Gender Differences in Adverse Childhood Experiences,Self-Control,and Delinquency

Adverse Childhood Experiences and Delinquency

Criminologists have increasingly used the ACEs conceptual framework to examine associations with delinquency among both incarcerated (Baglivio et al., 2015; Bonner et al., 2020; Fox et al., 2015; Muniz et al., 2019) and nonincarcerated youth (Brown & Shillington, 2017; Jones & Pierce, 2021; Leban & Gibson, 2020). Their studies have shown that children exposed to ACEs—including abuse, neglect, and chaotic home environments—have a higher risk of delinquent behavior than youth who did not experience ACEs (Hoffmann & Jones, 2020; Jones & Pierce, 2021; Muniz et al., 2019). Studies have also consistently shown that ACEs are more prevalent among correctional and high-risk samples than among those in the general population (for a systematic review, see Malvaso et al., 2021). Studies of juvenile offenders in Florida, for example, showed that most (96%–97%) had experienced at least one ACE, and 40% had experienced four or more ACEs (Baglivio et al., 2014). ACEs were also interrelated: exposure to one ACE was associated with a substantial increase (as much as 1,286 times higher) in the likelihood of experiencing additional ACEs (Baglivio & Epps, 2016).

Numerous studies have documented a dose-response dynamic between ACE exposure and delinquent behavior (e.g., Fox et al., 2015; Hunt et al., 2017; Pierce & Jones, 2022). For example, Fox et al. (2015) reported that each additional ACE was associated with a 35% increase in the likelihood of serious, violent, and chronic juvenile offending. Other studies have shown that youth who experience multiple types of ACEs are more likely to have substance use problems (Brown & Shillington, 2017), be involved in delinquent behavior (Jones & Pierce, 2021; Perez et al., 2018), be involved in the juvenile justice system (Baglivio et al., 2014), reoffend (Craig et al., 2020; Wolff & Baglivio, 2017), be involved in serious, violent, or chronic offending (Bonner et al., 2020; Fox et al., 2015; Wolff et al., 2020), and to experience early onset of offending (Baglivio et al., 2015). In contrast, some scholars have found that ACEs are unrelated to post-release supervision (Craig et al., 2020) and that the effects of ACEs on offending behaviors can vary considerably across racial and ethnic groups and across offense types (DeLisi et al., 2017). Overall, however, there is ample evidence to suggest that ACEs play a significant role in children’s pathways toward delinquent behavior.

Self-Control and Delinquency

The consistency of the association between ACEs and delinquent behavior is matched, or perhaps exceeded, by the association between low self-control and delinquency. Simply put, youth with low self-control are viewed as having short-term outlooks that value immediate benefits and generally fail to consider the costs of their actions. They are impulsive, hedonistic, and lack foresight, with risk taking tendencies that manifest in delinquent and several other maladaptive behaviors. Since Gottfredson and Hirschi (1990) pioneered their self-control model, hundreds of studies have examined its propositions. Systematic reviews and meta-analyses have identified low self-control as a consistent predictor of the incidence and frequency of delinquent and criminal behaviors (Hay & Meldrum, 2016; Pratt & Cullen, 2000; Vazsonyi et al., 2017). In a meta-analysis of almost 100 empirical studies, for example, Vazsonyi et al. (2017) reported effect sizes of self-control of about 0.415 in cross-sectional and 0.345 in longitudinal studies. Some scholars have called self-control one of the most reliable predictors of crime and delinquency and claimed that studies ignoring its effects risk misspecification bias that likely yields specious results (Pratt & Cullen, 2000). Others, though criticizing Gottfredson and Hirschi’s specific model by arguing that low self-control is better conceptualized as impulsivity or shortsightedness, nevertheless recognize the importance of impulse control that, though affected by various situational, opportunistic, affective, and perceptual factors, ultimately affects the likelihood of delinquent choices for some individuals (Burt, 2020).

