	New content
Youth Violence and Juvenile Justice

Free access

Research article

First published online July 9, 2020

Are the Effects of Adverse Childhood Experiences on Violent Recidivism Offense-Specific?

Jessica M. Craig https://orcid.org/0000-0001-5418-0834 [email protected] and Haley R. Zettler https://orcid.org/0000-0003-4319-7301View all authors and affiliations

Volume 19, Issue 1

https://doi.org/10.1177/1541204020939638

Abstract

A growing body of literature has found that exposure to child maltreatment and other forms of family dysfunction, often conceptualized as Adverse Childhood Experiences (ACEs), are associated with delinquent and criminal behavior. Recent research has indicated that the effects of ACEs on offending may differ not only by offense type but also by sex and race/ethnicity. However, no study to-date has investigated the effects of ACEs on violent-specific recidivism, nor how these effects differ by sex- and racial/ethnic-specific subgroups. The current study seeks to address this gap by examining a large, diverse sample of serious delinquents institutionalized in a large southern state. The results indicated that while ACEs increased the likelihood of being rearrested for any violent felony, the effects were particularly strong for domestic violence and sexual offenses among white females and minority males. A discussion of these findings are presented, along with the limitations of the study.

Introduction

A growing body of literature has found that exposure to child maltreatment and other forms of family dysfunction are associated with antisocial behavior, including violence (Crimmins et al., 2000; Fox et al., 2015). This literature, often conceptualizing trauma exposure as Adverse Childhood Experiences (ACEs), has joined a large body of literature starting in the public health domain that has found exposure to childhood trauma increases the likelihood of a large range of negative outcomes, such as chronic disease, reduced education and employment attainment, early onset of sexual promiscuity, and early death (Drury et al., 2017; Felitti et al., 1998; Hillis et al., 2001, 2004; Vaughn et al., 2017).

Traditionally, scholars include the following ten ACEs: physical abuse, emotional abuse, sexual abuse, physical neglect, emotional neglect, household substance abuse, violent treatment toward mother, parental separation or divorce, household mental illness, and having a household member with incarceration history (National Center for Injury Prevention and Control, 2020). An individual’s ACE score is calculated by summing the number of different events they have experienced where each type of ACE can only be counted once. Thus, an ACE score can range from 0, indicating the individual experienced no ACEs between birth and age 18, to 10, indicating they have experienced all ACEs at least once during that time. Scholars argue it is important to study all ACEs together, as opposed to separately, as the events have been found to be strongly correlated with one another and have lasting, cumulative effects on brain development (Anda et al., 2006, 2010; Cicchetti, 2013; Teicher et al., 2003). Among a sample of high-risk juvenile offenders, Baglivio and Epps (2016) found that among youth experiencing one ACE, 67.5% reported four or more ACEs and 24.5% reported six or more additional ACEs, demonstrating the interrelatedness of ACEs among juvenile offenders.

Though it has only been in the past few years criminologists have assessed the specific relationship between ACEs and crime, the most recent work has started to examine how the effects of ACEs on offending may differ not only by crime type but also by sex and race/ethnicity (DeLisi et al., 2017; Fagan & Novak, 2018). For example, DeLisi and his colleagues (2017) found ACEs tended to increase the likelihood of engaging in sexual offenses but had inconsistent effects with respect to homicide, particularly among racial/ethnic subgroups. However, they focused specifically on commitment offenses, not recidivism. While prior research has found an association between ACEs and recidivism in general, no study to-date has investigated the effects of ACEs on violent-specific recidivism, nor how these effects differ by sex- and racial/ethnic-specific subgroups. The current study seeks to address this gap by examining a large, diverse sample of serious delinquents, institutionalized in a large southern state. Prior to discussing the results of our study, we first provide an overview of the literature on the effects of trauma on later violence, as well as how these effects differ by race/ethnicity and sex.

Adverse Childhood Experiences/Trauma and Violence

A wealth of research has examined the relationship between trauma and aggressive or violent behavior (Crimmins et al., 2000; Fox et al., 2015; Orth & Wieland, 2006; Taft et al., 2011; Wolfe et al., 2004). Explanations of this relationship largely focus on the process by which trauma impacts affect regulation and impaired cognitive functioning (Baer & Maschi, 2003; Garrison & Stolberg, 1983; Ingram & Kendall, 1986). Complex trauma (i.e. experiencing multiple or chronic traumatic events) can lead to problems across various psychosocial domains and result in reactive aggression (Cook et al., 2003; Ford et al., 2012). Ford (2002) argued that early traumatic experiences in childhood disrupts central aspects of brain and personality development, including the ability to self-regulate.

It is proposed that early childhood victimization leads to dysregulation of emotional and social information processing, resulting in severe and persistent problems with oppositional-defiance and aggression, which is then compounded by post-traumatic reactivity and hypervigilance (Ford et al., 2006). Pollak and colleagues (2000) argue that adverse childhood experiences (ACEs) such as physical abuse and emotional neglect can result in hyper-reactivity and difficulties in discriminating emotional expressions. Thus, victimized children may rely on their detection of anger, putting them in a hyper-vigilant state. A national survey of 118 clinicians specializing in child trauma caseloads found that 50% of their clients reported disturbances in affect regulation, attention and concentration, as well as aggressive and risk-taking behavior (Spinazzola et al., 2005).

