Longitudinal Effects from Childhood Abuse to Bullying Perpetration in Adolescence: The Role of Mental Health and Social Problems

Abstract

Bullying has attracted increased attention due to the serious implications for perpetrators, victims, and schools. Recent studies have sought to identify factors that may contribute to bullying perpetration, and child abuse has been identified as one such factor. The mediating processes linking child abuse to bullying perpetration, however, are not well understood. The current study explored adolescent mental health problems, including symptoms of depression, anxiety, dissociation, and posttraumatic stress disorder, and poor social skills as pathways between childhood abuse and adolescent bullying perpetration. Data for the current study are from the Longitudinal Studies of Child Abuse and Neglect. The current study utilized a longitudinal and multi-informant design in which adolescents reported their history of childhood abuse, mental health problems, and social skills when they were 12 years old; bullying perpetration was reported by adolescent’s teachers when adolescents were 12 and 14. Results indicated childhood abuse was associated with higher levels of depression, anxiety, anger, dissociation, posttraumatic stress, and poor social skills. Only anxiety and poor social skills at age 12 were significantly associated with bullying perpetration when adolescents when were 14. Bootstrapped indirect effects from childhood abuse to bullying perpetration were significant for both anxiety and poor social skills, indicating full mediation. Addressing anxiety and poor social skills in early adolescence among children who have been abused may prevent bullying perpetration in mid-adolescence. Clinicians, teachers, and school administrators may desire to focus efforts on reducing anxiety and increasing social skills to mitigate bullying perpetration.

Introduction

Childhood abuse is a public health problem in the United States that affects millions of children and adolescents each year and has substantial effects on development across the life course (Sedlak et al., 2010; United States Department of Education, 2019). Further, children and adolescents who have been involved in the child protective system (CPS) or are at high risk for entry into the CPS experience higher rates of abuse than those who are not at risk (Author Citation; Finkelhor et al., 2015; Pecora et al., 2009). Among the problematic outcomes associated with childhood abuse, research has noted that interpersonal problems are particularly salient. Relationships with peers are especially important in adolescence and childhood abuse has been linked to problems with peers (Lancaster et al., 2018; Bolger et al., 1998; Bolger & Patterson, 2001).

Despite research widely acknowledging the association between childhood abuse and peer relationships, there have been surprisingly few studies examining the association between childhood abuse and bullying perpetration. A recent review of existing research found only six articles relating childhood abuse to bullying perpetration, yet a consistent, positive association between childhood abuse and bullying perpetration was found (Nocetini et al., 2019). It is of significant need to examine potential pathways linking childhood abuse to bullying perpetration so that trauma informed anti-bullying programs can be established and empirically supported. Scholars have theoretically proposed that mental health and social skill problems are possible mediators of the association between childhood abuse and bullying perpetration (Hong et al., 2012); however, there is little empirical research to substantiate those claims. Existing knowledge has demonstrated that mental health and social skill problems mediate the relationship between childhood abuse and adolescent’s interpersonal relationships (Author Citation; Kim & Cicchetti, 2010; Werkerle et al., 2001), yet much less is known about bullying perpetration specifically. To fill this gap, coupled with the importance of understanding factors related to bullying perpetration, the purpose of the current study was to longitudinally investigate the associations between childhood abuse and bullying perpetration over a two-year period and examine symptoms of depression, anxiety, anger, dissociation, and posttraumatic stress disorder, as well as social skill problems as mediators using a sample of at-risk adolescents.

Childhood Abuse and Psychosocial Outcomes

Childhood abuse is a form of interpersonal trauma in which the perpetrator, generally the primary caregiver(s) of the child, inflicts or threatens to inflict harm onto the child. There are three distinct types of childhood abuse: physical, sexual, and emotional abuse. Subtypes of abuse have been strongly correlated with each other indicating that children who experience one form of abuse are more likely to experience a second (or third) type of abuse (DiLillo et al., 2007). Using a national sample of adolescents, Finkelhor et al. (2015) found that 38.1% of adolescents aged 14-17 have experienced some form of abuse or neglect. More specifically, they reported that 18.1% of adolescents had experienced physical abuse, 23.9% had reported emotional abuse, and approximately 0.2% reported sexual abuse. Moreover, rates of childhood abuse are higher among at-risk adolescents. Using a sample of adolescents from the Longitudinal Studies of Child Abuse and Neglect (LONGSCAN; Runyan et al., 1998), a recent study (Fitzgerald & Ledermann, 2020) found that, by the age of 12, 39.8% of adolescents had experienced emotional abuse, 22.2% reported physical abuse, and 13.1% reported sexual abuse.

