Bullying is a relational problem involving repetitive, negative actions directed toward a peer and is characterized by a power imbalance—physical, social, or cognitive—between the victim and perpetrator.1 Bullying involvement encompasses the following 3 aspects of this relational problem: victimization, perpetration, and victimization/perpetration (ie, those who perpetrate and are victimized).1 Adolescents with developmental disabilities have higher rates of victimization (19%-94% vs 12%-41%) and perpetration (16%-83% vs 10%-44%) compared with typically developing peers.2-8 The prevalence of victimization/perpetration for typically developing adolescents is 6.8%, but, at present, comparable estimates are unavailable for adolescents with developmental disabilities.8 Bullying involvement is associated with (1) higher levels of depression, anxiety, and loneliness and (2) lower levels of academic performance and school commitment.8-10 The US Department of Health and Human Services has made bullying prevention a national priority through its Healthy People 2020 initiative, which aims to increase school safety and the adoption of antibullying policies during the coming decade.11
Relatively little research examines the prevalence and correlates of bullying involvement among adolescents with an autism spectrum disorder (ASD). This gap is concerning because adolescents with an ASD may be uniquely vulnerable to this form of aggression given the social and relational problems that are hallmarks of their condition.12,13 To examine this possibility and contextualize findings, prevalence estimates of bullying involvement are needed for adolescents with an ASD and their peers with impairments in some of the developmental areas affected by an ASD (eg, intellectual, speech, and learning). This area of study has important public health ramifications because larger numbers of adolescents are being identified as having an ASD with each passing year, and the number of adolescents served in the autism special category more than doubled from 2004 to 2010.14,15
Prevalence rates of bullying involvement for adolescents with an ASD vary by type of involvement and informant, with a range of 7% to 94% for victimization and 15% to 46% for perpetration.4,16,17 Prior studies have not reported rates of victimization/perpetration for this population. According to reports by their mothers, 94% of children with Asperger syndrome experienced some form of victimization within the previous year.4 Teachers reported the highest rates of bullying involvement (victimization, 30%; perpetration, 46%) for adolescents with an ASD compared with self-reports (victimization, 17%; perpetration, 19%) and peer reports (victimization, 7%; perpetration, 15%).17
Communication problems, fewer friendships, and lower income were found to be significant correlates of bullying involvement among adolescents with an ASD.16,17 Attention-deficit/hyperactivity disorder (ADHD) was found to be a significant correlate of perpetration.18 Among a nationally representative sample, parent reports indicated adolescents with an ASD and ADHD perpetrate at a higher rate (59.8%) compared with adolescents with an ASD only (28.4%).18 Attention-deficit/hyperactivity disorder is also a risk factor for victimization and victimization/perpetration among typically developing adolescents.19,20 The relationship of ADHD to rates of victimization and victimization/perpetration remains largely unexplored among adolescents with an ASD.
With one exception, previous studies were not nationally representative because the data were gathered through international websites or from the Netherlands and Canada.4,16-18 The study that used a nationally representative US sample, however, only examined perpetration.18 Other generalizability limitations include samples consisting primarily of adolescents with Asperger syndrome or drawn from special schools that only serve adolescents with an ASD.4,17
This study addresses these gaps by using a nationally representative US sample of adolescents with an ASD to investigate the following aims: (1) to identify the prevalence of bullying involvement, (2) to compare prevalence rates of bullying involvement with adolescents with developmental disabilities that overlap with the core deficits of ASD, and (3) to identify the social ecological correlates of bullying involvement. This study examines all 3 aspects of bullying involvement and their potential correlates, including sex, age, race, ethnicity, income, ADHD, social skills, conversational ability, interactions with friends, and classroom placement. This study contributes to a growing foundation of evidence that will inform school-based antibullying efforts and allow practitioners to address important questions from concerned parents regarding factors that may influence their children's bullying involvement.
The National Longitudinal Transition Study 2 was a 10-year, 5-wave prospective study of adolescents receiving special education services and was conducted by SRI International for the US Department of Education. The sampling plan was designed to produce nationally representative estimates that generalize to all students receiving special education services in the 7th through 12th grades or in ungraded programs who were ages 13 through 16 years on December 1, 2000. The National Longitudinal Transition Study 2 used a multistage, stratified, random sampling procedure and resulted in the identification of 1100 sample-eligible students in the autism category during wave 1.