A number of self-report studies lend support for the proposed relationship between trauma exposure and subsequent violent behavior. A study of high school students found that violence exposure and reported psychological trauma accounted for more than 50% of the variation in self-reported violence (Song et al., 1998). Similarly, Wolfe and colleagues (2004) found that trauma symptoms predicted incidents of dating violence in high schoolers. In a large-scale survey of middle and high school students (n = 136,549), Duke and colleagues (2010) found that each ACE category was significantly correlated with each category of antisocial behavior including: delinquency, bullying, physical violence, dating violence, weapon-carrying on school property, and self-directed violence.¹ Further, every additional ACE exposure increased the risk of violence from 35% to 144%.

There is also strong evidence of the trauma-violence association in samples of juvenile offenders. Youth remanded to the Office of Children and Family Services in New York for violent crimes reported an average of 8.57 traumatic life events (Crimmins et al., 2000). In a longitudinal study of children followed between the ages of 5 to 21, those who were physically abused during the first five years of their life had a greater risk of being arrested for violent offenses in adolescence (Lansford et al., 2007). Among detained male juveniles, 86% experienced a traumatic event and 71% experienced multiple types of prior trauma (Stimmel et al., 2014). Research on ACEs and juvenile offenders demonstrates the cumulative effect that ACEs have on becoming a serious, violent, and chronic offender (Baglivio et al., 2015; Fox et al., 2015). Further, research has consistently indicated juvenile offenders tend to have significantly higher trauma exposure than those in the general population (Baglivio & Epps, 2016; Grevstad, 2010). Overall, there is strong evidence that childhood trauma exposure and ACEs increase the likelihood of aggression and violent behavior in later adolescence and early adulthood.

Racial/Ethnic and Sex Effects of ACEs on Crime

Research has largely demonstrated that minorities experience ACEs at a higher rate than Whites (Gjelsvik et al., 2014; Reinert et al., 2015). Racial differences in ACE exposure may partially be explained by their overrepresentation in low-income communities (Adler & Rehkopf, 2008; Burke et al., 2011; Jackson et al., 2010; Nazroo, 2003; Shonkoff & Garner, 2012; William & Jackson, 2005). However, research provides mixed results surrounding the relationship between race/ethnicity and ACEs in the context of juvenile offending. ACEs significantly increased the likelihood of residential placement for Black and Hispanic males but had no effect on White males; while ACEs predicted residential placement for Black females but had no effect for White or Hispanic females (Zettler et al., 2018). In a longitudinal study of race differences in the prevalence and impact of ACEs on juvenile delinquency, Fagan and Novak (2018) found that the number of ACEs significantly increased the likelihood of alcohol use, marijuana use, and arrest at age 16 among Blacks but not Whites, though Whites had a higher ACE exposure than Black youth.

DeLisi and colleagues (2017) utilized a large sample of confined male juveniles (n = 2,520) to examine whether the effects of ACEs on commitment offense vary by race/ethnicity. The authors found that Blacks with one ACE were 39% less likely to be committed for homicide; Hispanics with 4 ACEs were 100% more likely to be committed for homicide, and Whites with 5 or 6 ACEs were 68% and 85% less likely to be committed for homicide (DeLisi et al., 2017). Further, Whites with less ACE exposure were more likely to commit a serious person or property offense, while Blacks and Hispanics with 3–5 ACEs were less likely to commit a serious person or property offense. It is important to note that the sample was limited to confined males, thus sex differences were unable to be examined. Further, as the study found significant racial and ethnic differences in commitment offense, the authors did not examine whether these effects held when considering recidivism as an outcome.

Several studies have considered sex differences in both the prevalence and consequences of ACEs, as females typically report more extensive trauma histories than males (Gavazzi et al., 2006; Johansson & Kempf-Leonard, 2009). In a sample of over 10,000 youth referred to juvenile court, 25% of females reported prior abuse or maltreatment as compared to 7% of males (Johansson & Kempf-Leonard, 2009). While there is evidence of a direct link between childhood maltreatment and externalizing behavior for males (rule-breaking and aggression), females’ internalizing behavior (affective and somatic problems) potentially mediate the link between childhood maltreatment and externalizing behavior (Maschi et al., 2008).

Scholars posit that trauma might be a risk factor of offending that might be especially salient among females. Child abuse is hypothesized to be a risk factor for a female-specific pathway to serious, violent, and chronic delinquency (Howell, 2003). According to Howell (2003), “the combination of all these experiences may have greater negative effects on girls than boys” (p. 68). Studies have examined sex differences in the effects of trauma on violent offending. Using a sample of convicted violent offenders, Rossegger and colleagues (2009) found that females were more likely to report ACEs than males. Using data from a sample of low-income minority participants, history of child maltreatment only increased the likelihood of violent arrest for females (Topitzes et al., 2012). However, while females were more likely to be victims of abuse and neglect than males, Asscher and colleagues (2015) found that abuse histories increased the likelihood to commit violent offenses for both males and females.

