The Effects of Adverse Childhood Experiences on Internalizing Versus Externalizing Outcomes

Abstract

This study examines the effects of adverse childhood experiences (ACEs) on the risk of internalizing or externalizing outcomes among juveniles. While myriad research has investigated the impacts of ACEs on internalizing and externalizing outcomes, it is unclear whether ACEs have a stronger link to one outcome over the other when controlling for other factors. Using a sample of 30,909 youth who exclusively exhibited internalizing (n = 1,030) or externalizing problems (n = 29,879), regression techniques and propensity score matching were utilized to evaluate the impact of each ACE on the risk of internalizing versus externalizing outcomes. Results indicate that the most pertinent factor for predicting externalized problems is emotional abuse. Household member incarceration, physical abuse, emotional neglect, and household violence or substance abuse also predicted externalizing outcomes. Sexual abuse was the only ACE predictive of internalizing, while physical neglect and parental mental illness did not have a correlation with either outcome.

Keywords

abuse trauma internalizing externalizing juveniles

Introduction

A landmark transdisciplinary work, the Adverse Childhood Experiences Study provided powerful empirical support for the notion that assorted forms of childhood abuse and neglect denote numerous risks for reduced physical, psychiatric, and behavioral health across adulthood (Dube et al., 2003; Felitti et al., 1998). Adverse childhood experiences (ACEs) work in a gradient manner in that individuals who sustain the greatest amount of abuse and neglect tend to have the worst health and behavioral outcomes. This finding also holds true in criminological research. Studies consistently indicate that juvenile offenders tend to have more ACEs and negative outcomes as compared with the general population (Baglivio & Epps, 2016; Dierkhising et al., 2013; B. H. Fox, Perez, Cass, Baglivio, & Epps, 2015). It is estimated that upward of 90% of juvenile offenders have experienced trauma, abuse, neglect, or maltreatment in childhood (Dierkhising et al., 2013), and childhood trauma and abuse increases the odds of violent behavior in juveniles by more than 200% (Maxfield & Widom, 1996).

Delinquency, however, is only one of the negative outcomes that may result from ACEs. Other mental health and behavioral issues also stem from these traumatic events. For example, severe, repeated, or long-term trauma in childhood has been shown to dramatically increase the risk of other conduct and behavioral problems (Craig, Piquero, Farrington, & Ttofi, 2017) such as risky sexual behavior and poor educational outcomes (Basto-Pereira, Miranda, Ribeiro, & Maia, 2016; Hillis, Anda, Felitti, & Marchbanks, 2001). ACEs have also been shown to increase the risk of mental health problems including depression, anxiety, posttraumatic stress disorder, lowered perceived quality of life, insomnia, psychological distress, eating disorders, substance abuse, and conduct disorder (Basto-Pereira et al., 2016; Chapman et al., 2004). Prolonged stress caused by trauma has also been associated with negative consequences such as telomere/chromosome damage (Shalev et al., 2013) and permanent changes to the developing brain (McEwen & Seeman, 2009; Teicher et al., 2003).

Less known is whether specific ACEs are differentially predictive of certain types of internalizing (i.e., psychological) or externalizing (i.e., criminological/antisocial) outcomes. The distinction between internalizing outcomes (IOs) and externalizing outcomes (EOs) has become well-known in the field of child psychology (Achenbach & Edelbrock, 1978). These constructs refer to problematic outcomes that are either manifested in children’s outward behavior through acting out on the external environment (Eisenberg et al., 2001) or problems that affect children’s internal psychological environment (Liu, 2004). However, as research consistently suggests that some children who are abused or maltreated show higher rates of internalizing problems such as anxiety, depression, and attempted suicide than nonabused children (McLeer, Callaghan, Henry, & Wallen, 1994; McLeer et al., 1998), while other abused youth are more likely to exhibit externalizing behaviors such as aggression, violence, and delinquency (Widom, 2000), it is unclear what (if any) traumas predict one type of problematic outcome over the other. In short, it is unclear why certain abused children internalize their trauma, while others externalize.

Although IOs and EOs can be consequences of various risk factors, no single theoretical explanation for these relationships currently exists. However, as the research and theories relevant to this inquiry are highly cross-disciplinary, drawing upon a multifaceted interdisciplinary theoretical framework appears most beneficial for the purposes of this study. As Developmental and Life-Course (DLC) theories emphasize the cumulative impact of various risk and protective factors for problematic outcomes across the life-course using a host of psychological, social, parental, community, structural, and biological measures, DLC theories are well-suited to explain the associations between childhood trauma and abuse and internalizing and externalizing behaviors among justice-involved youth (Farrington, 2003).

Specifically, DLC theories predict that risk factors such as high impulsivity, antisocial peers, weak parental supervision and attachment, poverty, early criminal onset, and abuse or neglect in childhood are related to negative outcomes such as crime, delinquency, and certain externalizing or internalizing behaviors (e.g., Lahey, Moffitt, & Caspi, 2003; Piquero & Moffitt, 2005). While DLC theories have primarily been utilized to examine the impact of risk and protective factors on criminogenic outcomes, given the predictive ability and success of DLC models across disciplines, we draw upon the concepts and research stemming from DLC theories to examine the impact of childhood trauma and abuse on the risk of internalizing or externalizing behavior among at-risk youth.

This study expands upon the existing literature by examining if specific ACEs can predict IOs versus EOs among at-risk youth. Despite a broad body of literature on the factors related to IOs and EOs, there has been no known study examining the ability of specific ACEs to predict either of these outcomes, particularly using a quasi-experimental analytical strategy. Because research indicates that various risk factors and ACEs are the major predictors of IOs and EOs, it is vital to find out how ACEs affect youth after taking into account other known correlates of these disorders. This research aims to examine this specific issue, and evaluate which specific ACEs significantly predict internalizing versus externalizing behavior using a sample of otherwise matched at-risk youth. To begin, we describe the extant research on the relationships between specific ACEs, criminological risk factors, demographic characteristics, and IOs and EOs.

ACEs and IOs and EOs

The ACE checklist was first developed by the medical community as a screening tool to evaluate the long-term effects of 10 types of childhood trauma and abuse on a host of serious and life-threatening ailments in adulthood. Based on a longitudinal study, Felitti et al. (1998) found that those who experienced multiple ACEs were at much higher risk for negative health outcomes such as heart disease, liver disease, high blood pressure, diabetes, obesity, chronic lung disease, bronchitis, emphysema, skeletal fractures, cancer, and even early death (Felitti et al., 1998; Flaherty et al., 2013). ACEs have since been also linked to negative IOs and EOs (Infurna et al., 2016; Lee, 2015).

Physical Abuse

Physical abuse is defined as “any act by a caregiver that results as a nonaccidental physical injury to a child” (McCoy & Keen, 2009, p. 63). Generally, victims of physical abuse exhibit more severe adverse behavioral and emotional problems compared with children who have not been physically abused (Johnsona et al., 2002). Physically abused youth are more likely than their counterparts to experience internalizing problems (Flisher et al., 1997) such as low self-esteem (Arata, Langhinrichsen-Rohling, Bowers, & O’Farrill-Swails, 2005), depression (Brown, Cohen, Johnson, & Smailes, 1999; Jaffe, Wolfe, Wilson, & Zak, 1986), anxiety (Hjorth & Ostrov, 1982), and suicidal ideations and attempts (Deykin, Alpert, & McNamarra, 1985). Physically abused children are also more likely to exhibit externalized outcomes such as conduct disorder (Flisher et al., 1997); serious, violent, and chronic offending (B. H. Fox et al., 2015; Widom, 2000); substance abuse (Arata et al., 2005); and aggression (Prino & Peyrot, 1994) compared with those who were not physically abused. Physical abuse also contributes to the co-occurrence of internalizing and externalizing behaviors (e.g., Arata et al., 2005; Johnsona et al., 2002). Conversely, some research has found no evidence of a unique relationship between externalizing issues and physical abuse when controlling for community violence exposure (Cecil, Viding, Fearon, Glaser, & McCrory, 2017).

Sexual Abuse

Child sexual abuse (CSA) can be broadly defined as “any sexual activity with a child where consent is not or cannot be given” (Berliner & Elliott, 2002, p. 55). CSA has been linked to IOs such as anxiety, depression, and suicidal behaviors (Arata et al., 2005; Brown et al., 1999; Mullen, Martin, Anderson, Romans, & Herbison, 1993). CSA has also been related to a number of externalizing behaviors such as greater risk of substance abuse (e.g., Rotheram-Borus, Mahler, Koopman, & Langabeer, 1996), involvement in the justice system (Cavaiola & Schiff, 1988), and future sexual offending (B. Fox, 2017).

Emotional Abuse

Emotional abuse typically involves verbal or emotional aggression where the victim is verbally assaulted, humiliated, rejected, terrorized, intimidated, or psychologically or emotionally harmed. There has been far less research on the impact of childhood emotional abuse on IOs and EOs; however, existing research suggests that emotional abuse may have a greater impact on psychological functioning, such as lower self-esteem (Mullen et al., 1993). One study examining the unique effects of maltreatment found that emotional abuse was the only type of maltreatment to be uniquely predictive of IOs (Cecil et al., 2017).

