enf ap 3

Article

Do Social Bonds Buffer the Impact of Adverse Childhood Experiences on Reoffending?

Jessica M. Craig 1 , Michael T. Baglivio

2 , Kevin T. Wolff

3 ,

Alex R. Piquero 4 , and Nathan Epps

5

Abstract Research from multiple disciplines has reported that exposure to childhood traumatic events, often referred to as adverse childhood experiences (ACEs), increases an individual’s chances of experi- encing a wide variety of negative consequences such as chronic disease, unemployment, and involvement in serious, violent, and chronic offending. The current study assesses how protective factors from social bonds may moderate the relationship between ACEs and future offending in a sample of high-risk adjudicated youth. While results showed that increased ACE exposure led to a higher likelihood of rearrest and more social bonds lowered the likelihood of rearrest, in contrast to expectations, the analyses revealed that stronger social bonds did not reduce the deleterious effects of exposure to more types of ACEs on recidivism. A discussion of these findings is offered, along with study limitations and future directions.

Keywords social bonds, childhood trauma, recidivism, adverse childhood experiences

Introduction

It has been well established that childhood victimization and trauma have an array of adverse

consequences in different life domains (cf. Felitti et al., 1998; Widom, 1989). Not surprisingly,

there has been an increased focus on the effect that these childhood traumatic events, often referred

to as adverse childhood experiences (ACEs), have on other outcomes later in life (Baglivio et al.,

2014; Baglivio, Wolff, Epps, & Nelson, 2015; Fox, Perez, Cass, Baglivio, & Epps, 2015; Hillis et al.,

2004). Recent studies have found that individuals who experience more types of ACEs are also more

1 University of North Texas, Denton, TX, USA

2 G4S Youth Services, LLC, Tampa, FL, USA

3 John Jay College of Criminal Justice, New York, NY, USA 4 University of Texas at Dallas, Richardson, TX, USA

5 Florida Department of Juvenile Justice, Tallahassee, FL, USA

Corresponding Author:

Jessica M. Craig, Department of Criminal Justice, University of North Texas, 410 S. Avenue C, Chilton Hall, 273L Denton,

TX 76203, USA.

Email: [email protected]

Youth Violence and Juvenile Justice 2017, Vol. 15(1) 3-20 ª The Author(s) 2016 Reprints and permission: sagepub.com/journalsPermissions.nav DOI: 10.1177/1541204016630033 yvj.sagepub.com

at UNIV NORTH TEXAS LIBRARY on December 1, 2016yvj.sagepub.comDownloaded from

likely to be serious, violent, and chronic offenders (Baglivio, Wolff, Piquero, & Epps, 2015; Fox

et al., 2015). Given the similar focus this work shares with developmental and life course perspec-

tives (cf. Cullen, 2011; Farrington, 2003; Moffitt, 1993), the current study extends this line of

research by assessing how the relationship between ACEs and future offending may be moderated

by protective factors from social bonds (Hirschi, 1969) in a sample of high-risk adjudicated youth.

Before presenting the results of the study, we provide a brief summary of existing research focusing

on ACEs, social control theory, and how ACEs and the attachment component of the social bond

may be related.

Evidentiary Backdrop

ACEs

The foundation for this study relies upon prior research that has assessed the role of ACEs on

outcomes later in life (Baglivio et al., 2014; Baglivio, Wolff, Epps, et al., 2015; Baglivio, Wolff,

Piquero, et al., 2015; Felitti et al., 1998; Fox et al., 2015; Hillis et al., 2004). ACEs as defined in

the original research refer to the following 10 specific experiences: emotional abuse, physical

abuse, sexual abuse, emotional neglect, physical neglect, violent treatment toward mother,

household substance abuse, household mental illness, parental separation or divorce, and having

a household member with a history of incarceration. Using a sample of privately insured, well-

educated adults, Felitti and his colleagues (1998) found that those who had been exposed to

more ACEs were more likely to be diagnosed with a chronic disease in adulthood. Similarly,

other studies have also reported that having higher ACE exposure is associated with a change in

brain functioning (Anda, Butchart, Felitti, & Brown, 2010), lower education and higher unem-

ployment (Bellis, Lowey, Leckenby, Hughes, & Harrison, 2013), and a higher prevalence of

risk-taking behaviors such as having 50 or more sexual partners, smoking, recent violence,

having spent at least one night in jail in the last 12 months, and drinking heavily (Bellis

et al., 2013; Hillis et al., 2004).

An individual’s ACE score is calculated by summing the total number of different types of ACEs

they report experiencing, with each exposure type counting once. Thus, ACE scores range between 0

(no ACEs experienced) and 10 (all ACEs experienced) for each individual. Even if a respondent had

reported being physically abused multiple times as a child, this experience would only count once.

This is an important point as research has indicated that ACEs are not only highly correlated and co-

occur nonrandomly but that they ‘‘have a powerful cumulative effect on human development’’

(Baglivio et al., 2014, p. 13) that can only be seen by studying the ACEs together as a collective

measure as opposed to studying one or two specific exposure types (see also Anda et al., 2010; Dong

et al., 2004).

Although the original study by Felitti and colleagues (1998) utilized a sample of well-educated

adults, results suggest, not surprisingly, that ACEs are more prevalent among youth involved with

the juvenile justice system (Baglivio et al., 2014; Dierkhising et al., 2013). There is a long tradition

of research examining childhood abuse and neglect and juvenile antisocial behavior, showing mal-

treatment is predictive of delinquency, increased recidivism, violent crimes, incarceration, and

earlier onset of antisocial behavior (Dembo et al., 1995; Maxfield & Widom, 1996; Rivera &

Widom, 1990; Widom, 1989). Extending the work on maltreatment, recent studies have examined

the relationship between ACE scores and offending. These studies have found that among justice

system-involved youth, those with exposure to more of these adverse and negative events are more

likely to be classified as higher risk to reoffend and engage in serious, chronic offending patterns

than their counterparts with lower ACE scores (Baglivio et al., 2014; Baglivio, Wolff, Piquero, et al.,

2015; Fox et al., 2015). Fox, Perez, Cass, Baglivio, and Epps (2015) recently reported that youth

4 Youth Violence and Juvenile Justice 15(1)

at UNIV NORTH TEXAS LIBRARY on December 1, 2016yvj.sagepub.comDownloaded from

with higher ACE scores were more likely to be serious, violent, and chronic juvenile offenders than

those having lower ACE scores. Further, Baglivio, Wolff, Piquero, and Epps (2015) found that,

controlling for related individual and family risk factors, ACE exposure distinguished between

offending trajectories such that those with higher exposure were more likely to be chronic offenders

than those with fewer ACEs. In the one study examining the ACE score and juvenile recidivism,

youth with higher ACE scores were not only more likely to reoffend, but reoffended in less time

postcompletion of community-based services such as probation supervision than youth with lower

ACE scores (Wolff, Baglivio, & Piquero, 2015).

In sum, these findings suggest that youth with higher ACE scores that are involved with the

juvenile justice system are not only more likely to be serious, chronic offenders but are also more

likely to experience other negative life outcomes such as chronic diseases and unemployment. From

a policy perspective, it is worthwhile to examine the extent to which some factors may be able to

buffer the relationship between ACE scores and reoffending. By identifying these factors, well-

tailored programs and strategies that target this group of adolescents can be implemented with the

objective of reducing a variety of adverse life outcomes, including subsequent offending. Given its

focus on protective factors, we draw upon Hirschi’s (1969) social control theory in this article in

order to assess some potential mechanisms. Next, we turn to a review of the key arguments and

findings surrounding this theory.

Social Control Theory

Hirschi’s (1969) social control theory is well suited for the identification of factors that are

protective against offending, as it focuses upon the elements that constrain people from acting

in their self-interest at the expense of others or the community. The theory’s premise states that

people are more likely to offend when their bond to society is weak or broken. Hirschi suggests

that this bond is composed of four elements: attachment, commitment, involvement, and belief.

