from the delinquent group placed in foster care (7 vs. 1.1 %), receiving CPS services (16.4 vs. 3.6 %) and diagnosed as having a disorder of aggression (12 vs. .9 %) than from the non-delinquent group. All four variables were included in the logistic regression model holding ethnicity as a constant. No significant interaction effects were identified. The predictor coefficients for the final model are presented in Table 4. As illustrated, female juveniles who had received CPS services were 3.2 times more likely to be delinquent than those who had not been placed in CPS (b = 1.171, p \ .001). Similarly, female juveniles who were placed in foster care were two times more likely to be involved in delinquency (b = .814, p \ .001). Eligibility for free lunch which partially indicated the family’s socioeconomic status also had a significant predictor effect. Female youths eligible for free or reduced lunch were almost 1.4 times more likely to become delinquent than those who were not eligible. The strongest predictor of delinquency was a mental health diagnosis related to a disorder of aggression. Female Table 3 Background characteristics for delinquent and non-delinquent groups Characteristics Delinquent group Non-delinquent group Phi coefficient .151** Free lunch eligibility Table 2 Cox proportional hazards regression with N = 34,580 Variable b SE Chi square Exp(B) 21,776 (62.9 %) 17,818 (51.5 %) No 12,838 (37.1 %) 16,796 (48.5 %) CPS services Yes No 5,693 (16.4 %) 1,236 (3.6 %) 28,921 (83.6 %) 33,378 (96.4 %) .096 .057 .094 1.100 Family delinquency .234** .056 17.510 \.001 1.264 Yes 2,439 (7 %) Drug use history .666** .056 140.835 \.001 1.947 No 32,175 (93 %) -.357** .064 31.081 \.001 .700 Age at 1st referral -.194** .015 170.801 \.001 .824 Yes 4,154 (12 %) 328 (.9 %) Family income -.027 .056 .973 No 30,460 (88 %) 34,286 (99.1 %) ** p \ .001 123 .233 .629 .124** Foster care African American Offense severity at 1st referral 2.801 p value Yes 389 (1.1 %) .078** 34,225 (98.9 %) Disorder of aggression ** p \ .001 .064** J Child Fam Stud (2015) 24:427–433 Table 4 Logistic delinquency regression 431 coefficients for prediction of Predictor R2 Free lunch eligibility .02** .358** CPS services .05** 1.171** .036 4.94** 3.226** Foster care Disorder of aggression .03** .09** .814** 2.415** .062 .059 6.828** 13.334** 2.257** 11.188** b SE .017 AORE 1.56** AORF 1.430** ** p \ .001. R2 refers to Nagelkerke’s R2 following this step in the equation and including the constant. Significance level for R2 is based on the change in the log likelihood of the outcome. Significance level for the Wald statistic is based on the final logistic regression equation. B refers to the logistic regression coefficient in the final equation. AORF refers to the adjusted odds ratio in the final equation. AORE refers to the adjusted odds ratio at the initial time of entry juveniles identified as aggressive were 11 times more likely to be delinquent (b = 2.415, p \ .001) than those without such a DSM-IV diagnosis. The SPSS analysis gave two measures of R2, which were .09 (Cox and Snell R2) and .13 (Nagelkerke’s adjusted R2). These effect sizes are reasonably similar values and represent medium effects, according to Cohen (1988) (i.e., .02 for ‘‘small’’, .13 for ‘‘medium’’ and .26 for large). Discussion The present study provides overwhelming evidence of the role of early social and psychological adversity in female delinquency. Particularly important is the impact of mental health disorders. In our study, females who had been diagnosed with a mental health disorder involving impulse control or aggression were approximately 11 times more likely to commit a criminal offense than females who had not been so diagnosed. While it is not possible to show a direct causal effect of mental illness on delinquency, it is important to recognize that in over 60 % of cases of female delinquents with a diagnosis of a disorder of aggression, the diagnosis of aggression preceded any involvement with the juvenile justice system. The second most powerful predictor of delinquency was child placement in CPS. The role of attachment problems in the development of psychopathology among females has been previously noted (Barrett et al. 2013c); in fact there is some evidence that early disruptions in parent–child relationships may have even more serious repercussions for females than for males (Benda 2002). Particularly important is the powerful influence of removal from the home, whether in CPS or foster care, even when child aggressiveness has been statistically controlled. According to current systems theories (Granic and Patterson 2011), child maltreatment (which is usually untreated or undertreated) is likely to precipitate coercive behaviors on the part of the child, behaviors that put the young female at risk for delinquent behaviors. To the extent that the pathways to delinquency may be somewhat different for boys and girls, the needs of girls in terms of prevention, treatment, and aftercare may also differ. According to Quinn et al. (2005), the juvenile justice must be prepared to address the needs of girls in a genderspecific, culturally competent manner. The overall success, however, will be dependent on the seamless service delivery of agencies such as child welfare, mental health, education (including special education), and juvenile justice (p. 137). With regard to recidivism, the present findings are largely supportive of previous research on this topic. It is interesting that the strongest predictor of female recidivism in our study was a history of drug use, particularly given the strong association between adolescent drug use and lack of parental control in the family (Lamborn et al. 1991; Steinberg 2011). Also, consistent with previous research (Barrett et al. 2010) are the findings that lower age at first referral and lower severity of first offense are predictors of repeat offending. While the latter finding might seem counter-intuitive, it may be due in part to the fact that the most common status offense is truancy and that failure to comply with mandatory attendance orders will automatically result in a referral for contempt of court. Interventions that help adolescent girls learn how to manage their risk (e.g., effectively dealing with the trauma of childhood physical and sexual assault) would be an important contribution to the delinquency prevention field (Ruffolo et al. 2004; see also Quinn et al. 2005). Additionally, interventions should focus on the protective factors that mitigate risk (Luthar 2006). Unfortunately, females not only experience higher levels of mental health problems than male peers, they are less likely to receive treatment, and more likely to abandon treatment (Caufmann 2008). In addition, there is a paucity of gender specific programs that have empirical support to address prevention and treatment-related challenges (Quinn et al. 2005). According to the General Accounting Office (2009), in 2004 the Office of Juvenile Justice and Delinquency Prevention (OJJDP) established a girls Study Group, funded by a $2.4 million multiyear contract with a research institute, to identify ‘‘effective or promising programs, program elements, and implementation principles’’ (p. 3). The Study Group reported that 44 out of 61 girls’ delinquency programs had no empirical support and that research findings on the remaining programs failed to establish evidence of their effectiveness in preventing or reducing girls’ delinquency (see also, Zahn et al. 2008b). The failure to address the problem of female delinquency 123 432 continues to have serious repercussions; particularly given the strong links between female delinquency and school failure, teen pregnancy and child bearing and later mental health problems (Barrett et al. 2013a). There are several limitations to the present study which may limit the generalizability of the findings. First, because the sample is drawn from one state, South Carolina, the sample may not be representative of a national sample. State agencies follow specific procedures for determining the need for CPS, referring children for mental health services, and referring cases to the criminal justice system; these procedures may differ from state to state (see for example U.S. Department of Health and Human Services 2003). Thus, the empirical relationships identified in this study between specific risk factors for delinquency and later delinquent behavior might be magnified or attenuated were the study to be conducted in other settings. A second limitation is that information regarding the background variables collected at intake on delinquent youth (including drug use, income, and criminal history) were not available for the control sample. Thus we were not able to include these variables in the logistic regression analysis in which we predicted membership in the delinquency group. Inclusion of these variables might have altered the magnitude of the coefficients reported for this analysis. Third, because of the very small number of females in the sample whose ethnicity was other than African-American or Caucasian, only these ethnic groups were included in the analyses. Generalizations to individuals from other population groups must be made cautiously. Finally, we emphasize that in the present study, in our examination of family-related variables related to delinquency, we included only the most extreme indicators of family dysfunction or disruption (CPS and foster care); closer examination of the influences of parent–child interaction (trust, consistency, coercive behavior) would be useful in helping us better understand how early social experiences contribute to later female delinquent behavior. With juvenile delinquency among females becoming more pervasive in the United States, the task of identifying the school, home and psychological factors that lead to female antisocial behavior has become more urgent. In the present study we have elucidated the role of early family disruption, and in particular removal from the home, in female delinquency. Also, consistent with research on male delinquents, the previous existence of mental health problems appears to be the strongest predictor of female delinquency, and, as indicated by the