Neglect

There are two types of neglect: physical and emotional. Physical neglect involves not properly feeding, clothing, or providing adequate medical care for a person, while emotional neglect is omission of interactions and/or responses to emotional needs (Barnett, Manly, & Cicchetti, 1993). Research on the impact of childhood neglect has been, ironically, neglected in the literature (Hildyard & Wolfe, 2002). Little is known about the effects of physical versus emotional neglect on future IOs and EOs. Generally, neglect has been found to relate to IOs such as low self-esteem (Loos & Alexander, 1997), depressive symptoms (Arata et al., 2005; Widom, DuMont, & Czaja, 2007), and withdrawn and submissive behaviors (Prino & Peyrot, 1994). Research has also associated childhood neglect with EOs such as aggression (Hildyard & Wolfe, 2002), violence (McGuigan, Luchette, & Atterholt, 2018), school issues (Eckenrode, Laird, & Doris, 1993), substance abuse (Campo & Rohner, 1992), and delinquency (Simons & Robertson, 1989).

Exposure to Household Violence

Research has indicated that children’s exposure to domestic violence can lead to both IOs and EOs (Sternberg, Baradaran, Abbott, Lamb, & Guterman, 2006). Childhood exposure to domestic violence leads to higher rates of cognitive, psychological, and emotional problems such as low self-esteem, depression, anxiety, aggression, and school failure (Edleson, 1999; Johnsona et al., 2002; McCloskey, Figueredo, & Koss, 1995). Exposure to household violence is believed to co-occur with other forms of abuse, and when these risk factors are combined, the likelihood of experiencing psychosocial problems increases (Moylan et al., 2010). It is currently unclear if exposure to violence alone leads to more IOs or EOs in children.

Exposure to Household Substance Abuse

Parental and caregiver substance abuse can lead to several negative consequences for children, including increased risk of internalizing and externalizing problems. For instance, children whose parents are substance abusers are more likely to also use drugs (Chassin, Rogosch, & Barrera, 1991); experience anxiety, depression (Johnson, Boney, & Brown, 1990; Tarter, Hegedus, Goldstein, Shelly, & Alterman, 1984), and aggression (Kandel, 1990); and display problematic behavior (Lieberman, 2000). Research suggests that children of substance users are more likely to express externalizing behaviors than internalizing symptoms (Chassin et al., 1991; Edwards, Leonard, & Eiden, 2001).

Parental Mental Illness

Research has consistently shown a relationship between parental psychopathology and children’s IOs and EOs (van Meurs, Reef, Verhulst, & van der Ende, 2009). Parental mental illness is associated with anxiety, depression, social withdrawal, aggression, conduct problems, and delinquency in children (Anderson & Hammen, 1993; van Meurs et al., 2009). It is unknown whether parental mental illness is more predictive of IOs or EOs.

Parental Incarceration

Research indicates that there are myriad consequences of parental incarceration on children, even when controlling for other risk factors (e.g., Huebner & Gustafson, 2007). The limited research on the direct effects of parental incarceration has shown relationships to both IOs and EOs in the forms of anxiety, depression, antisocial behavior, and delinquency (Murray & Farrington, 2005). Caregivers of children with an incarcerated parent often report that these children experience several IOs such as withdrawal (Fritsch & Burkhead, 1981), depression (Kampfner, 1995), and anxiety (Phillips, Burns, Wagner, & Barth, 2004). With regard to EOs, children with incarcerated parents were also more likely to be incarcerated (Murray & Farrington, 2005) and exhibit academic problems (Fritsch & Burkhead, 1981) and behavioral problems, especially aggression (Phillips et al., 2004; Sack, Seidler, & Thomas, 1976).

Criminological Risk Factors and IOs and EOs

There are a variety of risk factors related to IOs and EOs. Among these are DLC risk factors including neuropsychological deficits such as impulsivity, low empathy, association with antisocial peers, weak parental supervision/attachment, and structural issues such as poverty and neighborhood problems (Piquero, Jennings, & Barnes, 2012). While these issues have traditionally been related to crime, research has also found relationships between these risk factors and both IOs and EOs.

For instance, research indicates that low impulsivity (i.e., higher levels of self-control) is related to internalization issues, while externalization issues are related to high impulsivity (Vazsonyi, Mikuška, & Kelley, 2017). Similarly, research on empathy finds that it is negatively related to externalizing behaviors such as aggression (Miller & Eisenberg, 1988). Moreover, youth who associate with antisocial peers are more likely to exhibit antisocial behavior along with other externalizing behaviors (Gardner & Steinberg, 2005). As emphasized in DLC theories, parents play a significant role in the onset of offending (Baglivio, Wolff, Piquero, & Epps, 2015). Parenting style impacts externalizing behavior in children via the punitiveness, consistency, and warmth of the parent. Children who experience positive and consistent attachments to parents are generally less likely to engage in EOs than their counterparts and negative or inconsistent parenting behaviors have been associated with IOs among children (Cicchetti & Toth, 1995). Another DLC risk factor, age of onset, or age of first criminal activity, consistently predicts a host of criminal and externalizing behaviors. In addition, while poverty status has long been linked to crime and delinquency, it can also lead to internalizing and externalizing behaviors in adolescence which tend to worsen with prolonged exposure (Taylor, Dearing, & McCartney, 2004).

Demographic Characteristics and IOs and EOs

Demographic characteristics such as gender and race/ethnicity may also be related to IOs and EOs. For instance, gender has been shown to play a role in the manifestation of various psychopathologies (Nolen-Hoeksema & Girgus, 1994; Rutter, 1986), as boys tend to express themselves through externalized behaviors while girls tend to express internalized problems (e.g., Essex, Klein, Cho, & Kraemer, 2003). However, Sternberg and colleagues (2006) found no significant differences when comparing boys and girls in a meta-analysis on the relationship between violence, gender, and age. These inconsistencies for gender as a risk factor for IOs and EOs may be due to the different procedures used to determine adjustment problems based upon gender (Sternberg et al., 2006). With many mixed findings, it remains important to consider gender when addressing behavioral and psychological outcomes.

Racial and ethnic differences may also impact the expression of IOs or EOs, particularly when coupled with childhood maltreatment (Hatcher, Maschi, Morgen, & Toldson, 2009). There is some indication that certain cultures do not promote treatment of mental health issues, leading to significantly higher instances of expressing both IOs and EOs when compared with their counterparts (Hatcher et al., 2009). This disregard for treatment could potentially hinder a youth’s access to treatment that could prevent future negative outcomes.

Current Study

The current body of research shows a clear relationship exists between childhood trauma and abuse identified in the ACE assessment and increased risk of youth exhibiting IOs and EOs. However, research is mixed on whether specific ACEs increase the risk for IOs versus EOs, particularly when controlling for a host of other risk factors and types of maltreatment (see, for example, Cecil et al., 2017). Furthermore, prior research focuses primarily on just one or two ACEs as predictors of IOs and EOs, and do not take into account the variance explained by other ACEs and psychological, criminological, and demographic risk factors. Finally, as a growing body of literature demonstrates the benefits of using propensity score matching (PSM) as a quasi-experimental method to ethically examine the impact of a “treatment” (i.e., trauma/abuse) on outcomes among matched cases, this study is the first to do so to examine the effect of ACEs on IOs and EOs among at-risk youth. From this, we pose two research questions for this study:

Research Question 1: Do ACEs significantly predict whether at-risk youth will experience IOs versus EOs, after accounting for other ACEs and psychological, criminological, and demographic risk factors?

Research Question 2: Do ACEs significantly predict whether at-risk youth will experience IOs versus EOs, after being matched on other ACEs and other psychological, criminological, and demographic risk factors?

Method

Participants and Procedures

The data used in this study were acquired from the Florida Department of Juvenile Justice (FDJJ) and are comprised of high-risk youth who were referred to FDJJ for at least one misdemeanor or felony in Florida, administered the full Positive Achievement for Change Tool (PACT) risk assessment, and turned 18 (“aged out”) between January 1, 2007, and December 31, 2012. Data on the background, personality, and major risk factors were gathered using PACT, a risk assessment used to examine the risk of recidivism and customize treatment options for each case (B. H. Fox et al., 2015). Data collected from the PACT were used to identify if a juvenile experienced trauma, abuse, or adversity, and these measures were presented throughout the ACE items which allowed for two responses: (a) indication of the existence of an experience or (b) indication of the nonexistence of such an experience. While there are 10 ACEs measured in the PACT, the ACE items of interest in this study were emotional abuse, physical abuse, sexual abuse, emotional neglect, physical neglect, witnessing household violence, household substance abuse, household mental illness, and household member incarceration. One ACE, parental separation/divorce, was omitted from analyses as it is unclear from the data whether a child lives with one parent due to parental separation/divorce or parental death, incarceration, or other issues. The decision to omit this item is consistent with other studies using the PACT to evaluate ACEs (see, for example, B. Fox & DeLisi, 2018; B. H. Fox et al., 2015).

Measures

To examine IOs and EOs, multiple items measuring these constructs were incorporated from the FDJJ PACT assessment. Operationalization of IOs and EOs were based upon the Achenbach (1980) model and classifications used in prior research (e.g., Forns, Abad, & Kirchner, 2011; Frick et al., 1993), resulting in a mutually exclusive dichotomous variable where youth experiencing internalizing items were coded as “0” and those experiencing externalizing items were coded as “1.” Those categorized as internalizing only displayed internalizing behaviors and did not display the externalizing behaviors, and vice versa.