Attachment is the emotional element of the social bond and refers to the individual’s affection for

and sensitivity to prosocial others. How much an individual is rationally invested in conventional

society is referred to as their commitment to or their stake in conformity (Toby, 1957), while

involvement denotes the time and energy they spend engaging in prosocial activities. The final

element, belief, represents how much an individual accepts conventional society’s norms regard-

ing behavior. Hirschi puts forth this theory as an explanation for delinquency, arguing that youth

who were strongly bonded, or in other words more attached, committed, involved, and had a

higher belief in conventional norms, rules, and laws, were less likely to be delinquent than those

who were less bonded. He also argued that the attachment element is the most important element

of the social bond, as it insulates the youth from antisocial influences and motivates the youth to

act prosocially in order to avoid disappointing those they care about. It is this attachment element

of the social bond that is the focus of the current study.

In general, existing empirical evidence offers mixed support for Hirschi’s (1969) theory, with

most support emerging for the protective effects of attachment and commitment on offending

(Costello & Vowell, 1999; Gottfredson, 2006; Kempf, 1993). Further, scholars using both cross-

sectional and longitudinal analyses have found that juveniles who were more tightly bonded to their

parents were less likely to be delinquent than those with weaker parental attachment (Bell, 2009;

Craig, 2015; Demuth & Brown, 2004; Ingram, Patchin, Huebner, McCluskey, & Bynum, 2007).

Scholars have also found nonparental attachments, such as attachment to school (Anderson, Holmes,

& Ostresh, 1999) and nonrelative adults (Huebner & Betts, 2002), to be important for reducing

delinquency, though these relationships were only significant for females. Recent work examining

risk and protective effects and youth violence as part of the longitudinal Seattle Social Development

Craig et al. 5

at UNIV NORTH TEXAS LIBRARY on December 1, 2016yvj.sagepub.comDownloaded from

Project has found school attachment to be an essential factor for decreased violence (Herrenkohl,

Lee, & Hawkins, 2012).

The current study is focused on the impact of attachment among juvenile justice–involved

youth. Social control theory would suggest that these youth were low in attachment prior to the

delinquent act that led to their justice system involvement. This leads to the issue of the rela-

tionship between prior delinquent acts on subsequent attachment. A few studies have examined

this relationship and reported mixed results. Using the National Youth Survey, Agnew (1991)

found that earlier delinquency did not impact later parental attachment, while Jang and Smith

(1997) reported that prior delinquency reduced parental attachment among a sample of mid-

adolescent youth from an urban area. Thus, it is possible that relative to nondelinquent youth,

it is harder for delinquent youth to increase their attachment to conventional others. Additionally,

as Thornberry (1987) articulated in his interactional theory of delinquency, perhaps attachment,

particularly parental attachment, is more important in early adolescence than in mid- to late

adolescence, when peers become more important. It is possible that an adolescent could form

a close emotional bond with a conventional other after they have engaged in delinquency or other

antisocial acts. It is this premise that many mentoring programs are based upon (Curtis & Bandy,

2015; Tierney, Grossman, & Resch, 2000) and this study seeks to investigate.

ACE Scores and Social Control Theory

Similar to the effects of prior delinquency on an individual’s current social bonds, many ACEs seem

to be indicative of low parental attachment, particularly if the ACEs the youth were exposed to were

perpetrated by a parent or guardian. However, while these youth may initially have had weakened

social bonds due to their exposure to these negative events, which Hirschi (1969) would argue is a

key reason they committed their first offense, it does not necessarily prevent them from forming

social bonds later. In other words, it is possible that juvenile justice–involved youth who have also

experienced many negative and traumatic events early in life may be buffered from the negative

impacts of these experiences by increasing their attachment to other conventional sources (see Laub

& Sampson, 2003; Sampson & Laub, 1993). There is some evidence to support this claim. While

many studies have found individuals who were victims of child abuse were more likely to offend

later on than those who were not victims (Smith & Thornberry, 1995; Widom, 2003), others have

found evidence of possible buffering factors. For example, Teague, Mazerolle, Legosz, and San-

derson (2008) found that victims of childhood physical abuse who reported receiving maternal

support were less likely to offend as adults relative to those who did not receive such support.

Hypothesis

The theoretical and empirical evidence reviewed above leads us to hypothesize that the positive

relationship between ACE scores and reoffending will be moderated by a youth’s attachment to

conventional others. As we describe below, the measure of attachment to conventional others used in

the current study is comprised of several components that indicate the presence of close and trusting

relationships between the youth and prosocial others, including individuals to whom the youth is and

is not related. Given that many of the ACEs the youth was exposed to may have been perpetrated by

a family member, our attachment measure takes that into account and allows for the youth to be

attached to others outside of the family, such as a mentor or other prosocial adult. For example, a

youth who suffered parental neglect may be more likely to be attached to a nonfamily member later

in adolescence, so our measure of social bonds allows for this possibility. We investigate these issues

within a unique, policy-relevant sample of adjudicated delinquents from a large diverse state in the

Southeastern United States.

6 Youth Violence and Juvenile Justice 15(1)

at UNIV NORTH TEXAS LIBRARY on December 1, 2016yvj.sagepub.comDownloaded from

Method

Sample

This study analyzed data from the Florida Department of Juvenile Justice (DJJ). The sample con-

sisted of a 3-year cohort of all youth who received an official referral (equivalent to an adult arrest)

in Florida and subsequently completed a community-based placement (such as probation) between

July 1, 2009, and June 30, 2012. All youth referred to DJJ are administered the Positive Achievement

Change Tool (PACT) risk/needs assessment. The DJJ uses two versions of the PACT: the Pre-Screen

and the Full Assessment. Although both tools provide recidivism likelihood composite scores (low-,

moderate-, moderate-high-, or high-risk), the Full Assessment includes data about many additional

domains of direct relevance to this study, including the youth’s past and current relationships.

Several empirical studies have found the PACT to be a valid predictor of continued juvenile justice

system involvement (Baglivio, 2009; Baglivio & Jackowski, 2013; Baird et al., 2013; Winokur-

Early, Hand, & Blankenship, 2012), with a total N in excess of 130,000. Both the Pre-Screen and

Full Assessment are conducted through semi-structured interviews by a trained juvenile probation

officer, a review of the youth’s case file, and corroboration with available official child abuse and

education records from other state agencies.

When youth enter the DJJ system, they are assessed using the Pre-Screen Assessment. If they

evidence a moderate-high or high reoffending risk, then they are assessed using the Full Assessment.

While under supervision, all low- and moderate-risk youth are reassessed every 180 days, and all

moderate-high and high-risk youth every 90 days. In the event a low- or moderate-risk youth

becomes higher risk, they are reassessed using the PACT Full Assessment. 1

In order to construct

the ACE score and social bonds measures critical to this study, only youth administered a PACT Full

Assessment were included, as only the Full Assessment contains the necessary items. As a result, the

current study oversampled higher risk juvenile offenders resulting in the final sample of 28,169

youth being 39.1% low-risk, 16.4% moderate-risk, 30.3% moderate-high-risk, and 14.3% high-risk to reoffend. The total number of cases (N ¼ 28,169) represents 21.5% of all community-based placement completions over the study period. As the DJJ policy is to screen all youth scoring

moderate-high- and high-risk to reoffend with the Full Assessment, all youth excluded from the

current study were low or moderate risk to reoffend. 2

Measures: Dependent Variable

The outcome measure for this study was reoffending, measured as rearrest within 12 months after the

youth completed the community-based placement. This measure was coded dichotomously where

‘‘1’’ indicated the youth was arrested at least once and ‘‘0’’ indicated the youth was never arrested

during his or her 365-day follow-up period. 3

Forty percent of the youth in our sample were rearrested

within 12 months of being released from supervision.

Measures: Key Independent Variables

All independent measures were taken from the PACT Full Assessment closest to the date the youth

completed the community-based placement, which establishes the most accurate, up-to-date assess-

ment of the youth’s circumstances at the beginning of the follow-up period. This ensures we are

predicting rearrest (the intent of the study) and not success of the services delivered during place-

ment, which would require information related to dosage and fidelity of services. Using the exit

PACT assures that the study used the profile of the youth after services were completed.

