Journal of Youth and Adolescence

, Volume 42, Issue 3, pp 454–465

Adolescent Neglect, Juvenile Delinquency and the Risk of Recidivism

Authors

    • School of Social WorkUniversity of Michigan
  • Abigail B. Williams
    • School of Social WorkUniversity of Michigan
  • Mark E. Courtney
    • School of Social Service AdministrationUniversity of Chicago
Empirical Research

DOI: 10.1007/s10964-013-9906-8

Cite this article as:
Ryan, J.P., Williams, A.B. & Courtney, M.E. J Youth Adolescence (2013) 42: 454. doi:10.1007/s10964-013-9906-8

Abstract

Victims of child abuse and neglect are at an increased risk of involvement with the juvenile justice and adult correctional systems. Yet, little is known about the continuation and trajectories of offending beyond initial contact with law enforcement. Neglect likely plays a critical role in continued offending as parental monitoring, parental rejection and family relationships are instrumental in explaining juvenile conduct problems. This study sought to determine whether neglect is associated with recidivism for moderate and high risk juvenile offenders in Washington State. Statewide risk assessments and administrative records for child welfare, juvenile justice, and adult corrections were analyzed. The sample was diverse (24 % female, 13 % African American, 8 % Hispanic, 5 % Native American) and included all moderate and high risk juvenile offenders screened by juvenile probation between 2004 and 2007 (n = 19,833). Official records from child protection were used to identify juvenile offenders with a history of child neglect and to identify juvenile offenders with an ongoing case of neglect. Event history models were developed to estimate the risk of subsequent offending. Adolescents with an ongoing case neglect were significantly more likely to continue offending as compared with youth with no official history of neglect. These findings remain even after controlling for a wide range of family, peer, academic, mental health, and substance abuse covariates. Interrupting trajectories of offending is a primary focus of juvenile justice. The findings of the current study indicate that ongoing dependency issues play a critical role in explaining the outcomes achieved for adolescents in juvenile justice settings. The implications for improved collaboration between child welfare and juvenile justice are discussed.

Keywords

NeglectDelinquencyRecidivismAdolescence

Introduction

Juvenile offending and especially repeat juvenile offending is a serious public health concern. In 2009, law enforcement agencies in the United States arrested approximately 1.9 million persons under 18 years of age. Although this estimate represents a 17 % decrease from 2000, juvenile offenders continue to account for 15 % of all violent crimes and 24 % of all property crimes (OJJDP 2011a, b). Throughout the literature, child and adolescent maltreatment are consistently identified as powerful predictors of juvenile and adult crime (Widom 1989; Smith and Thornberry 1995; Ireland et al. 2002; Ryan and Testa 2005). Yet, little is known about whether various forms of maltreatment or ongoing involvement with child protection increases the risk of continued offending. This lack of knowledge is critical as state agencies seek to develop models of shared services between child welfare and juvenile justice and as program administrators seek to identify specific areas for targeted intervention. The current study addresses this important gap in the literature and helps advance the knowledge base as it relates to maltreatment and repeat juvenile offenders. We focus specifically on cases of neglect.

Physical abuse receives significantly more attention in the public discourse surrounding the broad category of child maltreatment as compared with neglect (Erickson and Egeland 2002). Yet, neglect is by far the most frequently investigated allegation by child protection and the most common reason cited for the placement of children in foster care. In the United States, approximately 78 % of the allegations reported to child protection involve neglect; as compared with 18 % for physical abuse, 10 % for sexual abuse and 8 % for psychological/emotional maltreatment (USDHHS 2009). Neglect refers to the “failure by the caregiver to provide needed, age-appropriate care although financially able to do so or offered financial or other means to do so” (USDHHS 2007). The Department of Health and Human Services (2011) defines neglect as “the failure of a parent or other person with responsibility for the child to provide needed food, clothing, shelter, medical care, or supervision to the degree that the child’s health, safety, and well-being are threatened with harm (p. 3).” Not surprisingly, poverty is a significant confounding factor in the conceptualization of neglect. To help differentiate the concepts, legal scholars and child welfare practitioners have long focused on the behaviors of parents rather than focusing only on the lack of resources within the family home (Wald 1976). One such behavior, lack of adequate supervision, is a critical component of neglect. Lack of supervision is defined as child(en) inadequately supervised for extended periods of time or allowed to remain away from the home overnight without the parent knowing (or more importantly attempting to determine) the child’s whereabouts. In early childhood, lack of supervision often results in traumatic injuries or the ingestion of hazardous materials (Kerr et al. 2010; Hymel 2006). As children move through adolescence and as families spend less time in close physical proximity, vigilant supervision is diminished. It is at this particular developmental transition that the terminology noted in the scientific literature shifts from neglect and parental supervision to parental monitoring. Although the terminology changes as children mature through adolescence, the concepts are similar and remain equally important.

