Abstract
Children with Attention-Deficit/Hyperactivity Disorder (ADHD) are at a higher risk of experiencing bully victimization compared with peers with and without disabilities. Yet the association between ADHD subtypes and bully victimization is not well understood. The current study examines which set of behaviors related to ADHD subtypes is influential in determining whether students are victimized. Using a latent class growth analysis, students with ADHD in a nationally representative special education sample (n = 291) were grouped by victimization trajectory. Latent class analysis revealed three victimization profiles. Approximately one third of participants (35%) experienced moderately high victimization decreasing over time. A slightly larger group reported consistently low victimization (39%), and one fourth (25%) reported consistently high victimization. Behaviors representing inattention and hyperactivity/impulsivity were examined as covariates in the model. Hyperactive/impulsive behaviors were predictive of latent class assignment, initial victimization, and the trajectory of victimization. Study limitations are discussed. Recommendations are made to inform the creation of interventions tailored for students who have ADHD.
Keywords
Bully victimization represents a highly problematic and prevalent form of aggression in schools that, in recent years, has gained widespread attention. As such, a variety of national and international initiatives have been undertaken to discover the nature of bully victimization, how it occurs, and who is typically involved in an effort to prevent its prevalence. The World Health Organization found that the rate of bullying among adolescents decreased in most countries between 1993 and 2006 (Molcho et al., 2009). This finding is likely due to greater awareness of the harmful effects of bullying, implementation of anti-bullying interventions, and enactment of anti-bullying legislation (Finkelhor, Ormrod, & Turner, 2009). Yet a high number of students remain at risk for victimization in the United States. The National Center for Education Statistics, U.S. Department of Education (2017) found that 21% of secondary school youth experienced some form of bullying during the 2015–2016 school year. Thus, it is crucial that studies continue exploring the social dynamics associated with bullying to inform intervention efforts.
Understanding Bully Victimization
Bullying is most commonly defined as a type of aggression that is (a) intended to harm others, (b) repetitive, and (c) characterized by an imbalance of power between the perpetrator and victim (Olweus, 1993). As such, it is considered a proactive form of aggression used for gaining and maintaining dominance or status over peers (Fanti & Henrich, 2015). It is widely recognized that bullies target different types of students based on a perceived imbalance of power (Juvonen & Graham, 2014). Typically, victims are dichotomously characterized as either “passive” or “aggressive,” with aggressive victims also engaging in bullying behaviors (Veenstra et al., 2005). Students in the latter category are commonly referred to as “bully-victims.” Bully-victims have distinct social characteristics and experiences when compared with pure bullies or victims. According to the “two social worlds of aggression” hypothesis, some aggressive students have high social influence and interact well with teachers and peers who are uninvolved in bullying, while others are highly disliked by peers and experience social marginalization (Farmer et al., 2010). Evidence for this hypothesis is supported by research suggesting that bullies tend to be socially competent, falling into the first category of aggressors, while bully-victims exhibit interpersonal difficulties more consistent with the second (Farmer et al., 2010). Thus, bully-victims struggle more so than bullies with demonstrating appropriate social behavior and their peer network is composed primarily of victims and other bully-victims. Famer and colleagues’ (2010) findings align with how Olweus (1994) initially described bully-victims. Olweus suggested that this group of victims exhibits poor social behaviors that irritate or annoy others, thereby eliciting bullying. He further described them as having poor concentration and lacking impulse control, resulting in classroom disruption. Based on these characteristics, he labeled them “provocative” victims. The social dynamics between bullies and provocative victims are distinct and worthy of study to improve bullying intervention efforts.
Bully Victimization and Children With Disabilities
Children with disabilities are particularly at risk of victimization due to the perception by peers as different and more socially vulnerable (Blake et al., 2016; Rose, Monda-Amaya, & Espelage, 2011). Research on bullying and students with disabilities has gained momentum in recent years, with studies revealing the unique experiences of these children (Blake, Lund, Zhou, Kwok, & Benz, 2012; Hebron, Oldfield, & Humphrey, 2017; Maiano, Normand, Salvas, Moullec, & Aime, 2016). An early study by Thompson, Whitney, and Smith (1994) found that students receiving special education were more likely to be victimized by peers than students without special needs. Thompson attributed this trend to special education students’ segregation from mainstream classrooms. In a nationally representative sample of students in special education, parents reported that 34% of children were involved in bullying as victims, 24% as bullies, and 10% as bully-victims (Van Cleave & Davis, 2006). Farmer et al. (2012) found that children receiving special education were significantly more likely to be involved as victims and bully-victims, but not perpetrators, contrary to prior studies. In fact, these students were 2 to 5 times more likely to be victimized than their nondisabled peers were.
