Abstract
Some research on childhood adversity is critiqued for emphasizing the experiences of white, middle/upper-middle-class people and not accounting for adversities faced by more diverse populations. Adversities are also often summed up in ways that are unhelpful for targeting interventions to reduce risk of poor outcomes. I examine adversities across ecological levels—child, parent, family, and neighborhood—to determine the risk of externalizing behavior problems (EBP) using a racially diverse sample from the Longitudinal Studies of Child Abuse and Neglect (N = 1058). I consider whether family social capital can offset the effects of adversity across ecological levels. Longitudinal models indicate that adversities across multiple levels and those at the child, parent, and neighborhood levels increase risk of EBP throughout childhood. Cross-sectional models yield that early family social capital is associated with fewer EBP for children with multiple levels of adversity and at the child, parent, family, and neighborhood levels.
Keywords
The original “ACE” study examined the impact of adverse childhood experiences—child maltreatment, parental substance abuse, mental illness, domestic violence, and household criminal behavior and found such adversities to have lingering effects for survivors decades later (Felitti et al., 1998). This study was groundbreaking, as it showed an important link between early exposure to adversity during childhood leading to an increased risk for many foremost causes of morbidity and mortality in adulthood such as cancer, and heart, lung, and liver disease. Moreover, Felitti et al. (1998) showed that the more ACEs someone was exposed to in childhood, the greater threat they had of poor health outcomes in adulthood. A meta-analysis with data from 253,000 + people finds that those with multiple ACEs are at risk of violence, mental illness, and substance abuse and can pass risk on to the next generation (Hughes et al., 2017).
Although ACE research often includes, “harms that affect children directly…and indirectly through their living environments” (Hughes et al., 2017, p. e357), there is need to more precisely identify adversities putting children at risk for hazardous outcomes. Such research will help target interventions to diminish and prevent negative effects.
If we are to understand child and human development, we must “consider the entire ecological system in which growth occurs” (Bronfenbrenner, 1994, p. 37). In other words, consider how adversities or experiences beyond the individual level can help or hinder children’s development. Childhood adversities increase the risk of poor outcomes, especially, externalizing behavior problems (EBP) through the accumulation of risk (or adversities) (Ferraro & Shippee, 2009). EBP are defined as behaviors that violate social norms in aggressive, delinquent, hyperactive, and disruptive ways (Keil & Price, 2006). Those who exhibit high levels of EBP face substantial risk of: receiving harsh physical punishment (Gershoff et al., 2012), being victimized by and perpetuating teenage dating violence (Spencer et al., 2019), and are more likely to engage in violent or criminal behaviors in adolescence and adulthood (Jung et al., 2017). Thus, preventing EBP among those facing multiple adversities is imperative.
In this study, using data from the Longitudinal Studies of Child Abuse and Neglect (Runyan et al., 2014), I consider a range of adversities that children directly face (child level) and those at higher ecological levels (parent, family, and neighborhood) that indirectly shape children’s environments. I raise four questions: (Q1) Does experiencing childhood adversities across multiple ecological levels increase risk of poor outcomes? (Q2) Does risk of poor outcomes vary by the ecological level (i.e., child, parent, family, and neighborhood) adversity is present? Cumulative Inequality Theory (CIT) says that risk of poor outcomes can be offset by available resources, perceived trajectories, and human agency (Ferraro & Shippee, 2009). Thus, I investigate if children’s perceptions of available family resources (e.g., family social capital) can (Q3) reduce the risk of EBP (a) when adversity is present across multiple ecological levels and (b) at specific ecological levels. I also examine (Q4) whether early family social capital is able to reduce or eliminate the risk of future EBP in middle childhood when adversities (a) are present across multiple ecological levels and (b) at each ecological level.
Background
Childhood Adversity and Externalizing Behavior Problems
In a nationally representative sample of adolescents, nearly 62% experienced a childhood adversity or traumatic event including physical abuse, intimate partner violence, witnessing domestic violence, or the death of a loved one (McLaughlin et al., 2013). The earlier children face adversities, especially multiple severe adversities; the likelihood for developing emotional and behavioral problems increases (Grasso et al., 2016). Indeed, youth with polyvictimization are at the greatest risk of developing EBP (Grasso et al., 2016).