Gottfredson and Hirschi (1990, 2020), as well as other scholars (e.g., Hay & Forrest, 2006; Nofziger, 2008), have argued that the source of high or low self-control among youth originates in parenting practices. Parents who consistently teach young children to delay gratification, for example, encourage the development of high self-control. Presumably, this means that contrary parenting practices, such as permissive, irregular, or aggressive styles, promote low self-control. Other research has traced the origins of self-control to genetic factors, neuropsychological deficits, and abnormalities in the frontal lobe and prefrontal cortex, areas of the brain responsible for impulse control, forward thinking memory, rational reflection, and related cognitive processes (Beaver et al., 2007; Meldrum et al., 2018). Regardless of the precise mechanism, self-control is a significant predictor of delinquency that also mediates the effects of several background factors on delinquency, including religious practices, neuropsychological deficits, MAOA activity, depressive symptoms, corporal punishment, and child abuse (Bunch et al., 2018; Evans et al., 2012; Janssen et al., 2016; Perez et al., 2018; Remster, 2014; Tomlinson et al., 2022).

Recent studies have shown that childhood ACEs are also related to self-control (Chapple et al., 2021; Craig, 2019; Jones, Pierce & Chapple, 2021; Meldrum et al., 2020). For example, using distinct samples Chapple et al. (2021) and Meldrum et al. (2020) determined that exposure to a greater number of ACEs was related to diminished self-control among youth, even after adjusting for the effects of a host of other factors. Research has also shown that experiencing a high number of ACEs is associated with higher levels of impulsivity, a trait closely related to low self-control (Lovallo, 2013; Shin et al., 2018). Additional studies have found that ACEs are associated with other concepts linked to self-control development, such as negative emotionality (Wolff & Baglivio, 2017), externalizing behaviors (Hunt et al., 2017; Muniz et al., 2019), empathy (Narvey et al., 2021), poor delay of gratification (Evans et al., 2012), and a lack of future orientation (Craig, 2019).

Though recent studies have provided compelling evidence that ACEs are linked to deficits in self-control, researchers have rarely taken the next logical step and assessed whether low self-control mediates the association between ACEs and delinquent behavior (Fava et al., 2022; Pechorro et al., 2021). In one of the few exceptions, Fava et al. (2022) examined longitudinal data from Michigan and found that low self-control in middle childhood mediated the relationship between early ACEs and delinquency. Other studies have suggested that a mediation model is viable, but have measured only specific types of ACEs, such as parental violence, child abuse or maltreatment, and coercive parenting (Bunch et al., 2018; Cheung & Cheung, 2010; Manzoni & Schwarzenegger, 2019). Given that these studies did not examine a full range of ACEs, however, additional research is needed to clearly assess the potential mediating role of self-control in the association between ACEs and delinquent behavior.

ACEs, Self-Control, and Delinquent Behavior: Differences Among Girls and Boys?

An additional issue is whether the proposed self-control mediation model, assuming it is viable, varies by gender. This issue is motivated by a small but growing line of research that documents gender differences in both ACE exposure and the development of self-control. Studies have found, for instance, that girls tend to experience more ACEs than boys and that gender differences exist in the types of ACEs youth experience (Baglivio et al., 2014; Dierkhising et al., 2019). Sexual abuse and assault victimization, for example, are more prevalent among girls (Dierkhising et al., 2019); whereas experiencing physical abuse tends to be more prevalent among boys (Teague et al., 2008). Research also indicates that the effects of ACEs differ for girls and boys, with ACEs associated with delinquency especially among boys (Gajos et al., 2022; Leban, 2021; Leban & Gibson, 2020). Other studies have found gender differences in the effect of ACEs on substance use, perpetration of violence, and internalizing and externalizing behaviors (Leban, 2021; Leban & Gibson, 2020).