The manner students with disabilities are targeted for bullying likely influences their response style to the bullying and how they are ultimately affected by bullying. While students with disabilities are more often victimized, this is largely because they are targets of physical victimization. Hartley, Bauman, Nixon, and Davis (2017) identified similar rates of verbal and relational victimization for students with and without disabilities but documented that students in special education experienced more frequent physical bullying. Their findings suggest that either physical aggression is considered more acceptable toward students with disabilities, due to perceived visible or nonvisible weaknesses, or students with disabilities are more likely to interpret physical cues as threatening. Gathering information about school culture, social norms, and students’ social skills is key to putting these findings in context and to understanding why physical bullying is more common among this population. However, it is important to note that students with disabilities who are targeted physically may respond with physical aggression, thus giving the appearance of being more “reactive” victims or bully-victims. Alternate factors that likely influence these social dynamics include victims’ mental health and disability status. It is proffered that students with emotional and behavioral disorders are better described as “reactive victims” or “provocative” victims rather than bully-victims due to the nature of their symptoms, which may include emotional dysregulation, hyperactivity/impulsivity, and hostility (Olweus, 1994; Rose & Espelage, 2012). Studies have found that depressive symptoms, in particular, may predict students’ involvement in bullying, as some respond aggressively to being victimized in an effort to minimize future attacks (Rose, Simpson, & Preast, 2016).
With regard to disability status, a number of researchers have sought to discover if certain disabilities place students more at risk of victimization than others. Twyman and colleagues (2010) conducted a study including children and adolescents with various disabilities. Those reporting the highest level of bully victimization were diagnosed with Attention-Deficit/Hyperactivity Disorder (ADHD), Autism, and Learning Disabilities. In another recent study, Blake et al. (2016) examined predictors of peer victimization for students in special education using a national sample. Similar to Twyman and colleagues, these authors found that the children most likely to be chronically targeted for bullying were those diagnosed with ADHD, followed by those classified with an Emotional Disturbance (ED). These results contrasted with prior research indicating that students with visible physical disabilities were more often victimized, suggesting instead that emotional and behavioral difficulties are the most significant indicators of chronic victimization risk, followed by history of prior victimization and low socioeconomic status (Blake et al., 2016; Rose et al., 2011).
Experiences of Children With ADHD
Although the number of studies related to bullying and children with disabilities has increased in recent years, relatively few have focused specifically on individuals with ADHD (Fite, Evans, Cooley, & Rubens, 2014; Taylor, Saylor, Twyman, & Macias, 2010). This is surprising given the academic, behavioral, and social difficulties commonly tied to the disorder (Bacchini, Affuso, & Trotta, 2008; Barkley, 1998). ADHD is often associated with behaviors that violate societal expectations, such as speaking or acting inappropriately given the time and place, disregarding others’ personal space, acting aggressively, and missing social cues (Holmberg & Hjern, 2008; Wiener & Mak, 2009). In general, children with ADHD exhibit underdeveloped social skills and have difficulty building age-appropriate friendships (Taylor et al., 2010). In addition to the strain that ADHD symptoms place on peer relationships, teacher relationships often suffer as well. While there is limited research specific to this area, existing studies suggest that teachers find instructing children with ADHD significantly more stressful than instructing students without ADHD (Greene, Beszterczey, Katzenstein, Park, & Goring, 2002). This is largely because children with ADHD are more likely to be disruptive, restless, and struggle with completing tasks. Using teacher nominations to explore teacher attitudes toward students with and without disabilities, Cook (2001) found that teachers were most likely to desire the removal of students identified with mild or “hidden” disabilities (e.g., ADHD and Asperger’s Syndrome) from their classroom because these students tend to violate expectations while appearing outwardly similar to classmates. Thus, the nonvisible nature of students with ADHD’s impairment leads to a reduction in others’ tolerance and they are more likely to be marginalized or rejected by peers (Bagwell, Molina, Pelham, & Hoza, 2001; Hinshaw, 2002; Wiener & Mak, 2009). In some cases, peer rejection leads to victimization, coinciding with Olweus’ description of the “provocative victim” who unintentionally elicits bullying (Hoza, 2007). Therefore, students with ADHD represent a unique population that experience rejection at a higher rate than other groups due to their disability.
