Abstract
Abstract
Background:
Child and adolescent overweight and obesity are independent risk factors for poor social outcomes. Whether children who are overweight display greater bullying behaviors than normal weight peers, controlling for demographic and other social-ecological factors, was examined. The influence of child (e.g., mental health), family (e.g., income, parent mental health, and exposure to domestic violence), and community (e.g., exposure to neighborhood violence, unfair treatment based on race or ethnic group) factors on bullying risk in the subset of children who are overweight and obese was explored.
Methods:
We conducted a secondary data analysis of the 2011–2012 National Survey of Children's Health, a cross-sectional study providing a nationally representative sample of participants, using a series of multinomial logistic regressions in Mplus software (Muthén & Muthén, Los Angeles, CA). Participants were 41,361 youth ages 10–17 years.
Results:
Despite statistically significant differences in unadjusted analyses, no differences were found in bullying behavior by weight status once controlling for other factors. Child, family, and neighborhood factors predicted bullying behavior in both the overall sample and the subset of overweight and obese youth. However, some risk factors were unique to children who were overweight or obese.
Conclusions:
Children who are overweight or obese are not at greater risk for engaging in bullying behavior than normal weight peers. Health professionals targeting bullying behavior should be aware of the impact child, family, and neighborhood factors have on bullying by overweight and obese children and adolescents.
Introduction
Bullying is a common poor psychosocial outcome among children and adolescents. 1 Bullying has been described as intentional and repetitious interpersonal aggression characterized by an imbalance of power between the perpetrator and victim. 2 Many estimates suggest that 10–33% of youth are the victims of peer bullying, 2 with some estimates suggesting as many as half of adolescent males and one third of adolescent females have been victims of bullying at school. 1 Overweight and obese youth are at increased risk of being the victims of bullying and peer victimization, compared to normal weight peers, and are also at increased risk of exposure to interpersonal violence.3–7 Further, overweight and obese youth may also be at increased risk of becoming perpetrators of peer bullying themselves.4,8 However, the increased risk of bullying behaviors among overweight and obese youth has been found to vary based on sex and age. 8 For example, weight status has been related to increased bully-perpetrating behaviors in younger children 4 and some ages of adolescents, 8 whereas other research suggests that adolescents may not be at increased odds of engaging in bully-perpetrating behaviors. 8 Several studies have demonstrated sex differences in engagement in bullying behaviors, with some studies indicating a relationship between bullying and weight status only in boys. 4 However, the finding of increased bully-perpetrating behavior among overweight and obese youth has not always been supported. 9
A social-ecological diathesis-stress model has been supported for explaining the complexity of bullying behaviors, with individual, family, peer, school, and community factors all influencing bullying behavior. 10 This model posits that bullying behavior is not simply the result of individual youth characteristics, but also the interaction between the individual youth characteristics and aspects of the youth's social context and environment (home, neighborhood, school, society, and community 10 ). Regarding individual youth characteristics, past research found that child mental health predicts bullying behaviors.11–13 For example, school-aged children with attention deficit hyperactivity disorder (ADHD) are almost 4 times more likely to bully other children, and behavior problems at school entry predict bully-perpetrating behaviors 3 years later. 12 Further, difficulties with emotional dysregulation have been related to bullying behavior in at-risk children, with emotional dysregulation related to greater bullying of peers. 14 Family factors, including parental health, parental marital relationship, family structure (e.g., no father in the home), parent-child relationship, family teasing, lack of supervision, maltreatment, and exposure to domestic violence, have been related to child conduct problems, delinquency, or bullying behavior.15–17 Additionally, community factors, such as children's exposure to violence in the community and neighborhood safety concerns, have also been related to aggressive behavior and bullying other children.18–20
The current study seeks to examine whether overweight and obese youth are more likely to engage in bully-perpetrating behaviors than normal weight peers in a nationally representative sample of US children and adolescents. Additionally, consistent with the social-ecological diathesis-stress model of bullying behavior, individual, family, and community risk factors for bullying behavior among overweight and obese youth were examined. This was done by assessing the following specific hypotheses:
1. Children who are overweight or obese will show greater engagement in bully-perpetrating behaviors when controlling for child demographic factors (i.e., sex, age). 2. Children who are overweight or obese will show greater engagement in bully-perpetrating behaviors when controlling for child and family demographics (i.e., child sex, child age, child ethnicity, family structure, and income), child factors (i.e., whether child currently has ADHD/anxiety/depression, self-control), family factors (i.e., family structure, caregiver relationship satisfaction, maternal/paternal mental and emotional health, family ability to “get by” on income, and exposure to domestic violence), and neighborhood/community factors (i.e., exposure to violence in the community, child being treated unfairly because of race/ethnicity). 3. Child, family, and neighborhood/community factors will predict bullying behavior in the subsample of overweight and obese children.