Parental monitoring is noted throughout the criminological literature as instrumental in the development of juvenile offending. Approximately three decades ago, Loeber and Stouthamer–Loeber (1986) reported that parental monitoring was one of the strongest predictors of juvenile conduct problems and delinquency, even stronger than parental criminality, marital relations, parental discipline and parental absence. Similar findings are reported in more recent studies. In a meta-analysis of 161 published and unpublished studies, parenting behaviors in gernal (including but not limited to parental supervision and monitoring) emerged as the strongest predictor of juvenile delinquency (Hoeve et al. 2009). In part, the literature on parenting behaviors and parental monitoring reflect aspects of the social control literature. Healthy development is dependent upon parents and other socializing agents making consistent investments in the care, education, and supervision of children. Consistent investments instill a sense of attachment and commitment that tie children to family members and conventional role models. Social control theorists posit that these investments and social bonds prevent delinquency. When confronted with opportunities to engage in nonconforming or undesirable behaviors, children with extensive and strong social bonds have a greater stake in conformity and are less likely to engage in delinquent behavior that might jeopardize those relationships (Furstenberg and Hughes 1995). Although a substantial proportion of the delinquency population likely experience low levels of parental investment, adolescents in the juvenile justice system with a substantiated and ongoing case of neglect seem especially vulnerable for continued involvement (i.e., recidivism) with the juvenile justice and even adult correctional systems. Thus, even after controlling for a wide range of important covariates, we hypothesize that the rate of recidivism will be significantly higher for neglected youth in the juvenile justice system as compared with similar delinquent offenders not involved with an official report of neglect.

Several terms are used throughout the literature to capture maltreated or neglected youth in the juvenile justice system. Examples of such terms include crossover, dually involved, dual jurisdiction and dual jacket youth. Unfortunately, there is very little consistency in the use of terms across studies. Crossover youth represent the broadest category and include youth involved with both child welfare and juvenile justice services at any point in time (Herz et al. 2010). Crossover youth may start in child welfare and move to juvenile justice or start in juvenile justice and move to child welfare. In the current study, we use the term crossover to represent neglect cases that were closed in child welfare prior to the juvenile arrest. We use the term dually involved to represent the population of youth that have simultaneously open and active cases in both child welfare and juvenile justice.

The Current Study

We investigate the likelihood of recidivism for both crossover and dually involved youth. We hypothesize that dually involved youth are at greatest risk for recidivism. Three arguments help to inform our hypothesis. First, a serious disconnect often exists between child welfare and juvenile justice systems. Three policy options exist for youth involved with child welfare and juvenile justice settings in the United States: concurrent jurisdiction, on-hold jurisdiction and separated jurisdiction. Concurrent jurisdiction refers to cases that are simultaneously open and serviced by both child welfare and juvenile justice systems. This is by far the most common and perhaps preferred model in the United States (Herz et al. 2010). On-hold jurisdictional models refer to the temporary suspension of dependency court proceedings—intended only to be a brief interruption until the delinquency court jurisdiction is terminated. Separate jurisdiction refers to the closing of a case, either on the dependency or delinquency side of the court. All services are then provided by either child welfare or juvenile justice. Washington State, the site for the current study operates with concurrent jurisdiction. Although concurrent jurisdiction is often the preferred model, conflicts surrounding information sharing and conflicting agency missions could result in unmet service needs (i.e., service gaps) and thus increase the complexity of these already difficult to manage cases. Second, cases that are simultaneously involved with both child welfare and juvenile justice are subjected to increased scrutiny from both probation officers and child welfare caseworkers. A likely consequence of increased scrutiny and surveillance is an increased risk of arrest, even if offending behaviors are similar across comparison groups. Third, the consequences associated with neglect are well documented. Studies of developmental traumatology indicate long lasting changes to a child’s biological system and cognitive functioning (Mills et al. 2010; Lanius et al. 2011). Such changes might result in equal risk of continued offending for all cases involving neglect, regardless of whether such cases are open or closed at the time of initial arrest. Yet, we argue that while children associated with closed cases may still experience some levels of neglect, a case closed by child welfare ought to indicate some level of improvement relative to cases that remain open. The open cases will likely report higher levels of family conflict and individual risk. From the intervention literature specific to neglect, indicators of improvement include increased attachment security, reduced child disorganization, lower levels of internalizing and externalizing behaviors and increased parental sensitivity (Mossal and Dubois-Comtois 2011). The analyses of the current article focus on this last argument and help inform our hypothesis that dually involved youth (specifically youth that are simultaneously neglected and delinquent) will be at the greatest risk for a subsequent arrest.