EOs consisted of six behaviors: (a) anger and irritability (Achenbach & Edelbrock, 1978), (b) violence (Achenbach & McConaughy, 1997; Vaughn, Salas-Wright, DeLisi, & Maynard, 2014), (c) runaway behavior (Frick et al., 1993), (d) alcohol and/or drug use (Vaughn et al., 2014), (e) chronic offending (defined as three or more felonies, see B. H. Fox et al., 2015), and (f) suspension/expulsion from school (Cairns & Cairns, 2000). Youth who exhibited any of the aforementioned behaviors, and did not exhibit any internalizing behaviors, were classified as “externalizers.”

Five items from the PACT were used to measure IOs including (a) documented or self-reported history of suicide attempts or ideation (Esposito & Clum, 2003), (b) depression (Achenbach & Edelbrock, 1979), (c) anxiety (Achenbach & Edelbrock, 1979), (d) somatic symptoms (Achenbach & Edelbrock, 1979; Hughes, Parkinson, & Vargo, 1989), and (e) thought disturbances (Verhulst, van der Ende, & Koot, 1997). Youth who exhibited any of these behaviors, and did not exhibit any externalizing behaviors, were classified as “internalizers.” This resulted in 1,030 youth categorized as internalizers and 29,879 youth categorized as externalizers. The 33,420 youth who exhibited both IOs and EOs were excluded from the final sample. It is important to note that while the internalizing youth may have exhibited an externalizing behavior (as they were referred to FDJJ), these youth were significantly different in composition across every single ACE and nearly all control variables (except for poverty status) from the externalizing youth. In other words, the internalizing youth are significantly different in makeup and background from the externalizing youth, despite the fact that the youth in both categories were referred to FDJJ.

As noted by Cecil and colleagues (2017), it is important to control for possible confounding effects of related factors when examining the effects of maltreatment on IOs and EOs. To guard against such confounding effects, several demographic and criminological factors were included as controls. Gender, race, and poverty are all dichotomized control variables. Gender is coded as “0” for female and “1” for male; race is coded as “0” for non-White and “1” for White; and poverty is coded as “0” for household income greater than US$15,000 and “1” for income less than US$15,000.

Youth categorized as “impulsive, often acts before thinking” or “highly impulsive, usually acts before thinking” are coded as “1” for impulsivity, while youth categorized as using self-control and thinking before acting are coded as “0.” Juveniles who displayed no empathy for their victims are coded as “1,” while those displaying empathy or some empathy are coded as “0.” Admiration and/or emulation of antisocial peers is included as a control, coded as “1” for juveniles who admire/emulate antisocial peers, while those who do not admire or emulate antisocial peers are coded as “0.” Age of onset is categorically coded as “2” for ages 12 and below, “1” for ages 13 or 14, or “0” for age 15 to 18, consistent with DLC literature (see B. H. Fox et al., 2015). The consistency of parental supervision and attachments was utilized as a control measure for the sample, coded as “1” for consistent parenting and “0” for inconsistent parenting.

Analytic Strategy

The analyses used to address this study’s research questions proceed in several stages. First, we present summary statistics and variation in the characteristics of juveniles in the FDJJ dataset designated as internalizing versus externalizing youth. Next, a series of diagnostic tests of multicollinearity are conducted, followed by a series of multivariate binary logistic regressions using the Firth method to account for the “rarity” of the internalizing (n = 1,030) youth in the sample (n = 30,909; King & Zeng, 2001). Similar to Poisson negative binomial regression models for count data, the Firth binary logistic regression model accounts for an unequally distributed (dichotomous) dependent variable through the use of penalized likelihood estimations to more accurately estimate the effects of the independent variables on the outcome (King & Zeng, 2001). Given the distribution of the IOs and EOs in the dataset, multivariate Firth binary logistic regression models are used throughout the study.

To address the first research question, an initial baseline regression model is conducted to estimate the effect of the nine ACEs items when predicting IOs versus EOs. A second model, which also includes the psychological, criminological, and demographic risk factors, evaluates the independent effects of each ACE on the risk of internalizing or externalizing while accounting for other ACEs and the control measures.

To address the second research question, we estimate the isolated independent effect of each ACE on the risk of IOs or EOs using the quasi-experimental method of PSM. This analytical technique, which has been growing in popularity in criminological research (e.g., Ridgeway, 2006), statistically removes the influence of systematic differences between cases using propensity scores. This allows for the evaluation of the isolated effect of a “treatment” on an outcome measure—in this case, internalizing versus externalizing among otherwise “matched” youth. This technique is particularly useful in a situation where it would be impossible or unethical to experimentally manipulate the treatment, such as childhood trauma and abuse (Austin, 2009; Rosenbaum, 2002; Rosenbaum & Rubin, 1983). By matching cases where all other circumstances are the same except for whether a child experienced a specific ACE, we are truly able to compare the effects of each ACE when comparing the likelihood of IOs versus EOs as a result. We therefore use PSM to remove systematic differences that may exist between the youth before assessing the independent effect of each ACE on risk of either internalizing or externalizing.

All youth in the sample were matched on nontreatment ACEs and control variables to include gender, age, age of onset, impulsivity, empathy, delinquent peer admiration, poverty, and parental supervision. We then estimate the effect of each ACE (as an experimental “treatment”) on internalizing and externalizing among statistically matched youth. PSM analyses were conducted using the PSCORE package in STATA 15.1 (Rosenbaum, 2002; Rosenbaum & Rubin, 1983). The restrictive 1:1 nearest neighbor PSM algorithm with a strict .05 caliper was used to exclude imperfect matches, while also increasing the precision in identifying ideal matches for each case (Austin, 2009). Following the PSM estimations, a series of Firth binary logistic regressions were run (King & Zeng, 2001). These models allow us to assess the effect of experiencing each ACE on risk of IOs versus EOs among the otherwise “matched” youth.

Findings

In Table 1, frequency statistics and chi-square associations are shown for the 17 measures used in the present study. Results show that there are significant variations in the prevalence of various demographic features, criminological risk factors, and ACEs for youth in the internalizing versus externalizing groups. A series of t-test analyses comparing mean differences for all ACEs, demographic features, and criminological risk factors across the internalizing versus externalizing groups were also conducted, and results indicate that significant variation exists across the internalizing and externalizing youth for all variables, except poverty status. These findings suggest that the internalizing and externalizing youth significantly differ in terms of their demographics, criminological risk factors, and the ACEs they experience.¹

Table 1:

Descriptive Statistics for Risk Factors Associated With Internalizing Versus Externalizing Behaviors Among At-Risk Youth

Risk factors, ACEs, and controls	Full sample %	Internalize %	Externalize %	χ²
Gender
Female	21.7	27.9	15.5	252.5***
Male	78.3	72.1	84.5	252.5***
Race/Ethnicity
Non-White (Black, Hispanic, Asian)	61.8	56.2	66.5	47.3***
White	38.2	43.8	33.5	47.3***
Age of onset
Age 12 and below	24.9	16.3	23.4	53.2***
Age 13-15	53.7	52.1	53.6
Age 16 and above	21.4	31.6	31.6
Antisocial peers
Does not admire/emulate antisocial peers	36.4	68.6	36.0	455.0***
Admires/emulates antisocial peers	63.6	31.4	64.0	455.0***
Impulsivity
Not impulsive, thinks before acting	54.0	81.5	57.9	228.2***
Highly impulsive, acts before thinking	46.0	18.5	42.1	228.2***
Empathy
Has empathy for victims	76.8	92.3	79.0	108.7***
Has no empathy	23.2	7.7	21.0	108.7***
Poverty status
Annual household income <US$15,000	80.5	80.8	80.9	0.0
Annual household income US$15,000+	19.5	19.2	19.1	0.0
Parent supervision
Inconsistent parental supervision	56.5	27.8	55.4	305.6***
Consistent parental supervision	43.5	72.2	44.6	305.6***
Emotional abuse
Never emotionally abused	67.5	89.0	90.3	157.8***
Emotionally abused	32.5	11.0	9.7	157.8***
Physical abuse
Never physically abused	70.5	85.0	80.0	15.5***
Physically abused	29.5	15.0	20.0	15.5***
Sexual abuse
Never sexually abused	88.1	90.3	93.9	21.5***
Sexually abused	11.9	9.7	6.1	21.5***
Emotional neglect
Never emotionally neglected	81.9	89.2	83.9	21.4***
Emotionally neglected	18.1	10.8	16.1
Physical neglect
Never physically neglected	86.7	93.4	92.1	2.2
Physically neglected	13.3	6.6	7.9
Household violence
No household violence	18.1	36.3	20.4	152.9***
Household violence	81.9	63.7	79.6	152.9***
Household substance abuse
No household substance abuse	75.0	87.4	36.3	38.2***
Household substance abuse	25.0	12.6	63.7	38.2***
Household mental illness
No household mental illness	91.2	93.5	94.7	2.9
Household mental illness	8.8	6.5	5.3	2.9
Household incarceration
No household members incarcerated	34.1	54.8	37.2	130.2***
Household members incarcerated	65.9	45.2	62.8	130.2***
% of sample	100.0	1.6	46.4
N	64,329	1,030	29,879

Note. Reference category is the last/bottom value presented for each measure. ACEs = adverse childhood experiences.

***

p < .001.

Research Question 1: Do ACEs Predict IOs Versus EOs?