The current study seeks to provide information useful for developing interventions for overweight and obese youth. Therefore, it is important to examine factors related to this behavior in these children specifically. Even if differences in bullying prevalence are not found based on weight status, it is possible that unique factors influence these behaviors among the subset of overweight and obese children. Therefore, understanding factors related to bullying in this subset of children specifically is necessary for tailoring interventions appropriately.
Methods
Participants
Participants were 41,361 youth ages 10–17 years drawn from the National Survey of Children's Health (NSCH) 2011–2012. 21 Anthropometric data were not collected for survey participants under the age of 10 years, which delimited our analysis to 10- to 17-year-olds. Given our desire to compare bullying behavior of overweight and obese youth to that of normal weight youth, children who were underweight (BMI standardized for sex and age [BMIz] <5th percentile) were excluded (n = 3856).
Data Source and Study Variables
The NSCH 2011–2012 is a national telephone-based survey funded by the US Department of Health and Human Services, Health Resources and Services Administration, and Maternal and Child Health Bureau. Surveys were completed regarding the health and development of 95,677 youth living in all 50 US states, the District of Columbia, and US Virgin Islands, and anthropometric data were collected among 45,217 of these youth. Families were contacted using a cross-sectional, list-assisted, random-digit-dial sample of landline telephone numbers and an independent random-digit-dial sample of cell-phone numbers. Families with a child under 18 years of age and those who were able to complete the interview in English, Spanish, Mandarin, Cantonese, Vietnamese, or Korean met inclusion criteria. If a family had multiple children, one child was randomly selected to be the focus of the interview, and the interview was completed by the adult with the most knowledge of that child's health. The majority of interviews were completed by mothers (either biological, step, foster, or adoptive; 68.6%), whereas 24.2% of interviews were completed by fathers and 7.2% were completed by other caregivers. Interviews lasted, on average, 33–34 minutes, and completion rates were 54.1% and 41.2% for the landline and cell-phone samples, respectively. Please reference the NSCH source guide 21 for additional details about the NSCH 2011–2012 methodology.
The current study was reviewed by the study authors' institutional review boards and was exempted from review and determined as not human subjects' research. A data use agreement was completed between the study authors and the Data Resource Center and the Child and Adolescent Health Measurement Initiative. Study analyses were completed using the following interview questions.
Demographics (Control Variables)
Sex
“Is (child) male or female?” (male, female)
Ethnicity
This variable was created using information from the following two questions: (1) “Is (child) of Hispanic, Latino, or Spanish origin?” (yes, no); and (2) “Please choose one or more of the following categories to describe (child's) race.” (White/Caucasian, Black/African-American, Other) For use in analyses, this variable was dummy coded with “White/Caucasian” serving as the reference group.
Age
This variable was recorded in years (whole numbers only).
Family income
Caregivers were asked a series of questions to assess family income. Specifically, caregivers were asked to “Please think about your total combined family income during the last year for all members of the family. Can you tell me that amount before taxes?” Families were told to include all sources of income including: money from jobs, child support, social security, retirement, unemployment, public assistance, and other sources of income (e.g., investments). If the caregiver was unable to state an income amount, then he or she was asked a number of questions to identify the family income range (e.g., “Would you say that the total combined family income, before taxes, was above or below $20,000?”). Each family's income was classified as a percentage of the federal poverty level (FPL; <100%, 100–199%, 200–399%, and >400%). This categorical variable was used in study analyses.