Methods

Data, Sample and Procedures

Several sources of data were made available via a unique statewide data sharing agreement. The data sharing agreement provided access to all administrative records for all youth associated with at least one official child protection investigation in the State of Washington between January 1, 2000 and December 31, 2009 (n = 252,057). The child welfare records include demographic information (birthdates, race, gender, geographic region), allegations of maltreatment (report date, type of maltreatment, finding), and child welfare services (placement dates, placement types, reasons for placement). The reasons for substitute care placement include sexual abuse, physical abuse, neglect, parental substance abuse, child disability, inadequate housing, abandonment, and child behavioral problems.

The delinquency records originate with the Washington State Center for Court Research. The Center records included all charges (n = 10,227,036) for all minors (n = 2,387,484) in the State of Washington between January 1, 2000 and June 30, 2009. The arrest records include youth demographics, arrest date, offense type, and judicial disposition. The arrest records capture both juvenile and adult offending. The child welfare and arrest records did not share a common unique identifier (e.g. social security number) and were thus linked by common identifiers (last name, first name, date of birth, race, ethnicity, gender) using probabilistic matching software. Probabilistic matching programs compare variables from diverse data sources to identify common individuals. A weight related to the probability that two records refer to the same individual was assigned to each pair. Probabilistic matching using likelihood ratio theory to identify the most likely comparison. This approach is believed to be superior in comparison with deterministic matching procedures (Schumacher 2007). Yet, probabilistic methods are not without limitation, as such methods can overestimate or underestimate the true overlap between data sources. The quality and success of the match are dependent on a number of factors including file size, number of shared variables and the discriminating power inherent in the matched variables (Cook et al. 2001). Given the relatively large number of observations in the analytic files, we are confident that probabilistic matching was the best approach for the available data.

Our sample was comprised of moderate and high risk offenders completing the full WSCJA between 2004 and 2007 (n = 19,833). Of this sample, 13,923 (70 %) were not associated with any formal record of neglect with the Washington State child welfare system, 3,900 (20 %) had at least one prior substantiated allegation of neglect but no open case at the time of arrest, and 2,010 (10 %) had at least prior substantiated allegation of neglect and an open child welfare case at the time of arrest. Of the 19,833 youth in the sample, 12 % were African American, 8 % Hispanic, 68 % White, 5 % American Indian, and 3 % Asian. On average, youth were 15.9 years old at the time of assessment and 76 % of the overall sample is male.

Measures

The measures of risk are part of the Washington State Juvenile Court Assessment (WSJCA). The WSJJCA was developed in 1997 by a group of international and interdisciplinary experts and was in part derived from a modified version of Baird’s 1984 Wisconsin Risk Scale (WSIPP 2004). Washington State uses risk scores to help determine placement security (minimum to maximum) and level of community supervision (low to intensive) (WSIPP 1999). The WSJCA is initiated when an adolescent is brought to court on a new offense. A juvenile probation counselor works with the youth and family to complete a 27 pre-screen. The pre-screen produces a score for social and criminal histories that are combined to determine low, moderate and high risk level. If the youth scores in the moderate or high range, the probation counselor completes the full risk assessment (127 items) via a structured motivational interview with both the juvenile offender and his/her family. In the current study, we analyze data associated with the full risk assessment and thus our sample is limited to moderate and high risk offenders.

The risk assessment is considered a process (rather than a single event) and often requires several meetings. The risk assessment captures risk factors that are both static and dynamic. Static factors represent historical life events that are not subject to change; the experience of child abuse is one example. In contrast, dynamic factors represent circumstances in life that have the potential for change; the level of academic achievement in school is one example. In the current study, we focus on the static and dynamic factors associated with the following domains: family, criminal history, school, peers, alcohol and drugs, mental health and attitudes and behaviors.

Family

The ten family items focused on level of support, closeness with mom/dad, verbal intimidation in the home, physical violence in the home, consistency of supervision, appropriateness of punishment, parental use of alcohol, parental drug use and family income. Sample items included “family willing to help and support youth” and “domestic and physical violence exist between parents and youth.”

Educational

The seven school items focused on behavioral problems, attendance, achievement, extra curricula activities, perceived value of school, probation officer prediction of graduation and expulsions. Sample items included “youth is associated with behavioral problems reported by teachers in the most recent term” and “probation officer thinks youth will likely graduate from high school or vocation school.”

Peers

The four peer items focused on number of friends, associated with pro-social and anti-social friends, and gang membership. Sample items included “youth has never had consistent friends or companions,” “youth currently spends time with pro-social friends” and “youth currently spends time with anti-social friends.”

Alcohol and Drugs

The alcohol and drug items focused on current use and the mental health item focused on the presence of a diagnosis. Sample items included “youth is currently using alcohol or drugs” and “youth currently has a mental health diagnosis.”