The high prevalence rates of childhood abuse are particularly troubling due to the associations between abuse and numerous mental and relational health problems. Adolescents who have been abused are at risk for developing depression (Paredes & Calvete, 2014), anxiety (McCullough et al., 2010), substance use (Yoon et al., 2017), post-traumatic stress symptoms (Fitzgerald & Ledermann, 2020), psychological distress (Weiss et al., 2011), and aggression (Auslander et al., 2016; Zurbriggen et al., 2010). In addition to mental health problems, adolescents who have been abused often experience interpersonal problems. Research has found that adolescents commonly experience difficulties with their parents (Fitzgerald & Ledermann, 2020), peers (Shields & Cicchetti, 2001), and romantic partners (Wekerle et al., 2009). Regarding peer relationships, numerous studies have shown that childhood abuse is associated with more negative peer relationships including peer rejection (Lancaster et al., 2018; Kim & Cicchetti, 2010) and bullying perpetration (Lereya et al., 2013). Childhood abuse is a form of interpersonal victimization and both social learning and attachment theories would suggest it leave adolescents who were abused vulnerable to being aggressive towards others.

Adolescence and Bullying

Bullying perpetration is a relatively common phenomenon in adolescence; however, rates of bullying vary considerably (National Center for Education Statistics, 2016). Bullying perpetration differs from typical peer conflict in that is it defined by intentional, repeated acts of physical, verbal, or relational aggression (Nansel et al., 2001; Raskauskas & Stoltz, 2007; Wang et al., 2009). Additionally, bullying can be reactive (e.g., a defensive reaction to a perceived external threat) or proactive (e.g., intentional and directed use of coercion towards a victim). Bullying perpetration can include physical aggression (e.g., punching), verbal (e.g., name-calling), or relational (e.g., social exclusion, spreading rumors; Voisin & Hong, 2012), such as making fun of others, spreading rumors, making threats, pushing or shoving others, and destroying property, among other behaviors (United States Department of Education, 2019). Bullying perpetration tends to decrease with age beginning with gradual increases between elementary and middle school, peaking in middle school, and then decreasing throughout high school (Kretschmer et al., 2017; Marsh, 2018; Nansel et al., 2001; Zych et al., 2020). A recent study using national data from the United States found that 6.4% of adolescents self-reported engaging in bullying perpetration (LeBrun-Harris et al., 2019). Other studies have found that bullying perpetration is far more common, with percentages approximately double those of LeBrun-Harris et al. (Hemphill et al., 2012; Inchley & Currie, 2016; Shetgiri et al., 2012). Rates of bullying perpetration among adolescents who are involved in CPS tend to be much greater. For example, a recent study by Sterzing et al. (2020) found that, among 236 adolescent girls involved in child protective services, over 50% of the sample reported bullying perpetration.

In addition to the detrimental effects on victims, bullying perpetrators often face short- and long-term consequences including academic (Nansel et al., 2004; National School Safety Center, 1995), substance use (Da Silva & Martins, 2020; Nansel et al., 2004), mental health concerns (Da Silva et al., 2020; Kaltiala-Heino et al., 2000; Klomek et al., 2007; Kretschmer et al., 2017) and suicidality (Kretschmer et al., 2017; Thomas et al., 2017). Given the commonality of bullying perpetration and its detrimental effects, it is essential to understand the risk factors that may lead some adolescents to perpetrate bullying; childhood abuse may be one such factor. Researchers have identified numerous risk factors for bullying perpetration including family violence (Voisin & Hong, 2012), childhood abuse (Shields & Cicchetti, 2001), emotional problems (Shetgiri et al., 2012), status as a racial or ethnic minority (Voisin & Hong, 2012), living in poverty (Shetgiri et al., 2012), and parenting factors (Shetgiri et al., 2012). Nocetini et al. (2019) conducted a review of family-level variables associated with bullying perpetration and childhood abuse was one factor documented in their review. Although they only found six studies examining the association between childhood abuse and bullying perpetration, all six found a positive association. Notably, a longitudinal study by Bolger and Patterson (2001) found that childhood abuse predicted peer rejection and aggressive behavior for both boys and girls as reported by parents, teachers, and the children themselves. In a more recent study, Espelage et al. (2014) examined the associations between family violence and fighting among students aged 10-15 years. Using three waves of data spaced approximately six months apart, they found that family violence in wave one was associated with higher rates of getting into fights in wave two and that wave two family violence was associated with more fights in wave three. Despite these contributions, there have been few studies examining the process which childhood abuse is associated with bullying perpetration over time.