Research has demonstrated that self-control varies by gender: girls generally exhibit higher self-control than boys (Jo & Bouffard, 2014). Girls also tend to score lower than boys on measures of impulsivity (Chapple & Johnson, 2007), risk-taking, and self-centeredness (Gibson et al., 2010), but higher on empathy (O’Neill, 2020). Some researchers have proposed that boys exhibit lower self-control due to early caregiving disturbances that are uniquely gendered and associated with traditional gender role socialization (Hayslett-McCall & Bernard, 2002). It seems likely that ACEs are one striking example of early caregiving disturbances. Yet only one study has examined gender differences in the ACEs—self-control association, finding that as ACEs increased, self-control tended to decline for both boys and girls, but that ACEs were also associated with increased impulsivity among boys (Chapple et al., 2021).

In sum, studies have identified an association between ACEs and self-control, ACEs and delinquent behavior, and self-control and delinquent behavior. But, with few exceptions (e.g., Fava et al., 2022), research has not directly examined how these factors might interrelate. Thus, this study seeks to test a more complete model by examining the following research questions: (1) does self-control mediate the association between early ACEs and delinquent behavior? (2) Assuming self-control does serve as a mediator, do its effects differ for girls and boys?

Delinquent Behavior—Year 15

We used adolescent self-reports to determine delinquent behavior at year 15. Each adolescent was asked if they had never, one to two times, three to four times, or five or more times participated in 14 different delinquent acts in the past 12 months. These acts included behaviors such as “deliberately damaging property that didn’t belong to you,” “taking something from a store without paying for it,” “getting into a serious physical fight,” and “going into a house or building to steal something” (see Supplemental Appendix A for a full list of items). We coded each act into a dichotomous variable (0 = never, 1 = 1–5+ times) and then created a summed variety scale that ranged from zero to 14 (Kuder-Richardson reliability = .72; Sweeten, 2012).

Adverse Childhood Experiences—Year 5

Mirroring the CDC-Kasier Studies (Felitti et al., 1998), we measured ACEs with eight categories reported during the early childhood waves (years 1, 3, and 5): physical abuse, emotional abuse, physical neglect, emotional neglect, household substance use, parental incarceration, parental intimate partner violence, and household depression and/or anxiety. Select measures from the Parent-Child Conflict Tactics Scale (CTS-PC; Straus et al., 1998) were used to determine physical and emotional abuse and neglect. In the FFCW, CTS-PC subscales were coded in the following scale: never happened, once, twice, 3 to 5 times, 6 to 10 times, 11 to 20 times, more than 20 times. To calculate the degree of maltreatment, we modeled our approach after previous studies (Hunt et al., 2017; Pierce & Jones, 2022) by considering the midpoint of each category when totaling each category together. These totals were then recorded into a dichotomous variable indicating whether a family scored in the top 10th percentile for the total number of acts toward the child separately. Physical abuse was measured at years 1, 3, and 5. At year 1 mother and father were asked: “have you spanked child in the past month.” At years 3 and 5 the mother and father were asked whether he or she: “shook child,” “hit child on bottom with a hard object,” “spanked him/her on the bottom with your bare hand,” “slapped child on hand, arm or leg,” and “pinched child.” Emotional abuse was measured at years 3 and 5 by asking the mother and father whether he or she had: “ever swore or cursed at child,” “shouted, yelled or screamed at child,” “said you would send child away or would kick child out of the house,” “called him/her dumb or lazy or some other name like that,” and “threatened to spank/hit child but did not actually do it.”

Physical neglect was measured at years 3 and 5 by asking the mother and father whether she or he “left child home alone, but thought some adult should be with him/her,” “was not able to make sure (child) got the food he/she needed,” “not able to make sure (child) got to a doctor or hospital when needed,” and “were so drunk/high that you had a problem taking care of your child.” Emotional neglect was measured at years 3 and 5 with one item from both mother and father asking if they were “caught up with your own problem that you were not able to show love to your child.”