Individual Attitudes and Beliefs

The items on beliefs focused on pro-social norms, level of optimism and perceptions of physical aggression as an acceptable means to resolve disagreement or conflict. Sample items included “youth currently is impulsive, acts before thinking,” “youth currently has trouble controlling aggression,” “youth believes physical aggression is an appropriate way to resolve disagreements” and “youth talks about the future in positive ways.”

The individual items and the possible responses are displayed in the Appendix. We recoded the individual items into a series of dichotomous variables, so that a value of “1” indicates an affirmative response. For example, a value of “1” associated with “anti-social friends” indicates the youth identified having anti-social friends.

Neglect

The current study focuses on neglect, the timing of neglect and recidivism. To capture and investigate the variation in the timing of neglect we used the date of the substantiated allegation of neglect, the child welfare case opening and case closing dates and the juvenile arrest date. The first group captured adolescents with no official history of neglect. The second group captured adolescents with a history of child neglect, but at the time of arrest, the case was closed with child protection. The third group captured adolescents with an ongoing and open case of neglect at the time of arrest. We labeled these youth as dually involved, as a way to capture their simultaneous involvement with both child welfare and juvenile justice. As the current study focuses specifically on neglect and delinquency, our sample does not include youth with a substantiated history of physical or sexual abuse.

Recidivism

Our primary interest was investigating the risk of continued offending (i.e., recidivism) across each of these unique groups. The dependent measure (recidivism) was the first arrest (as a juvenile or adult) subsequent to completion of the full risk assessment. We excluded status offenses, technical violations and traffic violations from all analyses. We recognize that all measures of recidivism have limitations. These limitations are well articulated in a recent study commission by the Council of Juvenile Court Administrators (Harris et al. 2009). Measuring recidivism via arrests both overestimates and underestimates the actual number of individuals committing new offenses, as all charges do not lead to adjudications/convictions and all crimes are not known to law enforcement. An ideal measure of recidivism would include interviews with offenders to determine the extent of subsequent crimes committed. Unfortunately such data were not available.

We used cross-tabulation, 2 and t tests to explore the differences between the three delinquency groups. We used event history analysis to examine the influence of individual variables on recidivism rates. This analytic technique is similar to logistic regression in that it enables one to calculate the odds of a particular event occurring. However, survival analysis considers the differential impact between groups on the timing of this event. In the current study, youth entered and remained in the observation period (2004–2009) for different length of time. Thus, their exposure to the risk of a subsequent offense varied. We calculated this risk period as the number of days between the placement start date and the final day of observation (June 30, 2009). The average time at risk of arrest was 3.7 years (1,357 days), and the minimum period of risk was 1.5 years (548 days).

Results

We compare youth demographics and risk assessment scores across the three delinquency groups (see Table 1). Delinquency only refers to youth with no official history of neglect. Crossover refers to youth with at least one substantiated allegation of neglect prior to the initial juvenile arrest, but the child welfare case is closed at the time of initial arrest. Dually involved refers to youth with an open and ongoing case with child welfare and a substantiated allegation of neglect at the time of initial arrest.
Table 1

Descriptive statistics for overall sample, crossover and dually involved subgroups (n = 19,833)

 

Overall

Delinquency only

Crossover

Dually involved

n (%)

n (%)

n (%)

n (%)

Child characteristics

 Age at arrest (M, SD)

15.9, 1.5

16.1, 1.4

15.6, 1.5

14.9, 1.5

 Female

4,775 (24)

2,814 (20)

1,243 (32)

718 (36)

 Male

15,051 (76)

11,105 (80)

2,656 (68)

1,290 (64)

 African American

2,412 (13)

1,711 (13)

471 (13)

230 (12)

 Hispanic

1,609 (8)

1,118 (8)

336 (9)

155 (8)

 Asian

671 (4)

535 (4)

88 (2)

48 (2)

 Native American

888 (5)

588 (4)

206 (5)

94 (5)

 White

13,555 (71)

9,404 (70)

2,704 (71)

1,447 (73)

 Prior misdemeanor

10,507 (53)

7,253 (52)

2,191 (55)

1,063 (53)

 Prior Felony

2,602 (13)

1,900 (14)

499 (13)

203 (10)

Family

 Supportive

11,941 (62)

9,093 (67)

1,956 (52)

892 (45)

 Close with mom

10,923 (57)

7,972 (59)

1,956 (52)

995 (51)

 Close with dad

5,059 (26)

3,862 (29)

806 (22)

391 (20)

 Verbal intimidation

7,873 (41)

5,438 (40)

1,639 (44)

796 (41)

 Physical violence

3,984 (21)

2,386 (18)

915 (24)

683 (35)

 Consistent supervision

8,534 (43)

6,424 (46)

1,444 (37)

666 (33)