Childhood Abuse, Adolescent Mental Health and Social Skill Problems and Bullying

Indeed, childhood abuse has been associated with bullying behavior, yet the underlying process by which they are associate remain unclear. Both mental health problems as well has social skills problems may help account for the relationship between childhood abuse and bullying perpetration. Mental health problems are common among adolescents who have been abused and have been previously down to mediate the association between childhood abuse and interpersonal functioning among adolescents (Kim & Cicchetti, 2010; Werkerle et al., 2001). In relation to bullying perpetration, specific mental health problems may either increase or decrease such behavior. Conceptually, depression and dissociation are forms of hypoarousal, which is characterized by increases in numbing, social disengagement, and restricted functioning, and these characteristics are likely to be associated with less bullying perpetration. Depression is characterized by sadness, emotional dysregulation, and social withdrawal, whereas dissociation is characterized by a disconnection from thoughts, feelings, and bodily sensations. Adolescents who are depressed are likely to socially withdraw and avoid their peers and minimize interaction. The increased isolation leaves fewer opportunities for adolescents to perpetrate bullying. Likewise, adolescents who are more dissociated are emotionally disconnected from themselves and others. In response to perceived threat, adolescents who are depressed or dissociated are more likely to retreat or withdrawal rather than be aggressive.

Conversely, symptoms of anxiety, anger, and PTSD reflect hyperarousal which are symptoms associated with being on “high alert.” Research has noted that hyperarousal symptoms have been associated with conflict, aggression, and violence (Author Citation, 2021; Auslander et al., 2016; Wekerle et al., 2001). Given the importance of peer relationships to adolescents, they may often worry about their social status. Thus, worry of negative evaluations from their peers may be a risk factor for aggressive behavior (Granic, 2014). Likewise, anger, hypervigilance, and intrusive memories may preclude adolescents from having positive peer interactions and lead to aggressive behavior (Auslander et al., 2016; Hanby et al., 2012; Wolf & Foshee, 2003). Given the high rates of mental health problems among adolescents who have been abused, it has been postulated that mental health symptoms may mediate the association between childhood abuse and perpetration of bullying perpetration in adolescence (Hong et al., 2012).

Poor social skills may also be a relevant predictor of bullying perpetration and mediate the link between childhood abuse and bullying perpetration. Voisin and Hong (2012) suggest that children from homes characterized by violence and abuse demonstrate poorer social competence, leaving those children vulnerable to more hostile and aggressive behavior. Children who demonstrate poor social skills may not recognize social cues, hold negative internal representations of relationships, and have low self-esteem (Voisin & Hong, 2012). Such problems can lead to more negative interactions and fewer positive interactions with peers, subsequently impeding their ability to form friendships. Continued, unsuccessful attempts to form and maintain friendships, as well as greater social exclusion (Lancaster et al., 2018; Kim & Cicchetti, 2010), is then likely to lead to bullying perpetration.

Theoretical Background

Numerous theoretical perspectives offer ways to explain the abuse-bullying perpetration link. Attachment theory and social learning theory have emerged as two of the most prominent perspectives. Attachment theory highlights the importance of children’s early relationships with their caregivers. It is suggested that children’s early experiences in relationships, or the extent to which caregivers are responsive, attuned, loving, and available, create the foundation for future relationships (Bowlby, 1969/1982). Attachment theorists would suggest that early interactions with caregivers create internal working models, or internalized representations of themselves, others, and the world. Children who receive consistent and attuned care then come to believe they are loved, others are a source of comfort, and the world is generally a safe place. On the other hand, children who have been abused develop more negative internal working models and subsequently come to believe they are unworthy of love, others are a source of pain and distress, and the world is a dangerous place (Cicchetti et al., 1992). Internal working models are suggested to govern adolescents’ cognitive, emotional, and behavioral functioning in relationships, with more negative internal working models being associated with more aggressive behavior (Grych & Kinsgoel, 2010; Hong et al., 2012).

On the other hand, social learning theorists would suggest that children do not inherently have a drive to be aggressive and bully others, rather they learned to be aggressive. One way in which aggression is learned is through observation. Children who are abused come to believe that aggression is an acceptable way of behaving (Bandura, 1978) and family members serve as the primary source of learning and socialization. More specifically, children may learn to be aggressive through observation other family members interact which may include physical, sexual, or psychological aggression as well as hostility and volatile conflict resolution strategies. Children may also learn that aggression is acceptable when they are directly victimized by others, which can include physical, sexual, and emotional abuse. Additionally, aggressive behavior may be reinforced within the family, leading to aggression outside of the family, such as towards their peers (Akers, 2017; Duncan, 2004). Indeed, research has well-established that adolescents who are abused are more likely to be aggressive (Auslander et al., 2016).