Several studies have assessed how sex might impact trajectories of juvenile offending. An examination of the subtypes of serious and violent female juvenile offenders using latent class analysis found that females who were categorized as violent and delinquent had significantly higher rates of internalizing disorders, affect dysregulation, family members with criminal histories, and exposure to violence (Odgers et al., 2007). Using data from a longitudinal study of matched maltreated and non-maltreated children, Widom and colleagues (2018) reported that offenders were more likely to be male and abused/neglected as compared to non-offenders. Sex-specific analyses revealed that maltreated females were more likely to be offenders well into adulthood as compared to the control group, suggesting the effects of maltreatment may be especially long-lasting for females.

Current Study

Prior research has established ACEs increase the likelihood of aggressive and violent behaviors, indicating those with higher trauma exposure are more likely to be serious, violent, and chronic offenders (Duke et al., 2010; Fox et al., 2015). However, additional evidence suggests the effect of ACEs may differ not only by sex and race/ethnicity but also by offense type (DeLisi et al., 2017; Fagan & Novak, 2018). To the best of our knowledge no one has yet examined the link between ACEs and violent recidivism specifically, nor how that varies by the intersections of race/ethnicity and sex. This gap is unfortunate given the strong association between ACEs and recidivism (see Craig et al., 2017); understanding potential subgroup differences in the ACE-recidivism relationship would allow us to identify those most in need of trauma-informed treatment. The current study seeks to address this gap by examining the impact of ACE exposure on violent recidivism among a sample of confined juvenile delinquents. It also considers how these effects may differ based on racial/ethnic and sex subgroups and by specific recidivism types (i.e. homicide, domestic violence, aggravated assault, and sexual offenses) as DeLisi et al. (2017) found substantive differences related to crime type.

Methods

Sample and Data

The sample for the current study consists of 11,788 indeterminate-sentenced juvenile offenders incarcerated in Texas state juvenile correctional facilities from 2005 to 2013. As our primary outcome of interest is recidivism, all of the offenders had been released from incarceration at the point the data were collected. All data were provided by the Texas Juvenile Justice Department (TJJD), including information across a number of domains such as delinquent history, demographics, family history, and institutional assessment. Information regarding the juveniles’ institutional behavior was also provided.

Measurement

Dependent variables

As our outcome of interest concerned violent recidivism, we included five different operationalizations of this concept. The first recidivism measure was a binary indicator of a rearrest for any violent offense. The remaining four measures indicated if the delinquent was rearrested for a specific violent offense. These included (1) murder/homicide, (2) domestic violence, (3) aggravated assault, and (4) sexual assault. As the offenders had different follow-up periods based upon their release dates, recidivism was measured as a standardized 3-year follow-up for all members of the sample. In other words, those who were rearrested for a violent offense within three years post-release were considered recidivists while those who did not re-offend within this 3-year period were considered non-recidivists. About thirty percent (30.3%) were rearrested for any violent felony within three years post-release. One percent of the offenders were rearrested for murder, 10% for domestic violence, 12% for aggravated assault, and 1.4% for sexual assault.

While TJJD provided the recidivism data, these data were originally collected by the Texas Department of Public Safety (DPS). DPS obtains the arrest information in the State of Texas and matched the arrest records to the offenders of focus in the current study. The arrest data collection effort followed the offenders into adulthood so there is no artificial right-censoring of the data as the youth mature into adulthood.

Independent variable

Several variables derived from risk assessments and other TJJD records were used to measure each youth’s ACE score. Dichotomous indicators of sexual abuse, emotional abuse, and physical abuse were included. While TJJD data included several variables related to neglect, specifically physical neglect, parental neglect, medical neglect, and neglectful supervision, these do not align with the ACE score paradigm of neglect which measures physical and emotional neglect separately. As a result, a conservative approach was taken and all potential neglect measures were combined to form an overall indicator of any form of neglect. Specifically, if the individual had experienced at least one of these four forms of neglect, they received a “1” for any neglect.

Next, several measures of family dysfunction were counted as part of the ACE score. Three of these measures represented if the youth had a mentally-ill, depressed, or suicidal family member, had a drug- or alcohol-addicted family member, and if at least one of their family members was incarcerated. An additional measure included a variable representing whether at least one of the youth’s parents either had abandoned their child or had their rights revoked. The last ACE measure represented if the youth had been exposed to domestic violence. Each of these variables were dichotomously coded where the presence of the trauma was coded as “1” and the absence was coded as “0.” To create the overall ACE score, each of the ACEs were summed to form an overall scale where “0” indicated the youth had not experienced any ACEs and “9” indicated the youth had experienced all ACEs at least once in their life. The average ACE score was 2.18 (SD = 2.01).²

Control variables

Several key demographic variables were included as control variables. The youth’s sex was controlled for, where males were coded as “1” and females were coded as “0.” In the current sample, 91% of the sample was composed of males. The delinquents’ racial/ethnic background was also included, specifically Black, Hispanic, White, and other. In the current sample, 35% of the sample was Black, 44% were Hispanic, 21% of the sample was White and less than one percent of the sample was of another race/ethnicity. In the full-sample analyses, those that were White or of another race/ethnicity were treated as the reference group. In the race/ethnicity-specific analyses, those that were of another race/ethnicity were combined with the White subgroup due to the small sample size. The final demographic variable controlled for was age at release, coded continuously. The mean age of the sample at release was 18.07 (SD = 1.09).