Three instruments from the National Longitudinal Transition Study 2—parent interviews, the Student's School Program Survey, and the School Characteristics Survey—were used for this analysis. Unweighted sample size numbers were rounded to the nearest 10 as required by the data use agreement with the US Department of Education. This study was approved as exempt by the Washington University institutional review board. Detailed information on the sample design and weighting procedure has been previously published.21
Adolescents were selected from the official special education enrollment-reporting category of autism. Adolescents were counted once in a primary disability category and were not designated into multiple disability categories. Schools do not necessarily use standardized Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition) (DSM-IV) criteria to assign the label of autism.22 Based on recent US epidemiological surveillance data, 99% of adolescents served under the autism educational designation also meet DSM-IV criteria for an ASD.23,24 Some adolescents who meet the DSM-IV criteria for an ASD may be served under another special education disability category.
Data collection procedures
Parent interviews were conducted in English (97.5%) and Spanish (2.5%). Parents who were unable to be reached by telephone were mailed a self-administered questionnaire (2.6%). Data were collected from 920 parents of students in the autism category (83.6% response rate). The Student's School Program Survey was mailed to a data collection coordinator with instructions to give it to the school staff person most familiar with each student's school program (580 completed; 52.7% response rate). The School Characteristics Survey was mailed to principals to gather information on student body demographics and other school characteristics (830 completed; 75.4% response rate).
The study included 3 dependent measures of bullying involvement: victimization, perpetration, and victimization/perpetration. Parents were asked whether the adolescent had ever had any of the following experiences during the 2000-2001 school year:
“Has [ADOLESCENT] been bullied or picked on by other students or made to do things like give them money, either at school or on the way to or from school?”
“Has [ADOLESCENT] been teased or called names at school?”
“Has [ADOLESCENT] bullied or picked on other students?”
The first 2 questions assessed the experience of victimization and were collapsed into 1 overall dichotomous indicator. The accepted definition of bullying encompasses the experience of verbal forms of victimization often specifically described as teasing and/or name-calling. To examine the prevalence of bullying involvement across its full continuum, the following 6 variables were constructed: any victimization, any perpetration, only victimization, only perpetration, victimization/perpetration, and no involvement.
Although the sample did not include a typically developing comparison group, prevalence rates of bullying involvement for adolescents with an ASD were compared with those groups who exhibit difficulties that often occur in autism: mental retardation (MR; ie, intellectual impairment), speech/language impairment (ie, communication problems), and learning disability (ie, academic impairment). Although the designation of MR is normally avoided, it was used to be consistent with special education legislative definitions.
Age, sex, race, ethnicity, ADHD, income level, social skills, and conversational ability of the adolescent were measured using parent report. A social skills scale was created by summing the following 3-category (never, sometimes, and very often) parent-report questions (α = .73): (1) joined group activities without being told, (2) made friends easily, (3) seemed confident in social situations, and (4) started conversations rather than waiting for others to initiate. The social skills scale was recoded into 4 categories (very low, low, medium, and high) with roughly equivalent group sizes to report stratified rates of bullying involvement. Conversational ability was measured on a 4-point ordinal scale (from converses as well as other children to does not converse at all) using the following question: “How well does [he/she] carry on a conversation?” The Student's School Program Survey provided information to determine the percentage of classes in general education. Frequency of friendship interaction was measured by dichotomizing the average number of days per week that parents reported their child got together with friends outside of school as less than once per week (includes never) and 1 to 7 days per week. Friendship interaction was dichotomized to facilitate interpretation of the multivariate models because the near absence of friendship interaction is a common reality for most adolescents with an ASD.25
We examined univariate percentage distributions for the independent variables and the stratified rates of the dependent variables for adolescents with an ASD. Prevalence estimates for bullying involvement were compared across adolescents with an ASD and from 3 other disability categories. Bivariate logistic regression was used to identify significant differences between these groups. Three multivariate logistic regression models estimated the correlates of victimization, perpetration, and victimization/perpetration among adolescents with an ASD. Multiple imputation with chained equations26 was performed using IVEware27 to create 50 data sets with no missing values. The multiply imputed data were analyzed using commercially available software (Stata, version 11; StataCorp), which combines estimates using well-established procedures.28 All adolescents who were not in school at wave 1 were excluded, yielding a final subsample size of 900. All estimates were weighted to the population level, and variances were adjusted in accordance with the complex sampling design. Therefore, unweighted subsample sizes were not reported alongside weighted point estimates.