CIT describes how disadvantages, such as adverse childhood experiences, shape life course trajectories and influence outcomes (Ferraro & Shippee, 2009). CIT is built on five axioms to explain the ways in which social systems create and perpetuate inequality. Inequality can start accumulating before a child is born and persist throughout childhood. Axiom 2 states, “Disadvantage increases exposure to risk but advantage increases exposure to opportunity” (Ferraro & Shippee, 2009, p. 335). As children face additional adversities, the risk for poor outcomes will increase unless forces intervene.
According to CIT, ecological forces influence inequality (Ferraro et al., 2009). To understand how adversities across ecological levels independently and in conjunction with others, exacerbate risk for EBP, this study examines exposures at child, parent, family, and neighborhood level. I hypothesize (H1) that as the number of ecological levels adversity is present increases, children's risk for EBP will increase. CIT asserts that magnitude, onset, and duration of adverse experiences affect risk of poor outcomes (Ferraro & Shippee, 2009). Therefore, in the present study, adversities vary across ecological levels and over time.
Child-Level Adversities. Polyvictimization is common among children who witness violence at home; a study using lifetime data reported over 56% of children in their nationally representative sample who saw violence at home also experienced maltreatment (Hamby et al., 2010). Children who witness violence at home and are also maltreated are at greater risk for suicidal, self-harming, and violent behaviors (Tossone et al., 2015). Seeing violence outside the home, including witnessing people attacked, shot, or killed, is associated with future criminal behavior (Eitle & Turner, 2002).
Harsh disciplinary methods at home are also associated with EBP. One study found that parents who used psychological aggression or physical punishment during early childhood had adolescents that were more likely to be aggressive with and bully their peers (Gómez-Ortiz et al., 2016). A longitudinal study using nationally representative data found that physical punishment in early childhood increased EBP, and those with higher levels of EBP received more physical punishment over time (Gershoff et al., 2012).
Parent-Level Adversities. Adversities at higher ecological levels can indirectly shape children’s environments and harm their physical, emotional, and mental health. As is the case with parental domestic violence, children are at risk for developing EBP when domestic violence occurs (McDonald et al., 2016). Domestic violence often co-occurs with other adversities such as single-mother households, poverty, and low parental educational attainment (Hornor, 2005). Research also finds that children with mothers who have a history of victimization and depression and parents with substance abuse problems had more EBP (Foster et al., 2008; Osborne & Berger, 2009; Thompson, 2007). Moreover, mothers’ poor self-rated health is strongly associated with children’s mental health (Waters et al., 2000).
Family-Level Adversities. The shared environmental context of families can serve as a source of inequality (Ferraro et al., 2009). Families under economic distress are likely to face negative life events and report higher levels of parenting stress, which predicts more physical punishment use and EBP in children (Baker & Brooks-Gunn, 2019). Even after major financial crises (e.g., the Great Recession), families of all income levels can be affected by economic strain and stress possibly harming children’s well-being (Leininger & Kalil, 2014).
Family poverty is among one of the strongest predictors of poor health outcomes and EBP for children (Dearing et al., 2006). A meta-analysis of 55 studies finds that children from low socioeconomic families were more likely to develop EBP than children from more affluent families (Reiss, 2013). Social welfare benefits (e.g., Medicaid, food stamps, etc.) are meant to mitigate resource deprivation due to poverty; however, such services do not seem able to eliminate the effects of poverty on children’s mental health (Reiss, 2013).
Neighborhood-Level Adversities. Research finds that neighborhoods with high crime and stressful events result in greater EBP among children (Roosa et al., 2005). Mothers that rated their neighborhoods as poor quality (e.g., unclean, unattractive, not a good place to live, etc.) had children with more EBP (Roosa et al., 2005). A study found that neighborhood disorder is associated with EBP among children, and that the effect was more pronounced for white children than black and Latino children (Wang et al., 2019). Residential instability also predicts EBP compared with established, stable neighborhoods (Beyers et al., 2003). Moreover, when teens feel their neighborhoods are stable, safe, and supportive, EBP declined (White & Renk, 2012). Based on CIT and research on adverse childhood experiences, I hypothesize (H2) that as the number of adversities increase within each ecological level (child, parent, family, and neighborhood), the risk of EBP will increase.