The Present Study

The current study examined the relationship between childhood abuse and adolescent bullying perpetration with adolescent mental health outcomes and social skills serving as possible mediators using two waves of data. It was expected that childhood abuse from birth to 12 will be positively associated with adolescent mental health problems, including anxiety, depression, anger, dissociation, and PTSD symptoms, as well as poorer social skills when adolescents were 12 years old. Next, it was hypothesized that greater mental health problems and poor social skills will be positively associated with bullying perpetration two years later, when adolescents were 14, while controlling for prior bullying perpetration, gender, race, and parental income (Voisin & Hong, 2012). Lastly, we hypothesized that mental health problems and poor social skills would mediate the relationship between childhood abuse to bullying perpetration over the two-year period. The theoretical multiple mediator model linking childhood abuse to bullying perpetration through mental health problems and poor social skills can be seen in Fig. 1.

Fig. 1.

Conceptual Diagram of Childhood Abuse, Mental Health Symptoms, and Bullying Perpetration Note. Adolescent’s age is presented in parenthesis following the variable name. Control variables included adolescent race, family income, adolescent gender, and bullying perpetration at 12

Methods

The current study is a secondary analysis of data from Longitudinal Study of Child Abuse and Neglect (LONGSCAN). The LONGSCAN study (Runyan et al., 1998) is a longitudinal examination of children who were predominantly recruited from CPS in the United States, including both substantiated and unsubstantiated cases. Participants were also recruited from a health care clinic in which the children experienced inadequate growth within the first two years or had parents who were HIV-positive or drug users. Children and their families who participated in the LONGSCAN data were recruited to represent a continuum of children based on maltreatment risk, ranging from children who were at-risk for maltreatment based on sociodemographic and socioeconomic factors to children who had substantiated cases of maltreatment (see Runyan et al., 1998 for a more detailed description). The LONGSCAN study followed a cohort of children starting when they were approximately four years old until they were 18. Data collection occurred when children were approximately 4, 6, 8, 12, 14, 16, and 18. Data were collected from interviews with children, parents, teachers, and investigators. In total, 1,354 children and their families participated. Considering bullying perpetration peaks at ages 12-14, the current study utilized data from two time points when adolescents were approximately 12 and 14. Participants were included in the current study if they participated in the LONGSCAN study when adolescents were both 12 and 14 and their teachers completed reports of adolescent bullying perpetration at both waves; the analytic sample included 239 adolescents. Within the analytic sample, adolescents tended to be female (50.6%), racial and ethnic minorities including African American, Hispanic, Asian, Native American, or multiracial (61.1%), and lived in a household making less than $30,000 per year (53.8%). The use of the LONGSCAN data was approved by the second author’s institutional review board.

Measures

Childhood Abuse

Childhood abuse was assessed using measures developed for the LONGSCAN study. Adolescents self-reported their history of emotional, physical, and sexual abuse when they were 12 years old. The emotional abuse measure consisted of 18 emotionally abusive behaviors (e.g., called you names or teased you in a way that made you feel really bad about yourself) that were dichotomously coded (yes/no). The physical abuse measure consisted of 15 dichotomous (yes/no) items measuring experiences of physical abuse (e.g., being kicked or punched). Sexual abuse was measured using four items including non-contact CSA, fondling or attempted fondling, oral contact or attempted oral contact, and penetrative or attempted penetrative sexual abuse. For each form of abuse, adolescents were coded as a 1 if they responded affirmatively to any of the abuse items and they were coded as a 0 if they had not experienced that type of abuse. This resulted in dichotomous codes for emotional abuse, physical abuse, and sexual abuse. The dichotomous codes, in turn, were then summed together for an indicator of how many types of abuse adolescents had experienced (range 0-3). Higher scores indicate a greater number of types of abuse experienced (Handley et al., 2019). Cronbach’s alpha was not computed because the childhood abuse measure is based on children’s lived experiences rather than a psychological construct.

Mental Health Symptoms

Adolescent’s self-reported symptoms of depression, anxiety, anger, and dissociation utilized the Trauma Symptom Checklist (TSC; Briere, 1996). The TSC has been established as a valid and reliable measure (Briere et al., 2001). Items on the TSC were measured using a 4-point Likert-type scale ranging from “Never” (0) to “Almost all the time” (3). The TSC has six subscales: anxiety, depression, PTSD, dissociation, anger, and sexual concerns; the current study utilized the depression, anxiety, anger, dissociation, and PTSD subscales. The items from each subscale were summed together to create an overall indicator for the specific domain of mental health problems. Higher scores reflect greater symptoms severity (e.g., anxiety). Cronbach’s alpha was acceptable (above .70) for all subscales