 Appropriate punishment

9,104 (48)

6,917 (51)

1,503 (40)

684 (35)

 Parental alcohol

4,630 (24)

2,895 (21)

1,089 (29)

646 (33)

 Parental drug

3,652 (19)

2,007 (15)

1,009 (27)

636 (32)

 Income <35 k

12,623 (67)

8,208 (62)

2,854 (77)

1,561 (81)

School

 Ex-activities

3,375 (18)

2,481 (19)

590 (16)

286 (15)

 Behavioral problems

12,916 (67)

8,721 (65)

2,693 (71)

1,502 (76)

 Expelled

9,717 (51)

1,064 (54)

2,000 (53)

1,064 (54)

 Good attendance

5,678 (30)

4,096 (31)

1,052 (28)

530 (27)

 Good performance “C”

5,840 (31)

4,273 (32)

1,053 (28)

514 (26)

 Likely graduate

5,239 (27)

4,105 (30)

795 (20)

339 (17)

 Value in school

7,641 (40)

5,654 (42)

1,319 (35)

668 (34)

Peer

 No friends

1,724 (9)

995 (7)

460 (12)

269 (14)

 Pro social friends

13,894 (70)

10,149 (73)

2,487 (64)

1,258 (63)

 Anti social friends

15,172 (77)

10,683 (77)

2,954 (76)

1,534 (76)

 Peers gang involved

2,622 (13)

1,852 (13)

512 (13)

258 (13)

Youth

 Alcohol/drug

13,243 (67)

9,399 (68)

2,547 (65)

1,297 (65)

 Mental health diag.

4,478 (23)

2,751 (20)

1.067 (27)

660 (33)

 Impulsive

9,218 (47)

5,928 (43)

2,085 (54)

1,205 (60)

 Ability to control

8,657 (44)

6,502 (47)

1,520 (39)

635 (32)

 Belief social norms

15,964 (81)

11,457 (83)

2,999 (76)

1,508 (75)

 Belief physical aggression

9,402 (48)

6,174 (44)

2,057 (53)

1,171 (58)

 Optimistic

13,739 (69)

10,053 (72)

2,493 (64)

1,193 (59)

In the comparison of individual demographics, gender and age varied between the three groups. The populations coming to juvenile justice with a history of child welfare involvement include a significantly greater proportion of girls. This finding is similar to previous studies of child welfare and juvenile justice involvement for girls (Ryan et al. 2007). Youth coming to the juvenile justice system with no official reports of neglect with child welfare were significantly older than both the crossover and dually involved youth (F(2, 19,831) = 541.9, p < .01). In the family domain, adolescents coming to juvenile probation with no formal child welfare involvement report higher levels of family support (χ2 (2, n = 19,833) = 546.9, p < .01), family income (χ2 (2, n = 19,833) = 199.3, p < .01), supervision (χ2 (2, n = 19,833) = 196.8, p < .01) and appropriate punishment (χ2 (2, n = 19,833) = 279.8, p < .01), and lower levels of physical violence (χ2 (2, n = 19,833) = 186.8, p < .01) and parental alcohol and drug use in the home (χ2 (2, n = 19,833) = 186.7, p < .01). In the academic domain, adolescents with no formal child welfare involvement reported fewer in-school behavioral problems (χ2 (2, n = 19,833) = 202.5, p < .01) and were perceived (from the probation officer perspective) as more likely to graduate high school (χ2 (2, n = 19,833) = 237.4, p < .01). There were few differences with regard to peers, but neglected youth reported fewer pro social friends (χ2 (2, n = 19,833) = 183.2, p < .01). Several important differences emerged with regard to health, attitudes and beliefs. Neglected youth were more likely to have a mental health diagnosis (χ2 (2, n = 19,833) = 237.6, p < .01), more likely to report impulsivity (χ2 (2, n = 19,833) = 209.1, p < .01) and more likely to believe physical aggression is an appropriate response to resolve disagreement (χ2 (2, n = 19,833) = 103.2, p < .01). Neglected youth were also less likely to report an ability to control anti-social behavior (χ2 (2, n = 19,833) = 109.5, p < .01), less likely to believe that pro-social rules/conventions apply to him/her (χ2 (2, n = 19,833) = 108.1, p < .01) and less likely to report a sense of optimism (i.e., sense of purpose, commitment to a better life) (χ2 (2, n = 19,833) = 209.2, p < .01).

For the most part, there were few differences when comparing the crossover and dually involved groups. Yet, some important differences emerged and seem to indicate greater risk associated with the youth that were simultaneously open in both child welfare and juvenile justice. Dually involved youth reported lower levels of family support (χ2 (1, n = 5,907) = 23.8, p < .01), optimism for the future (χ2 (1, n = 5,907) = 12.5, p < .01) and higher levels of physical violence (χ2 (1, n = 5,907) = 16.0, p < .01) and parental drug use (χ2 (1, n = 5,907) = 8.9, p < .01) in the home as compared with crossover youth.

