|Home | About | Journals | Submit | Contact Us | Français|
Our objectives with this study were to describe the prevalence of bullying involvement (ie, bullying and victimization) among children from a multigenerational study and to examine the relationship of these childhood behaviors and exposure to intimate partner violence.
A community-based cohort of 112 children (aged 6 to 13 years) was asked to self-report on physical, verbal, and relational types of bullying and victimization experienced in the past year. Parents reported on their child’s externalizing and internalizing behaviors during the previous 6 months using items from Achenbach’s Child Behavior Checklist. The frequency of parental experiences of intimate partner violence perpetration and victimization at 2 time points during the preceding 5 years was measured using Conflict Tactics Scale items. The association of intimate partner violence and parent-reported child behavioral problems was examined, followed by exposure to intimate partner violence and child-reported bullying or victimization. Parental risk factors (eg, race/ethnicity, education, problem drinking) that predispose to intimate partner violence were controlled for using propensity score statistical modeling.
Eighty-two (73.2%) children reported being victimized by peers, and 38 (33.9%) children reported bullying behaviors in the past year. More reports came from girls than from boys (55% for victimization and 61% for bullying). Almost all (97%) child bullies were also victims themselves. Intimate partner violence was reported by parent respondents in 53 (50.5%) households at any or both of the 2 time points. Exposure to intimate partner violence was not associated with child-reported relational bullying behaviors or victimization by peers, However, intimate partner violence–exposed children were at increased risk for problematic levels of externalizing behavior/physical aggression and internalizing behaviors.
In our sample, children who were 6 to 13 years of age reported a substantial amount of bullying and victimization; a large majority were bully-victims and female. Regression analyses did not show that children who were exposed to intimate partner violence were more likely to engage in relational bullying. However, children who are exposed to intimate partner violence have a higher likelihood of internalizing behaviors and physical aggression.
Childhood bullying and victimization are serious problems that can threaten a child’s socioemotional development. Bullying is defined as conscious, repeated acts of physical, verbal (eg, name-calling), or relational (eg, social exclusion, spreading rumors) aggression that causes injury or discomfort to the target1 between children of differing physical size and strength.2 Bullies, victims, and bully-victims are at risk for a variety of psychological, peer, and school problems.1,3,4 Victims also are at increased risk for suicide5 and school violence.6
Bullying behaviors can emerge as early as elementary school age and usually peak during middle school. A large US study found that 11% of children in grades 6 to 10 bullied others “sometimes,” with an additional 9% bullying more frequently.3 Former victims were more likely to have poorer self-esteem and experience depression at age 23; likewise, 70% of bullies who were identified in seventh and ninth grades were involved in criminal activity by age 24,7 yet little is known about risk factors that may predispose a child toward frequent bullying or victimization. Bullies tend to experience inconsistent authoritarian parenting styles and exhibit impulsive tendencies.8 Identification of predisposing factors in the home and environment (eg, school climate) can lead to timely identification of at-risk children and provide the basis for targeted interventions.
One possible risk factor for bullying is exposure to intimate partner violence (IPV). It is estimated that between 3.3 and 10 million children in the United States witness IPV in their homes annually.9 On the basis of theories of social learning10 and emotional dysregulation,11 children who are exposed to IPV in their homes can be at particular risk for learning negative relationship patterns. Through their early experiences with primary caregivers and siblings, children learn rules of relationships and begin to construct their views of the world. IPV is defined as physical, emotional, or sexual acts of aggression (actual or perceived) between 2 partners (eg, marital or nonmarital, current or past) that occur repetitively with the intent to harm.12 Exposure to IPV can influence a child’s perception of violence as an acceptable method of resolving conflict.
Children who are exposed to IPV exhibit both internalizing and externalizing behavior problems in the borderline to clinical range,13–15 yet little is known about the influence that witnessing IPV, either directly or indirectly, has on children’s peer relationships. Specifically, little is known about the impact of IPV on a child’s likelihood of becoming a bully or bully-victim. Given the paucity of research in this area, we examined the prevalence of bullying and victimization among children who were drawn from a longitudinal, multi-generational, community-based sample. In addition, we examined the relationship between exposure to IPV and subsequent bullying and victimization. Here, we differentiated between physical forms of bullying (eg, externalizing behavior/physical aggression) and those that are more relational in nature. We also examined the association between IPV and internalizing behaviors in children.