Bullying Perpetration

Adolescent bullying was measured using five items from the Child Behavior Checklist Teacher Report Form (CBCL-TRF; Achenbach, 1991). Items on the CBCL-TRF were rated on a 3-point Likert scale ranging from 0 = not true, 1 = somewhat or sometimes true, and 2 = very true or often true. These items were “Cruelty, bullying or meanness to others,” “Destroys others' property,” “Physically attacks people,” “Teases a lot,” and “Threatens people.” Items were summed together, and higher scores reflect greater bullying perpetration. The scale demonstrated adequate internal consistency; Cronbach’s alpha was acceptable for bullying perpetration at both time points (above .70).

Social Skills

Social skills were measured using the Child Behavior Checklist – Youth Self-Report (CBCL-YSR). The social skill checklist consists of 8 items that were rated on a 3-point Likert scale including 0 (Not at all true), 1 (Sometimes true), and 2 (Very often true). Example items of the CBCL-YSR include “I don't get along with other kids,” “I am not liked by other kids,” and “I keep from getting involve with other kids.” The items were summed together and higher scores reflect poorer social skills. Cronbach’s alpha was somewhat low for this measure (above .65).

Covariates

Race

Race was coded as a dichotomous variable (Racial Minority / White).

Gender

Gender was coded as a dichotomous variable (Male / Female).

Income

Income was coded as an ordinal variable ranging from 0-$5,000 to over $50,000.

Statistical Analysis

To examine the relationship between childhood abuse, mental health symptoms, poor social skills, and bullying perpetration, structural equation modeling (SEM) was utilized. SEM compares the proposed theoretical model to the empirical data and evaluates the extent to which the theoretical model fits the empirical data. Model-data fit is commonly evaluated through several fit statistics including the comparative fit index (CFI), Tucker-Lewis index (TLI), Chi-square statistic, and root mean square error of approximation (RMSEA). CFI and TLI values greater than .90, RMSEA values below .06, and a non-significant chi-square test indicate good model-data fit (Hu & Bentler, 1999). Results of the current study indicate the model data fit was good (χ² (1) = .601, p = .44; CFI = 1, TLI = 1, RMSEA = 0). The only path not estimated was the childhood abuse to bully perpetration path (measured when adolescents were 14).

To examine the longitudinal effects of mental health symptoms and poor social skills on bullying perpetration over a two-year period, numerous covariates were used. We controlled for teacher reports of adolescent bullying perpetration at age 12 as well as adolescents’ sociodemographic characteristics including gender, race, and family income. The mediating (indirect) effects were examined utilizing bootstrapping with 5,000 bootstrapped samples. Bootstrapping yields a bias-corrected point estimate and 95% confidence interval. If the confidence interval includes 0, the indirect effect is non-significant. There was little missing data in the current sample; only 2 participants did not respond to questions about sexual abuse. All other measures had complete data and thus the missing data were considered missing at random.

Results

Preliminary Analysis

Correlations, means, and standard deviations are reported in Table 1. Regarding the prevalence rates of childhood abuse, 41.2% of children reported emotional abuse before the age of 12, 23.4% reported physical abuse, and 16.3% reported sexual abuse. Regarding the number of abuse subtypes experienced, 49% reported no childhood abuse, 28.7% reported one type of abuse, 14.8% reported experiencing two types of abuse, and 7.8% reported experiencing all three types of abuse. Regarding bullying perpetration, teachers reported that 33.1% of adolescents engaged in some form of bullying perpetration when adolescents were 14.

Table 1.

Descriptive Statistics of Study Variables Including Correlations, Means, and Standard Deviations

	1.	2.	3.	4.	5.	6.	7	M(SD)
1. Childhood Abuse	-							.80 (.94)
2. Anxiety	.33***	-						3.21 (3.63)
3. Depression	.36***	.76***	-					3.05 (3.48
4. Anger	.39***	.70***	.75***	-				3.34 (4.06)
5. Dissociation	.35***	.78***	.80***	.76***	-			3.76 (3.75)
6. PTSDa	.39***	.83***	.74***	.76***	.79***	-		4.27 (4.46)
7. Social Problems	.25***	.46***	.51***	.43***	.50***	.47***	-	3.48 (2.54)
8. Bullying Behavior	.05	.05	.06	.06	–.02	.03	.08	.91 (1.84)