To provide a visual representation of the timing of recidivism, we developed a life table with the three unique groups (see Fig. 1). Several findings are worth noting. The trends indicate a steady accumulation of re-offending over time. Second, a relatively high proportion of youth recidivate within a fairly short period of time. In fact, the majority of all youth experience a subsequent arrest. Third, youth coming to the juvenile justice system with an open and active neglect case (dually involved) were at greatest risk of recidivism. Within 18 months from the date of initial arrest, approximately 61 % experience a subsequent arrest. In comparison, 51 % of crossover youth and 49 % of delinquent youth recidivated within the same time period. A majority (67 %) of the subsequent offenses occurred prior to the youth turning 18 years of age.
https://static-content.springer.com/image/art%3A10.1007%2Fs10964-013-9906-8/MediaObjects/10964_2013_9906_Fig1_HTML.gif
Fig. 1

Time between risk assessment and subsequent offense by neglect status (n = 19,833)

The results from the Cox regression are displayed in Table 2. The table includes the coefficient and standard error for each independent variable as well as the hazard ratio (Exp(β)). A hazard ratio greater than 1 indicates a higher likelihood of recidivism. If 1 is subtracted from the hazard ratio and the remainder is multiplied by 100, the resultant is equal to the percentage change in the hazard of arrest. We developed three cox regression models. The first model included only the neglect status. Delinquent youth with no official history of neglect served as the reference group. The second model included youth demographics and the third model included various family, school, peer and youth items associated with the risk assessment instrument. In the first and second model, youth associated with an official report of neglect were significantly more likely to experience a subsequent arrest. Yet, this association changed as we included the family, school, peer and youth covariates from the risk assessment. In the full model and controlling for a wide range of important factors, only dually involved youth were at an increased risk of recidivism. Crossover youth were no more likely to recidivate as compared with other juvenile offenders coming to the justice system with no documented history of neglect.
Table 2

Cox regression: neglect status and likelihood of continued offending (n = 19,833)

Independent variables

Model 1

Model 2

Model 3

β

SE

Exp(β)

β

SE

Exp(β)

β

SE

Exp(β)

Neglect status

 Crossover

0.05*

0.02

1.05

.06*

0.02

1.06

0.05

0.02

1.01

 Dually involved

0.18*

0.03

1.2

.24*

0.03

1.26

.16*

0.03

1.17

Child characteristics

 Age at arrest

   

.01

0.01

1.01

.01

0.01

0.99

 African American

   

.30*

0.03

1.35

.24*

0.03

1.27

 Hispanic

   

.41*

0.03

1.5

.28*

0.03

1.33

 Asian

   

−0.05

0.05

0.95

−.11*

0.05

0.89

 Native American

   

.10*

0.04

1.11

0.01

0.04

1.01

 Male

   

.36*

0.02

1.42

.40*

0.02

1.49

 Age 14

   

0.01

0.04

1.00

−0.3

0.04

0.97

 Age 15

   

0.02

0.04

1.01

−0.4

0.04

0.96

 Age 16

   

−0.01

0.04

0.99

−0.7

0.04

0.93

 Age 17

   

−0.02

0.04

0.98

−0.5

0.04

0.95

 Age 18

   

−0.07

0.04

0.93

−0.7

0.04

0.94

 Misdemeanor

   

.41*

0.02

1.5

.31*

0.02

1.37

 Felony

   

.34*

0.03

1.4

.27*

0.03

1.31

Family

 Supportive

      

−.05*

0.03

0.94

 Close w/mom

      

−0.02

0.02

0.98

 Close w/dad

      

−0.03

0.02

0.97

 Verbal intimidation

      

.06*

0.02

1.06

 Physical violence

      

.09*

0.02

1.1

 Consistent supervision

      

−.06*

0.02

0.93

 Appropriate punishment

      

−.06*

0.03

0.94

 Parental alcohol

      

0.02

0.02

1.02

 Parent drug

      

−0.03

0.03

0.97

 Income <35 k

      

.08*

0.02

1.08

School

 Ex-activities

      

−.07*

0.03

0.93

 Behavioral problems

      

.10*

0.03

1.11

 Expelled

      

.05*

0.02

1.05

 Good attendance

      

−.08*

0.03

0.93

 Good performance C

      

0.04

0.03

1.04

 Likely graduate

      

−.14*

0.03

0.87

 Believes value in school

      

0.03

0.02

1.03

Peer

 No friends

      

−.09*

0.05

0.91

 Pro social friends

      

−.07*

0.02

0.93

 Anti social friends

      

.14*

0.03

1.15

 Gang

      

.14*

0.03

1.15

Youth

 Alcohol/drug

      

.14*

0.02

1.15

 Mental health diag.