Table 1 reports demographics and stratified rates of victimization, perpetration, and victimization/perpetration. Based on parent report for the current school year, 46.3% of adolescents experienced victimization, 14.8% engaged in perpetration, and 8.9% experienced victimization/perpetration. The sample had a male to female ratio of 6.5:1. This ratio is within the range found in prior research.12,29 When we examined the different permutations of bullying involvement (Table 2), adolescents with an ASD had significantly higher rates of engaging in only perpetration compared with adolescents in the other 3 disability categories. Rates of any perpetration for adolescents with an ASD were significantly higher than those with speech/language impairment. Adolescents with an ASD had significantly lower rates of any victimization and victimization/perpetration compared with adolescents with MR. Rates of no bullying involvement were significantly higher for adolescents with an ASD compared with those with MR.
Table 1. Univariate Percentage Distributions of Independent Variables and Stratified Rates of Dependent Variables Among Adolescents With an Autism Spectrum Disordera
Table 2. Rates of Bullying Involvement Among Groupsa
In the multivariate model of victimization (Table 3), Hispanic adolescents had significantly lower adjusted odds of victimization compared with non-Hispanic adolescents (odds ratio [OR], 0.5). Adolescents with ADHD had significantly higher adjusted odds of victimization compared with adolescents without ADHD (OR, 1.7). Better social skills were associated with significantly lower adjusted odds of victimization (OR, 0.9). Compared with adolescents with no conversational ability, those with higher abilities had greater adjusted odds of victimization (OR, 3.0-6.1). Adolescents who had 76% or more of their classes in general education had significantly higher adjusted odds of victimization compared with adolescents who had 25% or fewer of their classes in general education (OR, 2.8).
Table 3. Logistic Regression Model of Bullying Involvement Among Adolescents With an Autism Spectrum Disordera
A second multivariate model examined the correlates of perpetration. African American adolescents had a significantly lower adjusted odds of perpetration compared with white adolescents (OR, 0.5). Adolescents with ADHD had significantly higher adjusted odds of perpetration compared with those without ADHD (OR, 2.1). Adolescents who got together with friends at least once per week had higher adjusted odds of perpetration compared with those with less frequent contact (OR, 1.9).
A third logistic regression examined the correlates of victimization/perpetration. Hispanic adolescents had lower adjusted odds of victimization/perpetration compared with non-Hispanic adolescents (OR, 0.3). African American adolescents had significantly lower adjusted odds of victimization/perpetration compared with their white counterparts (OR, 0.4). Adolescents with ADHD had significantly higher adjusted odds of victimization/perpetration compared with those without ADHD (OR, 2.6). Adolescents who got together with friends at least once per week had significantly higher adjusted odds of victimization/perpetration compared with those with less frequent contact with friends (OR, 2.2).
This study examined the prevalence and correlates of bullying involvement using a large national sample. The victimization rate for adolescents with an ASD (46.3%) was substantially higher than the national prevalence estimates for the general adolescent population (10.6%).8 The rates of perpetration (14.8%) and victimization/perpetration (8.9%), however, were roughly equivalent to national estimates found among typically developing adolescents (perpetration, 13%; victimization/perpetration, 6.8%).8 High prevalence rates of victimization were found for adolescents with an ASD, MR, speech/language impairment, and learning deficiency. Adolescents with MR, however, were significantly more likely to experience victimization compared with the other 3 groups. Adolescents with an ASD were significantly more likely to engage exclusively in perpetration compared with the other 3 groups. Tailored antibullying programs are needed to address the unique needs of these vulnerable adolescents given their social, communication, and academic impairments.