Family Social Capital
A review of the literature on social capital and health describes a missing level of social capital; specifically, family social capital has been understudied in favor of macro- and individual-level social capital. Carrillo Alvarez et al. (2017) explain that family social capital is a multidimensional construct that has clear benefits for health outcomes. CIT says that adverse trajectories can be offset or changed by “advantages,” or opportunities offered through resources, networks, and prestige (Ferraro et al., 2009). In keeping with CIT, this study uses children’s perceptions of available family networks and resources—family social capital, to determine if such perceptions of resources diminish or deter the risk of EBP among children with adversities across multiple ecological levels.
Succinctly stated, “family social capital is referred to here as the social capital that can be drawn from the family” (Carrillo Alvarez et al., 2017, p. 6). In order for family social capital to be effective, adults in the household must be present and engage in supportive adult-child interactions (Coleman, 1988). Family social capital combines family network, support, and cohesion. Family network includes the number of adults in the network. Family support is emotional and instrumental support and low levels of family conflict (Carrillo Alvarez et al., 2017). Emotional support includes adults caring or taking time to explain things the child needs to know, and instrumental support is helping with necessities. Family cohesion includes adults spending time with the child.
Drawing on CIT and family social capital, I hypothesize (H3) that children who perceive greater levels of family social capital at age six will have a lower risk of EBP at age six even when (a) adversity is present across multiple ecological levels and (b) at each adversity level. I hypothesize (H4) that children who perceived greater levels of family social capital at age six will have a lower risk of EBP at age eight even when (a) adversity is present across multiple ecological levels and (b) at each adversity level. Early adversity is a powerful force, but early advantages such as strong family networks, support, and cohesion may be able to offset such adversity.
Method
Data and Sample
Data for this study come from the Longitudinal Studies of Child Abuse and Neglect 0–18 (LONGSCAN) that was compiled across five research sites in the United States from 1991 to 2012 (Runyan et al., 2014). Data collection began when children were about age 4 with follow-up interviews every 2 years (except age 10). The purpose of the original LONGSCAN study was to gather information about the causes and consequences of child maltreatment. Purposive sampling methods were used to identify children who were known victims or deemed at-risk of maltreatment. An ecological framework was used to design the LONGSCAN study, which means measures are available at the child, parent, family, and neighborhood levels. The total enrolled sample of LONGSCAN participants was 1354. For more information on the original LONGSCAN study, see work by (Larrabee & Lewis, 2016; Runyan & Kotch, 2014).
For this study, inclusion criteria consists of children with family social capital measures at age 6 (N = 277 dropped), those with at least one score for EBP between the ages of 4–16 (N = 0 dropped), and limited to black, white, Latino, and multi-racial children. Too few children identified as “racially other,” Native American, and Asian and were omitted (N = 19 dropped). The final analytic sample includes 1058 children. Due to the purposive sampling methods in the original LONGSCAN study, the findings from this study are not generalizable. Despite the lack of generalizability, LONGSCAN is a powerful dataset with seven waves of data to study childhood adversities across multiple ecological levels and the impact on children’s mental health.
Measures
EBP Scores. The Child Behavior Checklist 4–18 (Achenbach, 1991) was used to measure EBP. Caregivers read a description of child behavior and said whether it was 0 = not true; 1 = somewhat/sometimes true; 2 = very true/often true. Aggressive behaviors include arguing a lot, being cruel, bullying, and destroying his/her own things. Delinquent behaviors included the child not feeling guilty after misbehaving, lying or cheating, or running away. See Achenbach (1991) for a full list of measures. Raw scores from the aggressive and delinquency scales were combined into a standardized EBP scale using T scores. LONGSCAN advises using T scores in all analyses (Hunter et al., 2001). T scores <60 are in the normal range.
Childhood Adversities. Childhood adversity was measured at the child, parent, family, and neighborhood levels with the number of adversities summed up—a higher number indicates more adversities at that level, as well as a summary measure to capture the number of ecological levels adversity was present. Adversities measured vary at each time point of the LONGSCAN study. For the number of ecological levels adversity was present, if a child had any adversity at a given level, it was coded = 1 and the maximum score possible = 4.