Our study was organized in 2 parts. First, we examined the relationship of IPV and externalizing and internalizing behaviors in children. We hypothesized that children who were exposed to IPV would exhibit problematic behaviors at the upper quartile. We then examined whether IPV exposure increases the risk for relational bullying as a stand-alone outcome.
Our study used data from 2 closely related longitudinal studies with first- (G1), second- (G2), and third-generation (G3) participants. We used a cohort of children (G3) who were between the ages of 6 and 13 years and living in Seattle, Washington. Parents (G2) of the selected children originally were enrolled in the Seattle Social Development Project (SSDP) in 1985 on entering the fifth grade. There were 808 participants from 18 Seattle public elementary schools that served diverse neighborhoods, including households (G1) within high-crime areas that consisted of lower socioeconomic status and mixed ethnicity/race,16 representing 77% of all fifth graders from these 18 schools. As G2 participants began having their own families, eligible parents (G2) who agreed to participate (N = 208) and their eldest biological child (G3) were enrolled in another longitudinal study (SSDP Intergenerational Study) in 2002. Recruited G2 parents (N = 208) did not differ from those who were eligible but not recruited (N = 73) in terms of gender; childhood neighborhood disorganization; childhood poverty; adolescent problem behavior; cigarette use or marijuana use in adolescence; binge drinking, cigarette use, or marijuana use at ages 21 to 24; educational attainment at age 24; marital status at age 27; or G1 binge drinking, cigarette use, or marijuana use.17 Of the 208 parents and children, 89 children were younger than 6 years and were unable to participate in the child survey. Seven children did not have complete data. Our data were derived from this sample of 112 G2 participants and their 6- to 13-year-old children. Study procedures were approved by the University of Washington Human Subjects Protection Committee.
We created 4 scales to measure more generalized acts of aggression, as well as other school-related and peer-related behaviors, using selected items from Achenbach’s Child Behavior Checklist (CBCL/6–18).18 A subset of CBCL items that have been shown to be reliable and valid indicators of Diagnostic and Statistical Manual of Mental Disorders diagnostic categories were used in the intergenerational study. Items that were used in these scales correspond with those that were used in various other studies19 based on original CBCL constructs, although we modified our scales to reduce overlap in the items and to capture the most salient characteristics of each behavior.20 Primary parent respondents were asked to rate how true each statement was for their child in the previous 6 months on a 3-point scale: not true, somewhat or sometimes true, or very or often true. In total, we used 13 items (Cronbach’s α = .812) to capture childhood externalizing behaviors; 3 items were used to capture internalizing behaviors (Cronbach’s α = .623). Other behavioral problems, such as attention (4 items; Cronbach’s α = .988) and social problems (3 items; Cronbach’s α = .603), were also included. Ratings for each child on each of the 4 subscales were computed. To isolate those who were at highest risk on each behavior, we separated cases in the highest 25% of each score distribution from those in the remaining 75%. This procedure was used in other studies from SSDP and elsewhere as an alternative to modeling highly skewed variables as continuous indicators.21–23
Bullying and relational aggression were measured with items from the revised Olweus Bully/Victim Questionnaire.1,24 Children were asked to report incidents of bullying and victimization that occurred in the past year. Children of age 6 to 9 years responded “yes” or “no” to a global indicator of bullying (“have you bullied others?”) and 4 specific indicators of relational bullying (“started rumors or told lies,” “teased others,” “did not let someone in their group of friends,” and “told someone they were not liked unless they did something I wanted”). Victimization experiences were assessed with similar measures (1 global indicator: “others have bullied me”; and 4 specific behaviors: “others started rumors or told lies about me,” “others teased me,” “someone told me I was not in their group of friends,” and “someone told me they wouldn’t like me unless I did something they wanted”). When a child responded “yes” to 1 or more of the indicators, he or she was classified as a bully or a victim, respectively. This method sought to capture children who perpetrate or experience bullying in any or multiple forms.
In addition to the above indicators, children who were 10 years and older responded to an additional question from the Olweus Bully/Victim Questionnaire on the experience of racism (“called someone names that made fun of his or her race” and “I was called names that made fun of my race”). The coding of the item was done similarly to other indicators of bullying. Response categories for children who were 10 years and older were based on frequency of incidents in the past 12 months: none, once or twice, sometimes, fairly often or almost always.