Anthropometrics
Weight status
Children were classified as underweight (BMIz < 5th percentile), normal weight (BMIz 5th–85th percentile), overweight (BMIz 85th–95th percentile), or obese (BMIz > 95th percentile). BMI percentile was computed using caregiver report of height and weight. This variable was dummy coded for use in analysis with normal weight serving as the reference group.
Risk Factors for Bullying Behavior
Risk factors for bullying were assessed by parent-report to the following questions listed below. Additional details about questions used in the NSCH are available through the 2011–2012 NSCH: Child Health Indicator and Subgroups SPSS Codebook (version 1.0; Available at www.childhealthdata.org).
Child Factors
Current attention deficit hyperactivity disorder, depression, and anxiety
A dummy-coded variable (yes, no) was used to represent current diagnosis with each of these disorders. These variables were developed from the following two questions regarding each disorder: (1) “Has a doctor or other health care provider ever told you that (child) had (disorder)?”; and (2) “Earlier you told me that (child) has been diagnosed with (diagnosis). Does (child) currently have (diagnosis)?”
Self-control
Self-control was assessed by the following question: “He/she stays calm and in control when faced with a challenge.” (never, rarely, sometimes, usually, always)
Family Factors
Family structure
This variable was developed from numerous questions assessing parent and guardian relationships. The final variable contained the following categories: (1) families consisting of 2 biological, step-, or adopted parents who were married and living together; (2) families consisting of 2 biological, step-, or adopted parents who were living together but not married; (3) single-parent families; and (4) other. This variable was dummy coded for analysis with “other” serving as the reference group.
Caregiver relationship satisfaction
Relationship satisfaction was assessed by the following question: “The next question is about your relationship with your spouse or partner. Would you say that your relationship is completely happy, very happy, fairly happy, or not too happy?”
Maternal/paternal mental health
Participants were asked: “Would you say that your/child's mother's/child's father's mental and emotional health is excellent, very good, good, fair or poor?” This question was completed regarding both caregivers.
Family ability to “get by” on income
This variable was assessed with the question: “Since (child) was born, how often has it been very hard to get by on your family's income, for example, it was hard to cover the basics like food or housing?” (very often, somewhat often, rarely, or never)
Domestic violence
Exposure to domestic violence was assessed by asking: “Did (child) ever see or hear any parents, guardians, or any other adults in his/her home slap, hit, kick, punch, or beat each other up?” (yes, no)
Neighborhood/Community Factors
Neighborhood violence
Exposure to neighborhood violence was assessed by the following question: “Was (child) ever the victim of violence or witness any violence in his/her neighborhood?” (yes, no)
Racial/ethnic discrimination
Caregivers were asked: “Was (child) ever treated or judged unfairly because of his/her race or ethnic group?” (yes, no)
Dependent Variable
Bully-perpetrating behavior
Bullying behavior was assessed by the following question: “He/she bullies or is cruel or mean to others.” (never, rarely, sometimes, usually, or always)
Statistical Analyses
All analyses were completed using Mplus software (version 7.31; Muthén & Muthén, Los Angeles, CA) under the instructions given in the NSCH codebook. 21 Because the NSCH uses a highly complex sampling design in order to establish a nationally representative sample of youth in the United States, appropriate weighting procedures were used in all analyses. Specifically, four variables provided in the database were used to estimate standard errors accurately: Two variables (“STATE” & “SAMPLE”) were specified as population stratification variables, one variable (“IDNUMR”) was specified as a cluster variable and included with a specific analysis command (“Type = complex”) in Mplus to avoid possible type 1 errors in hypothesis testing, and the final variable provided (“NSCHWT”) was included as a sampling weight variable to compensate for disproportionate sampling of subpopulations. Missing data were minimal (range, 0.7–2.5%) across the N = 41,361 participants and was handled by the default maximum likelihood estimation algorithm (i.e., MLR) in Mplus consistent with current practice. 22
Three multinomial logistic regression models using maximum likelihood estimation were performed. The first analysis examined whether child weight status (i.e., normal weight, overweight, or obese) predicted bullying behavior while controlling for child age and sex. The second analysis examined whether child weight status (i.e., normal weight, overweight, or obese) predicted bullying behavior while controlling for the factors entered in the first analysis, as well as child ethnicity, family structure, family income, whether the child currently had ADHD, whether the child currently had depression, whether the child currently had anxiety, whether the child stays calm and in control when faced with a challenge, caregiver relationship satisfaction, maternal/paternal mental health, family ability to “get by” on income, exposure to domestic violence, exposure to neighborhood violence, and discrimination based on race/ethnicity. Given the categorical nature of weight status, ethnicity, family structure, and income, these variables were dummy coded for inclusion in the analysis. The third analysis was performed with only the subsample of overweight and obese. All variables entered in the second analysis were entered in the third and final analysis to examine whether these variables predicted bullying behavior in overweight/obese children specifically.