Child-level adversities included exposure to violence in the home = 1 or community = 1 (age 6) (Richters & Martinez, 1992); ever witnessing violence = 1 (ages 12 and 14) (Knight et al., 2010); experiencing psychological aggression = 1 or physical assault = 1 as a means of discipline measured with the Conflict Tactics Scale (ages 4–16) (Straus, 1979); and allegations of child maltreatment (ages 4–16) (Larrabee & Lewis, 2016).
Parent-level adversities included parent/primary caregiver’s self-report of fair or poor health = 1, less than a high school education = 1, or being unemployed = 1 (ages 4–16); reporting 16+ depressive symptoms from the Center for Epidemiologic Studies Depression Scale = 1 (ages 4, 6, 12–16) (Radloff, 1977); having 2–4 positive responses on the CAGE Alcoholism Screening Tool = 1 (age 4) (Mayfield et al., 1974); history of physical and/or sexual abuse in childhood and/or adulthood = 1 (age 4) (Hunter & Everson, 1991); verbal = 1, psychological = 1, and/or physical = 1 aggression domestic violence (ages 6–16) (Straus, 1979).
Family-level adversities are reported by the parent/caregiver and include above average stressors = 1 from the Everyday Stressors Inventory (age 6) (Hall, 1983); not having enough money for food, clothing, rent, utilities, or medical care = 1 (ages 12–16); family income below the poverty line = 1 (ages 4–16); receiving social welfare such as Medicaid = 1, food stamps = 1, Women’s, Infants and Children’s (WIC) program = 1, and federal housing assistance = 1 (ages 4, 8–16); residing in a single, separated, divorced, or widowed-parent household = 1 (ages 4–16).
Neighborhood-level adversities reported by the parent/caregiver include above average neighborhood chaos = 1, below average neighborhood collective efficacy = 1, above average neighborhood instability = 1 (ages 12–16) (Coulton et al., 1996; Sampson et al., 1997); neighborhood drug abuse, dangers, and bad influences = 1, very little or no neighborhood support = 1, and very little or no neighborhood pride or morale (ages four and 8) from the Neighborhood Short Form (Hunter et al., 2001); little to no neighborhood attachment = 1, unsafe neighborhood = 1, neighborhood is not child-friendly = 1, and little to no tangible neighborhood support (age 6) from the Neighborhood Risk Assessment (Hunter et al., 2002).
Family Social Capital. Child-participants completed the Inventory of Supportive Figures at age six only (Hunter et al., 2002; Whitcomb et al., 1984). Family social capital includes family network, family support, and family cohesion, as described by Carrillo Alvarez et al. (2017). Family network is measured by the question: has there been any adult who has been especially [helpful] to you? 1 = yes and 0 = no; children were asked this question a maximum of three times or until they indicated no more helpful adults. To ensure only family network is captured, the responses are limited to family members only for this study—mother, stepmother, foster mother, grandmother, sister, other female relative, father, stepfather, foster father, grandfather, brother, and other male relative. Children who indicated a helpful adult that was an adult female friend, adult male friend, mother’s boyfriend, teacher, physician/nurse, pastor, or others were coded as 0 for that response to indicate “no” helpful family member for the survey question.
Reflecting on the helpful family members, emotional and instrumental family support was measured by children saying how much the family member(s) shows they care, explains things the child needs to know or helps solve a problem, and helps with necessities such as getting food and clothing. Family cohesion is measured by children reporting the amount of time spent with the helpful family member(s). Responding to the questions about caring, explaining, helping with necessities, and time spent, children said 0 = not at all, 1 = a little, 2 = some, and 3 = a lot. If a child had three helpful family members providing maximum support, then the highest score possible = 9. A support measure is included that sums up the caring, explaining, help with necessities, and time spent with the most helpful family member—the maximum possible score is 12. Then, a cumulative support measure from the most helpful family member (max. of three) is included—a maximum score of 36 is possible. Children who reported 0 helpful family members were coded as 0 for all measures.