The central independent variable for this analysis was an index composed of 4 items that each

measure a form of attachment to conventional others. The first measure reflects whether the youth

Craig et al. 7

at UNIV NORTH TEXAS LIBRARY on December 1, 2016yvj.sagepub.comDownloaded from

had any positive adult, nonfamily relationships at the time they completed their community super-

vision. This measure was coded such that ‘‘0’’ indicated having no positive adult relationships and

‘‘1’’ indicated having at least one such relationship. The second measure represented the degree to

which the youth had prosocial community ties at the time they exited supervision, where ‘‘0’’

indicated the youth had no prosocial ties to the community and ‘‘1’’ indicated they had at least

some prosocial ties. The third measure indicated whether the youth had prosocial friends they spent

time with at the point they were released. This measure was coded where ‘‘0’’ represented that they

did not spend time with any prosocial friends and ‘‘1’’ indicated they did spend time with prosocial

friends. The fourth and final indicator represented whether the youth felt close to their prosocial

parents or caretakers at the time they were released. This measure was coded where ‘‘0’’ indicated

they did not feel close to their prosocial parents/caretakers and ‘‘1’’ indicated they felt close to at

least one of their prosocial mother/female caretaker or father/male caretaker. These 4 items were

summed in order to create an index ranging from 0 to 4 where ‘‘0’’ indicated no prosocial bonds and

‘‘4’’ indicated they had all four possible bonds. The mean social bonding score was 3.10 on a range

from 0 to 4, indicating most of the respondents were fairly well bonded. 4

Measures: ACE Score

The respondents’ composite ACE score measure was constructed through a multipart process. First,

if the youth indicated they had ever experienced one of the 10 ACE indicators, they were coded as

‘‘1’’ for that ACE; otherwise, they were coded as ‘‘0.’’ Recall the 10 ACEs are emotional abuse,

physical abuse, sexual abuse, emotional neglect, physical neglect, violent treatment toward mother,

household substance abuse, household mental illness, parental separation or divorce, and having a

household member with a history of incarceration. The total number of ACEs each youth experi-

enced was then summed in order to create the overall ACE score, with a range of 0 to 10 where ‘‘0’’

indicated the youth had never experienced any of these events and ‘‘10’’ indicated they had expe-

rienced all of them at least once. The average ACE score was 2.64 on a range from 0 to 10 (a¼ .62).5

Measures: Control Variables

The current analysis also included several relevant control variables, including the respondents’ sex,

age, race/ethnicity, socioeconomic status (SES), impulsivity, and antisocial peer associations. The

youth’s sex was controlled for, where ‘‘0’’ indicated female and ‘‘1’’ indicated male. Seventy-seven

percent of the analysis sample was male. In order to control for age, the age the youth was when

completing the community-based placement was also included in the analysis (mean ¼ 17). The race/ethnicity variable was composed of two measures, each a dichotomous measure that indicated

whether the youth was Black or Hispanic, with White as the reference group. Forty-six percent of the

sample was Black, 15% of the sample was Hispanic, and the remaining 39% of the sample was White or of other race/ethnicity.

6 The youth’s SES was derived through a measure of neighborhood

disadvantage. The level of disadvantage present in the neighborhood in which each youth resided

was based on the socioeconomic conditions present in the youth’s census tract. American Commu-

nity Survey data were used to measure concentrated disadvantage using six variables of (1) propor-

tion of families living below the poverty line, (2) median family income (logged and reverse coded),

(3) proportion of female-headed households, (4) unemployment rate, (5) proportion with a high

school degree (reverse coded), and (6) proportion of households receiving public assistance. Con-

firmatory factor analysis indicated these measures loaded on a single factor. As such, they were

standardized and combined to form an additive index of concentrated disadvantage (a ¼ .879). These items have been used in other work (Baumer, Messner, & Felson, 1998; Kubrin & Stewart,

2006), including examining ACE scores and Florida juvenile offenders (Baglivio, Wolff, Epps,

8 Youth Violence and Juvenile Justice 15(1)

at UNIV NORTH TEXAS LIBRARY on December 1, 2016yvj.sagepub.comDownloaded from

et al., 2015). The mean concentrated disadvantage score was .39 on a scale from �2.30 to 3.90, indicating that a slightly larger proportion of youth in the analysis sample live in areas characterized

by marginally higher than average levels of disadvantage.

Two theoretically relevant individual-level risk factors were also included: impulsivity (Gott-

fredson & Hirschi, 1990) and antisocial peer associations (Burgess & Akers, 1966). The impulsivity

measure indicated whether the youth was impulsive and acted before thinking. This was scored on a

scale from 0 to 3, where ‘‘0’’ indicated they were not impulsive and ‘‘3’’ indicated they were highly

impulsive. The mean impulsivity score was 1.00 on a range from 0 to 3, indicating most of the

respondents scored low on this scale. In order to control for the influence of antisocial peers, we used

an item that gauged whether at the time the youth completed their supervision they spent time with

either antisocial peers and/or gang ‘‘members. If they indicated they had spent time with such peers,

they’’ were coded as ‘‘1’’ and if they did not spend time with such peers they were coded as ‘‘0.’’

Forty-seven percent of the sample indicated they had at least one antisocial or gang-affiliated friend.

Analysis Plan

After highlighting the bivariate relationships between the variables of interest, several logistic

regressions were estimated in order to examine the potential for social bonds to moderate the

relationship between ACEs and future reoffending. The first model represents our baseline model

and included the control variables predicting the juveniles’ rearrest. Next, the index for the respon-

dents’ social bonds was included in order to assess the independent effects of social bonds and ACEs

on rearrest. The third and fourth models represent our moderation analyses and rely upon split

samples in order to assess the role of ACEs on reoffending among those with lower social bonds

and those with higher social bonds. This approach has the benefit of providing a specific regression

estimate for each group and a method by which to calculate whether the regression coefficients are

significantly different between groups (cf. Paternoster, Brame, Mazerolle, & Piquero, 1998). Based

upon the distribution of the social bonds index, the sample was split such that the respondents

scoring 2 or below on the index were coded as having weaker bonds (the bottom 25% of the distribution) and those scoring 3 or 4 were coded as having stronger bonds (the remaining 75% of the distribution).

7 Prior to estimating the logistic regressions and as an initial investigation for

possible group differences, a t-test was conducted to assess the mean difference of ACE scores

between those with stronger social bonds and those with weaker bonds. Next, the third logistic

regression model estimated the effects of ACEs along with the other control variables on rearrest

among those with stronger bonds. The fourth model also estimated the effects of these same vari-

ables on rearrest, but among those with weaker social bonds. Given that prior research has found

discernable race and gender differences in social bonding (Craig, 2014; Huebner & Betts, 2002;

Piquero, MacDonald, & Parker, 2002) and offending (Sampson & Wilson, 1995; Steffensmeier &

Allen, 1996), all of the substantive models were also reestimated across split-sample analyses for

race and gender.

Results

Bivariate Correlations

The bivariate correlations between all of the variables in the current study are presented in Table 1. 8

Of note is the small but negative significant relationship between social bonds and rearrest. The

respondents’ ACE score also has a small but positive significant relationship with rearrest. There is

also a small and negative significant correlation between social bonds and ACE scores. Finally,

multicollinearity does not appear to be an issue, as no correlation was greater than r ¼ .5, and all postanalyses variance inflation factors were less than 1.19.

Craig et al. 9

at UNIV NORTH TEXAS LIBRARY on December 1, 2016yvj.sagepub.comDownloaded from

Effects of Control Variables on Rearrest

Model 1 in Table 2 presents the results of our baseline logistic regression for rearrest. This model

suggests the respondent’s ACE score, gender, age at release, race/ethnicity, impulsivity, and residing

in a disadvantaged environment significantly increase the likelihood of rearrest. The presence of

antisocial peers is a positive, marginally significant predictor of rearrest. Holding all else constant,

each additional type of ACE exposure results in a 9% higher odds of being rearrested. Black respondents had 52% higher odds of being rearrested than non-Blacks and those who were more impulsive had 9% higher odds of being rearrested than those who were less impulsive.

Effects of Social Bonds on Rearrest

The second model in Table 2 adds the measure of social bonds to the baseline model. These results

show that while having more ACEs had a positive and significant effect on rearrest (odds ratio

Table 1. Bivariate Correlations Between All Variables.