Several offending-related measures were included due to their known association with recidivism. The number of prior adjudications for any type of offense the delinquent had prior to their incarceration was included. The mean number of prior adjudications for the sample was 3.38 (SD = 1.81). A youth's gang affilitation at state-intake was also included, where “1” indicated the youth was gang-affiliated. Forty percent of the sample was believed to be gang-affiliated.³ The number of misconduct incidents the youth engaged in during their incarceration was also included, standardized to form a rate of misconduct incidents per 100 days in a TJJD facility. The mean misconduct rate was 12.24 (SD = 14.64). Finally, the length each youth was incarcerated was also controlled for, with the average length of incarceration equal to 1.97 years (SD = .70).

Analyses

After first assessing the bivariate correlations between the variables of interest, several logistic regression models were estimated to calculate the effects of ACE scores on violent recidivism. The first set of models were calculated among the full sample, estimating the impact of ACEs and the control variables initially on any violent re-arrest, followed by the specific offense types of murder, domestic violence, aggravated assault, and sexual assault. These five regression models were then estimated among the sex and racial/ethnic subgroups. First they were estimated among males by each racial/ethnic subgroup (i.e. White males, Black males, Hispanic males) and then among females by each racial/ethnic subgroup (i.e. White females, Black females, Hispanic females).⁴

Results

Bivariate Correlations

Table 1 presents the correlations between the variables of interest. As can be seen, ACE scores is weakly but significantly correlated with having at least one violent rearrests (r = .02). ACEs has a significant but weak, negative relationship with being rearrested for an aggravated assault (r = −.02) and a significant but weak, positive relationship with being rearrested for sexual assault (r = .03). Multicollinearity did not appear to be an issue as, with the exception of the two racial groups, none of the correlations were greater than r = .4 and all post-analyses variance inflation factors were less than 2.0.

Table 1. Bivariate Correlations Between All Variables.

Variable	1	2	3	4	5	6	7	8	9	10	11	12	13	14
Any Violent Rearrests	1.00
Any Murder Rearrests	.16^*	1.00
Any Domestic Violence Rearrests	.51^*	−.01	1.00
Any Aggravated Assault Rearrests	.56^*	.08^*	.05^*	1.00
Any Sex Offense Rearrests	.18^*	−.01	.02^*	.02^*	1.00
ACE Score	.02^*	−.02	.02	−.02^*	.03^*	1.00
Prior Adjudications	.04^*	.02	.02^*	.03^*	−.01	.01	1.00
Age at Release	−.04^*	−.01	.03^*	−.03^*	.01	−.03^*	−.06^*	1.00
Male	.09^*	.02^*	.04^*	.08^*	.04^*	−.17^*	−.07^*	.02	1.00
Black	.06^*	.05^*	.00	.07^*	.00	−.08^*	.05^*	−.02	.00	1.00
Hispanic	.01	−.02^*	.03^*	.00	−.02	−.09^*	.03^*	.01	.05^*	−.65^*	1.00
Gang Affiliation	.07^*	.04^*	.03^*	.06^*	.01	.00	.08^*	−.04^*	.09^*	−.03^*	.22^*	1.00
Misconduct Rate	.10^*	.00	−.01	.04^*	.01	.09^*	.09^*	−.30^*	−.02	.20^*	−.16^*	.08^*	1.00
Time Incarcerated	−.03^*	−.01	−.01	−.03^*	.01	.05^*	−.14^*	.27^*	.00	.05^*	−.05^*	−.02^*	.01	1.00

^*p < .05.

Effects of ACEs on Violent Rearrests

Our first set of logistic regression models were estimated among the full sample and can be seen in Table 2. The results from the first model indicated ACEs were found to increase the odds of being rearrested for a violent felony (OR = 1.05, p < .001). Those with more prior adjudications (OR = 1.03, p < .01) were also more likely to be rearrested for a violent felony. Males (OR = 2.35, p < .001), Blacks (OR = 1.62, p < .001), and Hispanics (OR = 1.41, p < .001) evinced higher odds of violent recidivism than their counterparts. Finally, those with a gang affiliation (OR = 1.20, p < .001), those with a higher rate of misconduct (OR = 1.01, p < .001) and those who spent less time incarcerated (OR = .90, p < .001) were significantly more likely to be rearrested for a violent felony.

Table 2. Logistic Regression Models Estimating Effects of ACE Score on Violent Rearrests Among the Full Sample.