relationship between drug use and repeat offending, a risk factor for later offending. While research on effective programing for females at risk for delinquent behavior has been limited, the present study suggests that multi-systemic programs— involving schools, communities, families and correctional 123 J Child Fam Stud (2015) 24:427–433 settings—that address the young female’s need for consistent and nurturing relationships and that provide models for academic and social success may be most helpful in reducing, if not preventing, female delinquency. References Acoca, L., & Dedel, K. (1998). No place to hide: Understanding and meeting the needs of girls in the California Juvenile Justice System. San Francisco: National Council on Crime and Delinquency. American Bar Association and the National Bar Association. (2001). Justice by gender: The lack of appropriate prevention, diversion and treatment alternatives for girls in the justice system. Washington, DC: Author. Barrett, D. E., Katsiyannis, A., & Zhang, D. (2010). Predictors of offense severity, adjudication, incarceration and repeat referrals for juvenile offenders: A multi-cohort replication study. Remedial and Special Education, 31, 261–275. Barrett, D. E., Katsiyannis, A., & Zhang, D. (2013a). Predictors of teen childbearing among delinquent and non-delinquent females. Manuscript submitted for publication. Barrett, D. E., Katsiyannis, A., Zhang, D., & Zhang, D. (2013b). Delinquency and recidivism: A multi-cohort, matched -control study of the role of early adverse experiences, mental health problems and disabilities. Journal of Emotional and Behavioral Disorders,. doi:10.1177/1063426612470515. Barrett, D. E., Zhang, D., Wang, Q., Katsiyannis, A., & Zhang, D. (2013c). A structural equation modeling analysis of influences on juvenile delinquency. Manuscript submitted for publication. Benda, B. B. (2002). A test of three competing theoretical models of delinquency using structural equation modeling. Journal of Social Service Research, 29, 55–91. Berlinger, L., & Elliot, D. M. (2002). Sexual abuse of children. In The APSAC, Handbook on child maltreatment (2nd ed., pp. 55–78). Thousand Oaks, CA: Sage. Bishop, D. M., & Frazier, C. E. (1992). Gender bias in juvenile justice processing: Implications of the JJDP Act. The Journal of Criminal Law and Criminology, 82(4), 1162–1186. Boesky, L. M. (2002). Juvenile offenders with mental health disorders: Who are they and what do we do with them?. Lanham, MD: American Correctional Association. Burman, M. (2003). Challenging conceptions of violence: A view from the girls. Sociology Review, 13(4), 2–6. Caufmann, E. (2008). Understanding the female offender. The Future of Children, 18(2), 119–142. Chesney-Lind, M., & Sheldon, R. (2004). Girls, delinquency, and juvenile justice (2nd ed.). Belmont, CA: Thompson/Wadsworth. Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum. Franke, T. M., Huynh-Hohnbaum, A. L., & Chung, Y. (2002). Adolescent violence: With whom they fight and where. Journal of Ethnic & Cultural Diversity in Social Work, 11, 133–158. Gaarder, E., Rodriguez, N., & Zatz, M. S. (2004). Criers, liars, and manipulators: Probation officers’ views of girls. Justice Quarterly, 21, 547–578. Granic, I., & Patterson, G. R. (2011). Toward a comprehensive model of antisocial development: A dynamic systems approach. Psychological Review, 11, 101–131. Hawkins, S. T., Graham, P. W., Williams, J., & Zahn, M. A. (2009). Resilient girls: Factors that protect against delinquency. Office of Juvenile Justice and Delinquency Prevention. Retrieved from http://girlsstudygroup.rti.org/docs/OJJDP_GSG_Resilience_ Bulletin.pdf. J Child Fam Stud (2015) 24:427–433 Hosmer, D. W., & Lemeshow, S. (2000). Applied logistic regression (2nd ed.). New York: John Wiley and Sons Inc. Lamborn, S. D., Mounts, N. S., Steinberg, L., & Dornbusch, S. M. (1991). Patterns of competence and adjustment among adolescents from authoritative, authoritarian, indulgent and neglectful families. Child Development, 62, 1049–1066. Lunn, M., & McNeil, D. (1995). Applying Cox regression to competing risks. Biometrics, 51, 524–532. Luthar, S. S. (2006). Resilience in development: A synthesis of research across five decades. In S. D. Cicchetti & D. J. Cohen (Eds.), Developmental psychopathology, Vol. 3: Risk, disorder, and adaptation (2nd ed., pp. 739–795). Hoboken, NJ: Wiley. Miller, D., Trapani, C., Fejes-Mendoza, K., Eggleston, C., & Dwiggins, D. (1995). Adolescent female offenders: Unique considerations. Adolescence, 30(118), 429–435. Puzzanchera, C., & Adams, B. (2011). Juvenile arrests 2009. Office of Juvenile Justice and Delinquency Prevention. Retrieved from http://www.ojjdp.gov/pubs/236477.pdf. Quinn, M. M., Poirier, J. M., & Garfinkel, L. (2005). Girls with mental health needs in the juvenile justice system: Challenges and inequities confronting a vulnerable population. Exceptionality, 13(2), 125–139. Ruffolo, M. C., Sarri, R., & Good-kind, S. (2004). Study of delinquent, diverted, and high-risk adolescent girls: Implications for mental health intervention. Social Work Research, 28(4), 237–245. Singer, J. D., & Willett, J. B. (2003). Applied longitudinal data analysis: Modeling change and event occurrence. Oxford, UK: Oxford University Press. 433 Steinberg, L. (2011). Adolescence (9th ed.). Boston: McGraw-Hill. Teplin, L. A., Abram, K. M., McClelland, G. M., Dulcan, M. K., & Mericle, A. A. (2002). Psychiatric disorders in youth in juvenile detention. Archives of General Psychiatry, 59, 1133–1143. U.S. Department of Health and Human Services and Administration for Children and Families. (2003). National systems of child protective service systems and reform efforts. Retrieved August 14, 2013, from http://aspe.hhs.gov/hsp/cps-status03/state-policy03/index.htm. Zahn, M. A., Agnew, R., Fishbein, D., Miller, S., Winn, D., Dakoff, G., et al. (2010). Causes and correlates of girls’ delinquency. Office of Juvenile Justice and Delinquency Prevention Retrieved form https://www.ncjrs.gov/pdffiles1/ojjdp/226358.pdf. Zahn, M. A., Brumbaugh, S., Steffensmeier, D., Feld, B. C., Morash, M., Chesney-Lind, M., et al. (2008). Violence by teenage girls: Trends and context. Washington, DC: U.S. Department of Justice, Office of Justice Programs, Office of Juvenile Justice and Delinquency Prevention. Retrieved from https://www.ncjrs.gov/ pdffiles1/ojjdp/218905.pdf. Zahn, M. A., Hawkins, S. R., Chiancone, J., & Whitworth, A. (2008). The girls study group—Charting the way to delinquency prevention for girls Washington, DC: U.S. Department of Justice, Office of Justice Programs, Office of Juvenile Justice and Delinquency Prevention. Retrieved from https://www.ncjrs. gov/pdffiles1/ojjdp/223434.pdf. Zhang, D., Barrett, D. E., Katsiyannis, A., & Yoon, M. (2011). Juvenile offenders with and without disabilities: Risks and patterns of recidivism. Learning and Individual Differences, 21, 12–18. 123 Copyright of Journal of Child & Family Studies is the property of Springer Science & Business Media B.V. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. Juvenile Delinquency Recidivism: Are Black and White Youth Vulnerable to the Same Risk Factors? David E. Barrett and Antonis Katsiyannis Eugene T. Moore School of Education, Clemson University ABSTRACT: Using large-sample, archival data from the state o f South Carolina's juvenile justice agency, we examine the question o f race differences in predictors o f repeat offending for a sample of approximately 100,000 youth who had been referred for criminal offenses. Independent variables relating to background, adverse parenting, mental health, school-related disabilities, and features o f first offenses contributed to more than 25% o f the variance in recidivism for both Black and White youth. Male gender, eligibility for free or reduced school lunch, diagnoses based on the Diagnostic and Statistical Manual of Mental Disorders (4th ed.; text rev.; American Psychiatric Association, 2000), placement in Child Protective Services, and school identification as having a classification of emotional or behavioral disorders or learning disabilities were all predictive o f juvenile recidivism. In addition, early age o f first offense, status offenses, and being prosecuted for a first offense were significant predictors. Magnitudes o f prediction were similar across racial groups, suggesting similar vulnerability o f both Black and White youth to these early adversities. Interactions between race and other independent variables accounted for only .001 o f the variance in recidivism. However, there were several significant interactions. Mental health history and characteristics o f the first offense were stronger predictors for White youth than for Black youth. Gender, poverty (free or reduced lunch), and school identification o f having a classification o f emotional or behavioral disorders were stronger predictors for Blacks than for Whites. Implications for prevention are addressed. ■ The overrepresentation of Black youth in the juvenile justice system has been a persistent concern and is evident across contact points (National Council on Crime and Delinquency, 2007). Research studies on m inority arrests and confinement attest to race effects, direct or indirect, at m ultiple stages of juvenile justice processing and across jurisdictions (Pope, Lovell, & Hsia, 2001; see also Huizinga et al., 2007). For example, in 2010, 1,642,600 juveniles were arrested in the United States for violent crimes. In the same year, the racial/ ethnic composition for juveniles ages 10 to 17 was 76% White, 17% Black, 5% Asian/Pacific Islander, and 1% American Indian. However, the racial/ethnic composition for juvenile arrests was 47% White, 51 % Black, 1% Asian, and 1% American Indian. Proportions of Black offenders were 56% for murder, 67% for robbery, 42% for motor vehicle theft, 41% aggravated assault, 38% simple assault, 36% for forcible rape, 36% for burglary, and 36% for weapons (Puzzanchera, 2013). Further­ more, data on cumulative prevalence arrest rates show that Black youth are almost 50% more likely than W hite males to have been arrested at least once by the age of 18 (Brame, Bushway, Paternoster, & Turner, 2014). 