Variable 1 2 3 4 5 6 7 8 9 10

Rearrest 1.00 Social bonds �0.04* 1.00 ACE score 0.04* �0.26* 1.00 Gender 0.14* 0.04* �0.19* 1.00 Age 0.04* �0.06* �0.04* 0.06* 1.00 Black 0.10* �0.01* �0.05* �0.02* 0.01* 1.00 Hispanic �0.02* �0.01 �0.10* 0.05* 0.02* �0.39* 1.00 Impulsivity 0.04* �0.25* 0.26* �0.02* �0.13* 0.02* �0.04* 1.00 Antisocial peers 0.02* �0.07* 0.10* �0.00 �0.03* 0.04* �0.01 0.23* 1.00 Concentrated disadvantage 0.07* �0.05* 0.01 �0.01 0.02* 0.41* �0.03* 0.03* 0.02* 1.00

Note. ACE ¼ adverse childhood experience. *p < .05.

Table 2. Logistic Regressions Predicting Rearrest.

Model 1 Model 2 Model 3

(Strongly Bonded) Model 4

(Weakly Bonded)

Variable OR SE OR SE OR SE OR SE

Social bonds — — .98y .01 — — — — ACE score 1.09*** .01 1.08*** .01 1.10*** .01 1.05** .01 Gender 2.31*** .03 2.31*** .03 2.42*** .04 2.05*** .06 Age 1.06*** .01 1.06*** .01 1.06*** .01 1.03y .02 Black 1.52*** .03 1.51*** .03 1.53*** .04 1.47*** .06 Hispanic 1.16*** .04 1.16*** .04 1.12* .05 1.28** .08 Impulsivity 1.09*** .02 1.08*** .02 1.09*** .02 1.08* .03 Antisocial peers 1.05y .03 1.05y .03 1.08* .03 0.97 .05 Concentrated disadvantage 1.08*** .02 1.08*** .02 1.07*** .02 1.11** .03 Constant 0.08*** .14 0.09*** .16 0.06*** .16 0.16*** .30 McKelvey and Zavoina’s R

2 .056 .056 .061 .042

n 25,461 25,461 19,114 6,347

Note. ACE ¼ adverse childhood experience. ***p < .001. **p < .01. *p < .05. yp < .10 (two-tailed).

10 Youth Violence and Juvenile Justice 15(1)

at UNIV NORTH TEXAS LIBRARY on December 1, 2016yvj.sagepub.comDownloaded from

[OR] ¼ 1.08, p < .001), the respondent’s social bonds had a negative, but marginally significant effect on rearrest (OR ¼ .98, p < .10). Similar to Model 1, all of the other control variables were positive, significant, or marginally significant predictors of rearrest.

Moderating Effects of Social Bonds on the Relationship Between ACEs and Rearrest

A means differences test indicates that those with weaker social bonds have significantly more

ACEs than those with stronger social bonds (3.28 vs. 2.43, p < .05). With that in mind, we now move

on to the analyses that test our central hypothesis that prosocial bonds moderate the effect of ACEs

on rearrest. Models 3 and 4 in Table 2 present the results from the moderating analyses. Model 3

includes the results for the effects of ACEs and the other control variables on rearrest among those

with stronger social bonds. Among respondents who are more strongly bonded, each additional ACE

was found to significantly increase the odds of rearrest by 10%, holding all of the other measures equal. As found in the prior two analyses, the other control variables also emerged as positive,

significant predictors of rearrest.

Model 4 shows the results of the effects of ACEs and the other variables on rearrest among those

that have weaker social bonds. Among this subsample, each additional ACE exposure increased the

odds of rearrest by 5%, holding all else equal. Further, although it had been significant or marginally significant in the prior three analyses, the measure indicating the presence of antisocial friends was

no longer significant. The remaining variables of gender, age at release, race/ethnicity, impulsivity,

and concentrated disadvantage were positive and significant predictors of rearrest. A visual repre-

sentation of these findings can be seen in Figure 1, which presents the predicted probabilities of the

effects of ACE scores on rearrest among individuals with stronger social bonds and those with

weaker social bonds. As can be seen, relative to those with weaker bonds, those with stronger social

bonds are protected from rearrest when they have experienced approximately five or fewer ACEs.

However, this protective effect no longer holds after they have experienced six or more ACEs.

Further, having an increased number of ACEs still increases the odds of rearrest, regardless of the

value of the social bonds index. 9

.3 .3

5 .4

.4 5

.5 .5

5 .6

R ea

rr es

t

0 2 4 6 8 10

ACE Score

Strongly Bonded Weakly Bonded

Figure 1. Predicted probabilities of rearrest by social bonds and ACE scores. ACE ¼ adverse childhood experience.

Craig et al. 11

at UNIV NORTH TEXAS LIBRARY on December 1, 2016yvj.sagepub.comDownloaded from

Racial and Ethnic Differences

Table 3 presents the logistic regression results examining the effects of social bonds and ACE scores on

rearrest by the three racial/ethnic groups included in the current study. Model 1 indicates that, among

Blacks, Hispanics, and Whites, those with more ACEs were more likely to be rearrested than those with

fewer such experiences (OR ¼ 1.07, p < .001; OR ¼ 1.08, p < .001; OR ¼ 1.10, p < .001, respectively). Hispanics with stronger social bonds were marginally less likely to be rearrested than Hispanics with

weaker social bonds (OR ¼ .94, p < .10). However, this same result did not hold for Black or White respondents. Gender, age, and impulsivity were consistently significant predictors across the three race/

ethnicities, while concentrated disadvantage was a significant predictor of rearrests only among Blacks.

The effects of ACEs among those with stronger bonds by racial/ethnic group are indicated in

Model 2. Similar to the results in the full sample, among those with stronger social bonds, having a

higher ACE exposure increases the odds of rearrest among Blacks by 9%, Hispanics by 6%, and Whites by 11%. Gender and age emerge as significant predictors of rearrest among all three groups, and impulsivity is a significant predictor of rearrests among Blacks and Hispanics. Antisocial peers

and concentrated disadvantage predict rearrests among Blacks.

Model 3 presents the results of the logistic regressions among those with weak social bonds.

While Hispanics (OR ¼ 1.13, p < .01) and Whites (OR ¼ 1.04, p < .10) with higher ACE scores are more likely to be rearrested than those with lower ACE scores, the same result does not hold for

Blacks. In other words, having an increased ACE exposure increases the odds of rearrest among

Hispanic and White youth who are weakly bonded. Black youth with weak bonds do not appear to

evidence this effect of ACEs on rearrests. Similar to the prior analyses, gender is a significant

predictor of rearrest among all racial/ethnic groups, as is impulsivity for Blacks and concentrated

disadvantage for Blacks and Hispanics.

Gender Differences

As can be seen in Table 4, there were very few gender differences in how ACEs and social bonds

predicted rearrests. Model 1 indicates that while social bonds decrease rearrests among both males

and females, social bonds are only significant among females (OR ¼ .94, p < .05). Similar to the full sample, having a higher ACE exposure increases rearrest odds by 8% for both male and female youth. Among males, age, being Black, being Hispanic, impulsivity, antisocial peers, and concen-

trated disadvantage significantly predict rearrest. Among females, however, only race and impul-

sivity are significantly predictive of rearrest.

Model 2 indicates that regardless of gender, strongly bonded males and females with higher ACE

exposure are more likely to be rearrested than those without such experiences (OR ¼ 1.09, p < .001 and OR ¼ 1.12, p < .001, respectively). All of the control variables emerge as significant predictors of rearrests for males, while only impulsivity and being Hispanic significantly predict rearrests

among females.

As can be seen in Model 3, having a higher ACE score significantly increases the odds of rearrest

among weakly bonded males by 7%. However, the significance of the ACE effect does not hold for females. Age, concentrated disadvantage, being Black, and being Hispanic are significant predictors

of rearrests for weakly bonded males while age, impulsivity, and being Black are significant pre-

dictors of rearrests among weakly bonded females.

Discussion and Conclusion

Prior research has established that individuals who experience multiple ACEs such as physical,

emotional, or sexual abuse are more likely to experience many negative and diverse outcomes at

12 Youth Violence and Juvenile Justice 15(1)

at UNIV NORTH TEXAS LIBRARY on December 1, 2016yvj.sagepub.comDownloaded from

T a b

le 3 .