Variable	Any Violent Rearrests		Murder		Domestic Violence		Aggravated Assault		Sexual Assault
Variable	OR	SE	OR	SE	OR	SE	OR	SE	OR	SE
ACE Score	1.05^***	.01	.96	.05	1.06^***	.02	1.01	.02	1.13^**	.04
Prior Adjudications	1.03^**	.01	1.06	.05	1.04^*	.02	1.03^†	.02	.93	.05
Age at Release	1.00	.02	.91	.08	1.10^**	.03	.96	.03	1.13	.09
Male	2.35^***	.21	3.33^*	1.97	1.99^***	.28	3.98^***	.69	--	--
Black	1.62^***	.10	3.40^***	1.13	1.32^**	.12	2.15^***	.20	.88	.19
Hispanic	1.41^***	.08	1.22	.43	1.47^***	.13	1.55^***	.14	.71	.15
Gang Affiliation	1.20^***	.05	2.16^***	.41	1.07	.07	1.27^***	.08	1.32^†	.22
Misconduct Rate	1.01^***	.00	.99	.01	1.00	.00	1.00^*	.00	1.01	.01
Time Incarcerated	.90^***	.03	.92	.13	.93	.04	.88^**	.04	1.05	.12
n	11,788		11,788		11,788		11,788		11,788

^***p < .001, ^**p < .01, ^*p < .05, ^†p < .10 (two-tailed).

The next model indicated ACEs did not have a significant effect on being rearrested for murder. The third model indicated those with more ACEs were more likely to be rearrested for domestic violence than those with lower ACE scores (OR = 1.06, p < .001). The fourth model indicated ACEs had no significant effect on the likelihood of being rearrested for aggravated assault. Finally, the fifth model indicated those with more ACEs were more likely to be rearrested for a sexual assault (OR = 1.11, p < .05).⁵ Across all of these models, with the exception of sexual assault, other control variables, particularly the demographic variables, had similar relationships with recidivism as the first model.

Racial/Ethnic Effects of ACEs on Violent Rearrests Among Males

Table 3 presents the results of the logistic regression models estimating the effect of ACEs and the control variables on violent rearrests among males, stratified by racial/ethnic group. The first three models indicate the effect of ACEs on being rearrested for any violent felony within three years post-release by racial/ethnic group. As can be seen, ACEs failed to have an effect on having a violent rearrest among White males, but did significantly increase the odds of being rearrested for a violent felony among Black (OR = 1.06, p < .01) and Hispanic males (OR = 1.06, p < .001). Across these models, the only control variable that had a consistent effect was misconduct rate, with those with a higher misconduct rate being significantly more likely to have a violent rearrest (White males OR = 1.02, p < .01; Black males OR = 1.01, p < .01; Hispanic males OR = 1.01, p < .001).

Table 3. Logistic Regression Models Estimating Effects of ACE Score on Violent Rearrests Among Males.

Variable	Any Violent Rearrests						Murder						Domestic Violence						Aggravated Assault						Sexual Assault
	W		B		H		W		B		H		W		B		H		W		B		H		W		B		H
	OR	SE	OR	SE	OR	SE	OR	SE	OR	SE	OR	SE	OR	SE	OR	SE	OR	SE	OR	SE	OR	SE	OR	SE	OR	SE	OR	SE	OR	SE
ACE Score	1.00	.02	1.06^**	.02	1.06^***	.02	.83	.14	.98	.06	.96	.09	.97	.03	1.08^**	.03	1.06^*	.03	1.03	.04	1.02	.02	1.00	.02	1.02	.07	1.24^***	.07	1.21^**	.08
Prior Adjudications	1.07^*	.03	1.01	.02	1.05^**	.02	.98	.20	1.06	.06	1.06	.08	1.07	.05	1.01	.03	1.05^†	.03	1.09^†	.05	.98	.02	1.05^*	.02	.93	.10	.87	.08	1.00	.07
Age at Release	1.01	.05	1.02	.03	.98	.03	.94	.28	.87	.10	.95	.15	.96	.07	1.24^***	.07	1.08^†	.05	1.01	.08	.99	.04	.91^*	.04	1.2	.21	1.15	.16	1.05	.14
Gang Affiliation	1.46^**	.18	1.09	.08	1.27^***	.08	4.08^*	2.63	1.98^**	.48	2.41^*	.86	1.23	.23	1.06	.12	1.08	.1	1.26	.24	1.18^†	.11	1.42^***	.13	1.57	.56	.96	.26	1.44	.39
Misconduct Rate	1.02^***	.00	1.01^***	.00	1.01^***	.00	1.01	.02	.99^†	.01	.99	.01	1.00	.01	1.00	.00	.99	.00	1.01	.01	1.00	.00	1.01^†	.00	1.00	.01	1.01^†	.01	.99	.01
Time Incarcerated	.94	.07	.91^†	.05	.89^*	.04	.54	.32	1.07	.19	.77	.21	1.01	.11	.97	.08	.87^†	.06	.76^*	.09	.87^*	.06	.94	.07	1.31	.25	1.09	.19	.79	.17
n	2,208		3,736		4,809		2,208		3,736		4,809		2,208		3,736		4,809		2,208		3,736		4,809		2,208		3,736		4,809

^***p < .001, ^**p < .01, ^*p < .05, ^†p<.10 (two-tailed).