184 / May 2015 In addition to race differences in arrest rates, there is also evidence of race disparities in adjudication (prosecution and referral to a fam ily court), w ith Black youth more likely than W hite youth to be prosecuted for serious crimes (Barrett, Katsiyannis, & Zhang, 2010). Black juveniles are more likely to be sent to secure confinement than W hite juveniles and are more likely to be transferred to adult facilities (The Sentencing Project, 2014). Fi­ nally, Black students are also disproportion­ ately represented in disciplinary proceedings in the schools, accounting for 27% of law enforcement referrals and 31% of schoolrelated arrests (U.S. Department of Education, 2014; U.S. Department of Education, Office of Civil Rights, 2014a, 2014b). To address disproportionate m inority con­ finement (DMC), the Office of Juvenile Justice and Delinquency Prevention (OJJDP) adminis­ ters the Formula Grants program under Title II, Part B, of the Juvenile Justice and Delinquency Prevention (JJDP) Act of 1974. Based on the most recent amendments to the JJDP Act of 1992, the need to address this problem was elevated to a core requirement, w ith 25% of funds linked to state compliance. Currently, based on the OJJDP's five-phase DMC Behavioral Disorders, 40 (3), 184-195 Reduction Model (identification, assessment/ diagnosis, intervention, evaluation, and mon­ itoring), 41 states have DMC subcommittees under their state advisory groups (18 full-time and 37 part-time DMC coordinators), 34 states have invested in targeted local DMC-reduction sites, 30 states have implemented nationally recognized models, and 39 states have pro­ vided a timeline for tracking and monitoring trends over time (OJJDP, 2012). Meta-analyses regarding the effectiveness of these programs have indicated positive intervention effects for minority youth; outcomes include a reduction in delinquent behavior and improvement in school participation, peer relations, academic achievement, and psychological adjustment. But particularly striking is the finding that the programs have been equally effective for minority and White juveniles. In their review of over 300 studies, including studies prior to 1992, Wilson, Lipsey, and Soydan (2003) compared method-adjusted effect sizes (in­ tervention effect sizes controlling for method­ ological differences between studies) for minority (primarily Black) and White youth; with method controlled, race differences in the magnitude of intervention effects were non­ significant. The finding of similar effects of interven­ tions on minority and nonminority delinquents raises the question of vulnerability and re­ silience among minority youth. The fact that targeted interventions have similar effects for Black and White youth suggests that youth of different backgrounds are vulnerable to the same risk factors and benefit from the same reparative experiences. Thus, we might expect that for both White and Black youth, early experiences such as parental neglect or abuse, mental health problems, and school failure would be predictive of delinquency (Barrett, Katsiyannis, Zhang, & Zhang, 2014) and that multisystemic interventions focusing on the family environment, the school experience, and the child's individual social adjustment would be ameliorative (Farmer & Farmer, 2001 ). Another possibility, however, is that mi­ nority youth, and in particular Black youth, are more resilient in the face of early adverse experiences. The concept of Black resilience is compelling (APA Task Force on Resilience and Strength in Black Children and Adolescents [APA Task Force], 2008) and has received both popular and empirical support. For example, family instability appears to be a stronger Behavioral Disorders, 40 (3), 184-195 predictor of behavioral and academic prob­ lems among White children than Black chil­ dren (Fomby & Cherlin, 2007). Also, there is evidence that family structure transitions (e.g., parental separation) have more deleterious consequences for White than Black children with regard to early sexual activity among females (Wu & Thomson, 2001). Finally, there is evidence that Black youth tend to have higher overall self-esteem than Whites, partic­ ularly with regard to their satisfaction with body image (Steinberg, 2014), a fact that has been linked with the lower rates of presenta­ tion of weight concerns (e.g., anorexia ner­ vosa) among Black Americans than White Americans (American Psychiatric Association [APA], 2013). Multiple explanations have been given to explain a lower vulnerability to certain social-environmental stressors among Black youth. Fomby, Mollborn, and Sennott (2010) suggested two bases for this resilience: inclusion in broader social networks (including church, peer group, and neighborhood) and importance of kin relationships (i.e., extended family). But the concept of resilience needs to be examined carefully. An assumption of greater resilience among minority populations may lead to an underestimation of the impact of the very factors that place a population "at risk." This is a particular concern if the early adverse experiences have the same or even greater impact on vulnerable populations. The com­ peting hypothesis (vs. resilience) is increased vulnerability. Vulnerability would result when weakened bonds between youth and main­ stream institutions, such as the school, and adverse early experiences result in poorer self­ perceptions and efficacy, variables which themselves are predictive of more persisting behavior problems among children in multi­ ple-risk families (Radke-Yarrow & Brown, 1993). A third possibility is that the magnitudes of effects of known risk factors for later behavior problems are very similar for White and minority youth. In the present study we examine the role of multiple categories of risk factors for juvenile delinquency recidivism for large samples of Black and White youth who had had a least one referral to a state juvenile justice system. Independent variables include parental maltreatment (foster care, referral to Child Protective Services [CPS]), mental health problems (referral to state department of mental health), and school-related disability May 201 5/1 8 5 (learning disability [ED], emotional/behavioral disorder [EBD]). The outcome of interest was juvenile recidivism: a second referral to the state juvenile justice system. In addition to testing for race differences in the above risk factors, we also tested for race differences in the role of gender and age of first offense, variables known to be highly predictive of juvenile offending and recidivism (Barrett et al., 2014). Method Source of Data Data for this study were obtained from two sources, the South Carolina Department of Juvenile Justice (DJJ) and the South Carolina Budget and Control Board's Office of Research and Statistics (ORS). DJJ data comprised in­ formation on approximately 100,000 youth who had been born in the period of 19811988 and who had been involved in de­ linquent activity. We linked the DJJ data with data obtained from the ORS. The ORS houses data from all of the state agencies in South Carolina, including, but not limited to, the South Carolina Department of Education (SDE), the South Carolina Department of Social Services (DSS), the South Carolina Department of Mental Health (DMH), and the South Carolina DJJ. These linkages enabled us to examine environmental influ­ ences on delinquency and recidivism using data that were not available in the original DJJ file. D/J Data Data were drawn from the South Carolina DJJ Management Information System. The DJJ sample consists of all juveniles born between 1981 and 1988 whose cases were referred to the South Carolina DJJ on at least one occasion ("referral"). The sample was part of a m ulti­ cohort, matched control study conducted in conjunction with the South Carolina Budget and Control Board (Barrett et al., 2014), a study that also included nondelinquent youth. In South Carolina cases are first processed at the fam ily court level by the DJJ. Intake workers from the DJJ assess risk and needs and forward cases to the Solicitor's Office w ith advisory recommendations (e.g., diversion or prosecu- 1 8 6 /M a y 2015 tion). If the case is prosecuted, the juvenile may be committed to the custody of the DJJ, given probation, or given another penalty, such as a school attendance order. The 1981-1988 cohorts include 99,602 individuals, 65,502 (65%) males and 35,100 (35%) females. The racial composition is 50,496 (51%) Black, 47,537 (48%) W hite, and 1,569 (2%) other (Asian and Hispanic). The average age of the juveniles when they were first referred to the system was 14.47 years (SD = 1.94), and the mean total number of referrals per juvenile was 2.21 (SD = 2.00). O f the 99,602 juveniles, 54% had one referral only, 19% had tw o referrals, and 27% had three or more referrals. Social demographic data were collected selectively by the DJJ and were available for only about half of the sample (see Barrett et al., 2010, for details). For purposes of the present study, only Black and W hite youth were considered in the analyses. Individual data on delinquency history were aggregated for each subject in the sample. Data available for each subject included age at first offense, severity of first offense, and severity of second offense, if applicable. Data on dispositions (penalties) were also collected. The determination of the seriousness of a crime was based on the coding scheme employed by South Carolina. The DJJ rates crimes on an ordinal scale, w ith lower ratings representing less serious offenses. For purposes of this analysis, we categorized offenses as status offenses (DJJ severity levels 1 and 1.5) and nonstatus offenses (rating levels of 2 and above). OKS Data For all individuals in the DJJ sample and also for the matched control group (described below), data from other state agencies (housed in the ORS) were made available. Files on each child in the DJJ file were linked w ith files of the other state agencies using a probabilistic matching algorithm. In the ORS linkage system, once a match is identified, an ID number is assigned. The same ID is used for all subsequent episodes of services, regardless of data source or service provider. Additional information about the key linkage system is