Few researchers have focused exclusively on bullying experiences of children with ADHD (Taylor et al., 2010). Unnever and Cornell (2003) were the first scholars to identify a direct relationship between self-reported stimulant medication use, which they used as a proxy for ADHD, and bullying involvement. Although those classified with ADHD in their sample were at a higher risk of being identified as both victims and perpetrators compared with non-ADHD peers, conclusions based on these findings are limited because a parent or a physician did not confirm participants’ diagnoses of ADHD. A later study by Bacchini et al. (2008) examined bullying among Italian elementary students with ADHD, finding that symptoms of ADHD were associated with increased risk of bullying behaviors in males and victimization in females. In addition, the study found that ADHD symptoms only had a direct influence on peer rejection when their model did not account for bully/victim status as a mediator. Therefore, it seems ADHD alone is not a predictor of peer rejection; it is only when these behaviors lead to bullying involvement that children with ADHD are at risk for social isolation. Holmberg and Hjern’s (2008) investigation of bullying in Sweden yielded similar results, with ADHD symptoms related to victimization and perpetration among fourth-grade students with confirmed ADHD diagnoses.
In an effort to improve on previous studies, Wiener and Mak (2009) attempted to identify children with ADHD using more formal criteria. Participants were required to have both a physician’s diagnosis of ADHD and elevated scores on the Conners Rating Scales (Conners, 1997). This study also expanded on others by examining different types of bullying, including verbal, physical, and relational. Consistent with previous results, Wiener and Mak found that 58% of students with ADHD between ages 9 and14 reported involvement in bullying as perpetrators, victims, or bully-victims, compared with only 14% of their non-ADHD peers. Victims with ADHD experienced physical, verbal, and relational bullying more frequently than other students, with verbal victimization being the most common form of bullying reported. In general, girls with ADHD reported a higher rate of victimization than their male counterparts.
Taylor and colleagues (2010) conducted a study focusing solely on children with ADHD after their more general exploration of bullying among those with special needs. Using the same sample of children, they examined the relationship between ADHD symptoms and self-reported bullying involvement. In contrast to other studies, their results showed that children with ADHD reported significantly higher rates of victimization than non-ADHD peers, but not bullying behaviors. The study also discovered that victims with ADHD experienced higher rates of internalizing and externalizing difficulties compared with peers, indicating significant psychological impairment. These findings echo the results of Humphrey, Storch, and Geffken (2007), who examined the psychological impairment of children with ADHD after experiencing bullying and found higher scores on subscales of the Child Behavior Checklist (CBCL; Achenbach & Rescorla, 2001) such as depression, anxiety, aggression, and interpersonal difficulties.
A limited number of studies have examined bullying trends specific to adolescents with ADHD, as most utilized a sample with a wide age range or focused solely on younger children. Considering the increased emphasis on peer relationships during adolescence in middle and high school, it is likely that any social difficulties previously experienced would be exacerbated, including those related to ADHD. Two recent investigations sought to isolate this population. Consistent with other studies, Timmermanis and Wiener (2011) found that secondary students with ADHD were twice as likely to be involved in bullying compared with their non-ADHD peers as both victims and perpetrators. In addition, students with ADHD who were victimized evidenced a higher level of interpersonal difficulty and a lower level of perceived social support, as reported by their parents, with effect sizes in the large range. Sciberras, Ohan, and Anderson (2012) conducted another study on adolescents with ADHD that focused on female students between ages 12 and 18, examining both overt and relational bullying. Results suggest that ADHD symptoms in females may be linked to greater social impairment and higher victimization rates when compared with peers without ADHD. These trends were more strongly related to victimization than bullying, according to both parent and self-report measures.
Purpose of the Current Study
Children with ADHD are more likely to be involved in bully victimization than those without ADHD and those with other disabilities. Unknown, however, is the influence of ADHD behavioral phenotypes on bully victimization trends (Taylor et al., 2010). This is surprising because inattentive and hyperactive-impulsive subtypes of ADHD are associated with a unique set of symptoms and impairment that could represent significant risk factors on their own (Chhabildas, Pennington, & Willcutt, 2001). While it is important to examine the effects of behavioral phenotypes on both bullying perpetration and victimization, the current analyses sought to begin research in this area by exploring how inattentive and hyperactive-impulsive behaviors may be linked to victimization alone and if shifts in victimization status occurred over time given developmental changes in student behavior as children matured.