Parental history of IPV perpetration or victimization was measured using 3 items from the Conflict Tactics Scale25 that measured verbal/relational aggression and physical violence tactics that were used between partners to resolve conflict (“pushed/grabbed/slapped/shoved,” “threatened to hit,” and “insulted/swore/cursed/yelled”). Parent respondents were asked to rate the frequency of incidents in the past year that occurred between themselves and their partner at 2 discrete time points (parent age 24 and parent age 27). Modified response scale categories were: never, rarely, sometimes, often, and very often. The cutoff of “sometimes” or greater was used to denote a positive response of IPV because we believed that it captured the recurrent nature of aggression often characteristic of violent intimate adult relationships. When a positive response was found for any of the 3 indicators for either perpetration or victimization, the respondent was classified as such.
Variables that are known to be highly correlated with IPV were examined and used in the statistical analysis26–28: maternal age at child birth (age 27), highest educational level completed (age 21), race/ethnicity (age 27), participation in Aid to Families with Dependent Children/Temporary Assistance to Needy Families/food stamps welfare programs (age 27), parental childhood history of home violence (age 21), alcohol use (age 24 and 27), and overall drug use (age 24). Alcohol use was assessed using 2 indicators: average quantity/frequency of drinking per week and binge drinking. The cutoff chosen was based on the World Health Organization standards for medium levels of regular drinking (average ≥4 glasses a day in a 1-week period for men and ≥2 glasses a day in a 1-week period for women). A composite variable denoting problem drinkers was defined as a positive response at either time point for either of the 2 indicators.
We examined the effects of IPV on children’s externalizing and internalizing behaviors to consider how our sample compared with previous studies.13–15,29 Bivariate analyses were conducted using each of the 4 constructs for measuring problem behaviors and IPV at each time point and combined.
For examination of the association between IPV and child problem behaviors, a composite variable for each time point (ages 24 and 27) was needed to denote a positive report of IPV perpetration and/or victimization. The 2 composite variables then were used to create a final composite variable to represent any report of IPV at either or both time points. The selected covariates were placed into the first part of the statistical model with IPV as the outcome. Predicted values were saved and used in the second step to represent a propensity for IPV. In the second step, the model was fitted with both the propensity score and the composite IPV variable with 1 of the parent-reported problem behavior constructs. Logistic regression then was performed using the fitted model. Using a propensity score for regression adjustment30 allowed us to control parsimoniously for parental risk factors that are known to be correlated with IPV. We repeated this procedure for each of the 4 behavior constructs and IPV.
We performed bivariate analyses of child-reported bullying involvement and IPV at each time point and combined time points. Frequencies of bullying and victimization were examined for the overall sample, as well as by age (ages 6–9 and 10–13) and gender.
For examination of the relationship between parental reports of IPV and child bullying and victimization, a separate, 2-step propensity score model was fit for each child outcome. Logistic regression was then performed using the fitted models for bullying and victimization.
The parent respondents in the current study predominantly were female (78.6%), with a median age of 27 years (Table 1). Half were single and never married, 56% reported incomes less than $31 000 a year, and almost one third (31.6%) were enrolled in programs such as Temporary Assistance to Needy Families or Aid to Families with Dependent Children in the past 12 months. One quarter of parents were younger than 18 years at childbirth. Slightly more than half (52.7%) of children in the final cohort were female.
A total of 42 (42.9%) of 98 households reported any form of IPV at parent age 24. This rate was similar at parent age 27: 41 (48.8%) of 84. Missing data required exclusion of 14 participants at age 24 and 28 participants at age 27 time points. Overlap of parental violence perpetrators and victims was high, 54.8% and 73.2% at parent age 24 and age 27 time points, respectively. IPV was reported in 53 (50.5%) households for at least 1 of the 2 time points with 7 excluded for missing data.
Overall, one third (33.9%) of children in our sample reported bullying others in the past year (Table 2). Prevalence of bullying was 34.6% for children who were 6 to 9 years of age and 32.4% for those who were 10 to 13 years of age. Only 5 (4.5%) children in our sample responded positively to “I bullied others,” yet positive responses to queries about specific acts of bullying led to a much higher overall prevalence of bullying. Girls had a higher prevalence of bullying for each indicator than did boys.