available on request. For the present analyses, individual data in the DJJ files were linked w ith data for the same individuals from the DSS, DM H, and SDE. Data obtained from the DSS included in­ formation about foster care placements and Behavioral Disorders, 40(3), 184-195 whether or not an individual had ever been placed in the custody of CPS. For foster care, information about age and duration of place­ ment and number of placements was obtained. With respect to CPS, we obtained information about the reason for and timing of CPS. Data obtained from the DMH included information about age at first, second, and most recent referrals and primary diagnosis based on the Diagnostic and Statistical Manual o f Mental Disorders (4th ed.; text rev.; DSM-IV-TR; APA, 2000) at each referral. Primary diagnoses were further categorized into seven major categories (described in Analyses section). Data from the SDE included information about the ages at which the student was eligible for free and/or reduced lunch and eligibility for special education services due to LDs or EBDs. After separate files were constructed for each agency (D)J, DMH, DSS, DOE), files were merged to create a new master file for the DJJ sample. Analyses We used a series of logistic regression analyses to examine the individual and com­ bined influences of selected categories of independent variables on juvenile delinquen­ cy recidivism; that is, presence of referral to the DJJ for a second offense. Our analysis involved three steps. First, we obtained de­ scriptive statistics on all independent variables and on the dependent variable for each of the two comparison groups, Black and White. We also conducted cross-tabulations and calculat­ ed chi-squared statistics to examine the re­ lationship between ethnic group and each of the categorical predictor variables (we con­ ducted an independent sample f test on the one continuous variable, age of first referral). Second, we carried out a multivariable logistic regression analysis using the whole sample, including interaction terms for the interaction of racial group and each in­ dependent variable. The inclusion of in­ teraction terms enabled us to test for the possibility of significant differences in the logistic regression coefficients for indepen­ dent variables for Black versus White youth. Youth incarcerated for the first referral were excluded from the analysis. In the logistic regression analysis we included six blocks of predictors. In predicting the variable recidi­ vist (vs. nonrecidivist), we first examined the role of demographic variables. Included in Behavioral Disorders, 40 (3), 184-195 this block were the variables eligible for free or reduced lunch (coded Yes or No), race, and gender. The second block of predictors included two measures of family background/dysfunction, placement in foster care (Yes or No) and placement in CPS (Yes or No). The third set of predictors focused on childhood psychopathology. In constructing these variables, all DSM-IV-TR diagnoses conferred by the DMH were assigned to one of seven categories. Category assignments were made by the first author, a licensed psychologist, in consultation with collea­ gues. The categories used were aggression and conduct problems; drug-related prob­ lems; attention and learning disorders, men­ tal retardation and, other problems starting in childhood; mood and anxiety disorders; psychotic disorders; adjustment and milder disorders; and other serious disorders. For the present analysis, subjects were first scored for presence or absence (at any time in development) of a primary diagnosis involving aggression and/or conduct prob­ lems. The DSM-IV-TR classifications that were used to define an aggressive behavior problem included antisocial personality dis­ order (DSM-IV-TR classification 301.7), im­ pulse control disorder (312.30), conduct disorders (312.81, 312.82, 312.89), disrup­ tive behavior disorder (312.9), oppositional defiant disorder (313.81), and child or adolescent antisocial behavior (V71.02; APA, 2000). They were then scored for presence or absence of a primary disorder involving any other type of disorder recog­ nized in the DSM-IV-TR. These two vari­ ables constituted the third block of predictor variables. The fourth set of vari­ ables included two indicators of eligibility for special education. Subjects were first scored for presence or absence of a schoolbased identification as eligible for special education services due to an LD. They were also scored for presence or absence of a school-based identification as eligible for services due to an EBD. The fifth block of predictors included the variables age at first offense (continuous variable), severity of first offense (status offense versus misdemeanor or felony), and prosecuted for first violation (Yes or No). The final block of predictors included all two-way interaction terms (e.g., Race X Gender). We also ran simple logistic regression analyses for each of the indepen­ dent variables to examine the relationship May 2015 / 187 between the independent variable and re­ cidivism with other independent variables uncontrolled. Third, we repeated the logistic regression analyses but this time for Black and White youth separately. We again entered variables in blocks. Blocks were the same as those described in the whole sample analyses, with two exceptions: (a) The first block did not include the variable race, and (b) interaction terms were not included. TABLE 1 Descriptive Statistics for Predictor Variables in Relation to Ethnicity Total Sample Black (N = 96,613 ) (n = W hite 49 ,7 1 0 ) (n = 46,903 ) Gender (Male) 64.58% 64.14% 65.04% Receives free or reduced lunchi 61.73% 75.26% 49.39% Variables Demographic Parenting Results Foster care CPS Descriptive Statistics Percentages of individuals manifesting different background characteristics and risk indicators are shown in Table I for the entire sample and for Black and White youth separately. As shown in Table 7, White offen­ ders were more likely than Black offenders to have been diagnosed with a DSM-IV-TR disorder not associated with aggressive or impulsive behavior. Black youth were more likely than White youth to receive free or reduced lunch in the school system, to have been in foster care or CPS, to have been diagnosed with a DSM-IV-TR disorder relating to aggression or impulse control, and to have been identified by a school as having a classifi­ cation of an EBD. White offenders were more likely than Black offenders to have been prosecuted for their first offense and were more likely to be male. Black youth had a lower mean age of first referral and were less likely than White youth to have been first referred for a status offense. 5.62% 5.84% 4.66% 12.27% 13.00% 11.49% DSM-IV-TR diagnoses Aggression 14.51% 16.56% 12.32% Other 25.59% 22.12% 29.24% 16.77% 16.66% 16.89% 5.62% 6.31% 4.90% Disabilities LD EBD Delinquency history M = 14.45 (1.95) M = 14.20 (2.06) M=14.73 (1.79) First offense adjudicated 21.02% 19.81% 22.31% Nonstatus offense 79.30% 79.87% 79.10% Age at first referral Note. Chi-squared analyses for differences between racial groups were conducted for all categorical variables. Significant differences at p < .001 were detected for all comparisons except Gender Male and LD. Comparison for Gender Male was significant at p < .01; LD was not significant. For Age at first referral, f(96/ 611) = 42.92 (p < .001), standard deviations are shown in parentheses. Sample sizes for Nonstatus Offense are n = 46,871 (Black) and n = 49,686 (White). CPS = Child Protective Services, LD = learning disability, EBD = emotional or behavioral disorder. Logistic Regression Whole Sample: No Interaction Terms Table 2 shows results of the multivariable logistic regression analysis with interaction terms not included, as well as results of simple logistic regression analyses. The multivariable analysis showed significant effects for race and gender with Black youth and males more likely than White youth and females to commit a second offense, x20 , N = 96,557) = 39.74, p < .001, and x2 = 494.08, p < .001, respectively. There was also an effect for free or reduced lunch with youth qualifying for free or reduced lunch more likely to commit a second offense (%2 = 709.10, p < .001). 188 / May 2015 While the effect for foster care was not significant, there was a significant effect for CPS (x2 = 279.04, p < .001). Youth who had been in CPS were approximately 50% more likely than those who had not been in CPS to commit a second crime. Mental health di­ agnosis was significantly related to the likeli­ hood of a second offense, with youth with either a diagnosis relating to aggressive behav­ ior or any other diagnosis more likely to be referred for a second offense (x2 = 2873.15, p < .001, and x2 = 1353.62, p < .001, respectively). The values for adjusted odds ratios (AORs) show that youth with mental Behavioral Disorders, 40 (3), 184-195 TABLE 2 Logistic Regression Analysis for Prediction of Recidivism (N = 96 ,557a) Block Block 1 Variable R 2 Block Race (Black) Gender (Male) Free or reduced lunch Block 2 Foster care CPS Block 3 .17** EBD LD Block 5 .08** DSM-IV-TR Aggressive DSM-IV-TR Other Block 4 .05** 1?