Covariates. Dates of data collection varied across the LONGSCAN sites; thus, children’s ages range around the targeted age (Larrabee & Lewis, 2016). The mean age at each time point is close to the target age of assessment, based on children’s birthday, and reported in years. Children’s sex is recorded from the age 4 or 6 parent-report. Children’s race/ethnicity is based on parent-report when the children were either age 4 or 6: black = 0; white = 1; Latino = 2; and multi-racial = 3. A regional variable is included in all analyses to adjust for the five research sites in the original LONGSCAN study: Eastern (Baltimore) = 1; Northwest (Seattle) = 2; Southwest (San Diego) = 3; Midwest (Chicago) = 4; and Southern (North Carolina) = 5.
Analytic Strategy
LONGSCAN data has correlated observations over seven waves of data, six of which are used in this study, an unbalanced sampling design, and participants that participated in some waves and then not others (Larrabee & Lewis, 2016). Multilevel mixed effects generalized linear models with random effects are used, which makes use of all available data and provides more efficient and consistent estimations (Little, 2008) and adjusts for EBP scores correlated over time (Edwards, 2000). Models for questions 1 and 2 include a random intercept and slope that allows EBP to vary for each child and over time as children age (StataCorp, 2017a). Log likelihood results, Wald
Research questions 3 and 4 test the effects of family social capital measured at age 6 on childhood adversity and EBP at age 6 and age 8, respectively. Cross-sectional generalized linear models are used and share many similar benefits as the multilevel mixed effects generalized linear models. Separate models are included for each type of childhood adversity measure and family social capital characteristic at ages 6 and 8, and interaction terms test for the impact of family social capital. Stata 15 is used for all statistical analyses (StataCorp, 2017b).
Results
Externalizing Behavior Scores and Childhood Adversities: Means, Standard Deviations, and Ranges by Age.
Family Social Capital and Study Covariates: Means/Percentages, Standard Deviations, and Ranges at Age 6 (N = 1045).
Childhood Adversities and Externalizing Behaviors throughout Childhood (N = 1058).
*p<0.05. **p<0.01. ***p<0.001. Standard errors in parentheses. 95% confidence intervals in brackets
Externalizing behavior scores by no. of ecological levels of adversity with 95% confidence intervals.
Question 2 asks whether the risk of EBP varies depending on the ecological level—child, parent, family, or neighborhood—that adversity is present. For every additional adversity at the child level, EBP increase by 1.33 (p < 0.001); however, the risk diminishes with each additional year of age by −0.09 (p < 0.001). Children whose parents face adversities have a 1.54 increase in EBP for every additional adversity at the parent level (p < 0.001), but a slight decrease over time −0.06 (p < 0.01). In the child and parent-level models, the main effect for age is not significant, and other covariates follow a similar pattern as those discussed in the overall model. Family-level adversities are not significant; the risk of EBP does not vary by family-level adversities among this sample. Neighborhood-level adversities have a 1.00 (p < 0.001) increase in EBP for each additional adversity, but this risk does not vary over time.
Childhood Adversities, Family Social Capital, and Externalizing Behaviors at Age 6 (N = 1045).
*p<0.05. **p<0.01. ***p<0.001 Note: Models control for children’s region, age, sex, and race
For question 3b, the main effect of child-level adversities is strengthened with family social capital measures included in the models. The main effects for family social capital measures and the interaction did not meet the standard threshold for statistical significance, but many were approaching significance. Looking at the marginal predicted mean scores from the model based on total amount of help from helpful family members by child-level adversity does show statistically significant differences. Figure 2 shows that among children with the highest child-level adversities and those with an average amount of adversity ( Externalizing behavior scores by child-level adversity across total amount of help at Age 6 with 95% confidence intervals.
At the parent and family levels, with family social capital characteristics in the model, the effects of these adversities are significantly weakened or eliminated (e.g., when the number of helpful family members is included with family-level adversities). Similar effects are observed for neighborhood-level adversities with the number of helpful family members in the model as well as children perceiving their family members care, help with necessities, and feel support from the most helpful family member.
Adversities, Family Social Capital at Age 6, and Externalizing Behaviors at Age 8 (N = 928).