The next set of results in Table 3 indicate that ACEs failed to predict being rearrested for a homicide-specific offense within three years post-release among all males, regardless of their race/ethnicity. The third set of columns indicated ACEs were significantly associated with an increased odds of being rearrested for domestic violence among Black (OR = 1.08, p < .01) and Hispanic males (OR = 1.06, p<.05). However these results did not hold for White males. The fourth set of columns demonstrated that the males’ ACE score was not associated with being rearrested for aggravated assault. Finally, the last set of columns indicate Black males with a higher ACE score were significantly more likely to be rearrested for sexual assault than those with a lower ACE score (OR = 1.24, p<.001). Hispanic males with a higher ACE score were significantly more likely to be rearrested for sexual assault (OR = 1.21, p < .01) than their counterparts. This relationship failed to hold for White males, though. The control variables did not have consistent, significant relationships with the outcomes of interest among these models.

Racial/Ethnic Effects of ACEs on Violent Rearrests Among Females

Our last set of logistic regressions estimated the impact of ACEs on violent rearrests among females, stratified by race/ethnicity, and the results can be seen in Table 4. The first set of results indicated White females with higher ACE scores were marginally more likely to be rearrested for any violent felony (OR = 1.16, p < .10) and Black females were significantly more likely to have a violent felony rearrest (OR = 1.15, p<.05). ACEs had no significant effect on violent rearrests among Hispanic females, however. Further, no control variable had a consistent effect across these models.

Table 4. Logistic Regression Models Estimating Effects of ACE Score on Violent Rearrests among Females.

Variable	Any Violent Rearrests						Domestic Violence						Aggravated Assault
	W		B		H		W		B		H		W		B		H
	OR	SE	OR	SE	OR	SE	OR	SE	OR	SE	OR	SE	OR	SE	OR	SE	OR	SE
ACE Score	1.16^†	.10	1.15^*	.07	1.09	.07	1.53^*	.26	1.20^†	.12	1.12	.10	.92	.21	1.00	.11	.93	.14
Prior Adjudications	.96	.09	1.03	.06	.97	.06	.80	.17	1.13	.10	1.00	.10	1.07	.25	1.03	.11	.98	.14
Age at Release	.83	.17	1.07	.15	.85	.12	1.32	.49	1.06	.25	.89	.19	1.35	.80	1.19	.30	1.38	.45
Gang Affiliation	1.42	.67	.84	.27	.81	.24	5.17^*	3.91	.25^†	.19	.72	.33	1.62	1.95	.51	.34	.60	.42
Misconduct Rate	1.03^**	.01	1.01^†	.01	1.01	.01	1.06^**	.02	1.01	.01	1.01	.01	1.05	.03	1.01	.01	1.03^†	.02
Time Incarcerated	.49^†	.20	.94	.20	1.01	.24	.11^**	.09	1.26	.43	1.22	.43	.62	.66	1.38	.49	.40	.24
n	296		362		377		296		362		377		296		362		377

^*** p < .001, ^** p < .01, ^* p < .05, ^†p<.10 (two-tailed).

Due to insufficient sample size, the only offense-specific outcomes we were able to assess for females were domestic violence and aggravated assault. As Table 4 indicates, ACEs were significantly associated with domestic violence rearrests among White females (OR = 1.53, p<.05) and marginally associated with domestic violence rearrests among Black females (OR = 1.20, p<.10). However ACEs had no significant effect on domestic violence rearrests among Hispanic females nor did it have an impact on aggravated assault rearrests among these females, regardless of race/ethnicity. The control variables also failed to emerge as significant predictors in most of these analyses.

Supplemental Analyses: All Females

As some of the racial/ethnic subsamples among females were small, supplemental analyses were estimated that included all females, regardless of racial/ethnic group (all supplemental analyses results are available upon request from the first author). These results were similar to those presented above; females’ ACE scores were significantly associated with having any violent rearrest (OR = 1.11, p < .01) and being rearrested for domestic violence (OR = 1.17, p < .05). However, the females’ ACE scores were not associated with being rearrested for homicide-related offenses or aggravated assaults specifically. As no female was rearrested for a sexual assault, this model was not estimated among females.

Supplemental Analyses: High ACE Score

Prior literature has indicated those with higher ACE scores are more likely to be serious, chronic, and violent offenders than those with no or fewer ACEs (Baglivio et al., 2015). The CDC (2015) identifies those with 4 or more ACEs as being at a particularly high risk so supplemental analyses were estimated using a cut-point of 4 ACEs to operationalize higher-risk youth. Specifically, those who had at least four ACEs were coded as “1” while those with three or fewer ACEs were coded as “0.” Among the full sample and the racial/ethnic and sex subsamples, the results were substantively similar to the ones presented above. All supplemental analyses results are available upon request from the first author.

Discussion and Conclusion

Though prior research had indicated those with more ACEs were not only more likely to recidivate but also more likely to be serious, violent, and chronic juvenile offenders (Craig et al., 2017; Fox et al., 2015), additional evidence suggested there may be racial/ethnic and sex differences in the effects of ACEs on offending (DeLisi et al., 2017; Fagan & Novak, 2018). Further, while DeLisi and colleagues (2017) found important distinctions in the impact of ACEs on violent commitment offenses, it was unknown the extent to which this would extend to violent recidivism as well. Using a large sample of confined juveniles, the current study examined the effects of ACEs on violent recidivism and whether the effects varied by race/ethnicity and sex. Taken as a whole, our findings suggested ACEs increase the likelihood of being rearrested for any violent felony, for domestic violence, and for sexual assault. ACE exposure was not found to be associated with being rearrested for murder or aggravated assault.