** Age first offense Severity Adjudicated .26** B AORE AORf 0.10 39.74** 1.33** 1.10** 0.35 494.08** 1.34** 1.42** 0.42 709.10** 2.09** 1.53** 0.07 3.41 3.03** 1.07 W ald (x2i) 0.43 279.04** 2.70** 1.54** 1.21 2873.15** 4.79** 3.34** 0.64 1353.62** 2.51** 1.91** 0.36 110.77** 3.19** 1.44** 0.10 22.72** 1.55** 1.10** -0 .3 0 5263.94** 0.71** 0.74** -0 .3 4 367.04 0.62** 0.71** 852.36** 1.96** 1.70** 0.53 Note. R2 block refers to Nagelkerke R2 following this step in the equation and including the constant. Significance level for R2 block is based on the change in the log-likelihood of the outcome. Significance level for the Wald statistic is based on the final logistic regression equation. B refers to the logistic regression coefficient in the final equation. AORE refers to the adjusted odds ratio if entered alone in the equation. AORF refers to the adjusted odds ratio in the final equation. CPS = Child Protective Services, DSM-IV-TR = Diagnostic and Statistical Manual of Mental Disorders, 4th edition, text revision, LD = learning disability, EBD = emotional or behavioral disorder. “ Sample size is reduced to 96,557 for multivariable analysis due to missing observations on Severity. **p < .001. health diagnoses relating to aggressive behav­ ior were more than 3 times more likely to commit a second offense than other first offenders and that youth with another mental health diagnosis were almost twice as likely as nondiagnosed youth to commit a second offense. Youth identified as eligible for special education services due to an EBD or LD were more likely to commit a second offense than youth without these special education classi­ fications (x2 = 110.77, p < .001 and x2 = 22.72, p < .001, respectively). Finally, there was a significant relationship between age of first offense and recidivism (x2 = 5263.94, p < .001). The AOR of .76 shows that for each year of reduced age of first offense, the odds of a second offense increase by approximately 25%. In addition, youth who had been referred for status offenses and youth adjudicated for their first offense were more likely to commit a second offense (x2 = 367.04, p < . 001 and X2 = 852.36, p < . 001, respectively). Foster care was not significantly related to recidivism. The total adjusted R2 was .26; model x2(12, /V = 96,557) = 20482.90, p < .001. Also, as shown in Table 2, simple logistic regression analyses showed that each independent vari­ able when considered alone was a significant predictor of recidivism; AOREshows values of Behavioral Disorders, 40 (3), 184-195 adjusted odds ratios with only one variable in the equation. Whole Sample: Interactions When all two-way interaction terms were added to the equation at step six, the total value of R2 was only slightly increased from .256 to .257, x2(11, N = 96,557) = 159.99, p < .001. Six interactions were significant: Race X Gender (X2 = 11 -83, p = .001), Race X Free Lunch (x2 = 34.30, p < .001), Race X DSM-IV-TR NonAggressive (x2 = 22.38, p < .001), Race X EBD (X2 = 21.43, p < .001), Race X Severity (x2 = 14.99, p < .001), and Race X Adjudication (X2 = 31.60, p < .001). These interactions are explained in the following section. Split Files: Analyses by Racial Group Tables 3 and 4 show the results of the logistic regression analyses for White and Black youth considered separately. Values of R2 were .25 for White youth and .26 for Black youth. As with the analysis for the total sample, for both groups all variables except foster care were significant predictors of recidivism in the multivariable analysis, and all variables were significant in the simple logistic regression May 2 0 1 5/1 8 9 TABLE 3 Logistic Regression Analysis for Prediction of Recidivism for White Youth (N = 46,871a) Block Block 1 Variable Free or reduced lunch Block 2 Block 4 AORe 1.16** 1.35** 0.34 240.88** 2.16** 1.40** 0.08 2.03 3.21** 1.08 .08** 0.42 122.07** 2.90** 1.52** 1.17 1208.22** 4.62** 3.23** .16** 0.72 917.57** 2.79** 2.06** 0.20 15.31** 2.65** 1.22** 0.08 8.78* 1.55** 1.09* .05** EBD LD Block 5 AORe 164.13** DSM-IV-TR Aggressive DSM-IV-TR Other B 0.30 Foster care CPS Block 3 R2 Block Gender (Male) .15** W ald ( * 2,) Age first offense -0 .3 0 2234.72** 0.71** 0.75** Severity -0 .4 0 262.27** 0.56** 0.67** 0.63 594.21** 2.37** 1.87** Adjudicated .25** Note. R2 block refers to Nagelkerke R2 following this step in the equation and including the constant. Significance level for R2 block is based on the change in the log-likelihood of the outcome. Significance level for the Wald statistic is based on the final logistic regression equation. B refers to the logistic regression coefficient in the final equation. AOR e refers to the adjusted odds ratio if entered alone in the equation. AORF refers to the adjusted odds ratio in the final equation. CPS = Child Protective Services, DSM-IV-TR = Diagnostic and Statistical Manual of Mental Disorders, 4th edition, text revision, LD = learning disability, EBD = emotional or behavioral disorder. aSample size is reduced to 46,871 for multivariate analysis due to missing observations on Severity. *p < .01. **p < .001. analyses. In addition, the relative magnitudes of the different predictors were very similar across groups; in fact, using independent variable as the unit of analysis, the correlation between adjusted odds ratios (across the two groups) was .95 (f9 =9.13, p < .001). Significant interactions in the total sample analysis indicated differences in the magnitude of six predictor variables between the two groups. Significance tests are based on differ­ ences in the magnitude of logistic regression coefficients (Paternoster, Brame, Mazerolle, & Piquero, 1998). Three variables were stronger predictors for White youth than for Black. Having a DSM-IV-TR diagnosis for a disorder not described as relating to aggression or impulse control was a stronger predictor for White youth than for Black (Z = 4.72, p < .001). Also, having been referred first for a status offense and having been prosecuted for the first offense were stronger predictors for Whites than for Blacks (Z = 3.88, p < .001 and Z = 5.55, p < .001, respectively). In contrast, being male, being eligible for free or reduced lunch, and having been classified by a school as eligible for special education due to an EBD were all stronger predictors of recidivism for Black than White youth. Results of significance tests were Z = 3.46, p < .001 for gender; Z = 5.81, p < .001 for free or 190 / May 2015 reduced lunch; and Z = 4.65, p < .001 for EBD. Summary of Results Independent variables relating to back­ ground, adverse parenting, mental health, school-related disabilities, and features of first offenses contributed to more than 25% of the variance in recidivism for both Black and White youth. Male gender, eligibility for free or reduced school lunch, DSM-IV-TR diagno­ ses, placement in CPS, and identification as qualifying for special education services due to an EBD or LD were all predictive of juvenile recidivism. In both groups, a mental health diagnosis relating to aggression or impulse control was the strongest predictor of recidi­ vism. In addition, early age of first offense, status offenses, and being prosecuted for a first offense were significant predictors. Magni­ tudes of prediction were similar across ethnic groups. While interactions between race and other independent variables accounted for only .001 of the variance in recidivism, there were several significant interactions. Mental health history and characteristics of the first offense were stronger predictors for White youth than for Black youth. However, gender, Behavioral Disorders, 40 (3), 184-195 TABLE 4 Logistic Regression Analysis for Prediction of Recidivism for Black Youth (N = 49,710) Block Block 1 Variable I f Block Free or reduced lunch Block 2 Block 3 1.51** .04** 0.52 483.02** 1.88** 1.69** 0.08 2.37 2.82** 1.09 .08** 0.45 163.52** 2.50’ * 1.58** 1.24 1669.04** 4.80** 3.45** 0.56 458.06** 2.44** 1.75** 0.52 117.39** 3.66** 1.68** .16** EBD LD Block 5 AORf 1.62** DSM-IV-TR Aggressive DSM-IV-TR Other Block 4 AOR e 356.42** Foster care CPS .17** Age first offense Severity Adjudicated B 0.41 Gender (Male) .26** W ald (x2i) 0.11 16.01** 1.56** 1.12** -0.31 3022.93** 0.71** 0.74** -0 .2 6 100.57** 0.64** 0.77** 0.42 269.48** 1.66** 1.52** Note. R2 block refers to Nagelkerke R2 following this step in the equation and including the constant. Significance level for R2 block is based on the change in the log-likelihood of the outcome. Significance level for the W ald statistic is based on the final logistic regression equation. B refers to the logistic regression coefficient in the final equation. AORE refers to the adjusted odds ratio if entered alone in the equation. AORF refers to the adjusted odds ratio in the final equation. CPS = Child Protective Services, DSM-IV-TR = Diagnostic and Statistical Manual of Mental Disorders, 4th edition, text revision, LD = learning disability, EBD = emotional or behavioral disorder. “Sample size is reduced to 49,710 for multivariate analysis due to missing observations on Severity. * * p < . 001. poverty (free or reduced lunch), and school classification of an EBD were stronger pre­ dictors for Blacks than for Whites. Discussion It is w ell recognized that early adverse experiences in the fam ily and in school are strongly linked to later child behavior prob­ lems (Dodge, Greenberg, Malone, & Conduct Problems Prevention Research Group, 2008). Recognized also is the higher prevalence of fam ily systems disruption and school failure among Black children (Fomby et al., 2010; Steinberg, 2014), factors that help explain in part the disparities in behavioral outcomes, including both prosocial behaviors, such as school achievement, and antisocial outcomes, such as juvenile delinquency, between W hite and Black youth. The findings of race dispar­ ities in achievement, mental and physical well-being, and antisocial behavior have motivated both scholars and clinicians to examine further the sources of strength and resilience among Black children and families (APATask Force, 2008). Empirical studies have identified a number of protective factors that may lim it the deleterious effects of early adversity, including institutional barriers to Behavioral Disorders, 40 (3), 184-195 social and emotional well-being. Protective factors include individual factors, such as emotional regulation (Mendez, Fantuzzo, & Cicchetti, 2002); fam ily influences, such as closeness to parents (Bynum & Kotchik, 2006); school variables, such as authoritative teachers (Ladson-B i 11i ngs, 1994); and comm unity fac­ tors, such as quality childcare (Gottfredson, Gerstenblith, Soule, Womer, & Lu, 2004). The APA Task Force (2008) provided a detailed review of the literature. But studies on individual differences in resilience do not address the broader question of whether known risk factors for later de­ velopmental problems— including risk factors relating to early adversity in the home envi­ ronment, mental health, school learning prob­ lems, and demographic factors such as gender and social-economic status— have similarly deleterious effects for W hite and m inority youth. Are Black youth as a group protected against social-environmental and personal risk factors by virtue of having experienced social