Victimization by bullies was a common experience among our sample, reported by 73.2% of children (Table 3). The prevalence of victimization was 78.2% for children who were younger than 10 years and 61.8% for those who were older. Similar to the indicators for bullying, a discrepancy existed between the reports of being bullied (32.1%) and specific questions about victimization. As with bullying, girls were more likely to report victimization than were boys.
Bullies who also are bullied have been previously defined and referred to as “bully-victims.”3,31 Of the 38 child bullies, 37 (97.4%) reported concomitant incidents of victimization by peers. Conversely, 45.1% of victims also reported being bullies.
Of the 112 parent respondents, 22 were not the child’s primary caregiver and therefore were excluded. The unadjusted association of IPV at parent age 27 with externalizing (aggressive) behaviors revealed a relative risk (RR) of 5.2 (95% confidence interval [CI]: 1.6–16.1); with internalizing (withdrawn) behaviors, RR of 1.2 (95% CI: 0.5–2.6); with attention problems, RR of 1.6 (95% CI: 0.7–3.7); and with social problems, RR of 1.9 (95% CI: 1.0–3.7).
Logistic regression using propensity score modeling with each secondary dependent variable showed a statistically significant association between IPV and child’s externalizing behaviors (odds ratio [OR]: 3.1; 95% CI: 1.0–9.5; Table 4). There were similar trends for IPV and internalizing behaviors (OR: 1.6; 95% CI: 0.5–4.6) and attention (OR: 2.2; 95% CI: 0.7–6.6). We did not find a statistically significant association with parent-reported social problems and IPV (OR: 1.0; 95% CI: 0.4–2.6). Analyses using IPV reports at parent age 24 and combined time point revealed similar results.
In unadjusted analysis, the risk for child bullying was increased in those who were exposed to IPV that was reported by the parent at age 27, although the CI included 1 (RR: 1.31; 95% CI: 0.7–2.5). Similarly, the risk for bullying victimization was higher in those whose parents reported IPV at age 27, although the CI included 1 (RR: 1.24; 95% CI: 0.9–1.6). Analysis with parent IPV reports at age 24 and combined reports at both time points yielded similar results. We did not find an association between parental IPV and child-reported bullying (OR: 0.7; 95% CI: 0.3–1.8) or victimization by peers (OR: 0.9; 95% CI: 0.4–2.3) after adjusting for our selected covariates using propensity score analysis (Table 5).
Bullying others and being victimized by peers were highly prevalent in this sample of children. Girls in our sample were more likely to report bullying others and engaged more frequently in relational aggressive acts when compared with boys. When parents were asked to report on their child’s behaviors, children who were exposed to IPV displayed a higher rate of both physically aggressive and internalizing behaviors, when compared with children whose parents did not have any history of IPV. Exposed children were more likely to display problems with attention.
Involvement in bullying is a common phenomenon during school-age years.3,7,32,33 Our overall reported prevalence of bullying is higher than national estimates reported by Nansel et al34 in 2002, most likely for 3 reasons. First, our study included children who were 6 to 13 years of age, whereas previous studies focused on middle school–aged students.29,31,35 Little is known about the incidence of bullying in children who are younger than 9 years. Schools across the country are of varying grade compositions and types (eg, alternative, K–8, K–12), making it harder to isolate the influence of student age on the effect on bullying involvement and overall school climate. In addition, the bullying indicators that were used in our study emphasized relational bullying behavior, whereas previous studies have used both relational and physical measures. A subset of relational type behaviors are covert (eg, spreading rumors, social exclusion) and were recently shown to be harder to detect by both teachers and parents.36,37 Children are less likely to report incidents of relational aggression when compared with direct physical or direct verbal incidents.35 Therefore, prevalence of bullying is highly dependent on the behaviors studied and how questions about bullying are posed. Thirdly, our cohort resided in diverse neighborhoods, with a large proportion of households of lower socioeconomic status, mixed ethnicity, and single teen parents. Other studies found that these types of environmental factors are associated with an increased prevalence of bullying behaviors.4,38
Our rates of IPV and female-to-male family violence are similar to other longitudinal community-based samples.37 We recognize that these findings have been challenged because the Conflict Tactics Scale39 does not elicit information about relationship context, motivation of the perpetrator, or potential of injury to the victim. However, others have confirmed female individuals can be aggressive, especially within the context of a relationship,40,41 using moderately severe forms of aggression as measured by Conflict Tactic Scale items.