However, this pattern did not hold when the sex-specific subsamples were examined. Among males, ACE exposure increased the odds of being rearrested for any violent felony, domestic violence, and sexual assault among Blacks and Hispanics but failed to be significant for Whites. These results are somewhat similar to those reported by DeLisi et al. (2017) who found ACEs increased the likelihood of being committed for a sexual offense for all males, regardless of race/ethnicity. There is additional theoretical and empirical evidence to support a connection between traumatic experiences and later sexual offending. Developmental psychopathology theories hypothesize that households characterized as traumatic produce inappropriate models of emotional and behavioral regulation which may result in the adoption of maladaptive coping behaviors, including sexualized coping (Bloom & Farragher, 2013; Cicchetti & Banny, 2014; Levenson & Socia, 2016; Patterson et al., 1990; Rutter & Sroufe, 2000; Young et al., 2003). In a sample of adult sexual offenders, Levenson and Socia (2016) found that childhood sexual abuse, emotional neglect, and exposure to domestic violence significantly predicted the total number of adult sex crime arrests. To the best of our knowledge there has not been sufficient research into the potential for racial/ethnic disparities in these effects so future studies should further explore the relationship between childhood trauma and later sexual offending, particularly across racial/ethnic groups.

The current study also found that ACEs increased the odds of domestic violence recidivism for Black and Hispanic males but not among White males. Prior research has consistently found that witnessing domestic violence in childhood increases the likelihood of later domestic violence perpetration (Gil-González et al., 2008; Milaniak & Widom, 2015; Miller et al., 2011; Pournaghash-Tehrani & Feizabadi, 2009). One potential explanation for our finding regarding ACE and domestic violence perpetration for minority males is that minorities are more likely to witness domestic violence during childhood, thus potentially increasing the risk of later perpetration (Roberts et al., 2011). However, in our sample 29% of White males, 15% of Black males, and 20% of Hispanic males were reported as having been exposed to domestic violence, suggesting the presence of domestic violence may not be a salient factor. Perhaps the severity and/or frequency of exposure may be more important; unfortunately, neither the available data nor the ACE construct is able to account for this contextual information, leaving it an important area for future research.

Our next subsample analyses focused among female delinquents. Among females, ACEs significantly increased the odds of recidivism for having any violent felony rearrest among Black females and marginally among White females. ACEs were also found to significantly predict domestic violence among White females and marginally for Black females. ACEs failed to be associated with any violent recidivism outcome for Hispanic females, however. In sum, these findings echoed those of the males that suggested ACEs increase the likelihood of recidivism for some females for some, but not all, offense types. To the best of our knowledge, this is the first study to examine the impact of ACEs on offending among female racial/ethnic subsamples. Nonetheless, it is telling that there is a consistent link between the effects of ACEs on domestic violence rearrests among both males and females, though the racial/ethnic subgroups differ between the two sexes. As previously mentioned, prior research has established a link between witnessing domestic violence as a child and later perpetrating the same behavior (Gil-González et al., 2008; Milaniak & Widom, 2015; Miller et al., 2011; Pournaghash-Tehrani & Feizabadi, 2009). Further studies have found this effect may be particularly strong among females, as female domestic violence perpetrators are more likely to have a history of domestic violence exposure as well as childhood physical abuse than their male counterparts (Goldenson et al., 2007; Henning et al., 2006).

Although prior research has found racial differences in domestic violence recidivism, this may be an artifact of racial bias in the criminal justice system as recidivism is usually defined as any official arrest (Kingsnorth, 2006; Ménard et al., 2009). For example, Maxwell et al. (2002) found that while minority males were more likely to recidivate according to official records, White men were more likely to recidivate using victim’s reports. Similarly, data from the National Crime Victimization Survey found that African American and Hispanic victims were more likely to contact the police about a domestic violence incident than Whites (Rennison & Welchans, 2000). Therefore, it is possible that the racial and ethnic differences in domestic violence recidivism are a result of systemic racial bias or underreporting of victimizations by White victims.

This study’s limitations should be kept in mind when considering its results. First, the measures of child maltreatment and family dysfunction were missing important contextual details, such as the timing, severity, and frequency of such events. This is an issue common to all studies relying upon the traditional conceptualization of ACEs though having richer data may have clarified some of the sex and racial/ethnic differences our study revealed. Second, as our analyses relied upon secondary agency data collected from TJJD, some of the specific information regarding the source of the variables were missing. TJJD collects a large amount of data from several sources, including self-repots, official records, staff observations, and diagnostic assessments. Unfortunately, the source(s) for some of the variables included in the current study were not provided which can make it challenging to further contextualize the results. Third, the juveniles included in the sample served time in a number of different institutions where their experiences and quality of treatment may have varied, potentially leading to systematic bias within our results. Future studies could utilize hierarchical models to investigate the potential for facility-specific effects.