and economic adversity in the past and learning how to adapt to m ultiple stressors? O r are Black youth even more vulnerable than W hite youth to these early environmental risk factors, perhaps due to the same early experi­ ences and unique socio-cultural contexts that are often seen as protective? May 2 0 1 5 / 191 The results of our study provide a mix of conclusions. The most striking finding in the present study is the extraordinary concordance in the effects (and magnitudes of effects) of early risk indicators for later antisocial behav­ ior, measured in this study by the presence of juvenile delinquency recidivism. For large samples of both Black and White youth who had been referred to the DJ) on at least one occasion, juvenile recidivism (referral for a second offense) was significantly predicted by the same set of variables: gender, poverty, being referred to CPS, having a DSM-IV-TR diagnosis for a psychological disorder, being identified as eligible for special education services due to an EBD or LD, having an earlier age of first offense, committing a status offense as a first offense, and being prosecuted for a first offense. In addition, the amount of variance in recidivism accounted for by these predictor variables was nearly identical: 25% of the variance in the White sample and 26% of the variance in the Black sample. Finally, AORs for the two groups were highly corre­ lated; with type of predictor variable as the unit of analysis, the correlation in the magni­ tudes of AORs was .95. But there was also evidence of differential effects of individual risk factors for Black versus White youth. Recognizing that interac­ tions between race and individual predictor variables accounted for only .001 of the variance in recidivism in the total sample, it was still the case that the increase in variance accounted for by interaction was significant. Comparison of logistic regression coefficients for the Black and White samples showed six variables where the magnitudes of effects differed for the two groups: gender, poverty, and EBD classifications were stronger predic­ tors for Black youth; DSM-IV-TR diagnoses (nonaggressive), status offending, and prose­ cution for first offense were stronger predictors among Whites. Several of the interactions identified are difficult to interpret. Earlier analyses (Barrett, Katsiyannis, & Zhang, 2006; Barrett et al., 2010) indicated that status offenses are more likely than nonstatus offenses to be adjudicat­ ed and to lead to a second offense. Why the relationship between status offending and recidivism should be stronger for one ethnic group than another is not clear. Similarly the findings of a stronger link between nonaggres­ sive (i.e., internalizing) disorders and recidi­ vism among White youth and a stronger 192 / May 2015 relationship between gender and recidivism among Black youth merit further attention. On the other hand, the interactions in­ volving poverty (free or reduced school lunch) and EBD are more readily interpretable. For Black youth in particular, qualifying for free or reduced lunch was a risk factor for recidivism. Since poverty is also associated with lower family and neighborhood stability, this finding is inconsistent with a view that due to adaptation to difficult environments, Black youth may develop a special resilience in dealing with poverty. Rather, poverty appears to place a particularly heavy burden on Black youth. Findings relating to EBDs are notewor­ thy. A school's identification of a child as eligible for special education due to an EBD is a stronger predictor of juvenile recidivism for Black youth than for White, and is also a stronger predictor of recidivism than identi­ fication as eligible due to an LD. It is important to recognize that Black students are dispro­ portionately represented in special education, particularly in the disability categories of EBDs, a finding supported by the present data, and intellectual disabilities (ID; Kauffman, Simpson, & Mock, 2009). Black students receiving special education services are also less likely to be placed in the general education classroom, resulting in resegregation of the population (Artiles, Kozleski, Trent, Osher, & Ortiz, 2010). Further, Black students with (and without) disabilities are at greater risk than White peers to be expelled or suspended (U.S. Department of Education, Office of Civil Rights, 2014a); in fact, one out of six Black students in public school has been suspended at least once (Losen & Gillespie, 2012) and 59% of Black males have been suspended or expelled from school, compared to 24% of White males (Toldson, 2011). Importantly, Black students with disabilities account for a disproportionate­ ly high percentage of disciplinary inclusions totaling 10 days or more (IDEA Data Center, 2014). Finally, Black students, particularly males, are also dropping out of school at almost twice the rate for White students (Chapman, Laird, & KewalRamani, 2010; National Center for Education Statistics, 2013). Regarding academ­ ic performance, in 2012 the percentage of 17year-old Black students at or above the National Assessment of Educational Progress (NAEP) reading score level of 300 (able to find, understand, summarize, and explain relatively Behavioral Disorders, 40 (3), 184-195 complicated literary and informational mate­ rial) was 22% versus 47% for White students. Similarly, percentages of mathematics scores of 300 or higher (able to perform reasoning and problem solving involving fractions, dec­ imals, percentages, elementary geometry, and simple algebra) were 33.8% for Black students versus 70.3% for Whites; the disparity was even more pronounced (1.1% for Black students vs. 9.1% for White) for scoring 350 or above (able to perform reasoning and problem solving involving geometry, algebra, and beginning statistics and probability; Na­ tional Center for Education Statistics, 2013). Given these disparities in academic and behavioral outcomes, it is important that schools examine closely the methods used to identify students' educational needs, including the procedures that are used to identify children as having EBDs. Limitations and Conclusions A major limitation in the present study was that while we were able to include a general measure of poverty in our analyses, more specific indicators of socioeconomic disadvan­ tage, such as parental characteristics, family history, and family income were not included. As noted previously, the South Carolina DJI does obtain more specific socio-demo­ graphic data (e.g., persons living in the family, family income) on youth referred to the DJJ; this information is collected at intake. However, this background information is collected selectively; in general, background data are more likely to be collected in cases where the DJJ intake worker sees the situation as more serious. Because of this potential bias, we chose to include an SDE variable— eligibility for free or reduced lunch—that was available for all subjects. In addition, our definition of recidivism—presence of a second referral to the DJJ—was very broad. W hile it would have been possible to use a more stringent measure of recidivism (e.g., pres­ ence or absence of a felony at some time after the first violation), we recognized that be­ cause more serious violations are less com­ mon, we would probably account for a lower percentage of the variation in recidivism if we made this decision; for this reason (the decision to increase variability in the out­ come variable), we chose to examine only whether there was or was not a repeat offense. Behavioral Disorders, 40 (3), 184-195 Despite these limitations, our findings suggest a number of conclusions and lead to several directions for further research. First, the risk factors for Black and White youth for juvenile delinquency recidivism are remark­ ably similar. Many scholars have noted the ways in which Black youth appear, in some ways, more resilient than White youth in dealing with the challenges of adolescence. Black youth have higher self-esteem than White youth; this is particularly true for girls (Biro, Striegel-Moore, Franko, Padgett, & Bean, 2006). Further, as discussed earlier, Black youth appear to tolerate variations in family structure that White youth do not. But these findings should not lead us to underestimate the impact of early adverse experiences— in the family, neighborhood, or school— on Black children. Individual children, regardless of ethnic/racial group, often show great resilience in coping with difficult circumstances. Chil­ dren with strong peer relationships, realistic and positive views of themselves, and good decision-making skills are more able than others to overcome multiple stressors and adversities (Radke-Yarrow & Brown, 1993), and among today's Black youth it is critical that these abilities and relationships be sup­ ported (APA, 2008). But all children, both minority and nonminority, are likely to be negatively affected by the types of early adverse experiences we have examined in this study. Second, one area where Black youth may be particularly impacted by adversity is the area of special education. Not only are Black children overrepresented in special education classifications, identification as having an EBD for Black children is a stronger predictor of juvenile delinquency recidivism than it is for White children. This is true when factors such as poverty, family disruption, and mental health history are controlled. While the iden­ tification of a child as having a disability should set in motion a chain of events resulting in increased attention and support for the child, the resulting experiences for the child may not always be positive. It is imperative that we continue to monitor the ways in which school-based classifications are made, inter­ ventions are implemented, and children are evaluated following the introduction of special education services. Finally, it is important that future research­ ers recognize that while adversity in the early years of development poses significant threats May 2015 / 193 to the individual's later social and behavioral health, there are evidence-based interventions that may help to reduce the likelihood of problem behavior for children and