*p<0.05. **p<0.01. ***p<0.001 Note: Models control for region and child’s age, sex, and race. Standard errors in parentheses
Discussion
Drawing on CIT, using an ecological framework, and research on family social capital and health, I sought to determine (1) if having childhood adversities across multiple ecological levels increased the risk of EBP throughout childhood; (2) if the risk of EBP varied depending on the ecological level that adversity is present; (3) whether family social capital characteristics in early childhood could offset unfavorable EBP trajectories associated in early and (4) middle childhood.
I hypothesized (H1) that children’s risk of EBP would increase as the number of ecological levels of adversity increased. Consistent with CIT and previous research, I find as adversities increased so too did the risk of EBP. Previous ACE research is criticized for focusing too narrowly on the experiences of white, middle and upper-middle-class people (Cronholm et al., 2015). This research contributes to the literature by expanding on ACEs previously considered, uses an ecological framework called for in CIT (Ferraro et al., 2009) by considering adversities at the child, parent, family, and neighborhood levels, and uses a racially and ethnically diverse sample of 1058 children from ages 4 to 16 years old.
Although the cumulative impact of adversities across multiple ecological levels for children’s EBP is not surprising, this study adds to our understanding of how risk varies by the ecological level where adversities are present. I hypothesized that (H2) as adversities increase at each ecological level, the risk of EBP would increase. In the longitudinal models, hypothesis 2 is confirmed at the child, parent, and neighborhood levels and in cross-sectional models at ages 6 and 8 at the family level.
In the longitudinal models and cross-sectional model at age 6, the coefficient for parent-level adversity is greater than other levels. Studies confirm the harmful effects that parental depression, alcoholism, history of victimization, low levels of education, unemployment, and domestic violence have for children’s mental health (Foster et al., 2008; McDonald et al., 2016; Osborne & Berger, 2009; Thompson, 2007). These indirect experiences can and do shape children’s environments in deleterious ways. Better support services are imperative for parents struggling with addiction, mental health problems, a history of trauma, and improved educational and employment opportunities to improve child, parent, and family well-being.
Another contribution of this study is identifying the role family social capital can play in offsetting the effects of childhood adversity. I hypothesized (H3) that when children perceived greater levels of family social capital at age 6, they will have a reduced risk of externalizing behaviors at age 6 and (H4) age 8 even when (a) adversities are present across multiple ecological levels and (b) at each adversity level subtype. Many forms of family social capital were associated with diminishing or eliminating the effects of childhood adversity when it was present at multiple ecological levels and by adversity subtype in early and middle childhood.
I caution that this study is based on observational data, thus causation cannot be determined. However, encouraging early interventions that support positive parent-child interactions fostering caring and communicative relationships, making sure that children’s needs are met, and that children have time with helpful family members will likely benefit all children, and especially those with polyvictimization (McPherson et al., 2014). Indeed, in a meta-analysis of 55 studies on family and community social capital, parent-child relationships and relationships with other family members protected children’s mental health (McPherson et al., 2014).
Although this study makes several contributions to the existing literature, there are limitations to note. First, the purposive sampling methods in the original LONGSCAN design means that the findings from this study are not generalizable. Future research should examine childhood adversity across multiple ecological levels, family social capital, and EBP using a more representative sample. Second, family social capital appears to have a positive impact in reducing or eliminating the effects of adversity across ecological levels, but causation cannot be determined. Further, family social capital measures are only available at one time point in this data. Longitudinal research with family social capital measured across time is needed to form a clearer understanding about the relationship between adversity, family social capital, and EBP. Third, family social capital is an important missing level in research on social capital and mental health (Carrillo Alvarez et al., 2017). Yet, family social capital may work in collaboration with other individual or macroforms of social capital. Future studies should consider how individual, family, and community social capital works together to improve children’s mental health.
Reducing EBP among children and adolescents has lifelong benefits for youth, their peers, families, and society. Disrupting the intergenerational transmission of EBP is needed to help all children (Hughes et al., 2017). Some scholars have even suggested that reducing such behaviors is imperative for reducing violence in society (Jung et al., 2017). This study improves our understanding of childhood adversity across multiple ecological levels as well as at the child, parent, family, and neighborhood levels and provides insight into the ways family social capital appears to improve EBP among the most vulnerable children.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This project received support from the Dr. Charles Hill Faculty Research Award, College of Social and Behavioral Sciences, University of Northern Iowa.