*** p < .001

^aPTSD = Posttraumatic Stress Disorder Symptoms

Primary Analysis

Direct Effects

Results of the SEM model indicate the childhood abuse was associated with mental health symptoms and poor social skills (See Fig. 2). Specifically, childhood abuse was associated with greater anxiety (ß = .35, p < .001), depression (ß = .38, p < .001), anger (ß = .43, p < .001), dissociation (ß = .36, < .001), and PTSD symptoms (ß = .40, p < .001). Additionally, childhood abuse was associated with poorer social skills (ß = .25, p = .001). In terms of the effects from adolescent mental health symptoms and poor social skills to bullying perpetration two years later, anxiety (ß = .23, p = .03) and poor social skills (ß = .11, p = .049) were associated with higher levels of bully behavior, whereas depression (ß = –.12, p = .12), anger (ß = .10, p = .39), dissociation (ß = –.14, p =.16), and PTSD symptoms (ß = –.05, p =.62) were not significant. Overall, the model accounted 17.4% of adolescent’s anxiety symptoms, 27.6% of depressive symptoms, 21.6% of anger, 25% of dissociation, 14.2% of PTSD, 5.1% of poor social skills, and 26.3% of variance of bullying perpetration at 14. Among the covariates, bullying perpetration at 12 (ß = .45, p < .001) was a significant predictor of bullying perpetration two years later indicating that adolescents who bullied others at 12, also bullied others two years later. Gender was marginally significant in predicting bullying perpetration where males were more likely to engage in bullying perpetration compared to females (ß = –.14, p = .05); race (ß = .09, p = .25) and income were not significant predictors of bullying perpetration (ß = –.03, p = .65).

Fig. 2.

Results of the Structural Equation Model Linking Childhood Abuse to Bullying Perpetration Through Mental Health Symptoms, and Poor social skills Note. Bolded effects are significant. Ages of adolescents are presented in parenthesis following the variable name. PTSD = Posttraumatic Stress Disorder. Control variables included adolescent race, family income, adolescent gender, and bullying perpetration at 12

Indirect Effects

To examine the potential mediating effects of mental health symptoms and poor social skills, we used bootstrapping procedures to calculate indirect effects (See Table 2). Childhood abuse was indirectly associated with bullying perpetration through anxiety symptoms, such that more severe anxiety symptoms resulting from more types of abuse experienced then increased bullying perpetration (β = .08, 95% CI [.016, .187]). Likewise, adolescents who reported a greater number of abuse types also reported poorer social skills, which was then associated with more bullying perpetration (β = .02, 95% CI [.002, .068]). On the other hand, childhood abuse was not associated with adolescents’ bullying perpetration through depression (β = –.07, 95% CI [–.184 .004]), dissociation (β = –.05, 95% CI [–.150, .009]), anger (β = .04, 95% CI [–.050, .153]), or PTSD (β = –.02, 95% CI [–.106, .066]).

Table 2.

Results of the Indirect Effects Linking Childhood Abuse to Bullying Perpetration Over a 2-Year Period

Indirect Effect	Estimate	95% CI
Childhood Abuse -> Anxiety-> Bullying Perpetration	.08	[.016, .187]
Childhood Abuse -> Depression-> Bullying Perpetration	–.07	[–.184 .004]
Childhood Abuse -> Anger-> Bullying Perpetration	.04	[–.050, .153]
Childhood Abuse -> Dissociation-> Bullying Perpetration	–.05	[–.150, .009]
Childhood Abuse -> PTSDa-> Bullying Perpetration	–.02	[–.106, .066]
Childhood Abuse -> Social Skills -> Bullying Perpetration	.02	[.002, .068]

Bolded effects are significant

^aPTSD = Posttraumatic Stress Disorder Symptoms

Discussion

The purpose of the current study was to examine adolescent mental health and social skill problems as possible pathways through which childhood abuse is associated with bullying perpetration over a two-year period. We used a sample of at-risk adolescents, which is a subgroup of adolescents who demonstrate disproportionately high rates of both childhood abuse and bullying perpetration, to determine if mental health and social skills mediate the association between childhood abuse and bullying perpetration. We found partial support for our hypothesis. Although childhood abuse was associated with each of the mental health problems, only anxiety and poor social skills were identified as significant mediators linking childhood abuse and bullying perpetration over a two year period.