Results
Study Sample
The weight status of included youth was as follows: 28,573 (69%) normal weight; 6495 (16%) overweight (95th > BMIz > 85th percentile); and 6293 (15%) obese (BMIz ≥ 95th percentile). Approximately 33% of the overall sample engaged in at least some bully-perpetrating behaviors, and approximately 10% of youth engaged in these behaviors sometimes, usually, or always. Participant demographics and frequency of bullying behavior are presented in Table 1.
Participant Characteristics
FPL, federal poverty level; ADHD, attention deficit hyperactivity disorder.
Analyses 1 and 2: Association Between Weight Status and Bully-Perpetrating Behavior
In analysis 1 (when controlling for child sex and age), overweight children were more likely than normal weight children to always versus never bully (b = 0.68, p = 0.04; odds ratio [OR] = 1.97). Obese children were more likely than normal weight children to always (b = 0.84, p < 0.01; OR = 2.31), usually (b = 0.64, p = 0.03; OR = 1.90), or sometimes (b = 0.30, p < 0.01; OR = 1.35) bully versus never bullying.
However, the association between weight status and bullying behaviors was no longer significant in the second analysis, when controlling for key child, parent, family, and neighborhood-/community-level factors. Although weight status was not related to bullying behavior, several child, family, and neighborhood factors were related to bullying risk in the total sample of children regardless of weight status.
Child factors
Current depression was related to greater likelihood of always (b = 1.50, p < 0.001; OR = 4.50) or sometimes (b = 0.79, p < 0.001; OR = 2.21) bullying versus never bullying. Current anxiety and ADHD diagnoses were not related to bullying behavior. Decreasing frequency of staying calm when faced with a new challenge was significantly associated with children being more likely to always (b = −1.08, p < 0.001; OR = 2.94), usually (b = −0.84, p <0.001; OR = 2.31), sometimes (b = −0.58, p < 0.001; OR = 1.78), and rarely (b = −0.38, p < 0.001; OR = 1.47) engage in bullying behavior versus never bullying.
Family factors
Children who had seen or heard any parents or adults in the home slap, hit, punch, or beat each other were more likely to always (b = 0.66, p = 0.03; OR = 1.94), sometimes (b = 0.39, p = 0.002; OR = 1.48), or rarely (b = 0.32, p = 0.005; OR = 1.37) bully versus never bully. Children in homes where the parent was less satisfied with their marital or significant other relationship were more likely to rarely versus never bully (b = −0.15, p = 0.003; OR = 1.16). Conversely, children living in homes where the two caregivers were married were more likely to never bully versus always bully (b = −1.38, p = 0.02; OR = 4) or sometimes bully (b = −0.61, p = 0.005; OR = 1.85). Children living in single-parent homes were also more likely to never bully versus sometimes (b = −0.66, p = 0.003; OR = 1.92) bully. Maternal mental health was related to bullying behavior, but paternal mental health was not. Children whose mothers had poorer emotional and mental health were more likely to never bully versus always (b = −0.64, p < 0.001; OR = 1.90), usually (b = −0.35, p = 0.002; OR = 1.42), sometimes (b = −0.13, p = 0.01; OR = 1.14), or rarely (b = −0.11, p = 0.01; OR = 1.12) bully. A family's ability to “get by” on their income was not related to bullying behavior.