A couple of sample-specific limitations should also be acknowledged. As the sample only included serious juvenile offenders confined in Texas, this limits its generalizability to other less serious delinquents and in other locations. Additionally, some of the female-specific analyses relied upon small subsamples, potentially limiting the analyses.

Future research directions could include seeking to address these limitations. A promising avenue would be adding contextual details on childhood trauma. Though our findings indicated trauma-exposed minority males were more likely to engage in later domestic violence than their White male counterparts, this may be due to differences in exposure severity, frequency, or timing. A second area for future study includes assessing potential mediators and moderators of the effect of ACEs on subsequent offending. Craig and her colleagues (2019) recently found substance use and mental health issues partially mediated the effect of ACEs on recidivism but there were some racial/ethnic and sex differences in these effects. Though the researchers only considered recidivism broadly, not violence-specific, it is possible that variables such as mental illness and substance use, known consequences of trauma exposure, may explain some of the differences found here.

In closing, our results indicate exposure to family dysfunction and child maltreatment increase the likelihood of serious juvenile offenders engaging in violent offenses following confinement in a juvenile facility. However, these effects seem to be offense-specific and differ based upon the juvenile’s sex and race/ethnicity. This study represents an important first step in understanding how the effects of cumulative trauma are associated with later violent behavior, though more research is needed to clarify the mechanisms leading to the group differences. This will help increase the targeting of trauma-informed interventions to gear them toward those who pose the highest risk of violently recidivating.

Research on trauma-informed interventions in juvenile residential facilities provide promising results for reducing violent behavior. For instance, evaluations of programs such as Trauma and Grief Component Therapy (TGCTA), Trauma Affect Regulation: A Guide for Education and Therapy (TARGET), Sanctuary, and Think Trauma report success in reducing violence among committed youth (Ford et al., 2005; Ford & Hawke, 2012; Marrow et al., 2012; Olafson et al., 2018). Future research is needed to evaluate the effectiveness of such programs in reducing violent recidivism. In particular, as the current results indicated ACEs increased the likelihood of engaging in future sexual assaults and domestic violence, programs such as Project BUILD should be adopted by juvenile justice institutions. Project BUILD incorporates presentations on domestic violence and rape by a local sexual assault victim’s advocacy group and has been found to significantly reduce recidivism compared to nonparticipants in a control group (Lurigio et al., 2000).

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

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

ORCID iDs

Jessica M. Craig https://orcid.org/0000-0001-5418-0834

Haley R. Zettler https://orcid.org/0000-0003-4319-7301

Footnotes

1. Delinquency included: past-month truancy, past-year running away from home, vandalism, theft from store, gambling (summed, range: 0–10).

2. Some may be surprised at the relatively low prevalence of ACEs in the current sample, given that prior research using data from FL Department of Juvenile Justice found a mean ACE score of 3.61 (Baglivio et al., 2015). Some of this discrepancy may be due to the current study’s measurement of neglect. TJJD records included four types of neglect and thus the traditional measures of emotional and physical neglect were not able to be used. Instead all forms of neglect were consolidated into one overall indicator.

3. Unfortunately, TJJD did not provide the definitions utilized when deciding if a particular juvenile had a history of gang membership. While it is a limitation that these contextual details were missing, this is not unusual in the use of agency data.

4. The data were assessed for univariate and multivariate outliers. The misconduct rate variable appeared to have four outliers so these were dropped from the analyses. While dropping these cases did not substantively change the results, the fit statistics of the models were improved. There were no other univariate outliers. 495 cases (4.03% of the sample) were identified as multivariate outliers using Mahalanobis Distance scores. Supplemental analyses were estimated with and without the outliers and as the results were substantively different, they were dropped from the analyses. The data were also assessed for missing data. One respondent, less than 1% of the entire sample, was missing data for prior adjudications so they were dropped from all analyses.

5. No females were rearrested for sexual assault so the control variable for sex was removed from this analysis.

References

Adler N. E., Rehkopf D. H. (2008). U.S. disparities in health: Descriptions, causes, and mechanisms. Annual Review of Public Health, 29, 235–252.

Abstract

Introduction

Adverse Childhood Experiences/Trauma and Violence

Racial/Ethnic and Sex Effects of ACEs on Crime

Current Study

Methods

Sample and Data

Measurement

Dependent variables

Independent variable

Control variables

Analyses

Results

Bivariate Correlations

Effects of ACEs on Violent Rearrests

Racial/Ethnic Effects of ACEs on Violent Rearrests Among Males

Racial/Ethnic Effects of ACEs on Violent Rearrests Among Females

Supplemental Analyses: All Females

Supplemental Analyses: High ACE Score

Discussion and Conclusion

Declaration of Conflicting Interests

Funding

ORCID iDs

Footnotes

References

Biographies

Cite article

Cite article

Download to reference manager

Share

Share this article

Share with email

Share on social media

Share access to this article

Information

Published In

Keywords

Rights and permissions

Authors

Affiliations

Notes

Metrics

Journals metrics

Article usage*

Articles citing this one

Figures & Media

Tables

View options

PDF/ePub

Get access

Access options

Sign in

Also from Sage

Article usage^*