The primary contribution of the current study is documenting that adolescent anxiety and poor social skills are longitudinal risk factors for bullying perpetration among children who have been abused. This is congruent with prior studies which have found that children who are abused struggle with peer relationships (Lancaster et al., 2018; Shields & Cicchetti, 2001) and have more aggressive and conflictual relationships with others (Auslander et al., 2016; Wekerle et al., 2001). Not surprisingly, we found that bullying perpetration is quite common in the current sample, finding that 33.1% of adolescents engaged in bullying perpetration. These numbers are significantly higher than many studies, including national samples from across the United States (Hemphill et al., 2012; LeBrun-Harris et al., 2019; Shetgiri et al., 2012), but are somewhat lower than other studies with adolescents involved in child protective services (Sterzing et al., 2020). Differences in prevalence rates between our study and those of Sterzing and colleagues could be attributed to our sample using both male and female adolescents and the use of teacher reports compared to their use of only female adolescents and self-report bullying perpetration. These findings indicate that at-risk adolescents engage in more bullying perpetration compared to those in the general population; however, the methodological differences may account for a lower percentage compared to other studies on at-risk adolescents.

Partially supporting our hypothesis, which was that mental health and poor social skills would mediate childhood abuse to bullying perpetration, findings of our study indicate that anxiety is one potential pathway. Adolescents who experienced a greater number of childhood abuse subtypes reported more severe anxiety symptoms in early adolescence which, in turn, was associated with higher levels of teacher-reported bullying perpetration. In addition to informing the well-established associations between childhood abuse and adolescent anxiety (Fitzgerald, 2021; Bomysoad & Francis, 2020), the results of the current study suggest that there may be significant implications for bullying perpetration (Granic, 2014). Childhood abuse and anxiety have been associated information processing problems which poses significant problems for successful navigating peer relationships (Teicher & Samson, 2016). That is, children who were abused are hypervigilant of potential threats from peers and interpret benign or innocuous stimuli as a threat, thereby increasing state-level anxiety. To alleviate the increases in anxiety, adolescents engage in aggressive behavior in an attempt to protect themselves from additional victimization. Additionally, dolescents with a history of abuse and high anxiety may come to expect rejection, hostility, or aggressive behavior from others (Gibb et al., 2009). Granic (2014) proposed that unpredictable parenting, including childhood abuse, increases anxiety in children, leaving them vulnerable to aggressive behavior. More precisely, children with anxiety use their available psychological resources to override aggressive impulses; however, children who have consistently high levels of anxiety become less able to regulate aggressive impulses and when their psychological resources are depleted aggressive behavior occurs. In this way, engaging in aggressive behavior serves to decrease or regulate anxiety (Granic, 2014).

As expected, we also found that adolescents who were abused and experienced subsequent poorer social skills were also at an enhanced likelihood of engaging in bullying perpetration. From an attachment perspective, adolescents who have been abused tend to have negative internalized representations of relationships (e.g., internal working models) and may subsequently approach relationships with suspicion, anger, and mistrust (Greenberg et al., 1993; Webster & Hackett, 2011). There are several consequences to negative internal working models, including greater emotional dysregulation and higher levels of behavioral and emotional problems that may make it difficult to form strong peer relationships and increase peer rejection (Lancaster et al., 2018; Kim & Cicchetti, 2010). Adolescents may not be accepted by their peers if they engage in deviant behavior or struggle to effectively resolve conflict. Adolescent’s inability to effectively regulate their emotions and successfully navigate the increasingly complex nature of peer relationships may leave adolescents at risk for engaging in bullying perpetration (Hong et al., 2012).

In contrast to our expectations, our findings also suggest that depression, dissociation, anger, and PTSD symptoms did not mediate the relationship between childhood abuse and bullying perpetration. Although we found that childhood abuse was associated with higher levels of depression, dissociation, anger, and PTSD, these mental health problems were not associated with bullying perpetration. It was surprising to find that anger and PTSD were not associated with bullying perpetration, as they have been previously linked to aggressive behavior (Auslander et al., 2016; Maneta et al., 2012). Anger may not be associated with bullying perpetration over time because it covaries with prior bullying perpetration and may account for the same variance in future bullying perpetration. We also found that PTSD did not predict more bullying perpetration. In contrast to the current study, Auslander et al. (2016) found that PTSD mediated the relationship between childhood abuse and aggression. One explanation is that their study was cross-sectional, and the current investigation utilized longitudinal data where we controlled from prior levels of bullying. These findings suggest that PTSD does not contribute to aggressive behavior when considering prior levels of aggression. A second difference is that Auslander and colleagues utilized self-reports of physical, psychological, and relational aggression, whereas the current study utilized teacher reports of bullying perpetration and only included psychological and physical behavior. Research has noted that teachers may not be aware of all of adolescent’s bullying perpetration (Bradshaw et al., 2007) and, in actuality, PTSD may influence adolescents’ bullying perpetration over time. Future research is needed to determine if this interpretation is valid.