L o gi

st ic

R e gr

e ss

io n s

P re

d ic

ti n g

R e ar

re st

b y

R ac

e an

d E th

n ic

it y.

M o d e l 1

M o d e l 2

(S tr

o n gl

y B o n d e d )

M o d e l 3

(W e ak

ly B o n d e d )

B la

ck H

is p an

ic W

h it e

B la

ck H

is p an

ic W

h it e

B la

ck H

is p an

ic W

h it e

V ar

ia b le

O R

S E

O R

S E

O R

S E

O R

S E

O R

S E

O R

S E

O R

S E

O R

S E

O R

S E

S o ci

al b o n d s

0 .9

7 .0

2 0 .0

4 y

.0 3

1 .0

0 .0

2 —

— —

— —

— —

— —

— —

— A

C E

sc o re

1 .0

7 **

* .0

1 1 .0

8 **

* .0

2 1 .1

0 **

* .0

1 1 .0

9 **

* .0

1 1 .0

6 *

.0 3

1 .1

1 **

* .0

1 1 .0

3 .0

2 1 .1

3 **

.0 4

1 .0

4 y

.0 2

G e n d e r

2 .6

7 **

* .0

5 2 .5

9 **

* .1

0 1 .8

4 **

* .0

5 2 .8

0 **

* .0

6 2 .8

9 **

* .1

2 1 .9

0 **

* .0

6 2 .3

5 **

* .0

9 2 .1

3 **

* .1

8 1 .6

8 **

* .1

1 A

ge 1 .0

5 **

* .0

1 1 .1

0 **

* .0

2 1 .0

5 **

.0 1

1 .0

5 **

* .0

1 1 .1

3 **

* .0

3 1 .0

5 **

* .0

1 1 .0

3 .0

2 1 .0

5 .0

4 1 .0

2 .0

3 Im

p u ls

iv it y

1 .1

0 **

* .0

3 1 .1

1 *

.0 5

1 .0

5 y

.0 3

1 .1

1 **

* .0

3 1 .1

7 **

.0 6

1 .0

4 .0

3 1 .1

0 *

.0 5

0 .9

8 .0

9 1 .0

9 .0

5 A

n ti so

ci al

p e e rs

1 .0

5 .0

4 1 .0

3 .0

7 1 .0

5 .0

4 1 .0

8 y

.0 4

1 .1

3 .0

8 1 .0

5 .0

5 0 .9

5 .0

7 0 .7

9 y

.1 4

1 .0

6 .0

9 C

o n ce

n tr

at e d

d is

ad va

n ta

ge 1 .1

1 **

* .0

2 1 .0

6 .0

5 1 .0

3 .0

3 1 .1

1 **

* .0

3 1 .0

2 .0

5 1 .0

2 .0

4 1 .1

2 **

.0 4

1 .2

0 *

.0 9

1 .0

4 .0

6 C

o n st

an t

0 .1

3 **

* .2

2 0 .0

5 **

* .4

3 0 .1

1 **

* .2

6 0 .1

0 **

* .2

2 0 .0

2 **

* .4

7 0 .0

9 **

* .2

8 0 .2

1 **

* .4

3 0 .1

4 *

.8 0

0 .2

2 **

.5 1

M cK

e lv

e y

an d

Z av

o in

a’ s

R 2

.0 5 8

.0 5 2

.0 2 8

.0 6 2

.0 6 2

.0 3 1

.0 4 5

.0 4 0

.0 1 8

n 1 1 ,8

8 4

3 ,8

3 9

9 ,7

3 8

8 ,8

6 4

2 ,8

5 5

7 ,3

9 5

3 ,0

2 0

9 8 4

2 ,3

4 3

N o te

. A

C E ¼

ad ve

rs e

ch ild

h o o d

e x p e ri

e n ce

. **

*p <

.0 0 1 . **

p <

.0 1 . *p

< .0

5 . y p

< .1

0 (t

w o -t

ai le

d ).

T a b

le 4 .

L o gi

st ic

R e gr

e ss

io n s

P re

d ic

ti n g

R e -a

rr e st

b y

G e n d e r.

M o d e l 1

M o d e l 2

(S tr

o n gl

y B o n d e d )

M o d e l 3

(W e ak

ly B o n d e d )

M al

e F e m

al e

M al

e F e m

al e

M al

e F e m

al e

V ar

ia b le

O R

S E

O R

S E

O R

S E

O R

S E

O R

S E

O R

S E

S o ci

al b o n d s

0 .9

8 .0

1 .9

4 *

.0 3

— —

— —

— —

— —

A C

E sc

o re

1 .0

8 **

* .0

1 1 .0

8 **

* .0

2 1 .0

9 **

* .0

1 1 .1

2 **

* .0

2 1 .0

7 **

* .0

2 1 .0

2 .0

3 A

ge 1 .0

7 **

* .0

1 0 .9

7 .0

2 1 .0

8 **

* .0

1 0 .9

8 .0

2 1 .0

6 **

.0 2

0 .9

3 *

.0 4

B la

ck 1 .6

1 **

* .0

3 1 .1

7 *

.0 7

1 .6

4 **

* .0

4 1 .1

3 .0

8 1 .5

3 **

* .0

7 1 .2

8 y

.1 3

H is

p an

ic 1 .2

2 **

* .0

4 0 .8

8 .1

0 1 .2

0 **

* .0

5 0 .7

8 *

.1 3

1 .3

1 **

.0 9

1 .1

9 .1

8 Im

p u ls

iv it y

1 .0

6 **

.0 2

1 .1

9 **

* .0

4 1 .0

7 **

.0 2

1 .1

5 **

.0 5

1 .0

2 .0

4 1 .2

9 **

* .0

7 A

n ti so

ci al

p e e rs

1 .0

5 y

.0 3

1 .0

2 .0

6 1 .0

7 *

.0 3

1 .0

8 .0

7 0 .9

8 .0

6 0 .9

0 .1

1 C

o n ce

n tr

at e d

d is

ad va

n ta

ge 1 .1

0 **

* .0

2 1 .0

1 .0

4 1 .0

8 **

.0 2

1 .0

5 .0

5 1 .1

6 **

* .0

4 0 .9

3 .0

7 C

o n st

an t

0 .1

4 **

* .1

7 0 .4

4 *

.3 6

0 .1

2 **

* .1

8 0 .2

5 **

* .3

9 0 .2

1 **

* .3

4 1 .0

1 .6

5 M

cK e lv

e y

an d

Z av

o in

a’ s

R 2

.0 3 1

.0 2 5

.0 3 2

.0 2 6

.0 2 7

.0 2 2

n 1 9 ,6

4 7

5 ,8

1 4

1 4 ,8

4 7

4 ,2

6 7

4 ,8

0 0

1 ,5

4 7

N o te

. A

C E ¼

ad ve

rs e

ch ild

h o o d

e x p e ri

e n ce

. **

*p <

.0 0 1 . **

p <

.0 1 . *p

< .0

5 . y p

< .1

0 (t

w o -t

ai le

d ).

13 at UNIV NORTH TEXAS LIBRARY on December 1, 2016yvj.sagepub.comDownloaded from

later stages in life (Baglivio et al., 2014; Baglivio, Wolff, Epps, et al., 2015; Baglivio, Wolff,

Piquero, et al., 2015; Felitti et al., 1998; Fox et al., 2015; Hillis et al., 2004; Wolff et al.,

2015). These outcomes range from increased chronic disease diagnoses to higher unemployment

to exhibiting more risk-taking behaviors (Bellis et al., 2013). These findings highlight the impor-

tance of intervening with these individuals early in life in order to not only improve their future

health and life outcomes but also decrease delinquent or criminal behavior. This study sought to

examine the independent effects of both ACEs and social bonds on the likelihood of reoffending,

as well as the potential moderating effect of social bonds on the relationship between ACEs and

reoffending among a sample of adjudicated, high-risk youth. While results showed that higher

ACE scores led to a higher likelihood of rearrest and more social bonds lowered the likelihood of

rearrest, in contrast to our expectations, the analyses revealed that having a higher ACE score

significantly increased the chances of rearrest among those with both weaker and stronger social

bonds. In other words, ACEs were found to increase a youth’s chance of rearrest, regardless of

the strength of attachment they had to conventional others such as prosocial parents or commu-

nity members.