To begin, diagnostic tests were run to evaluate multicollinearity among the variables. No multicollinearity was identified in the bivariate correlations or variance inflation factor (VIF) statistics. Next, a baseline logistic regression using the Firth method of penalized likelihood estimation was run. Results of Model 1 indicate that there are six ACEs that significantly predict Eos versus Ios. These significant ACEs include emotional abuse (odds ratio [OR] = 2.54, p < .001), physical abuse (OR = 1.39, p < .001), emotional neglect (OR = 1.30, p < .001), household violence (OR = 1.51, p < .001), household substance abuse (OR = 1.31, p < .001), and incarcerated household member (OR = 1.75, p < .001). Two ACEs significantly increased the risk of IOs versus EOs in the model: sexual abuse (OR = 0.44, p < .001) and household member mental illness (OR = 0.54, p < .001). Model 1 produced an area under the receiver operating characteristic curve (AUC), an indication of model accuracy, of .674. This indicates the model was better than chance at predicting EOs versus IOs.

Model 2 included the ACE risk factors and the psychological, criminological, and demographic control measures. Results (see Table 2) indicate that five ACEs significantly predicted externalization (coded as “1”) and two predicted internalization (coded as “0”), after including the control factors. This model yielded an improved AUC value, at .759, indicating a high accuracy in the model’s predictive ability. The strongest significant predictor of externalizing was emotional abuse (OR = 1.77, p < .001), as this increased the odds of EOs by 77%. Other ACEs that significantly increased the odds of EOs include physical abuse (OR = 1.32, p = .003), experiencing household violence (OR = 1.23, p = .003), household substance abuse (OR = 1.29, p = .012), and parental incarceration (OR = 1.48, p < .001). Sexual abuse was found to decrease the odds of externalizing behavior by 50% (OR = 0.50, p < .001), and household mental illness decreased the odds of externalizing behavior by 48% (OR = 0.52, p < .001). Emotional and physical neglect did not significantly distinguish between IOs and EOs in the model.

Table 2:

Multivariate Firth Binary Logistic Regression Models of ACE Predicting Internalizing Versus Externalizing Outcomes Among At-Risk Youth (n = 30,909)

Risk factors, ACEs, and controls	Model 1ACE risk factors			Model 2ACE risk factors + controls
Risk factors, ACEs, and controls	B	SE	OR	B	SE	OR
Emotional abuse	0.93	0.10	2.54***	0.57	0.11	1.77***
Physical abuse	0.33	0.11	1.39***	0.28	0.11	1.32**
Sexual abuse	−0.83	0.12	0.44***	−0.70	0.13	0.50***
Emotional neglect	0.26	0.10	1.30***	0.06	0.11	1.06
Physical neglect	−0.15	0.13	0.86	−0.18	0.14	0.83
Household violence	0.41	0.07	1.51***	0.21	0.08	1.23**
Household substance use	0.27	0.10	1.31***	0.25	0.10	1.29*
Household mental illness	−0.61	0.13	0.54***	−0.65	0.14	0.52***
Household incarceration	0.56	0.07	1.75***	0.40	0.07	1.48***
Gender				0.62	0.07	1.86***
Race/ethnicity				−0.28	0.07	0.75***
Age of onset				−0.11	0.05	0.90*
Antisocial peers				0.79	0.08	2.20***
Impulsivity				0.47	0.09	1.59***
Empathy				0.38	0.12	1.47**
Parental supervision				−0.48	0.08	0.62***
Poverty				−0.11	0.08	0.90

Note. Model 1: pseudo R² = .044, area under ROC curve = 0.674, p < .001. Model 2: pseudo R² = .099, area under ROC curve = .759, p < .001. ACE = adverse childhood experience; OR = odds ratio; ROC = receiver operating characteristic.

p < .05. **p < .01. ***p < .001.

Research Question 2: Do ACEs Predict IOs Versus EOs among Statistically Matched Cases?

To address the final research question, a series of analyses were conducted to determine the impact of each individual ACE on the likelihood of experiencing IOs versus EOs, among youth who are matched across all other ACEs and psychological, criminological, and demographic risk factors. To do this, PSM analyses were utilized to statistically match cases based upon all relevant control measures and ACEs, while each ACE of interest was allowed to vary. Firth binary logistic regressions with penalized likelihood estimations (to account for the relative rarity of internalizers vs. externalizers) were then conducted to assess the likelihood that each ACE increases the risk of exhibiting IOs or EOs among the otherwise “identical” youth.

To evaluate the effectiveness and impact of the PSM technique, statistical analyses of the reduction in bias were conducted. Results show that after the application of PSM, considerable bias reductions (up to 1,252%) were seen across every covariate (see the appendix). This suggests that the PSM was effective and beneficial in generating a matched sample of youth in the dataset.

To illustrate the impact of utilizing the PSM technique when estimating the effects of each ACE on risk of IO versus EO, a series of Firth binary logistic regressions after matching youth across all ACEs and control variables was conducted (see Table 3). All models produced AUC values greater than .500, indicating the models are better than chance at predicting EOs versus IOs using the ACE items after matching cases across all remaining covariates. Sensitivity analyses using the Mantel-Haenszel (1959) statistic where gamma (γ) = 1 show that each model had a “true” treatment effect, and there was no significant overestimation of treatment effects due to hidden bias in the models. In each case, the models were insensitive to bias as γ increased to 2 (and in some cases, 3), suggesting that unobserved variables could double (or triple) in effect size and not reduce the impact of each ACE on the outcome. In other words, the odds of the current models overestimating the treatment effects due to omitted variables is very low.

Table 3:

PSM Multivariate Firth Binary Logistic Regression Models Predicting Internalizing Versus Externalizing Outcomes Among At-Risk Youth

Adverse childhood experiences	Results with PSM			Change in effect post-PSM
Adverse childhood experiences	B	SE	OR	ΔOR
Emotional abuse	1.21	0.11	3.37***	+1.60
Physical abuse	0.59	0.10	1.80***	+0.48
Sexual abuse	−0.45	0.15	0.63**	+0.13
Emotional neglect	0.43	0.13	1.53**	—
Physical neglect	0.15	0.16	1.16	—
Household violence	0.93	0.05	2.52***	+1.29
Household substance abuse	0.70	0.11	2.02***	+0.73
Household mental illness	−0.14	0.17	0.87	—
Household incarceration	0.31	0.06	1.36***	−0.12

Note: Before matching: n = 30,909, after matching: Model 1: n = 8,759, AUC = 0.634; Model 2: n = 22,734, AUC = 0.569; Model 3: n = 10,829, AUC = 0.556; Model 4: n = 15,956, AUC = 0.553; Model 5: n = 12,212, AUC = 0.519; Model 6: n = 24,249, AUC = 0.611; Model 7: n = 20,393, AUC = 0.584; Model 8: n = 8,759, AUC = 0.513; Model 9: n = 20,270, AUC = 0.538. PSM = propensity score matching; OR = odds ratio; ROC = receiver operating characteristic; ACE = adverse childhood experience; AUC = area under the ROC curve. PSM models conducted after matching cases on all demographic and criminological risk factors and nontreated ACEs, with each specified ACE used as the statistical treatment. Change in effect post-PSM calculated using the difference in statistically significant odds ratios for each ACE in the models before matching and after matching.

p < .01. ***p < .001.

Results of these analyses indicate that seven ACEs remain significant predictors of IOs or EOs after the youth are statistically matched across all other ACEs and the psychological, criminological, and demographic controls. When compared with the results of the models before matching, effect sizes in the post-PSM model appear to substantially increase. For instance, after matching, results showed that youth who experienced emotional abuse in childhood were 237% more likely to show negative externalizing versus internalizing behaviors in adolescence (OR = 3.37, p < .001) compared with the youth who did not experience emotional abuse. This represents a change in effect size of 1.60, compared with the standard regression technique before matching.

Physical abuse was also significantly more likely to predict externalizing behaviors in the post-PSM models, with an increased odds of 80% (OR = 1.80, p < .001). Again, this effect is substantially larger than the effect size seen through standard regression techniques utilizing only statistical controls versus statistically matching based upon controls. Youth who experienced emotional neglect in childhood showed a 53% higher risk of externalizing behavior (OR = 1.53, p = .001), while witnessing household violence (OR = 2.52, p < .001) and household substance abuse (OR = 2.02, p < .001) raised the risk of externalizing behaviors by 152% and 102%, respectively. The change in effect size from before to after matching was OR = 1.29 for household violence and 0.73 for household substance abuse. Finally, incarceration of a household member raised the odds of externalizing by 36% after matching (OR = 1.36, p < .001), which is a 12% decrease in risk estimated through the prematching logistic regression.

Discussion

This study examined whether specific ACEs are more predictive of IOs or EOs among a sample of delinquent youth. To do this, we utilized a series of logistic regressions and PSM analyses on a sample of 30,909 youth referred to the FDJJ. To address our first research question, whether ACEs were able to significantly predict either IOs or EOs, two separate analyses were conducted. The first baseline model (with no control measures) indicated that all ACEs, except for physical neglect, significantly predicted either IOs or EOs.

Interestingly, after adding the control measures into the regression model in the second analysis, significant effects still held for all ACEs except emotional neglect. In this model, neither form of neglect significantly predicted internalizing or externalizing, while sexual abuse and household mental illness both significantly increased the risk of internalizing problems. The remaining ACEs (emotional abuse, physical abuse, household violence, household substance use, and household member incarceration) all significantly increased the odds of externalizing behaviors in the full model.