Parental substance use was measured in years 1, 3, and 5, and was measured differently for mothers, biological fathers, and mothers’ current partner (when applicable). For maternal substance use at year 1, mothers were asked: (1) if they have smoked marijuana or pot; (2) or used cocaine or other hard drugs in the past month; (3) if drinking or drugs has interfered with how they manage daily since the birth of their child; (4) if drinking or drugs interfered with personal relationships since the birth of their child; (5) and if they’ve sought help for drug or alcohol problem since the birth of their child. Lastly, to measure paternal substance use during the first year, mothers were asked if the child’s biological father has problems keeping a job or getting along because of alcohol or drugs. If the mothers responded yes to any of these items, parental substance use was coded “yes” for that year.

For maternal substance use in years 3 and 5, mothers were asked if they had used a variety of drugs or heavy alcohol in the past year: sedatives, tranquilizers, amphetamines, analgesics, inhalants, marijuana, cocaine, LSD, and heroine. Heavy drinking was measured by asking if the mother often had four or more drinks in 1 day almost “every day,” “a few times a week,” or “a few times a month.” Data for fathers and current partners had significantly less detail about drug and alcohol use. To determine child’s biological father and mother’s current partner’s alcohol and drug use we used mother’s affirmative responses to whether the father or current partner had “problems with job/family/friends because of alcohol/drug use.” A dichotomous variable representing parental substance use was created to indicate whether child’s biological mother, biological father, or mother’s current partner had a substance use problem between survey years 1 and 5 (0 = no drug or heavy alcohol use, 1 = some drug and/or heavy alcohol use).

Parental incarceration was identified if the mother, father, or mother’s current partner had spent any time in prison or jail at years 1, 3, and 5. At year 1, mothers reported if the biological father was in jail or had ever spent time in jail. If so, children were categorized as exposed to parental incarceration. At years 3 and 5, mothers reported whether the father had spent time or was in jail since the prior interview. Additionally, at year 5 mothers reported if they have spent any time in jail in the last 2 years.

Parental interpersonal violence (IPV) was measured at years 1, 3, and 5 by a combination of sexual, physical, and emotional violence experienced by the mother. While this violence is not directly aimed at the focal child, the idea is that a child would be exposed to domestic violence that their mother experienced. The following five sets of questions make up exposure to IPV: (1) if the child’s biological father or mothers’ current partner “tried to make you have sex or do sexual things;” (2) if the mother was “kicked, slapped, or hit with a fist or object”; (3) if the child’s biological father or mothers’ current partner “tried to keep you from seeing or talking with your friends or family;” (4) “tried to prevent you from going to work or school;” and/or (5) “withholds money, makes you ask for it, or takes it.” If the woman reported any of the emotional, physical, or sexual abuse measures at years 1, 3, or 5, parental IPV was coded as yes (1 = yes, 0 = no).

Parental anxiety was measured at years 1 and 3 from the Composite Interview Diagnostic Interview (CIDI; Kessler et al., 1998). The CIDI is a standardized assessment of mental disorders containing 20 items to measure generalized anxiety disorder and 15 items to assess major depression. This scale has been found to have strong reliability and validity for depression and anxiety (Patten et al., 1994). The constructed scale determined if the mother or biological father (data was not available for mother’s current partner) meets anxious criteria. Similarly, the CIDI was used to determine mother’s and/or biological father’s depression at years 1, 3, and 5. These results were dichotomized to create overall parental anxiety and/or depression (0 = no parental anxiety or depression, 1 = parent had anxiety and/or depression).

After creating eight dichotomous variables for ACEs by year 5, individual ACE measures were summed to show a range of ACE exposure for each child varying from 0 (no exposure to ACEs) to 8 (exposure to all categories of ACEs).