Community/neighborhood factors
The child being a victim of violence or witnessing violence in his her neighborhood was related to greater likelihood of usually (b = 1.17, p < 0.001; OR = 3.22), sometimes (b = 0.33, p = 0.003; OR = 1.39), and rarely (b = 0.24, p = 0.01; OR = 1.27) bullying versus never bullying. The child being treated or judged unfairly because of his or her race or ethnic group was related to greater likelihood of rarely versus never bullying (b = 0.36, p = 0.003; OR = 1.43).
Analyses 3: Predictors of Bullying Behavior among Overweight and Obese Youth
Analyses examining predictors of bullying behavior in youth with a BMIz > 85th percentile were completed controlling for child sex, child age, child ethnicity, and family income.
Child factors
Regarding individual child characteristics related to bullying, currently being diagnosed with ADHD was related to greater likelihood of always (b = 1.19, p = 0.002; OR = 3.27), sometimes (b = 0.57, p = 0.001; OR = 1.77), and rarely (b = 0.35, p = 0.04; OR = 1.42) bullying versus never bullying. Current depression was related to greater likelihood of always bullying versus never bullying (b = 1.75, p = 0.003; OR = 5.73). Current anxiety diagnosis was not related to bullying behavior. Decreasing frequency of staying calm when faced with a new challenge was significantly associated with children being more likely to always (b = −1.01, p < 0.001; OR = 2.75), usually (b = −0.77, p < 0.001; OR = 2.16), sometimes (b = −0.53, p < 0.001; OR = 1.70), and rarely (b = −0.39, p < 0.001; OR = 1.47) engage in bullying behavior versus never bullying.
Family factors
Children who had seen or heard any parents or adults in the home slap, hit, punch, or beat each other were more likely to sometimes bully versus never bully (b = 0.49, p = 0.01; OR = 1.64). Children in homes where the parent was less satisfied with their marital or significant other relationship were more likely to sometimes versus never bully (b = −0.23, p = 0.02; OR = 1.26). Family structure (e.g., single-parent home, two-parent married home) did not have an influence on bullying behaviors. Children in homes where it was hard to “get by” more frequently were more likely to never versus always bully (b = −0.43, p = 0.04; OR = 1.54). Maternal and paternal mental health was not related to bullying behavior.
Community/neighborhood factors
The child being the victim of violence or witnessing violence in his or her neighborhood was related to greater likelihood of usually (b = 1.37, p < 0.001; OR = 3.93) and rarely (b = 0.32, p = 0.03; OR = 1.37) bullying versus never bullying. The child being treated or judged unfairly because of his or her race or ethnic group was related to greater likelihood of rarely versus never bullying (b = 0.53, p = 0.005; OR = 1.70).
Discussion
Results of the current study suggest that bullying perpetrating for overweight and obese youth is related to a host of child, family, and community factors. Although rates of bullying were not higher than those of children of average weight status, nonetheless it is important to note that approximately one third of all children engaged in bullying behaviors, although only approximately 2% of the sample engaged in bullying behavior on a frequent basis. Hence, our findings demonstrate that bullying behaviors may be frequent in children and adolescents irrespective of their weight status, and that prevention programs remain important to stem bullying perpetration among our nation's youth.
We applied the social-ecological diathesis-stress model 10 in this study and examined individual, family, and community risk factors for bullying behavior among overweight and obese youth. Our results add important quantification of characteristics related to bullying perpetration in overweight youth. Youth who are overweight and had ADHD were more likely to bully, perhaps attributable to poor impulse control, but this child factor did not predict bullying behavior in the total sample. Although previous findings are not always consistent, recent evidence posits a strong association between ADHD and bullying involvement11,12 and ADHD and obesity. 23 Our results suggest the possibility of screening for or targeting symptoms of impulsivity, inattention, and impaired executive functions that may impede social functioning in bullying interventions for overweight youth. When addressing bullying in overweight children, intervention messages may also need to include techniques to assist children in mediating impulsive reactions with positive self-talk that will allow them to stop and think about the welfare of others. These ideas are speculative, however, and more research will be needed to determine what types of interventions are effective for youth who are overweight and also have ADHD.