We also investigated internalizing processes such as depression and dissociation as possible mediators. Although neither depression nor dissociation were significant, both were negatively associated with bully perpetration which was in the expected direction. Depression and dissociation are characterized by social withdrawal and being emotionally disconnected from inner and outer experiences. Thus, depression and dissociation may lead adolescents to withdraw from their peers rather than engage with them, thereby decreasing the likelihood of bullying others. On the other hand, it may be that dissociation and depression may be more strongly related to bullying victimization (Hong et al., 2012). Adolescents who are bullied may develop depression and dissociation in response to victimization. Thus, adolescents who report high levels of depression and dissociation may be victims, but not perpetrators, of bullying perpetration.

Limitations and Future Directions

Despite the strengths of our study, which include a multi-informant, longitudinal design using a sample of racially diverse adolescents, our study was not without limitations. First, we did not examine gender differences due to issues related to statistical power. There are established associations between gender and bullying perpetration in adolescence, with boys being more likely to perpetrate bullying and be involved in physical or verbal bullying and girls being more likely to be involved in relational bullying (Álvarez-García et al., 2015; Wang et al., 2009). Future studies may want to examine gender differences for bullying perpetration in adolescents with histories of abuse. Second, when measuring bullying perpetration, we did not differentiate the specific forms of bullying; rather, we measured general bullying perpetration. Future research should address which type of bullying (e.g., physical, psychological, relational) abused adolescents may be more or less prone to engage in (Shetgiri, 2013). Furthermore, we controlled for prior bullying perpetration, but did not control for prior or simultaneous bullying victimization. Bully-victims, or adolescents who are both perpetrators and victims of bullying, will likely have experiences that differ from those of the exclusive perpetrator or victim. For example, Sterzing et al. (2020) found that bullying perpetration was far more common in the context of bidirectional bullying, or being both a victim and a perpetrator. Future research should consider the role of adolescent mental health and social skills in the context of both bullying perpetration and victimization.

Another limitation is that we utilized teacher reports rather than adolescent self-reports or peer nominations to measure bullying perpetration. Previous research indicates teachers and school staff underestimate the frequency of bullying in elementary, middle, and high schools (Bradshaw et al., 2007), therefore it is likely that our study underestimated the prevalence of bullying perpetration and there may be stronger associations between mental health, poor social skills, and bullying perpetration. Similarly, the current study’s sample was recruited mainly from CPS and health clinics; thus, this study should be replicated in the general student population to obtain data on the experiences of abuse and bullying perpetration of adolescents from different backgrounds. Further, our study was limited in that we examined the number of abuse subtypes experienced (Handley et al., 2019), so the specific effects of emotional, physical, and sexual abuse on mental health, social skills, and bullying perpetration cannot be discerned. Another limitation is that our measure of social skills reported a Cronbach’s alpha just below the commonly accepted cutoff of .70.

Conclusion and Clinical Implications

Results of the study indicate that anxiety and poor social skills link childhood abuse to bullying perpetration. These results may have important implications for clinical practice and school-based anti-bullying programs. First, it is important that clinicians, particularly school counselors, assess for childhood abuse in adolescents, especially those who demonstrate mental health and poor social skills. Second, recent research has shown that bullying perpetration may have a bidirectional relationship with mental health concerns (Da Silva et al., 2020); thus, school-based interventions should aid adolescents in developing strategies to address negative feelings and mental health concerns as a means to lessen bullying perpetration and mitigate mental health challenges. Additionally, clinicians working with adolescents who have poor social skills and more severe anxiety should inquire about their relationships with their peers in an attempt to reduce future bullying perpetration. It is possible that adolescent anxiety may be related to their peer relationships and helping reduce anxiety may lead to the cultivation of stronger peer relationships.

More generally, when implementing anti-bullying programs, schools should strongly consider the role trauma and mental health may have on bullying perpetration and adopt trauma-informed practices. For example, mindfulness-based practices may be particularly effective. Mindfulness interventions have been shown to decrease adolescent mental health problems among traumatized adolescents (Sibinga et al., 2016) and, consequently, scholars have argued that mindfulness interventions are likely to be effective for reducing bullying perpetration (Foody & Samara, 2018). Further, social skill interventions (e.g., making friends, expressing emotions, handling conflict) can be a crucial intervention point for clinicians, teachers, and school staff in targeting bullying perpetration (Da Silva et al., 2018). Social skill interventions may help adolescents adopt the strategies needed for positive peer engagement that may be essential in reducing bullying perpetration. In all, the results of our study highlight the relationships between childhood abuse and mental health, social skills, and bullying perpetration in adolescence. These findings show promise in identifying mental health, specifically anxiety, and social skills as possible points for intervening and preventing bullying in adolescence.

Declarations

Conflict of Interest

The authors declare no conflict of interest in the publication of the current manuscript.