Still, the moderating effects may be more nuanced. For example, among youth with zero to

approximately five ACEs, those who are strongly bonded are less likely to be rearrested than those

with weak bonds. However, once the youth have been exposed to six or more ACEs, this relationship

no longer holds and those with stronger social bonds are more likely to be arrested than those with

weaker bonds. Youth who are strongly bonded are more protected with fewer ACE exposures.

However, once a strongly bonded youth surpasses a certain number of ACEs, then the protective

effect of bonds is not strong enough to counteract the deleterious effect of ACEs. Additionally, those

with weak bonds do not evidence the protective effect of social bonds contributing to a higher

likelihood of rearrest. Further, as those with weak bonds have more ACEs, each additional event

further increases the probability of recidivism.

It is important to note as well that these general effects held across racial/ethnic and gender

subsample analyses. The majority of results from the subsample analyses mirrored the results from

the full sample. For example, among the strongly bonded youth in the full sample, having a higher

exposure to ACEs significantly increased the odds of rearrest. This same significant relationship

held across the three racial/ethnic groups as well as both genders. Additionally, while having a

higher exposure to ACEs did not significantly increase rearrests among weakly bonded Blacks and

weakly bonded females, the direction of both ORs were in the expected, positive direction. The

consistency of the results across these subgroup analyses indicate that ACEs, social bonds, and the

relationship between the two have stable effects, regardless of an individual’s racial/ethnic back-

ground or gender.

Limitations and Future Directions

Several limitations should be acknowledged. First, the data were collected for the purposes of risk

assessment, and only measures collected as part of the Full PACT Assessment could be used to

generate an ACE score. Although the original conceptualization of ACE scores proposed the expo-

sures be summed in a binary fashion, other child neglect and abuse scholars have argued that factors

such as timing, duration, and severity of exposure are important (Smith & Thornberry, 1995). While

these measures were not available in the current data, future studies should aim to include these

concepts in order to assess whether there are differences in how they interact with social bonds.

Relatedly, as only one item was available to measure each of these constructs, the full battery of

items to assess impulsivity and antisocial peer associations was not available. Future studies should

incorporate more thorough measures of these constructs that might capture additional nuances that

the available measures are unable to show.

14 Youth Violence and Juvenile Justice 15(1)

at UNIV NORTH TEXAS LIBRARY on December 1, 2016yvj.sagepub.comDownloaded from

A second limitation includes the possibility that some youth may have underreported ACEs such

as sexual abuse or neglect. As many in the sample were not evaluated until they were adolescents,

they may have forgotten, concealed, or otherwise blocked out such early childhood experiences.

While the caseworker conducting the structured interview is trained to identify possible cues of

abuse and this professional has access to the state’s child abuse registry database in order to

corroborate any possible reports, these behaviors may still be underreported. Nevertheless, unless

there is systematic underreporting of abuse, this would not impact our results. A third and related

limitation is that many of the variables were skewed, indicating many of the youth were positively

adjusted. This again might reflect possible underreporting or minimizing of undesired traits such as

antisocial peers. Fourth, the generalizability of the sample might be limited as the sample contained

only adjudicated Florida Department of Juvenile Justice–involved youth who received the Full

PACT assessment and are generally considered to be more serious offenders. Further, the sample

only consisted of the youth who received a placement to a form of community correction such as

juvenile probation, as opposed to all youth, regardless of their disposition outcome. However, over

50% (55.5%) of the sample was assessed as either low- or moderate-risk to reoffend, which, while lower than the 87% of all youth assessed with the PACT (Pre-Screen plus Full), allows for over- sampling of the most policy-relevant youth: those at higher risk to reoffend and ethnic minorities. To

be sure, given the large sample size and largely significant effects, caution should be used when

interpreting the substantive effects that emerged. While the resulting ORs from the multivariate

models provide support for substantive effects, future research should corroborate our study and its

findings across a range of sampling frames.

Fifth, the study relied upon an official recidivism measure of rearrest as opposed to self-reported

delinquency. Although 40% of the youth were rearrested within 1 year of completing their community-based sanction, this probably is not inclusive of all juveniles who reoffended. Future

studies should aim to address these limitations with self-reported delinquency measures. Addition-

ally, while this study did not find its attachment measure buffered the relationship between ACEs

and rearrest as would be predicted from social control theory, other measures of attachment or social

bonds may be relevant. For instance, as the literature has demonstrated that mentoring programs can

be effective in reducing recidivism among juvenile justice–involved youth (Lipsey, 2009; Lipsey,

Howell, Kelly, Chapman, & Carver, 2010), perhaps the individual the youth is attached to matters

(e.g., a mentor as opposed to a prosocial parent, for instance), or the strength of that attachment

matters more. Further, as ACEs have been shown to lead to several negative life outcomes along

with their effects on delinquency (Felitti et al., 1998; Hillis et al., 2004), more attention should be

paid to other social control variables that may buffer this effect, such as involvement in prosocial

activities and having prosocial beliefs (Hirschi, 1969).

Policy Implications

This study underscores the importance of targeted interventions for children who are at risk for or

have already experienced ACEs. However, for justice-involved youth who have experienced many

traumatizing and abusive events, such interventions aimed specifically at enhancing attachment and

increasing social bonds may not be as effective, though this possibility should be empirically

assessed in future research. This suggests efforts should be made toward preventing ACEs and

intervening in cases where they are experienced as early as possible and before the individual is

involved with the justice system. For instance, school counselors, pediatricians, and other practi-

tioners in contact with children could use ACE scores as an early screening tool and then direct these

children and their families toward programs such as Functional Family Therapy and Multisystemic

Therapy, both crime prevention programs that have been found to be cost effective and reduce crime

(Lee et al., 2012). Focusing efforts on these individuals early in life would serve to not only reduce

Craig et al. 15

at UNIV NORTH TEXAS LIBRARY on December 1, 2016yvj.sagepub.comDownloaded from

crime but also help these individuals avoid other negative outcomes in physical and mental health

domains as well and reduce the social and financial burdens taken on society by the most chronic

offenders (cf. Cohen & Piquero, 2009).

In sum, the current study has a number of theoretical and policy strengths. First, it extended our

knowledge of criminological theory by evaluating the extent to which social bonds exert some

degree of protective role among a sample of deep-end juvenile offenders. Second, we examined

the extent to which attachment to prosocial others moderated the relationship between adverse

childhood experiences and reoffending. This particular feature of our work demonstrated the dele-

terious persistence of traumatic childhood events, highlighting the need for both prevention and

early intervention.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or pub-

lication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

1. A low- or moderate-risk youth might become higher risk due to referral for a new offense or through

exposure to social history risk factors such as initiation of substance use, truancy, or antisocial peer

associations.

2. In comparison with the current sample, the full sample of all community-based completions was composed

of 131,261 youth, with 114,502 receiving any Positive Achievement Change Tool assessment (Pre-Screen or

Full), of which 75.5% were low risk, 11.0% moderate risk, 9.2% moderate-high risk, and 4.4% high risk to

reoffend. The current sample also differs from the full sample of 131,261 youth in terms of demographic

variables with the study sample youth being older at age of release (t ¼�68.5, p < .001), a higher proportion male (t ¼�41.8, p < .001), higher proportion Black (46.1% compared to 38.2% in full sample, t ¼�23.6, p < .001), and slightly less Hispanic (15.2% compared to 16.0% in the full sample, t ¼ 3.3, p ¼ .001).

3. This measure does not include probation violations, as all youth completed Department of Juvenile Justice

supervision prior to the follow-up period.

4. A supplemental factor analysis indicated that the social bond measure was unidimensional, with all items

loading strongly on one factor.

5. To provide context, only 22% of individuals in the original ACE study had ACE scores over 2 (Centers for

Disease Control and Prevention, 2016). Further, while some may be concerned over the low a, the measure operates in the manner in which both prior theory and research anticipated and was created as an additive

index in an identical fashion to that used by the Centers for Disease Control and Prevention of the same 10

exposure types.