To answer our second research question, we evaluated whether ACEs would predict IOs or EOs after statistically matching youth across all relevant psychological, criminological, and demographic factors using the PSM technique. Results indicate that emotional abuse, physical abuse, household violence, household substance abuse, emotional neglect, and household member incarceration all significantly increased the risk of EOs, when these items were the only thing to differ among otherwise “matched” youth. Sexual abuse was found to significantly increase the risk of IOs, while physical neglect and household incarceration were not significant predictors in the models. These results largely support previous DLC research, while also expanding the literature on the outcomes associated with ACEs in several ways.

First, it was notable to find that out of all the ACEs examined, the strongest predictor of externalizing behavior was consistently emotional abuse. This ACE was also the most prevalent form of abuse (nearly a third of the sample were emotionally abused in childhood), and consistently yielded large effects. Prior research has shown several negative outcomes of emotional abuse, including externalizing behavior (e.g., Mullen et al., 1993; Ney, Fung, & Wickett, 1994). Other studies have found that emotional abuse is a stronger predictor than physical abuse for IOs and EOs (McGee, Wolfe, & Wilson, 1997; Mullen et al., 1993). However, no known research has examined whether emotional abuse is more predictive of EOs versus IOs. In light of the present findings, we surmise that emotional abuse may lead to a lack of positive coping strategies and difficulty handling stressful situations, resulting in a higher risk of externalizing behaviors. Given the unfortunate prevalence of this form of abuse, researchers should continue to examine the impact of emotional abuse with an eye toward intervention and prevention.

Other salient predictors of EOs include physical abuse, exposure to household violence, and household substance use. These findings are not surprising, given the research establishing that these ACEs have been shown to relate to delinquency (Chassin et al., 1991; Widom, 2000), aggression (Kandel, 1990; Prino & Peyrot, 1994), and other problematic externalizing behaviors (Edwards et al., 2001; Lieberman, 2000). DLC theories would suggest that physical abuse and witnessing violence could increase externalizing behavior as the child learns from his or her abusers to deal with problems using violence (e.g., Felson & Lane, 2009).

Emotional neglect was also a significant predictor of externalizing behavior in this study. Prior research has shown a relationship between emotional neglect and certain externalizing factors such as poor academic achievement (Basto-Pereira et al., 2016), running away (Jeanis, Fox, & Muniz, 2018), and criminal involvement (Assink et al., 2015). Future research should examine if emotionally neglected children have trouble responding to others’ emotion (Pollak, 2008), have poor emotional regulation skills (Shipman, Edwards, Brown, Swisher, & Jennings, 2005), or experience peer rejection (Kim & Cicchetti, 2010) which may explain the increased risk of externalizing behaviors.

Household member incarceration is an established correlate of delinquency in the DLC literature (Murray & Farrington, 2005), yet the effects on other aspects of externalizing behavior have not been well examined. The limited research on parental incarceration and externalizing behaviors suggests the relationship may be mediated by factors such as poverty and parenting style (Kjellstrand & Eddy, 2011). Given that both poverty and consistency of parental supervision are controlled for in this study, and youth were matched on both factors, the results show that parental incarceration has deleterious effects on youth regardless of these factors.

Sexual abuse was the only ACE predictive of IOs over EOs. Prior research has established a link between CSA and IOs such as anxiety, depression, and suicidal behaviors (Arata et al., 2005; Brown et al., 1999; Mullen et al., 1993). It is interesting to find that sexual abuse significantly increased the risk of internalizing over externalizing among the at-risk youth. Perhaps the psychological trauma and stigmatization of CSA lead to clinically significant feelings of powerlessness, guilt, and shame, which affect these victims differently than youth who experience other ACEs (Finkelhor & Browne, 1985) Furthermore, CSA victims may be more likely to internalize due to coercion to keep the victimization secret. Not disclosing may inhibit access to counseling services allowing the manifestation of this trauma as internalizing behaviors (Lewis, McElroy, Harlaar, & Runyan, 2016). Although based on prior research, these explanations are speculative as the present study did not test these potential mechanisms. Future research should specifically test for such pathways among at-risk youth.

It is worth noting that physical neglect was not predictive of IOs or EOs in any model. While it is unclear why physical neglect is not predictive of either outcome, preliminary descriptive statistics indicated that levels of physical neglect did not significantly differ among the internalizing and externalizing youth. This finding may indicate that those who are physically neglected are more likely to exhibit IOs and EOs at similar rates, and therefore physical neglect is unable to significantly predict either outcome. Future research should examine the effects of physical neglect on the co-occurrence of IOs and EOs. Similarly, the only significant predictor of internalizing problems was sexual abuse; however, this does not mean that internalizing issues are not problematic. Externalizing behaviors are more likely to be noticed and responded to, whereas internalizing behaviors may go undiagnosed and untreated.

An alarming finding is how underestimated the effects of specific ACEs were before matching. Standard regression techniques largely underestimated the effects of these ACEs on future behavior, which underscores the fact that several forms of abuse and trauma in childhood are even more impactful than we originally surmised. To that end, it is critical to recognize that results of this study consistently indicate many ACEs significantly increased the risk of negative outcomes, even when considering the effects of other impactful psychological, criminological, and demographic factors. This indicates that ACEs significantly impact children from a variety of backgrounds and situations, and the long-term downstream impact is manifested in psychological or antisocial problems later in life.

These findings have important practical implications. Not only do IOs and EOs affect youth during adolescence, but these behaviors typically carry on to adulthood (Hofstra, van der Ende, & Verhulst, 2002) often predating both substance use and other mental disorders (Kim-Cohen et al., 2003). Moreover, internalizing problems during adolescence are associated with IOs in adulthood, such as major depressive disorders, suicidal ideations/suicide, psychiatric hospitalization (Pine, Cohen, Gurley, Brook, & Ma, 1998), and increased physical health issues (Schraedley, Gotlib, & Hayward, 1999). Left untreated, both IOs and EOs have severe ramifications for both the individual and society. Specifically, the detrimental effects of emotional maltreatment should be alarming to practitioners. While the consequences of physical abuse have rightfully garnered the attention of scholars and practitioners, paying attention to sometimes less obvious signs and symptoms of emotional maltreatment could help youth get the necessary treatment to alleviate effects of such abuse. In sum, using ACEs to screen children can allow for early and specially tailored interventions and policies to reduce problematic behaviors. Being proactive in treating children before their trauma can manifest psychologically or behaviorally should be of utmost importance for criminologists and psychologists aiming to decrease the negative impacts of crime and psychological disorders.

There are also limitations to consider. First, the data were collected on a sample of high-risk delinquent youth who are unique in many ways from the average adolescent population. Future research should replicate the current study with a community sample, which may indicate higher rates of IOs and lower rates of abuse and childhood trauma in general. Second, the measures used in this study were obtained through the PACT risk assessment, which was not intended for the purpose of the present research. Certain measures in the PACT, such as the parental separation/divorce measure, were not suitable for inclusion and such exclusion may affect the results of this study. Ideally, future studies will utilize measures specifically created and designed for the purpose of evaluating the impact of ACEs on various behaviors and outcomes. For instance, personality data may be collected, as this may help partially explain why youth who experience a certain ACE tend to externalize their trauma, while others are more likely to internalize. In this vein, while these findings were generally supportive of the tenets of DLC theories, future studies should explore the proposed risk factors of DLC in regard to IOs specifically. Furthermore, future research should expand the scope of this study by investigating variations in the co-occurrence of IOs and EOs. In other words, future research should explore the relationships between ACEs and predominantly externalizers who also internalize and vice versa. Similarly, this study examined ACEs independently. More research is needed to explore the effects of combinations of ACEs as there may be meaningful differences in variations of co-occurrence. Second, gender differences in externalizing/internalizing symptomology may be explained by ACEs differentially affecting boys and girls. This potential relationship should be explored in future research. Finally, a prospective longitudinal design could be used to more clearly understand the development and relationship between ACEs and future IOs and EOs.

In conclusion, it is axiomatic that ACEs cast a long shadow on the lives of children that incur them, and the negative consequences manifest in multitudinous ways. Using a large sample of juvenile offenders, some of whom engaged exclusively in externalizing ways and still others whom engaged exclusively in internalizing ways, the current study shed new light on ACEs with propensity score methods. This method empirically demonstrated that the assorted effects of abuse and neglect are even more pernicious than prior research has found, but fortunately identified specific ACE-internalizing/externalizing relationships to inform treatment and correctional interventions for delinquent youth.