youth who are otherwise at risk (Brody, Breach, Philibert, Chan, & Murry, 2009). These interventions generally involve a multilevel, systemic approach to prevention, including individual-level, family-level, and schoollevel components (Farmer & Farmer, 2001). Future research studies should continue to address the efficacy of such preventative programs, beginning always with the assump­ tion that all children and youth—regardless of race or background— have the same needs for security, support from others, and belief in their own competence and worth. REFERENCES American Psychiatric Association. (2000). Diagnos­ tic and statistical manual o f mental disorders. (4th ed., text rev.) Washington, DC: Author. American Psychiatric Association. (2013). Diagnos­ tic and statistical manual o f mental disorders. (5th ed.) Washington, DC: Author. APA Task Force on Resilience, Strength in Black Children, & Adolescents (2008). Resilience in African American children and adolescents: A vision for optimal development. Washington, DC: Author. Retrieved from http://www.apa.org/ pi/cyf/resil ience.html Artiles, A. J., Kozleski, E. B., Trent, S. C., Osher, D., & Ortiz, A. (2010). Justifying and explaining disproportionality, 1968-2008: A critique of underlying views of culture. Exceptional Chil­ dren, 76(3), 279-299. doi: 10.1177/001440291 007600303. Barrett, D. E., Katsiyannis, A., & Zhang, D. (2006). Predictors of offense severity, adjudication, in­ carceration and repeat violations for adolescent male and female offenders, journal o f Child and Family Studies, 15, 708-718. doi: 10.1007/ s10826-006-9044-y. Barrett, D. E., Katsiyannis, A., & Zhang, D. (2010). Predictors of offense severity, adjudication, in­ carceration and repeat referrals for juvenile offenders: A multi-cohort replication study. Re­ medial and Special Education, 31, 261-275. doi: 10.1177/0741932509355990. Barrett, D. E., Katsiyannis, A., Zhang, D., & Zhang, D. (2014). Delinquency and recidivism: A multi­ cohort, matched-control study of the role of early adverse experiences, mental health prob­ lems and disabilities, journal o f Emotional and Behavioral Disorders, 22, 3-15. doi: 10.1177/ 1063426612470515. Biro, F., Striegel-Moore, R., Franko, D. L., Padgett, J., & Bean, J. A. (2006). Self-esteem in adolescent females, journal o f Adolescent Health, 39, 501-507. doi: 10.1016/j.jadohealth.2006.03.010. 194 / May 2015 Brame, R., Bushway, S. D., Paternoster, R., & Turner, M. G. (2014). Demographic patterns of cumu­ lative arrest prevalence by ages 18 and 23. Crime and Delinquency, 60, 471-486. doi: 10.1177/0011128713514801. Brody, G., Beach, S., Philibert, R., Chen, Y., & Murry, V. (2009). Prevention efforts moderate the association of 5-HTTLPR and youth risk behavior initiation: G X E hypotheses tested via a randomized prevention design. Child De­ velopment, 80, 645-661. doi: 10.1111/j.14678624.2009.01288.x. Bynum, M. S., & Kotchick, B. A. (2006). Motheradolescent relationship quality and autonomy as predictors of psychosocial adjustment among African American adolescents, journal o f Child and Family Studies, 15, 529-542. doi: 10.1007/ s10826-006-9035-z. Chapman, C., Laird, J., & KewalRamani, A. (2010). Trends in high school dropout and completion rates in the United States: 1972-2008. Wash­ ington, DC: National Center for Education Statistics, Institute of Education Sciences, U.S. Department of Education. Dodge, K., Greenberg, M., Malone, P., & Conduct Problems Prevention Research Group. (2008). Testing an idealized dynamic cascade model of the development of serious violence in adoles­ cence. Child Development, 79, 1907-1927. doi: 10.1111/j. 1467-8624.2008.01233.x. Farmer, T. W., & Farmer, E. M. Z. (2001). De­ velopmental science, systems of care, and pre­ vention of emotional and behavioral problems in youth. American Journal o f Orthopsychiatry, 71, 171-181. doi: 10.1037/0002-9432.71.2.171. Fornby, P., & Cherlin, A. J. (2007). Family instability and child well-being. American Sociological Review, 72, 181-204. doi: 10.1177/000312240 707200203. Fomby, P., Mollborn, S., & Sennott, C. A. (2010). Race/ethnic differences in effects of family instability on adolescents' risk behavior. Journal o f Marriage and the Family, 72, 2 34-253. doi: 10.1111/j.1 741 -3737.2010.00696.x. Gottfredson, D. C., Gerstenblith, S. A., Soule, D. A., Womer, S. C., & Lu, S. (2004). Do after-school programs reduce delinquency? Prevention Sciences, 5, 253-266. Huizinga, D, Thornberry, T., Knight, K., Lovegrove, P., Loeber, R., Hill, C., & Farrington, D. P. (2007). Disproportionate minority contact in the juvenile justice system: A study o f differential minority arrest/referral to court in three cities. Retrieved from https://www.ncjrs.gov/pdffiles1/ ojjdp/grants/219743.pdf IDEA Data Center. (2014). 2011-2012 IDEA Part B Discipline. Retrieved from https://explore.data. gov/dataset/2011-2012-IDEA-Part-B-Discipline/ xxsy-7tda Kauffman, J. M., Simpson, R. L., & Mock, D. R. (2009). Problems related to underservice: Behavioral Disorders, 40 (3), 184-195 A rejoinder. Behavioral Disorders, 34(3), 172-180. Ladson-Billings, C. (1994). The dreamkeepers: Suc­ cessful teachers o f African American children. San Francisco: Jossey-Bass. Losen, D. J., & Gillespie, J. (2012). Opportunities suspended: The disparate impact of disciplinary exclusion from school. Retrieved from http:// civilrightsproject.ucla.edu/resources/projects/ center-for-civil-rights-remedies/school-to-prisonfolder/federal-reports/upcoming-ccrr-research Mendez, J. L., Fantuzzo, J., & Cicchetti, D. (2002). Profiles of social competence among low- income African American preschool children. Child De­ velopment, 73, 1085-1100. doi: 10.1111/14678624.00459. National Center for Education Statistics. (2013). Digest o f education statistics. Retrieved from http://nces.ed.gov/programs/digest/2013 menu_ tables.asp National Council on Crime, & Delinquency (2007). And justice for some: Differential treatment of youth o f color in the justice system. Retrieved from http://www.nccdglobal.org/sites/default/ files/publication_pdf/justice-for-some.pdf Office of Juvenile Justice and Delinquency Pre­ vention. (2012). Disproportionate minority con­ tact. Retrieved from http://www.ojjdp.gov/pubs/ 239457. pdf#page=1 &zoom=auto,612,800 Paternoster, R., Brame, R., Mazerolle, P., & Piquero, A. (1998). Using the correct statistical test for the equality of regression coefficients. Criminology, 36, 859-866. doi: 10.1111/j.l 745-9125. Pope, C. E., Lovell, R., & Hsia, FI. M. (2001). Disproportionate minority confinement: A re­ view o f the research literature from 1989 through 2001. Retrieved from http://www. ojjdp.gov/dmc/pdf/dmc89_01 .pdf Puzzanchera, C. (2013). Juvenile arrests 2010. Office of juvenile Justice and Delinquency Prevention. Re­ trieved from http://www.ncjj.org/pdf/242770.pdf Radke-Yarrow, M., & Brown, E. (1993). Resilience and vulnerability in children of multiple-risk families. Development and Psychopathology, 5, 5 81 -5 92. doi: org/10.1017/S09545 79400006179. The Sentencing Project. (2014). Policy brief: dispro­ portionate minority contact in the juvenile justice system. Washington, DC: Author. Re­ trieved from http://sentencingproject.org/doc/ publications/jj_Disproportionate%20Minority% 20Contact.pdf Skiba, R. J., Simmons, A. B., Poloni-Staudinger, L., Feggins-Azziz, L. R., & Chung, C. (2005). Unproven links: Can poverty explain disproportionality in special education? journal o f Special Behavioral Disorders, 40 (3), 184-195 Education, 39, 130-144. doi: 10.1177/0022466 9050390030101. Steinberg, L. (2014). Adolescence. (10th ed.) New York: McGraw-Hill. Toldson, I. A. (2011). Breaking barriers 2: Plotting the path away from juvenile detention and toward academic success for school-age african american males. Washington, D.C.: Congressio­ nal Black Caucus Foundation, Inc. Retrieved from http://www.cbcfinc.org/oUploadedFiles/ BreakingBarriers2.pdf U. S. Department of Education. (2014). Guiding principles: A resource guide for improving school climate and discipline. Retrieved from http://www2.ed.gov/poIicy/gen/guid/schooldisci pi ine/guidi ng-principles.pdf U.S. Department of Education, Office of Civil Rights. (2014a). Civil rights data collection data snap­ shot: School discipline. Retrieved from http:// www2.ed.gov/about/offices/list/ocr/docs/crdcdiscipline-snapshot.pdf U.S. Department of Education, Office of Civil Rights. (2014b). Dear colleague letter: Nondiscriminatory administration o f school discipline. Re­ trieved from https://www2.ed.gov/about/offices/ list/ocr/letters/colleague-201401-title-vi.pdf Wilson, S. J., Lipsey, M. W., & Soydan, H. (2003). Are mainstream programs for juvenile delin­ quency less effective with minority youth than majority youth? A meta-analysis of outcomes research. Research on Social Work Practice. 13, 3-26. doi: 10.1177/1049731502238754. Wu, L., & Thomson, E. (2001). Race differences in family experiences and early sexual initiation: Dynamic models of family structure and family change, journal o f Marriage and the Family, 63, 682-696. doi: 10.1111/j.1 741-3737.2001. 