6. Of the 28,169 youth, 121 youth (0.4%) were classified as ‘‘other’’ race/ethnicity, with 1 additional youth

having race/ethnicity ‘‘unknown.’’ Based on these low sample sizes, the current study includes the Other and

the 1 unknown youth with the White youth sample.

7. Supplemental split-sample analyses were also estimated, where the sample was split at approximately 50%

(54.85%) of the social bonds index distribution. Specifically, those scoring 3 or less were coded as having

weaker social bonds and those scoring 4 were coded as having stronger social bonds. The results from these

logistic regressions were virtually identical to the ones presented in this study (results available upon

request).

8. Heteroscedasticity was examined after each logistic regression model, and none was detected. A missing

data analysis found the concentrated disadvantage measure was missing at random. Two hundred sixty-six

multivariate outliers (approximately 1% of the sample) were identified using Mahalinobis distance scores.

16 Youth Violence and Juvenile Justice 15(1)

at UNIV NORTH TEXAS LIBRARY on December 1, 2016yvj.sagepub.comDownloaded from

Supplemental analyses were estimated with and without these outliers, and the results were found to be

substantially similar, so these cases were retained in the analysis.

9. A supplemental moderation analysis was also estimated where an interaction of the mean-centered social

bond and ACE score measures was used in the logistic regression (Jaccard, 2001). Similar to the results

presented here, the interaction variable was positive and significant.

References

Agnew, R. (1991). A longitudinal test of social control theory and delinquency. Journal of Research in Crime

and Delinquency, 28, 126–156.

Anda, R. F., Butchart, A., Felitti, V. J., & Brown, D. W. (2010). Building a framework for global surveillance of

the public health implications of Adverse Childhood Experiences. American Journal of Preventive Medi-

cine, 39, 93–98.

Anderson, B. J., Holmes, M. D., & Ostresh, E. (1999). Male and female delinquents’ attachments and effects of

attachments on severity of self-reported delinquency. Criminal Justice and Behavior, 26, 435–452.

Baglivio, M. T. (2009). The assessment of risk to recidivate among a juvenile offending population. Journal of

Criminal Justice, 37, 596–607.

Baglivio, M. T., Epps, N., Swartz, K., Huq, M. S., Sheer, A., & Hardt, N. S. (2014). The prevalence of Adverse

Childhood Experiences (ACE) in the lives of juvenile offenders. Journal of Juvenile Justice, 3, 12–34.

Baglivio, M. T., & Jackowski, K. (2013). Examining the validity of a juvenile offending risk assessment

instrument across gender and race/ethnicity. Youth Violence and Juvenile Justice, 11, 26–43.

Baglivio, M. T., Wolff, K. T., Epps, N., & Nelson, R. (2015). Predicting adverse childhood experiences: The

importance of neighborhood context in youth trauma among delinquent youth. Crime & Delinquency, 1–23.

doi:10.1177/0011128715570628

Baglivio, M. T., Wolff, K. T., Piquero, A. R., & Epps, N. (2015). The relationship between Adverse Childhood

Experiences (ACE) and juvenile offending trajectories in a juvenile offender sample. Journal of Criminal

Justice, 43, 229–241.

Baird, C., Healy, T., Johnson, K., Bogie, A., Dankert, E. W., & Scharenbroch, C. (2013). A comparison of

risk assessment instruments in juvenile justice. Madison, WI: National Council on Crime and

Delinquency.

Baumer, E. P., Messner, S. F., & Felson, R. (1998). Influence of crack cocaine on robbery, burglary, and

homicide rates: A cross-city, longitudinal analysis. Journal of Crime and Delinquency, 33, 316–340.

Bell, K. E. (2009). Gender and gangs: A quantitative comparison. Crime & Delinquency, 55, 363–387.

Bellis, M. A., Lowey, H., Leckenby, N., Hughes, K., & Harrison, D. (2013). Adverse childhood experiences:

Retrospective study to determine their impact on adult health behaviors and health outcomes in a UK

population. Journal of Public Health, 36, 81–91. doi:10.1093/pubmed/fdt038

Burgess, R. L., & Akers, R. L. (1966). A differential association-reinforcement theory of criminal behavior.

Social Problems, 14, 128–147.

Centers for Disease Control and Prevention. (2016). Prevalence of individual adverse childhood experiences.

Retrieved January 5, 2016, from http://www.cdc.gov/violenceprevention/acestudy/prevalence.html

Cohen, M. A., & Piquero, A. R. (2009). New evidence on the monetary value of saving a high risk youth.

Journal of Quantitative Criminology, 25, 25–49.

Costello, B. J., & Vowell, P. R. (1999). Testing control theory and differential association. Criminology, 37,

815–842.

Craig, J. M. (2014). The effects of marriage and parenthood on offending levels over time among juvenile

offenders across race and ethnicity. Journal of Crime and Justice, 38, 163–182. Retrieved from http://dx.doi.

org/10.1080/0735648X.2014.884468

Craig, J. M. (2015). Which bond matters more? Assessing the differential strengths of parental bonding

measures on adolescent delinquency over time. Youth Violence and Juvenile Justice. doi:10.1177/

1541204014565670

Craig et al. 17

at UNIV NORTH TEXAS LIBRARY on December 1, 2016yvj.sagepub.comDownloaded from

Cullen, F. T. (2011). Beyond adolescence-limited criminology: Choosing our future. The American Society of

Criminology 2010 Sutherland Address. Criminology, 49, 287–330.

Curtis, A., & Bandy, T. (2015). The Quantum Opportunities Program: A randomized controlled evaluation.

Washington, DC: The Eisenhower Foundation.

Dembo, R., Turner, G., Sue, C. C., Schmeidler, J., Borden, P., & Manning, D. (1995). Predictors of recidivism

to a juvenile assessment center. Substance Use & Misuse, 30, 1425–1452.

Demuth, S., & Brown, S. L. (2004). Family structure, family processes, and adolescent delinquency: The

significance of parental absence versus parental gender. Journal of Research in Crime and Delinquency,

41, 58–81.

Dierkhising, C. B., Ko, S. J., Woods-Jaeger, B., Briggs, E. C., Lee, R., & Pynoos, R. S. (2013). Trauma histories

among justice-involved youth: Findings from the National Child Traumatic Stress Network. European

Journal of Psychotraumatology, 4. Retrieved from http://dx.doi.org/10.3402/ejpt.v4i0.20274

Dong, M., Anda, R. F., Felitti, V. J., Dube, S. R., Williamson, D. F., Thompson, T. J., . . . Giles, W. H. (2004).

The interrelatedness of multiple forms of childhood abuse, neglect, and household dysfunction. Child Abuse

& Neglect, 28, 771–784.

Farrington, D. P. (2003). Developmental and life-course criminology: Key theoretical and empirical issues—

The 2002 Sutherland Award Address. Criminology, 41, 221–256.

Felitti, V. J., Anda, R. F., Nordenberg, D., Williamson, D. F., Spitz, A. M., Edwards, V., . . . Marks, J. S. (1998).

Relationship of childhood abuse and household dysfunction to many of the leading causes of death in adults:

The Adverse Childhood Experiences (ACE) Study. American Journal of Preventive Medicine, 14, 245–258.

Fox, B. H., Perez, N., Cass, E., Baglivio, M. T., & Epps, N. (2015). Trauma changes everything: Examining the

relationship between adverse childhood experiences and serious, violent, and chronic juvenile offenders.

Child Abuse & Neglect, 46, 163–173. Retrieved from http://dx.doi.org/10.1016/j.chiabu.2015.01.011

Gottfredson, M. R. (2006). The empirical status of control theory in criminology. In F. T. Cullen, J. P. Wright &

K. R. Blevins (Eds.), Taking stock: The status of criminological theory (pp. 77–100). New Brunswick, NJ:

Transaction.

Gottfredson, M. R., & Hirschi, T. (1990). A general theory of crime. Stanford, CA: Stanford University Press.

Herrenkohl, T. I., Lee, J., & Hawkins, J. D. (2012). Risk versus direct protective factors and youth violence:

Seattle Social Development Project. American Journal of Preventive Medicine, 43, S41–S56.