Footnotes

Appendix

Reduction of Bias in Risk Factors After Propensity Score Matching

	Emotional abuse	Physical abuse	Sexual abuse	Emotional neglect	Physical neglect	Household violence	Household substance abuse	Household mental illness	Household incarceration
Emotional abuse	—	−60.8%	−84.6%	−96.9%	−98.1%	−60.6%	−98.3%	−97.3%	−97.7%
Physical abuse	−87.4%	—	−98.4%	−99.3%	−97.8%	−96.7%	−97.0%	−99.4%	−99.7%
Sexual abuse	−64.5%	−99.5%	—	−95.5%	−95.8%	−92.3%	−98.1%	−92.5%	−77.9%
Emotional neglect	−94.1%	−84.4%	−89.1%	—	−94.2%	−97.0%	−88.1%	−92.3%	−82.4%
Physical neglect	−90.1%	−95.6%	−95.8%	−99.1%	—	−38.1%	−99.4%	−94.6%	−89.6%
Household violence	−100.0%	−100.0%	−97.3%	−95.5%	−98.6%	—	−98.5%	−98.5%	−98.1%
Household substance abuse	−98.9%	−89.9%	−92.6%	−99.5%	−98.8%	−21.6%	—	−99.5%	−95.2%
Household mental illness	−85.4%	−98.7%	−90.7%	−86.7%	−96.8%	−30.6%	−89.8%	—	−83.6%
Household incarceration	−92.0%	−90.0%	−92.5%	−81.0%	−96.3%	−59.3%	−99.1%	−94.8%	—
Gender	−84.8%	−95.9%	−97.1%	−94.6%	−89.5%	−61.7%	−84.7%	−94.0%	−83.6%
Race/ethnicity	−99.8%	−94.0%	−98.3%	−47.0%	−95.5%	−331.2%	−97.1%	−97.6%	−44.4%
Age of onset	−97.8%	−99.6%	−95.8%	−91.1%	−98.6%	−15.2%	−97.1%	−99.8%	−97.9%
Antisocial peers	−97.0%	−99.4%	−77.7%	−84.7%	−86.9%	−85.2%	−93.8%	−84.7%	−94.4%
Impulsivity	−97.8%	−95.3%	−94.2%	−99.1%	−93.3%	−79.0%	−93.8%	−95.7%	−96.9%
Empathy	−99.9%	−91.3%	−99.4%	−99.6%	−98.4%	−61.0%	−98.6%	−98.2%	−97.4%
Parental supervision	−98.4%	−98.6%	−85.5%	−96.5%	−95.8%	−54.8%	−98.1%	−94.4%	−96.1%
Poverty	−1,252.4%	−44.9%	−67.2%	−91.8%	−81.7%	−346.5%	−67.2%	−85.3%	−84.6%
n	25,248	22,734	10,829	15,956	12,212	24,249	20,393	8,759	20,270

Note. Valid percentages within each group are shown. Total sample N = 30,909.

Notes

Authors’ Note:

The authors thank Mark Greenwald, Sherry Jackson, and the Florida Department of Juvenile Justice (FDJJ) for the continued support and opportunity to work together on research aimed at helping young people achieve a safe, healthy, and prosperous future. They also thank the anonymous reviewers for their extremely thoughtful and beneficial suggestions on a prior draft of this article. The datasets analyzed during the current study are not publicly available due FDJJ standards of use and guidelines, but may be available from the authors on reasonable request.

ORCID iDs

Caitlyn N. Muniz

Bryanna Fox

Matt DeLisi

Caitlyn N. Muniz is an assistant professor of criminal justice at the University of Texas at El Paso. She received her PhD in criminology from the University of South Florida (USF) where she was the lead research supervisor of SPRUCE Lab. Her research interests focus on various aspects of abuse and victimization.

Bryanna Fox is an assistant professor in the Department of Criminology, faculty affiliate in the Florida Mental Health Institute, and director of the SPRUCE research lab at the USF. She earned her PhD from the University of Cambridge and is a former FBI Special Agent. Her research interests relate to the identification of risk factors for criminal behavior, developing evidence-based tools for law enforcement, and conducting experimental research.

Lauren N. Miley is a doctoral student in the Criminology Department at the USF. She received her MA in criminology from USF. She is a research supervisor of SPRUCE lab. Her research interests include mental health in the criminal justice system, biosocial criminology, and developmental and life-course criminology.

Matt DeLisi is College of Liberal Arts and Sciences Dean’s professor, coordinator of Criminal Justice Studies, and faculty affiliate of the Center for the Study of Violence at Iowa State University. A fellow of the Academy of Criminal Justice Sciences, he is the author of nearly 400 scholarly publications.

Gerald P. Cigarran II is a master’s student in criminology at the University of Tampa. He received bachelor’s degrees in criminology and philosophy from the USF. He is a former research assistant in SPRUCE Lab at the USF Department of Criminology.

Aliya Birnbaum is a master’s student in forensic psychology at John Jay College of Criminal Justice CUNY where she studies juveniles’ intersection with the law. She received bachelor’s degrees in psychology and criminology from the USF. She is a former research assistant in SPRUCE Lab at the USF Department of Criminology.

References

Achenbach

T. M.

(1980). DSM-III in light of empirical research on the classification of child psychopathology. Journal of the American Academy of Child Psychiatry, 19, 395-412.

Achenbach

T. M.

Edelbrock

C. S.

(1978). The classification of child psychopathology: A review and analysis of empirical efforts. Psychological Bulletin, 85, 1275-1301.

Achenbach

T. M.

Edelbrock

C. S.

(1979). The Child Behavior Profile: II. Boys aged 12–16 and girls aged 6–11 and 12–16. Journal of Consulting and Clinical Psychology, 47(2), 223-233.

Achenbach

T. M.

McConaughy

S. H.

(1997). Empirically based assessment of child and adolescent psychopathology: Practical applications. Newbury Park, CA: SAGE.

Anderson

C. A.

Hammen

C. L.

(1993). Psychosocial outcomes of children of unipolar depressed, bipolar, medically ill, and normal women: A longitudinal study. Journal of Consulting and Clinical Psychology, 61, 448-454.

Arata

C. M.

Langhinrichsen-Rohling

Bowers

O’Farrill-Swails

(2005). Single versus multi-type maltreatment: An examination of the long-term effects of child abuse. Journal of Aggression, Maltreatment & Trauma, 11(4), 29-52.

Assink

van der Put

C. E.

Hoeve

de Vries

S. L. A.

Stams

G. J. J. M.

Oort

F. J.

(2015). Risk factors for persistent delinquent behavior among juveniles: A meta-analytic review. Clinical Psychology Review, 42, 47-61.

Austin

P. C.

(2009). Balance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity-score matched samples. Statistics in Medicine, 28, 3083-3107.

Baglivio

M. T.

Epps

(2016). The interrelatedness of adverse childhood experiences among high-risk juvenile offenders. Youth Violence and Juvenile Justice, 14, 179-198.

10.

Baglivio

M. T.

Wolff

K. T.

Piquero

A. R.

Epps

(2015). The relationship between adverse childhood experiences (ACE) and juvenile offending trajectories in a juvenile offender sample. Journal of Criminal Justice, 43(3), 229-241.

11.

Barnett

Manly

J. T.

Cicchetti

(1993). Defining child maltreatment: The interface between policy and research. In Cicchetti

Toth

S. L.

(Eds.), Child abuse, child development, and social policy: Advances in applied developmental psychology (Vol. 8, pp. 7-73). Norwood, NJ: Ablex.

12.

Basto-Pereira

Miranda

Ribeiro

Maia

(2016). Growing up with adversity: From juvenile justice involvement to criminal persistence and psychosocial problems in young adulthood. Child Abuse & Neglect, 62, 63-75.

13.

Berliner

Elliott

D. M.

(2002). Sexual abuse of children. In Myers

J. E. B.

Berliner

Briere

Hendrix

C. T.

Jenny

Reid

T. A.

(Eds.), The APSAC handbook on child maltreatment (2nd ed., pp. 55-78). Thousand Oaks, CA: SAGE.

14.

Brown

Cohen

Johnson

J. G.

Smailes

E. M.

(1999). Childhood abuse and neglect: Specificity of effects on adolescent and young adult depression and suicidality. Journal of the American Academy of Child & Adolescent Psychiatry, 38, 1490-1496.

15.

Cairns

R. B.

Cairns

B. D.

(2000). The natural history and developmental functions of aggression. In Sameroff

A. J.

Lewis

Miller

S. M.

(Eds.), Handbook of developmental psychopathology (pp. 403-429). Boston, MA: Springer.

16.

Campo

A. T.

Rohner

R. P.

(1992). Relationships between perceived parental acceptance-rejection, psychological adjustment, and substance abuse among young adults. Child Abuse & Neglect, 16, 429-440.

17.

Cavaiola

A. A.

Schiff

(1988). Behavioral sequelae of physical and/or sexual abuse in adolescents. Child Abuse & Neglect, 12(2), 181-188.

18.

Cecil

C. A.

Viding

Fearon

Glaser

McCrory

E. J.

(2017). Disentangling the mental health impact of childhood abuse and neglect. Child Abuse & Neglect, 63, 106-119.

19.

Chapman

D. P.

Whitfield

C. L.

Felitti

V. J.

Dube

S. R.

Edwards

V. J.

Anda

R. F.

(2004). Adverse childhood experiences and the risk of depressive disorders in adulthood. Journal of Affective Disorders, 82, 217-225.

20.

Chassin

Rogosch

Barrera

(1991). Substance use and symptomatology among adolescent children of alcoholics. Journal of Abnormal Psychology, 100, 449-463.

21.

Cicchetti

Toth

S. L.

(1995). A developmental psychopathology perspective on child abuse and neglect. Journal of the American Academy of Child & Adolescent Psychiatry, 34, 541-565.

22.

Craig

J. M.

Piquero

A. R.

Farrington

D. P.

Ttofi

M. M.

(2017). A little early risk goes a long bad way: Adverse childhood experiences and life-course offending in the Cambridge study. Journal of Criminal Justice, 53, 34-45.