Self-Control—Year 9

Child self-control at year 9 was measured using a scale comprised of 25 questions drawn from the behavioral, emotional, and social problems scales of the Child Behavior Checklist (Greenfield et al., 2004; Haynes & O’Brien, 2000). The survey items included responses from the PCG about whether the youth “is impulsive or acts without thinking,” “shows off or clowns around,” and “is confused or seems to be in a fog” (see Supplemental Appendix A for a full list of items). Each question was coded according to the following responses: “not true” = 0, “somewhat or sometimes true” = 1, and “very true or often true” = 2. The variables were then summed to create a single variable measuring self-control that was scaled from 0 to 50, with children scoring higher on the scale manifesting lower self-control (alpha reliability coefficient = .89).

Control Variables

The control variables used in the analysis included age and race/ethnicity of the youth, household socioeconomic status, mother’s age at wave 1, mother’s education at baseline, mother’s marital status at baseline, closeness to mother and father, peer delinquency, perceived neighborhood violence, and neighborhood collective efficacy. Age ranged from 14 to 19 at year 15. Race was self-reported by the focal child at year 15 and coded as White (reference group), Black, Hispanic, or Other. A dichotomous measure of wealth was created at baseline to compare families whose income placed them at two times the poverty level (coded 1) to those whose incomes placed them at less than two times the poverty level (coded 0). Mother’s age at wave 1 was self-reported and ranged from 15 to 48. Mother’s education at baseline was coded into three categories: “less than high school” (reference group), “high school or equivalent,” and “greater than high school.” Mother’s marital status at baseline was coded as 0 if not married to the biological father of the child and 1 if married to the biological father. Closeness to mother was measured by the following question, “How close do you feel to your biological mother” (coded 0 = not very close, 1 = fairly close, 2 = quite close, 3 = extremely close), with higher scores indicating more closeness. Closeness to father was measured in the same manner.

Peer delinquency was measured by asking youth in year 15 if their friends “often,” “sometimes,” or “never” participated in the following nine delinquent acts: “deliberately damaging property that didn’t belong to them,” “taking something from a store without paying for it,” “stealing something worth more than $50,” “going into a house or building to steal something,” “threatening to use a weapon to get something,” “selling marijuana or other drugs,” “stealing something worth less than $50,” “drinking alcohol more than two times without parents,” and “trying illegal drugs.” We recoded these responses such that no participation was coded as 0 and sometimes or often was coded as 1 and then added them so that the scores ranged from 0 (no peer delinquency) to 9 (peers often or sometimes delinquent for all nine acts; Kuder-Richardson reliability = .82).

Using the primary caregiver data from year 15, perceived neighborhood violence was measured as a dichotomous variable: “Have you ever been afraid to let your child outside due to neighborhood violence?” (yes = 1, no = 0). Responses to seven items were summed together to create an index of neighborhood collective efficacy (alpha reliability coefficient = .76). The items included: “People around here are willing to help their neighbors,” “This is a close-knit neighborhood,” and “People in this neighborhood generally don’t get along with each other” (reverse coded). Responses ranged from “strongly disagree” (coded 0) to “strongly agree” (coded 3; see Supplemental Appendix A for a full list of items).

Analytical Strategy

We first estimated weighted descriptive statistics for the full sample and by gender. The analysis was weighted to the baseline national weights and thus the estimates are representative of youth born in large U.S. cities (the 77 U.S. cities with populations over 200,000 in 1994) between 1998 and 2000 (Kennedy & Gelman, n.d.).

To examine the mediation model, we used a structural equation model (SEM) estimated with maximum likelihood (Cheong & MacKinnon, 2012). ACEs serve as the key explanatory variable, self-control as the mediator, and delinquent behavior as the outcome variable, along with the control variables. The SEM was first estimated with the full sample (see Table 2) and then stratified by gender (see Table 3) to assess whether any mediation effect differed for girls vs. boys, with no equality constrains on the error variances, the variances of the variables, or the covariances of the variables. The SEM included the raking variables employed to create the weights in the model, thereby considering important aspects of the complex sampling design. We also estimated direct, indirect, and total effects of the variables in the model (see Table 4). The figures display the standardized coefficients representing the associations between early ACEs, self-control, and delinquency, controlling for all other variables specified in Tables 2 to 4. Figure 1 looks at the whole sample, while Figure 2 looks at boys and girls separately. All analyses were conducted with the Stata statistical software (version 17.0).