      

−0.03

0.02

0.98

 Impulsive

      

.04*

0.02

1.04

 Ability to control

      

−.08*

0.02

0.93

 Belief social norms

      

−.07*

0.03

0.93

 Belief phys. aggression

      

.06*

0.02

1.06

 Optimistic

      

−0.05*

0.02

0.95

p < .01

Regarding additional main effects, the findings indicate that males were significantly more likely to experience a subsequent arrest. African American and Hispanic youth were also more likely to experience a subsequent arrest, even after controlling for a wide variety of important covariates such as income, prior arrests, supportive home environment, school attendance, peer relationships and alcohol/drug use. Relative to other covariates in the model, the coefficients associated with race were fairly large. Within the family domain and related to our primary questions of interest, consistent supervision was a significant predictor of recidivism. The hazard decreased by 7 % when youth reported consistent supervision in the home. Peer associations and alcohol/drug use also emerged as relatively important predictors of recidivism. Youth associated with anti-social peers and youth associated with gangs were significantly more likely to recidivate. Similarly, youth associated with alcohol and drugs were significantly more likely to recidivate.

Discussion

A substantial proportion of adolescents enter the juvenile justice system with a history of neglect and approximately one-third of these adolescents are still associated with an active child welfare case at the time of the initial arrest. Identifying youth with an active child welfare case is critical for justice systems as they seek to understand and address specific family dynamics that may directly interfere with efforts to interrupt offending behaviors. We report that crossover youth—that is closed neglect cases with child welfare—were at no greater risk of recidivism as compared with juvenile offenders without an officially documented history of neglect. However, open neglect cases at the time of arrest were at an increased risk of future offending. Dually involved adolescents were significantly more likely to re-offend, even after controlling for a wide range of individual, school, family and peer characteristics. These findings raise questions about the role of ongoing child maltreatment in offending, about the need for collaboration between child welfare and juvenile justice agencies and about specific areas of risk and resilience that ought to be targets for intervention.

Regarding the role of ongoing maltreatment in offending, there is an accumulating body of knowledge indicating that the timing of maltreatment matters in the etiology of delinquency. Using data from the Rochester Youth Development Study, the findings generally indicate that persistent maltreatment (that is, from childhood to adolescence) or maltreatment that only occurs during adolescence increases the risk of juvenile offending (Smith et al. 2004; Thornberry et al. 2001). Similar findings were reported in an Australian study of children born in Queensland (Stewart et al. 2008). Several hypotheses are proposed to help explain why the timing of maltreatment might matter so much in the development of delinquency. The authors of the Rochester studies offer two possible interpretations. First, the authors of the Rochester studies hypothesize that children may be more “developmentally resilient” as compared with adolescents. That is, young victims of abuse/neglect may experience short term negative outcomes, but the long term effects are limited once the abusive/neglectful conditions and behaviors are addressed. Second, interventions associated with child protection may be less available and less effective with adolescents as compared with the interventions targeted toward younger children (Smith et al. 2004). The authors associated with the Australian study suggest that perhaps adolescence is an especially vulnerable time because youth are also experiencing a host of other transitions—in particular school transitions (Stewart et al. 2008). When maltreatment occurs during such transitions, important academic and peer relationship are disrupted, thus increasing the risk of violence and delinquency (Bolger and Patterson 2001). All three of these hypotheses seem reasonable. Yet, we offer an alternative explanation that focuses less on the developmental response to maltreatment and more on the conceptual differences between child and adolescent maltreatment. These differences likely exist at the individual level and at the systems level.

We posit that dually-involved cases represent a distinct part of the child welfare services population and that neglect allegations associated with adolescents really mean something different than neglect allegations associated with younger children. At the individual level, adolescent neglect is not so much about inadequate supervision (as it would be for very young children) but perhaps more about parent–child conflict, incidents that might not involve physical or sexual abuse but that does nonetheless elicit the attention of child welfare authorities. The conceptualization of neglect thus varies by age. The neglect of a young child is often considered an act of omission—a term that captures parents not doing something to meet the child’s individual needs. In adolescence however, neglect might be better conceptualized as an act of commission. A heated argument resulting in the parent locking the adolescent out of the family home is an example of an incident that would likely attract interest from child protection and is a representative act of commission (Stein et al. 2009). At the agency level, social service systems would respond to these scenarios differently, as young children are often viewed as troubled and older children are more often viewed as troublesome (Cashmore 2011).