23.

Deykin

E. Y.

Alpert

McNamarra

J. J.

(1985). A pilot study of the effect of exposure to child abuse. American Journal of Psychiatry, 142, 1299-1303.

24.

Dierkhising

C. B.

S. J.

Woods-Jaeger

Briggs

E. C.

Lee

Pynoos

R. S.

(2013). Trauma histories among justice-involved youth: Findings from the National Child Traumatic Stress Network. European Journal of Psychotraumatology, 4(1), Article 20274.

25.

Dube

S. R.

Felitti

V. J.

Dong

Chapman

D. P.

Giles

W. H.

Anda

R. F.

(2003). Childhood abuse, neglect, and household dysfunction and the risk of illicit drug use: The adverse childhood experiences study. Pediatrics, 111, 564-572.

26.

Eckenrode

Laird

Doris

(1993). School performance and disciplinary problems among abused and neglected children. Developmental Psychology, 29, 53-62.

27.

Edleson

J. L.

(1999). Children’s witnessing of adult domestic violence. Journal of Interpersonal Violence, 14, 839-870.

28.

Edwards

Leonard

K. E.

Eiden

R. D.

(2001). Temperament and behavioral problems among infants in alcoholic families. Infant Mental Health Journal: Official Publication of the World Association for Infant Mental Health, 22, 374-392.

29.

Eisenberg

Cumberland

Spinrad

T. L.

Fabes

R. A.

Shepard

S. A.

Reiser

. . . Guthrie

I. K.

(2001). The relations of regulation and emotionality to children’s externalizing and internalizing problem behavior. Child Development, 72, 1112-1134.

30.

Esposito

C. L.

Clum

G. A.

(2003). The relative contribution of diagnostic and psychosocial factors in the prediction of adolescent suicidal ideation. Journal of Clinical Child and Adolescent Psychology, 32, 386-395.

31.

Essex

M. J.

Klein

M. H.

Cho

Kraemer

H. C.

(2003). Exposure to maternal depression and marital conflict: Gender differences in children’s later mental health symptoms. Journal of the American Academy of Child & Adolescent Psychiatry, 42, 728-737.

32.

Farrington

D. P.

(2003). Developmental and life-course criminology: Key theoretical and empirical issues. Criminology, 41, 221-225.

33.

Felitti

V. J.

Anda

R. F.

Nordenberg

Williamson

D. F.

Spitz

A. M.

Edwards

. . . 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.

34.

Felson

R. B.

Lane

K. J.

(2009). Social learning, sexual and physical abuse, and adult crime. Aggressive Behavior: Official Journal of the International Society for Research on Aggression, 35, 489-501.

35.

Finkelhor

Browne

(1985). The traumatic impact of child sexual abuse: A conceptualization. American Journal of Orthopsychiatry, 55(4), 530-541.

36.

Flaherty

E. G.

Thompson

Dubowitz

Harvey

E. M.

English

D. J.

Proctor

L. J.

Runyan

D. K.

(2013). Adverse childhood experiences and child health in early adolescence. JAMA Pediatrics, 167, 622-629.

37.

Flisher

A. J.

Kramer

R. A.

Hoven

C. W.

Greenwald

Alegria

Bird

H. R.

. . . Moore

R. E.

(1997). Psychosocial characteristics of physically abused children and adolescents. Journal of the American Academy of Child & Adolescent Psychiatry, 36, 123-131.

38.

Forns

Abad

Kirchner

(2011). Internalizing and externalizing problems. In Encyclopedia of Adolescence (pp. 1464-1469). New York, NY: Springer.

39.

Fox

(2017). What makes a difference? Evaluating the key distinctions and predictors of sexual and non-sexual offending among male and female juvenile offenders. Journal of Criminal Psychology, 7, 134-150.

40.

Fox

DeLisi

(2018). From criminological heterogeneity to coherent classes: Developing a typology of juvenile sex offenders. Youth Violence and Juvenile Justice, 16, 299-318.

41.

Fox

B. H.

Perez

Cass

Baglivio

M. T.

Epps

(2015). Trauma changes everything: Examining the relationship between adverse childhood experiences and serious, violent and chronic juvenile offenders. Child Abuse & Neglect, 46, 163-173.

42.

Frick

P. J.

Lahey

B. B.

Loeber

Tannenbaum

Van Horn

Christ

M. A. G.

. . . Hanson

(1993). Oppositional defiant disorder and conduct disorder: A meta-analytic review of factor analyses and cross-validation in a clinic sample. Clinical Psychology Review, 13, 319-340.

43.

Fritsch

T. A.

Burkhead

J. D.

(1981). Behavioral reactions of children to parental absence due to imprisonment. Family Relations, 30, 83-88.

44.

Gardner

Steinberg

(2005). Peer influence on risk taking, risk preference, and risky decision making in adolescence and adulthood: An experimental study. Developmental Psychology, 41, 625-635.

45.

Hatcher

S. S.

Maschi

Morgen

Toldson

I. A.

(2009). Exploring the impact of racial and ethnic differences in the emotional and behavioral responses of maltreated youth: Implications for culturally competent services. Children and Youth Services Review, 31, 1042-1048.

46.

Hildyard

K. L.

Wolfe

D. A.

(2002). Child neglect: Developmental issues and outcomes. Child Abuse & Neglect, 26, 679-695.

47.

Hillis

S. D.

Anda

R. F.

Felitti

V. J.

Marchbanks

P. A.

(2001). Adverse childhood experiences and sexual risk behaviors in women: A retrospective cohort study. Family Planning Perspectives, 33, 206-211.

48.

Hjorth

C. W.

Ostrov

(1982). The self-image of physically abused adolescents. Journal of Youth and Adolescence, 11, 71-76.

49.

Hofstra

M. B.

van der Ende

Verhulst

F. C.

(2002). Child and adolescent problems predict DSM-IV disorders in adulthood: A 14-year follow-up of a Dutch epidemiological sample. Journal of the American Academy of Child & Adolescent Psychiatry, 41(2), 182-189.

50.

Huebner

B. M.

Gustafson

(2007). The effect of maternal incarceration on adult offspring involvement in the criminal justice system. Journal of Criminal Justice, 35, 283-296.

51.

Hughes

H. M.

Parkinson

Vargo

(1989). Witnessing spouse abuse and experiencing physical abuse: A “double whammy”? Journal of Family Violence, 4, 197-209.

52.

Infurna

M. R.

Reichl

Parzer

Schimmenti

Bifulco

Kaess

(2016). Associations between depression and specific childhood experiences of abuse and neglect: A meta-analysis. Journal of Affective Disorders, 190, 47-55.

53.

Jaffe

Wolfe

Wilson

Zak

(1986). Similarities in behavioral and social maladjustment among child victims and witnesses to family violence. American Journal of Orthopsychiatry, 56, 142-146.

54.

Jeanis

M. N.

Fox

B. H.

Muniz

C. N.

(2018). Revitalizing profiles of runaways: A latent class analysis of delinquent runaway youth. Child and Adolescent Social Work Journal. Advance online publication. doi:10.1007/s10560-018-0561-5

55.

Johnson

J. L.

Boney

T. Y.

Brown

B. S.

(1990). Evidence of depressive symptoms in children of substance abusers. International Journal of the Addictions, 25, 465-479.

56.

Johnsona

R. M.

Kotch

J. B.

Catellier

D. J.

Winsor

J. R.

Dufort

Hunter

Amaya-Jackson

(2002). Adverse behavioral and emotional outcomes from child abuse and witnessed violence. Child Maltreatment, 7, 179-186.

57.

Kampfner

C. J.

(1995). Post-traumatic stress reactions in children of imprisoned mothers. In Katherine

Denise

(Ed.), Children of incarcerated parents (pp. 89-100). New York, NY: Lexington Books.

58.

Kandel

D. B.

(1990). Parenting styles, drug use, and children’s adjustment in families of young adults. Journal of Marriage and the Family, 2, 183-196.

59.

Kim

Cicchetti

(2010). Longitudinal pathways linking child maltreatment, emotion regulation, peer relations, and psychopathology. Journal of Child Psychology and Psychiatry, 51(6), 706-716.

60.

Kim-Cohen

Caspi

Moffitt

T. E.

Harrington

Milne

B. J.

Poulton

(2003). Prior juvenile diagnoses in adults with mental disorder: Developmental follow-back of a prospective-longitudinal cohort. Archives of General Psychiatry, 60(7), 709-717.

61.

King

Zeng

(2001). Logistic regression in rare events data. Political Analysis, 9(2), 137-163.

62.

Kjellstrand

J. M.

Eddy

J. M.

(2011). Mediators of the effect of parental incarceration on adolescent externalizing behaviors. Journal of Community Psychology, 39, 551-565.

63.

Lahey

B. B.

Moffitt

T. E.

Caspi

(2003). Causes of conduct disorder and juvenile delinquency. New York, NY: Guilford Press.

64.

Lee

M. A.

(2015). Emotional abuse in childhood and suicidality: The mediating roles of re-victimization and depressive symptoms in adulthood. Child Abuse & Neglect, 44, 130-139.

65.