Consistent with past study findings, 13 a current diagnosis of depression was related to greater likelihood of engagement in bully-perpetrating behaviors. For overweight and obese youth, this is particularly concerning because this group may experience greater depression.24,25 Additionally, one key finding that was consistent with our hypothesis is that children and adolescents who stay calm when faced with a new challenge were at lower risk for bully-perpetrating behaviors. Children who remain calm in the face of challenges may have an ability for emotional regulation and coping skills for managing emotions that leads them to engage in fewer bullying behaviors. Similarly, it may be the case that children experiencing depression show poorer emotional regulation and coping skills, which may result in greater bullying behavior.
The current study demonstrated a relationship between bully-perpetrating behavior and several aspects of the family environment (e.g., parental satisfaction in romantic/marital relationship, exposure to violence in the home). In past research, overweight and obese children who witness family violence were more likely to report both bullying victimization and bully-perpetrating behaviors,26,27 potentially as a result of these youth socially learning that violence is an acceptable method for resolving conflict. 28 These findings suggest that violence in the home may exacerbate bullying behaviors among overweight and obese youth. Hence, interventions should incorporate parents and other adult family members living with the child. Adults need to be educated on the impact violence in the home can have on youth living in their household during the childhood and adolescent years. Interestingly, the results of this study did not support our hypothesis of a significant association between bullying behavior and caregiver relationship satisfaction, maternal and paternal mental health, and family structure among overweight and obese youth. Research is needed to understand why these relationships may not have impacted bullying behaviors in overweight and obese youth.
In addition to family factors, neighborhoods may influence bullying behaviors in overweight and obese youth. 26 Our sample of youth was more likely to engage in bully-perpetrating behaviors if the child was a victim of violence or witnessed violence in his or her neighborhood regardless of weight status. Community-based projects and support services may prevent bullying among youth who live in areas with neighborhood violence, and screening for bullying behaviors among youth living in these communities may be important. In addition, youth who had been treated or judged unfairly because of their ethnic group were more likely to rarely versus never bully. These associations were found both among overweight/obese children and the overall study sample.
Limitations
Our study is limited by the caregiver-report nature of the child anthropometric data used for calculating BMI percentiles. Research has documented that parents may not always accurately report child anthropometric data,29,30 and it has been suggested that parent-report of child anthropometrics should be used with caution. 31 Caregiver report of child bullying behavior is also a limitation, given that caregivers may have been unaware that their child was engaging in bullying behavior or may have been hesitant to report this behavior because of social desirability. Further, only a single item was used to assess bullying behavior. Use of a single item may not have adequately assessed the complex construct of child bullying. Additionally, our study was delimited to 10- to 17-year-olds because BMI data were only collected among this age group, so caution should be used when generalizing the present study's findings to other age groups. The NSCH data were also cross-sectional in nature, and we are unable to determine causal associations. Moreover, reasons for bullying and children's attitudes about bullying were not assessed, and these are important areas for future research. Additionally, although analyses examined factors related to bullying behaviors among the overall sample and the subset of overweight and obese children, no analyses were completed examining factors related to bullying in the subset of normal weight youth. Therefore, it cannot be determined whether there are significant differences in the factors related to bullying between these two groups. Finally, all study constructs were assessed using single-item measures.
Conclusions
In a nationally representative sample of youth, overweight and obese youth were not found to engage in more bully-perpetrating behavior than normal weight peers when controlling for demographic, child, family, and community factors. Bullying behaviors should be assessed and interventions developed for children across all weight statuses, because approximately one third of children overall engaged in at least some bullying behaviors. Future research using longitudinal designs to examine bullying over time and assess characteristics related to bullying would provide information about what characteristics matter when, in terms of key developmental periods. Moreover, qualitative research examining child and caregiver perceptions of the linkage between child and family characteristics and bullying behavior may provide information about why certain characteristics of the child and his or her family are related to bullying.
Footnotes
Author Disclosure Statement
No competing financial interests exist.