OPEN IN VIEWER

Figure 1.

Mediation results for the full sample, standardized coefficients.

OPEN IN VIEWER

Figure 2.

Mediation result for boys (top coefficients) and girls (bottom coefficients), standardized coefficients.

Limitations and Future Research

Although the current study’s findings add to existing literature on ACEs, self-control, and delinquency, there are several limitations worth noting. First, although the present findings are derived from national data, the nature of the FFCWS sampling strategy (i.e., overrepresentation of low income, unmarried couples, and their children) does not allow the results to be generalized to the U.S. population of youth. Second, there were two measures of ACEs included in the CDC-Kaiser ACE study that were not included in the current study. We were unable to include sexual abuse as an ACE as there were no measures of this behavior in the FFCW study. We chose not to include parental divorce or separation because we did not consider it to be a relevant event for families participating in the FFCW, which significantly oversampled for nonmarital births. Third, there are likely other types of ACEs that we were unable to capture in this study but may be important predictors of self-control and involvement in delinquency, such as the death of a parent or loved one, serious accidents or injuries, or community-level adverse events. Future research should continue to explore more types of ACEs to determine their relationship to self-control and delinquency.

Fourth, the CDC-Kaiser ACE studies assessed information on mental illness, substance use problems, and criminal behaviors among all members of the household. Unfortunately, we only had these measures for the child’s mother, with limited information on the child’s biological father or the mother’s current partner. Consequently, our measure of ACEs largely characterizes the child’s biological parents rather than the characteristics of all members of the household. Fifth, the FFCW did not administer a consistent inventory across all time points. Although repeated measures of adversity were obtained, some of the measures were not included in each wave. Sixth, other parental socialization strategies such as parenting style, consistent parenting practices, and parental warmth—or the lack thereof—may also affect the development of self-control and the risk of involvement in delinquent behavior, but were not included in this study. Future research should explore these parenting strategies to determine how they are affected by ACEs and how they affect the development of self-control. Finally, even though our modeling of the effect of ACEs on self-control and delinquency provides us with a better sense of these relationships among adolescents, this study did not capture the frequency, timing, or seriousness of each ACE event. Future studies that better capture the overall complexity of ACE exposure would be informative.

Implications for Practice

Notwithstanding these limitations, findings from the current study indicate that ACEs present a serious challenge to the development of self-control and indirectly affect the involvement in delinquency during adolescence. Understanding the long-term effects of ACEs on children and adolescents should motivate trauma prevention and intervention efforts to regularly screen for ACEs. As our study points toward the importance of early ACE exposure (by age 5) in self-control development during middle-childhood, intervention efforts that target ACEs very early in a child’s life may be the most advantageous at reducing deficits in self-control. Moreover, because children who are exposed to ACEs are at an increased risk of experiencing additional ACEs, it is imperative that children who have a history of ACEs are identified and referred to services that can reduce the risk of exposure to future ACEs. Trauma-informed health care approaches that recognize the need for ACE and trauma screening practices may help clinicians and other healthcare professionals monitor, identify, and address the psychological and behavioral repercussions of ACEs (Finkelhor, 2018). The Substance Abuse and Mental Health Services Administration (SAMHSA, 2018), for example, implements a “system of care” approach that utilizes screening, assessment, and trauma treatment services for children and youth with a history of ACEs who are also at risk of developing serious emotional disturbances. Such an approach emphasizes the importance of trauma-informed care which is coordinated across a network of community-based services. A “system of care” approach can aid children in developing healthy strategies for coping with ACEs, which may significantly reduce the risk of delinquency.