Our explanation is consistent with the current findings as the crossover cases were, among other differences, significantly younger at the time of their first allegation. The dually involved youth were not entirely comprised of cases that opened at a very young age and simply never closed out. These are cases that opened, on average, at later points in time as compared with crossover youth. Previous studies clearly demonstrate that older youth entering the child welfare system, and older youth entering for reasons other than maltreatment are at great risk of contact with the juvenile justice system (Ryan in press). The current study extends this knowledge base by demonstrating that older youth involved with both child welfare and juvenile justice are also significantly more likely to continue offending. This is an important finding as much of the maltreatment—delinquency literature simply focuses on a documented history of abuse/neglect and fails to consider whether or not the child welfare and juvenile justice cases are simultaneously open. In short, timing matters for both the initiation of delinquency (Ireland et al. 2002) and for the continuation of offending. Future investigations of child welfare and juvenile justice ought to pay close attention and in fact disentangle crossover and dually involved populations.

Regarding the need for collaboration between child welfare and juvenile justice systems, the current findings clearly indicate that ongoing dependency issues matter in the continuation of offending. This is consistent with much of the empirical evidence focused on positive family relationships. Healthy development requires parents making consistent investments in the care of children. A substantiated allegation of neglect might be an indicator of low investment. Although there exist a variety of evidence based family focused interventions, it is currently unclear how much emphasis is placed on family relationships and parental investments in juvenile justice, especially with offenders classified as moderate and high risk, as such offenders often experience placement in secure settings. In secure placements, and perhaps as a result of geographic proximity, the individual youth rather than the family is perceived as the sole client. This client orientation is problematic because the empirical evidence clearly indicates that many individually focused interventions (e.g. boot camps, individual counseling, behavioral token programs, wilderness challenge programs) are ineffective with high risk juvenile offenders (Greenwood 2006). In contrast, programs that motivate families to engage and participate in services (e.g. functional family therapy, multisystemic therapy and multidimensional treatment foster care) are likely to decrease the risk of recidivism (Greenwood 2006). The family context and parent–child relationships are critical because they serve as a foundation for the provision of quality care and represent an important predictor of healthy psychological development (Vuchinich et al. 2002; Kelly and McSherry 2002; Committee on Early Childhood, Adoption, and Dependent Care 2000). The findings of the current study indicate that the family system, and the manifestation of problems within that family (i.e., neglect) are indeed important predictors for future offending.

An important finding noted in the current study, and a finding with clear implications for policy and practice, is that juvenile justice systems ought to target interventions at individual youth and their respective family system. Specific to dually involved youth, this would require child welfare caseworkers and juvenile probation officers to work together across professional boundaries—as the child welfare issue (neglect) plays a critical role in the outcomes achieved in the juvenile justice system (desistance from offending). The main effects from the regression model support this approach. At the family level, consistent supervision, verbal intimidation and the perception of a supportive home environment were some of the largest contextual predictors of recidivism. Yet, the family system is not the only context that matters. The academic experiences (e.g., attendance, extra-curricular activities) and peer relationship (e.g., friends with pro or anti-social youth) also emerged as critical contextual mechanisms associated with continued offending.

Although outside the scope of the current study, a concerning finding continues to emerge in studies of child welfare youth and subsequent contact with the juvenile justice system. African American and Hispanic youth experience significantly higher rates of repeat contact with the juvenile justice system. In the current study, the hazards associated with recidivism increase by 27 % for African American youth and 33 % for Hispanic youth. This disparity cannot be explained by income, family status, academic status, peer networks or child welfare history—as we controlled for a wide range of covariates in each domain. Perhaps there are other more salient domains of risk and protective factors that matter for African American and Hispanic youth. Recidivism is an important predictor for long term criminal justice involvement. Thus, it is important for future studies of crossover and dually involved youth to focus specific attention on race and ethnicity.

Conclusion

The current study extends the literature by investigating the association between child welfare status and continued offending. To date, the vast majority of studies in the area of maltreatment and delinquency are limited to initial offending. Neglect, and in particular ongoing neglect, during adolescence play a critical role in the development of delinquency and in the continuation of offending trajectories for moderate and high risk adolescents involved with the juvenile justice system. Looking forward, cross system innovations are desperately needed that target youth and families straddling multiple state agencies. While there is a rapidly growing body of literature focused on identifying the risk of delinquency and adult crime for victims of abuse and neglect, the complimentary literature focused on interventions with this population remains thin. The development, implementation and rigorous evaluation of such interventions are priority.

Acknowledgments

This work was funded by a grant from the John D. and Catherine T. MacArthur Foundation as part of the Models for Change research initiative.

Author Contributions

JR conceived the study, helped secure the data, contributed to the analyses and writing; AW contributed to data analysis and writing; MC helped secure the data, contributed to our understanding of the child welfare system in Washington State and contributed to the writing. All authors read and approved the initial and revised manuscripts.

Copyright information

© Springer Science+Business Media New York 2013