00682.x. AUTHORS' NOTE Address correspondence to David E. Barrett, Eugene T. Moore School of Education, Clemson University, 101 Tillman Hall, Clemson, SC 29634; E-mail: bdavid@clemson.edu. MANUSCRIPT Initial Acceptance: 12/14/2014 Final Acceptance: 4/03/2015 May 2 0 1 5 /1 9 5 Copyright of Behavioral Disorders is the property of Council for Children with Behavioral Disorders and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. Measuring the Influence of Juvenile Arrest on the Odds of Four-Year College Enrollment for Black Males: An NLSY Analysis Royel Montel Johnson Spectrum: A Journal on Black Men, Volume 4, Number 1, Autumn 2015, pp. 49-72 (Article) Published by Indiana University Press For additional information about this article https://muse.jhu.edu/article/613203 Access provided by University Of South Florida Libraries (22 Dec 2016 18:13 GMT) Measuring the Influence of Juvenile Arrest on the Odds of Four-Year College Enrollment for Black Males: An NLSY Analysis Royel Montel Johnson ABSTRACT: Black youth make up just 16% of public school students in the United States, though they constitute 31% of all juvenile arrests, with Black males outnumbering females. Very little is known from research about the long-term consequences of such contact on their odds of college enrollment. Thus, the purpose of this study was to measure the relationship between Black males’ early contact with the criminal justice system through arrest and their probability of enrolling in a four-year college using a nationally representative sample of approximately 1,100 Black males who participated in the National Longitudinal Study of Youth (1997). Survey data were analyzed using descriptive, chi-square, and hierarchical binomial logistic regression techniques. Results expose pervasive limits on Black males’ college enrollment, reveal the statistically significant influence of early arrest on college entry, and have far-reaching implications for research, policy, and outreach. In light of mounting concerns about the United States’ diminishing ability to compete globally, state and federal policy makers have adopted several education reform strategies designed to increase the number of students graduating from Spectrum, 4(1), 49–72. Copyright © 2016, Trustees of Indiana University and The Ohio State University. doi:10.2979/spectrum.4.1.04 50 SPECTRUM 4.1 one of the more than 4,200 colleges and universities in the country. For instance, in his very first State of the Union address, President Barack Obama (2009) outlined an ambitious goal to once again have the world’s largest share of college graduates by 2020. His justification was simple: “A good education is no longer just a pathway to opportunity—it is a prerequisite. . . every American will need to get more than a high school diploma.” Indeed, higher education has become the most direct pathway to ensuring economic and social mobility for the individual, and increasing the number of Americans with college degrees has significant benefits for society (Haveman & Smeeding, 2006). To meet President Obama’s “2020 Completion Goal,” significantly more students must enroll in higher education, particularly those from traditionally underrepresented populations (Complete College America, n.d.). And while national figures show increases for some groups, enrollment for Black males has stalled for several years.1 Data from the National Center for Education Statistics (NCES) indicate that Black males accounted for only 5.3% of all students at undergraduate institutions in 2012 (US Department of Education, 2013)—nearly the same proportion as in 1976 (Harper, 2006; Strayhorn, 2008c). Low enrollment rates cannot be explained by the talent pool alone, since national data indicate that upward of 15% of Black men in the US are college-age and many never attempt higher education (U.S. Census Bureau, 2012). Scholarly research on Black male college enrollment is replete with references to factors that may drive “achievement gaps” for Black males, such as pre-college preparation (e.g., Palmer & Young, 2009; Polite, 1999), academic tracking (e.g., Palmer, Davis, & Hilton, 2009; Oakes, Gamoran, & Page, 1992), and low expectations from teachers (Ferguson, 2003; Kunjufu, 1986; Strayhorn, 2008d), to name a few. One area that has yet to be sufficiently examined in prior research is the experiences of Black males who come into contact with the criminal justice system prior to entering or enrolling in college, which may significantly reduce their odds of enrolling or finding success in higher education (Strayhorn, Johnson, & Barrett, 2013). Recent data indicate that Black youth represent just 16% of all public school students in the US but constitute 31% of all arrested youth, with Black males outnumbering females (Rovner, 2014). Existing research, though limited, suggests that early contact with the criminal justice system, such as an arrest, has negative educational consequences for all youth generally, and Black youth specifically, although relatively little is known about the relationship between arrest and college enrollment (e.g., Hirschfield, 2009; Kirk & Sampson, 2013). Studies that do focus on college outcomes offer limited insight into the condition of Black males, who are overrepresented among juvenile arrests, school suspensions, and expulsions, Johnson / Influence of Juvenile Arrest 51 yet make up only 5% of collegians in the US (Irvine, 1990; Noguera, 1997; Palmer, Wood, Dancy, & Strayhorn, 2014). Though scholarly research is limited, there is theoretical support for hypothesizing a relationship between juvenile arrest and college enrollment. For instance, life course theory of cumulative disadvantage (LCTCD), which draws on the assumptions of both social control theory (Hirschi, 1969) and labeling theory (Becker, 1963; Lemert, 1951), suggests that an arrest could serve as a negative turning point in one’s life course, leading to a series of detachment processes that increase one’s likelihood of school dropout. Thus, it is reasonable to believe that Black males’ arrest experiences as juveniles have negative consequences for their likelihood of enrolling in college, which was the focus of this study. PURPOSE OF STUDY The purpose of this study was to test the relationship between Black males’ early contact with the criminal justice system through juvenile arrest and four-year college enrollment using a nationally representative sample of approximately 1,100 Black males who participated in the National Longitudinal Study of Youth (NLSY:97). Specifically, a battery of statistical controls were employed to isolate and test the predictive validity of Black male arrest history on their probability of enrolling in a four-year college in 2003. Two central research questions guided this study: 1. Are there significant differences between Black males who report being arrested as a youth and those who do not in terms of four-year college enrollment? 2. Controlling for a battery of background and family factors, does Black male youth arrest status significantly predict enrollment in a four-year college? BRIEF REVIEW OF LITERATURE To conduct this study, it was necessary to review literature in two areas of inquiry: (a) what we know from research about the influence of juvenile arrest on education outcomes and (b) what we know from research about Black males’ pathways to college. This organization served as a useful frame for reviewing existing literature. Influence of Juvenile Arrest on Educational Outcomes The weight of empirical evidence offers compelling arguments about the negative influence of early criminal justice contact, such as juvenile arrest, on 52 SPECTRUM 4.1 educational outcomes—the majority of which focus on high school (e.g., De Li, 1999; Hirschfield, 2009; Lochner, 2004; Tanner, Davies, & O’Grady, 1999). For instance, Hjalmarsson (2008) analyzed data from a nationally representative sample (N = 7,417) of youth who participated in NLSY:97 to test the relationship between juvenile justice system interactions and high school graduation. The author reported that arrested youth were 11 times less likely to graduate high school than non-arrested youth—results were not disaggregated by race and sex. Sweeten (2006) drew similar conclusions for Black and Latino youth using NLSY:97 data to test the effect of first-time arrest and court involvement during high school on educational attainment. He noted, “first-time arrest during high school nearly doubles the odds of high school dropout, while a court appearance nearly quadruples the odds of dropout” (p. 473). Though none of the studies report results disaggregated by both race and sex, a growing number of authors have focused attention on at-risk populations such as those from large cities or economically disadvantaged backgrounds—factors that significantly influence one’s likelihood of arrest and educational achievement (e.g., McDonough, 1997; Sampson & Laub, 1997; Strayhorn, 2009a). Hirschfield (2009) analyzed a sample of nearly 4,900 students from predominantly minority neighborhoods in Chicago, controlling for a number of confounding factors such as race, their parents’ educational background and expectations, delinquency, and prior academic achievement. Results from his study indicate that students who were arrested during the 9th or 10th grade were six to eight times more likely to dropout of high school than those who reported no arrests. Likewise, Bernburg and Krohn (2003) found that juvenile arrest experiences significantly reduced students’ odds of graduating from high school by more than 70%, examining data from a sample (N= 605) of seventh- and eight-grade males from Rochester, New York, public schools. Though their results were not disaggregated by race, nearly 70% of participants in their sample identified as African American. Research related to juvenile arrest and college outcomes is virtually nonexistent. Kirk and Sampson (2013) provided an initial foray into this line of inquiry, estimating the direct effect of arrest on later high school dropout and college enrollment for youth, suggesting that arrest has “severe consequences for the prospects of [college] educational attainment” (p. 47). Youth with arrest records in their study had only a 0.18 probability of enrolling in a four-year college compared to nonarrestees, who had a probability of college enrollment equal to 0.34. The authors concluded, “. . . arrest in adolescence hinders the transition to adulthood by undermining pathways to educational attainment” (p. 19). The following section reviews what we know from research about Black males’ pathways to higher education. Johnson / Influence of Juvenile Arrest 53 Black Males’ Pathways to Higher Education Black males encounter significant challenges along the education pipeline that collectively reduce their odds of college enrollment (e.g., Cuyjet & Associates, 2006; Ferguson, 2003; Ford, 1998; Irvine, 1990; Jackson & Moore, 2008; Jenkins, 2006; Polite, 1999; Steele, 1997; Strayhorn, 2008b). One line of inquiry directs attention to the role of teachers, who often maintain low or negative expectations of Black males (e.g., Kunjufu, 1986; Wood, Kaplan, & McLoyd, 2007). Strayhorn (2008d) examined the relationship between teacher expectations and academic achievement among urban Black males who responded to the National Education Longitudinal ...
Purchase answer to see full attachment