Hispanic adolescents with an ASD were significantly less likely to experience victimization and victimization/perpetration compared with their non-Hispanic counterparts. African American adolescents with an ASD were significantly less likely to engage in perpetration and victimization/perpetration compared with white adolescents. Prior studies with typically developing adolescents have also found lower rates of bullying involvement for minority youth.30 Prior research suggests these racial/ethnic differences may be due to variations in how bullying is defined and measured.30 Minority adolescents underreport victimization compared with their white counterparts when using definition-based single-item measures vs multiple-item behavior-based measures of bullying.30 Further research is needed to address variations in the definition and measurement of bullying involvement across racially and ethnically diverse samples.
Consistent with prior research, adolescents with greater social skills were significantly less likely to experience victimization.31 Conversational ability was also a significant correlate for victimization. Compared with adolescents with no conversational ability, adolescents with some level of conversational ability were significantly more likely to experience victimization. Adolescents with no conversational ability may be institutionally protected (eg, lower teacher-student classroom ratios), or school staff are not informing parents about their child's bullying involvement. Conversational ability is a multidimensional construct (eg, clear speech, appropriate gestures and expressions, responsiveness to questions and changes in topic), and adolescents with an observable disability are often victimized at a higher rate.32,33 Even most adolescents with no trouble conversing may still possess noticeable differences in their conversational abilities compared with typically developing peers, placing them at greater risk for victimization.32
Adolescents who had most of their classes in general education were significantly more likely to experience victimization compared with those in segregated settings. This finding contradicts previous research, which found that adolescents in segregated classrooms reported significantly more victimization compared with their peers in inclusive classrooms.34,35 The integration of students with and without disabilities into inclusive classroom settings has been conceptualized as a protective factor because of the greater likelihood of developing social skills through behavioral modeling, increasing acceptance and social participation, and reducing negative stereotypes.3,31 However, if adolescents with a developmental disability are not fully integrated into peer groups, inclusion may increase social isolation and worsen rates of victimization.3,31 This study suggests schools need to examine their existing inclusive practices and adopt strategies that educate students about ASD and explicitly target inclusion of these students into protective peer groups.3,5
Adolescents who got together with friends at least once a week were significantly more likely to experience perpetration and victimization/perpetration. These findings may be partially explained by the increased social opportunity these adolescents had to experience these forms of bullying involvement compared with their counterparts who were almost completely isolated from friends. This study did not find a significant relationship between the frequency of getting together with friends and victimization. Prior research indicates friendships are an important protective factor in reducing victimization among all adolescents.34,36,37 A possible explanation for this nonsignificant finding may relate to the simplicity of this dichotomous measure and its inability to account for other important dimensions, such as friendship quality and support.
Adolescents with an ASD and ADHD were significantly more likely to experience victimization, perpetration, and victimization/perpetration compared with those with only an ASD. This study supports previous research that found an association between ADHD and perpetration among adolescents with an ASD.18 Future studies should continue to include ADHD as an important predictor for bullying involvement, whereas antibullying programs should focus their intervention efforts on the needs of this vulnerable subgroup.
This study has some limitations. Bullying involvement was not defined for the respondents and may have led to reporting bias. Bullying involvement was measured using dichotomous indicators, precluding the assessment of frequency and duration. The study also lacked multiple informants to measure bullying involvement. Parent report may underestimate prevalence rates of bullying involvement compared with self-reported measures because parents may be unaware of the full extent of bullying involvement.38,39
Bullying has garnered an increasing amount of public attention in the United States, with most states adopting antibullying legislation.35 Universal bullying prevention programs (eg, the Olweus Bullying Prevention Program) are considered the criterion standard for reducing bullying in schools, but these programs were not developed to meet the needs of adolescents with an ASD.40 Future interventions should incorporate content that addresses the core deficits of adolescents with an ASD, which limits their verbal ability to report bullying incidents. Schools should incorporate strategies that address conversational difficulties and the unique challenges of those with comorbid conditions. Inclusive classrooms need to increase the social integration of adolescents with an ASD into protective peer groups while also enhancing the empathy and social skills of typically developing students toward their peers with an ASD and other developmental disabilities.