Previous studies reported substantial consequences for children (witnesses and victims) who live in homes with IPV. Children experience a higher number of health problems, such as asthma, gastrointestinal complaints, headaches, and cold or flu,29,42 and are at risk for development of posttraumatic stress among children who are 6 to 12 years of age.43 Child victims of physical and sexual abuse are more likely to engage in bullying or be victimized by peers44 and have more emotional problems over time when abuse is long term. Our results are consistent with previous work showing that children from violent homes exhibit high rates of externalizing behavior problems and total behavioral problems.13–15,45
Our study is the first in the United States to examine the association of child witnesses to IPV and bullying involvement specifically. We attempted to capture the different facets of bullying and examine its association with IPV through our 2-part analysis. A previous Italian study found an association of IPV and bullying among Italian elementary and middle school students46 but was cross-sectional and relied on children reporting bullying incidents in the previous 3 months and whether their parents ever engaged in domestic violence. Our sample was asked to report on bullying and domestic violence incidents in the past year. We did not find the hypothesized association between IPV exposure and relational bullying or bullying victimization. Rather, we found that children who were exposed to IPV engaged in higher levels of generalized aggression as measured by specific items that often are viewed and used to represent physical bullying from the aggression subscale of Achenbach’s CBCL.47 Our findings are important in light of the work of Veenstra et al38 that used peer ratings to classify children as bullies, bully-victims, victims, or uninvolved in bullying. Teachers reported bullies and bully-victims as having higher levels of aggression when compared with other students who were thought of as victims or uninvolved in bullying. This highlights the correlation between peers’ perception of bullies in the classroom and adult ratings of highly aggressive children who use physical means to overpower others. Our results, having the advantage of a longitudinal study design, indicate that IPV exposure is more related to physical, more aggressive, acts of bullying than to relational acts of bullying.
There are several limitations to the study. Our sample size was small compared with previous studies on bullying. The measurement of bullying is imprecise and not yet standardized, although there is general consensus of specific features that constitute bullying. There are very few instruments that measure bullying specifically.48 The bullying measures that were available in our study represented examples of relational bullying and therefore may elicit greater reporting of relational bullying and victimization than of physical bullying and victimization. In addition, child respondents were not given a definition of what constitutes bullying as did previous researchers.1,3,29,35,49 Another limitation is sole reliance on child self-reporting of bullying behavior. Children who are exposed to IPV may assimilate “acceptable” relationship patterns that are learned in the home and thereby do not recognize their own negative peer interaction style (eg, bullying involvement). Moreover, children involved in bullying exhibit differential types of aggression (eg, reactive versus proactive) depending on the situation and whether they are mainly bullies, victims, or bully-victims.50 Our measures did not capture children’s perceptions of peer intentions or ascertain specifics on the circumstances surrounding bullying incidents or whether there were any triggers. Given that we found an association between IPV and parent-reported externalizing and internalizing child behaviors similar to past studies, the systematic measurement of bullying may be problematic, especially in those who come from violent homes. Lastly, possible residual confounding by factors we did not examine, such as exposure to general community violence, child maltreatment, and harsh parenting styles, could bias the result toward the null.
Exposure to IPV in childhood can have both immediate and far-reaching effects on a child’s development. The importance of identifying significant psychosocial factors, both acute and chronic, cannot be understated. Children rely on parents to model socially acceptable ways of behavior; therefore, households with IPV pose a significant risk to a child’s socioemotional development. Children who are exposed to IPV have a higher risk for displaying physically aggressive acts of bullying (eg, pushing or shoving others, fighting) but are not more likely to engage in relational bullying. Traditional instruments that are used to capture bullying should separate relational and physical acts of bullying because of gender trends. Large longitudinal studies are needed to evaluate the full extent of IPV on a child’s socioemotional development and peer relations in those who witness violence but are themselves not victims.
This study was supported by grants R01DA12138 and R01DA09679 from the Health Resources and Service Administration, National Institute of Drug Abuse, and a National Research Service Award Primary Care Research Fellowship.
We acknowledge Jennifer Bailey, PhD, and Sabrina Oesterle, PhD, Social Development Research Group, and James F. Lymp, PhD, Child Health Institute, for consultation on statistical methods.
The authors have indicated they have no financial relationships relevant to this article to disclose.