Lewis

McElroy

Harlaar

Runyan

(2016). Does the impact of child sexual abuse differ from maltreated but non-sexually abused children? A prospective examination of the impact of child sexual abuse on internalizing and externalizing behavior problems. Child Abuse & Neglect, 51, 31-40.

66.

Lieberman

D. Z.

(2000). Children of alcoholics: An update. Current Opinion in Pediatrics, 12(4), 336-340.

67.

Liu

(2004). Childhood externalizing behavior: Theory and implications. Journal of Child and Adolescent Psychiatric Nursing, 17, 93-103.

68.

Loos

M. E.

Alexander

P. C.

(1997). Differential effects associated with self-reported histories of abuse and neglect in a college sample. Journal of Interpersonal Violence, 12, 340-360.

69.

Mantel

Haenszel

(1959). Statistical aspects of the analysis of data from retrospective studies of disease. Journal of the National Cancer Institute, 22, 719-748.

70.

Maxfield

M. G.

Widom

C. S.

(1996). The cycle of violence: Revisited 6 years later. Archives of Pediatrics & Adolescent Medicine, 150, 390-395.

71.

McCloskey

L. A.

Figueredo

A. J.

Koss

M. P.

(1995). The effects of systemic family violence on children’s mental health. Child Development, 66, 1239-1261.

72.

McCoy

M. L.

Keen

S. M.

(2009). Child abuse and neglect. New York, NY: Psychology Press.

73.

McEwen

Seeman

(2009). Research: Allostatic load notebook. MacArthur Research Network on SES and Health. Retrieved from https://macses.ucsf.edu/research/allostatic/

74.

McGee

R. A.

Wolfe

D. A.

Wilson

S. K.

(1997). Multiple maltreatment experiences and adolescent behavior problems: Adolescents’ perspectives. Development and Psychopathology, 9, 131-149.

75.

McGuigan

W. M.

Luchette

J. A.

Atterholt

(2018). Physical neglect in childhood as a predictor of violent behavior in adolescent males. Child Abuse & Neglect, 79, 395-400.

76.

McLeer

S. V.

Callaghan

Henry

Wallen

(1994). Psychiatric disorders in sexually abused children. Journal of the American Academy of Child & Adolescent Psychiatry, 33, 313-319.

77.

McLeer

S. V.

Dixon

J. F.

Henry

Ruggiero

Escovitz

Niedda

Scholle

(1998). Psychopathology in nonclinically referred sexually abused children. Journal of the American Academy of Child & Adolescent Psychiatry, 37, 1326-1333.

78.

Miller

P. A.

Eisenberg

(1988). The relation of empathy to aggressive and externalizing/antisocial behavior. Psychological Bulletin, 103, 324-344.

79.

Moylan

C. A.

Herrenkohl

T. I.

Sousa

Tajima

E. A.

Herrenkohl

R. C.

Russo

M. J.

(2010). The effects of child abuse and exposure to domestic violence on adolescent internalizing and externalizing behavior problems. Journal of Family Violence, 25(1), 53-63.

80.

Mullen

P. E.

Martin

J. L.

Anderson

J. C.

Romans

S. E.

Herbison

G. P.

(1993). Childhood sexual abuse and mental health in adult life. The British Journal of Psychiatry, 163, 721-732.

81.

Murray

Farrington

D. P.

(2005). Parental imprisonment: Effects on boys’ antisocial behaviour and delinquency through the life-course. Journal of Child Psychology and Psychiatry, 46, 1269-1278.

82.

Ney

P. G.

Fung

Wickett

A. R.

(1994). The worst combinations of child abuse and neglect. Child Abuse & Neglect, 18, 705-714.

83.

Nolen-Hoeksema

Girgus

J. S.

(1994). The emergence of gender differences in depression during adolescence. Psychological Bulletin, 115, 424-443.

84.

Phillips

S. D.

Burns

B. J.

Wagner

H. R.

Barth

R. P.

(2004). Parental arrest and children involved with child welfare services agencies. American Journal of Orthopsychiatry, 74, 174-186.

85.

Pine

D. S.

Cohen

Gurley

Brook

(1998). The risk for early-adulthood anxiety and depressive disorders in adolescents with anxiety and depressive disorders. Archives of General Psychiatry, 55, 56-64.

86.

Piquero

A. R.

Jennings

W. G.

Barnes

J. C.

(2012). Violence in criminal careers: A review of the literature from a developmental life-course perspective. Aggression and Violent Behavior, 17(3), 171-179.

87.

Piquero

A. R.

Moffitt

T. E.

(2005). Explaining the facts of crime: How the developmental taxonomy replies to Farrington’s invitation. In D. Farrington (Ed.), Integrated developmental & life-course theories of offending: Advances in criminological theory (pp. 51-72). New York, NY: Routledge.

88.

Pollak

S. D.

(2008). Mechanisms linking early experience and the emergence of emotions: Illustrations from the study of maltreated children. Current Directions in Psychological Science, 17, 370-375.

89.

Prino

C. T.

Peyrot

(1994). The effect of child physical abuse and neglect on aggressive, withdrawn, and prosocial behavior. Child Abuse & Neglect, 18, 871-884.

90.

Ridgeway

(2006). Assessing the effect of race bias in post-traffic stop outcomes using propensity scores. Journal of Quantitative Criminology, 22, 1-29.

91.

Rosenbaum

P. R.

(2002). Observational studies. In Rosenbaum

P. R.

(Ed.), Observational studies (pp. 1-17). New York, NY: Springer.

92.

Rosenbaum

P. R.

Rubin

D. B.

(1983). The central role of the propensity score in observational studies for causal effects. Biometrika, 70, 41-55.

93.

Rotheram-Borus

M. J.

Mahler

K. A.

Koopman

Langabeer

(1996). Sexual abuse history and associated multiple risk behavior in adolescent runaways. American Journal of Orthopsychiatry, 66, 390-400.

94.

Rutter

(1986). The developmental psychopathology of depression: Issues and perspectives. In Rutter

Izard

C. E.

Read

P. B.

(Eds.), Depression in young people: Developmental and clinical perspectives (pp. 491-519). New York, NY: Guilford.

95.

Sack

W. H.

Seidler

Thomas

(1976). The children of imprisoned parents: A psychosocial exploration. American Journal of Orthopsychiatry, 46, 618-628.

96.

Schraedley

P. K.

Gotlib

I. H.

Hayward

(1999). Gender differences in correlates of depressive symptoms in adolescents. Journal of Adolescent Health, 25, 98-108.

97.

Shalev

Moffitt

T. E.

Sugden

Williams

Houts

R. M.

Danese

. . . Caspi

(2013). Exposure to violence during childhood is associated with telomere erosion from 5 to 10 years of age: A longitudinal study. Molecular Psychiatry, 18, 576-581.

98.

Shipman

Edwards

Brown

Swisher

Jennings

(2005). Managing emotion in a maltreating context: A pilot study examining child neglect. Child Abuse & Neglect, 29, 1015-1029.

99.

Simons

R. L.

Robertson

J. F.

(1989). The impact of parenting factors, deviant peers, and coping style upon adolescent drug use. Family Relations, 38, 273-281.

100.

Sternberg

K. J.

Baradaran

L. P.

Abbott

C. B.

Lamb

M. E.

Guterman

(2006). Type of violence, age, and gender differences in the effects of family violence on children’s behavior problems: A mega-analysis. Developmental Review, 26, 89-112.

101.

Tarter

R. E.

Hegedus

A. M.

Goldstein

Shelly

Alterman

A. I.

(1984). Adolescent sons of alcoholics: Neuropsychological and personality characteristics. Alcoholism: Clinical and Experimental Research, 8, 216-222.

102.

Taylor

B. A.

Dearing

McCartney

(2004). Incomes and outcomes in early childhood. Journal of Human Resources, 39, 980-1007.

103.

Teicher

M. H.

Andersen

S. L.

Polcari

Anderson

C. M.

Navalta

C. P.

Kim

D. M.

(2003). The neurobiological consequences of early stress and childhood maltreatment. Neuroscience & Biobehavioral Reviews, 27, 33-44.

104.

van Meurs

Reef

Verhulst

F. C.

van der Ende

. (2009). Intergenerational transmission of child problem behaviors: A longitudinal, population-based study. Journal of the American Academy of Child & Adolescent Psychiatry, 48, 138-145.

105.

Vaughn

M. G.

Salas-Wright

C. P.

DeLisi

Maynard

B. R.

(2014). Violence and externalizing behavior among youth in the United States: Is there a severe 5%? Youth Violence and Juvenile Justice, 12, 3-21.

106.

Vazsonyi

A. T.

Mikuška

Kelley

E. L.

(2017). It’s time: A meta-analysis on the self-control-deviance link. Journal of Criminal Justice, 48, 48-63.

107.

Verhulst

F. C.

van der Ende

Koot

J. M.

(1997). Youth Self-Report (YSR). Rotterdam, The Netherlands: Afdeling Kinder-en Jeugdpsychiatrie, Sophia Kinderziekenhuis/Academisch Ziekenhuis Rotterdam/Erasmus Universiteit Rotterdam.

108.

Widom

C. S.

(2000). Childhood victimization: Early adversity, later psychopathology. National Institute of Justice Journal, 242, 3-9.

109.

Widom

C. S.

DuMont

Czaja

S. J.

(2007). A prospective investigation of major depressive disorder and comorbidity in abused and neglected children grown up. Archives of General Psychiatry, 64, 49-56.