Hillis, S. D., Anda, R. F., Dube, S. R., Felitti, V. J., Marchbanks, P. A., & Marks, J. S. (2004). The association

between adverse childhood experiences and adolescent pregnancy, long-term psychosocial consequences,

and fetal death. Pediatrics, 2, 320–327.

Hirschi, T. (1969). Causes of delinquency. Berkeley: University of California Press.

Huebner, A. J., & Betts, S. C. (2002). Exploring the utility of social control theory for youth development:

Issues of attachment, involvement, and gender. Youth & Society, 34, 123–145.

Ingram, J. R., Patchin, J. W., Huebner, B. M., McCluskey, J. D., & Bynum, T. S. (2007). Parents, friends, and

serious delinquency: An examination of direct and indirect effects among at-risk early adolescents. Criminal

Justice Review, 32, 380–400.

Jaccard, J. (2001). Interaction effects in logistic regression. Thousand Oaks, CA: Sage.

Jang, S. J., & Smith, C. A. (1997). A test of reciprocal causal relationships among parental supervision,

affective ties, and delinquency. Journal of Research in Crime and Delinquency, 34, 307–336.

Kempf, K. L. (1993). The empirical status of Hirschi’s control theory. In F. Adler & W. S. Laufer (Eds.), New

directions in criminological theory: Advances in criminological theory (Vol. 4, pp. 143–185). New Bruns-

wick, NJ: Transaction.

Kubrin, C. E., & Stewart, E. A. (2006). Predicting who reoffends: The neglected role of neighborhood context

in recidivism studies. Criminology, 44, 165–197.

Laub, J. H., & Sampson, R. J. (2003). Shared beginnings, divergent lives: Delinquent boys to age 70. Cam-

bridge, MA: Harvard University Press.

18 Youth Violence and Juvenile Justice 15(1)

at UNIV NORTH TEXAS LIBRARY on December 1, 2016yvj.sagepub.comDownloaded from

Lee, S., Aos, S., Drake, E., Pennucci, A., Miller, M., & Anderson, L. (2012). Return on investment: Evidence-

based options to improve statewide outcomes April 2012 update. Olympia, WA: Washington State Institute

for Public Policy.

Lipsey, M. W. (2009). The primary factors that characterize effective interventions with juvenile offenders: A

meta-analytic overview. Victims and Offenders, 4, 124–147.

Lipsey, M. W., Howell, J. C., Kelly, M. R., Chapman, G., & Carver, D. (2010). Improving the effectiveness of

juvenile justice programs: A new perspective on evidence-based practice. Washington, DC: Georgetown

University, Center for Juvenile Justice Reform.

Maxfield, M. G., & Widom, C. S. (1996). The cycle of violence revisited 6 years later. Archives of Pediatrics

and Adolescent Medicine, 150, 390–395.

Moffitt, T. E. (1993). Adolescence-limited and life-course-persistent antisocial behavior: A developmental

taxonomy. Psychological Review, 100, 674–701.

Paternoster, R., Brame, R., Mazerolle, P., & Piquero, A. (1998). Using the correct statistical test for the equality

of regression coefficients. Criminology, 36, 859–866.

Piquero, A. R., MacDonald, J. M., & Parker, K. F. (2002). Race, local life circumstances, and criminal activity.

Social Science Quarterly, 83, 654–670.

Rivera, B., & Widom, C. S. (1990). Childhood victimization and violent offending. Violence and Victims, 5,

19–35.

Sampson, R. J., & Laub, J. H. (1993). Crime in the making: Pathways and turning points. Cambridge, MA:

Harvard University Press.

Sampson, R. J., & Wilson, W. J. (1995). Toward a theory of race, crime, and urban inequality. In J. Hagan & R.

D. Peterson (Eds.), Crime and inequality (pp. 37–54). Stanford, CA: Stanford University Press.

Smith, C. A., & Thornberry, T. P. (1995). The relationship between childhood maltreatment and adolescent

involvement in delinquency. Criminology, 33, 451–481.

Steffensmeier, D., & Allan, E. (1996). Gender and crime: Toward a gendered theory of female offending.

Annual Review of Sociology, 22, 459–487.

Teague, R., Mazerolle, P., Legosz, M., & Sanderson, J. (2008). Linking childhood exposure to physical abuse

and adult offending: Examining mediating factors and gendered relationships. Justice Quarterly, 25,

313–348.

Thornberry, T. P. (1987). Toward an interactional theory of delinquency. Criminology, 25, 863–891.

Tierney, J. P., Grossman, J. B., & Resch, N. L. (2000). Making a difference: An impact study of Big Brothers/

Big Sisters. Philadelphia, PA: Public/Private Ventures.

Toby, J. (1957). Social disorganization and stake in conformity: Complementary factors in the predatory

behavior of hoodlums. Journal of Criminal Law, Criminology, and Police Science, 48, 12–17.

Widom, C. S. (1989). The cycle of violence. Science, 244, 160–166.

Widom, C. S. (2003). Understanding child maltreatment and juvenile delinquency: The research. In J. Wiig, C.

S. Widom & J. A. Tuell (Eds.), Understanding child maltreatment and juvenile delinquency: From research

to effective program, practice, and systemic solutions (pp. 1–10). Washington, DC: Child Welfare League of

America.

Winokur-Early, K., Hand, G. A., & Blankenship, J. L. (2012). Validity and reliability of the Florida

Positive Achievement Change Tool (PACT) risk and needs assessment instrument: A three-phase

evaluation (validation study, factor analysis, inter-rater reliability). Tallahassee, FL: Justice Research

Center.

Wolff, K. T., Baglivo, M. T., & Piquero, A. R. (2015). The relationship between adverse childhood

experiences and recidivism in a sample of juvenile offenders in community-based treatment. Interna-

tional Journal of Offender Therapy and Comparative Criminology, 1–33. doi:10.1177/030

6624X15613992

Craig et al. 19

at UNIV NORTH TEXAS LIBRARY on December 1, 2016yvj.sagepub.comDownloaded from

Author Biographies

Jessica M. Craig is an assistant professor at the University of North Texas. Her research interests

include criminological theory, life-course criminology, and white-collar crime. Some of her recent

works can be seen in Criminal Behaviour and Mental Health, Crime & Delinquency, and Journal of

Criminal Justice. She earned her PhD in criminology from the University of Texas at Dallas.

Michael T. Baglivio is the director of Research and Program Development for G4S Youth Services,

LLC, tasked with examining the effectiveness of juvenile treatment programming on short- and

long-term performance measures and outcomes. For the past 10 years, he has evaluated the effec-

tiveness of juvenile justice reform initiatives throughout Florida. His research interests include

criminological theory, risk assessment, and life-course criminology.

Kevin T. Wolff is an assistant professor at John Jay College of Criminal Justice in New York City.

He earned his PhD from the College of Criminology and Criminal Justice at Florida State Univer-

sity. His research interests include the spatial patterning of crime, juvenile justice, criminological

theory, and quantitative methods. He is also the chief methodologist at the Research and Evaluation

Center at John Jay and is part of the project devoted to evaluating the Cure Violence program,

funded by the Robert Wood Johnson Foundation.

Alex R. Piquero is the Ashbel Smith professor of Criminology and Associate Dean of Graduate

Programs in the School of Economic, Political and Policy Sciences at the University of Texas at

Dallas. His research interests include criminal careers, criminological theory, and quantitative

research methods. He has received several research, teaching, and service awards and is a fellow

of both the American Society of Criminology and the Academy of Criminal Justice Sciences. In

2014, he was awarded the University of Texas System Regents’ Outstanding Teaching Award.

Nathan Epps, MS, completed his graduate studies in criminology in May 1986 at the Florida State

University, School of Criminology. He currently works for the Bureau of Research and Data

Integrity at the Florida Department of Juvenile Justice (FDJJ). His research interests include

trauma-informed care, adverse childhood experiences, and the application of geographical informa-

tion systems (GIS) software to juvenile justice data.

20 Youth Violence and Juvenile Justice 15(1)

at UNIV NORTH TEXAS LIBRARY on December 1, 2016yvj.sagepub.comDownloaded from