We hypothesized that students with ADHD exhibiting a high level of hyperactive/impulsive behaviors would be more likely to experience bully victimization. This was expected because students who exhibit hyperactive and impulsive behaviors best fit Olweus’ definition of the provocative victim and would likely experience a higher level of social difficulty, placing them at risk for peer rejection (Hoza, 2007; Olweus, 1994). Second, we hypothesized that the trajectory of bully victimization would shift over time for students with ADHD, due to research indicating that hyperactive and impulsive behaviors tend to lessen in frequency and intensity as children mature (Willcutt et al., 2012). Thus, it was expected that longitudinal analysis would reveal different trends based on the manner in which symptoms change and interact with other factors.
Method
This investigation was conducted using a nationally representative sample of children with disabilities from the Special Education Elementary Longitudinal Study (SEELS). The SEELS dataset was collected by SRI International through funding from the Office of Special Education Programs (OSEP) to examine the characteristics, academic success, and experiences of children enrolled in special education, including exposure to bully victimization. The study followed students from 2000 to 2006; data were collected in three waves. Information was gathered from multiple sources, including parents, language arts teachers, and the school staff most knowledgeable about each student’s educational program. School administrators also provided information regarding school characteristics and services offered in each setting. Surveys were completed via mail or telephone interview every 2 years.
SEELS Sampling Procedures
The sample utilized for the SEELS study was gathered in two stages. First, a sample of 1,124 local education agencies (LEAs) was randomly selected from those serving special education students. Two hundred forty-five LEAs and 32 special schools provided rosters of their special education students between ages 6 and 12. Although the sample of LEAs was smaller than anticipated, analyses of their characteristics based on variables of size, region, and socioeconomic status found that they were nationally representative. The rosters provided were stratified by disability category. Students were selected for participation from each category. The original sample was composed of 9,747 randomly selected students in first through seventh grade.
Participants
At the onset of the study in Wave 1, 35.4% of participants fell between ages 7 and 9, 50.2% between 10 and 12, and 14.5% between 13 and 14. The sample was 59.4% male and 30.4% female, with 10% of respondents not indicating their gender. Regarding ethnicity, 59.1% of students were White, 22.5% African American, 13.4% Hispanic, 2.6% Asian/Pacific Islander, 0.7% American Indian/Alaska Native, and 0.4% Multiracial. School staff and parents provided information about students’ disability categorization. The primary disabilities of the full sample in Wave 1 were as follows: Learning Disability (12.1%), Emotional Disturbance (10.9%), Hearing Impairment (10.8%), Orthopedic Impairment (10.3%), Mental Retardation (10.2%), Speech Impairment (9.7%), Autism (8.7%), Visual Impairment (8.3%), Other Health Impairment (7.3%), Multiple Disabilities (6.8%), Traumatic Brain Injury (4.0%), and Deaf/Blindness (0.5%).
Latent class analysis sample
Participants in the present study were selected based on ADHD diagnosis and inclusion in the Other Health Impairment (OHI) disability category. We limited the sample to students with ADHD and placement only in the Other Health Impairment (OHI) disability category to reduce the potentially confounding influence of secondary characteristics associated with intellectual, learning, and emotional or behavioral disabilities that might affect victimization risk. An ADHD diagnosis was reported by either parents or school staff for 3,468 students (35.6%). Of these students, 684 were served under the OHI category as reported by either parents or teachers, suggesting that ADHD was likely their primary disability. Due to portions of missing teacher and/or parent surveys in Wave 1, 57.5% of students (n = 393) were excluded from analyses. Chi-square analyses determined that students who were included and excluded did not differ significantly based on gender, χ2(1) = 1.01, p = .32, race/ethnicity, χ2(5) = 5.20, p = .39, or age, χ2(7) = 12.95, p = .07. A significant difference was found in parental income, χ2(2) = 7.91, p = .02. Cramer’s V was obtained as a post hoc measure to determine the strength of this association, and a weak effect size was found (v = 0.11). Thus, 291 students were selected for inclusion in the present study. Gender representation for this group was 74.0% male (n = 217) and 25.1% female (n = 73) with one student missing information regarding gender. Race/ethnicity as reported by parents was 82.1% White (n = 239), 12.4% African American (n = 36), 4.5% Hispanic (n = 13), 1.0% other race/ethnicity (n = 3).
As students were followed over a 6-year period, attrition was present within Waves 2 and 3. Missing data values for the bully victimization composites ranged from 1.7% in Wave 1 to 25.1% in Wave 2, and 30.9% in Wave 3. Students with and without missing data in Wave 2 did not differ based on age, χ2(7) = 3.20, p = .87, gender, χ2(1) = 0.04, p = .85, race/ethnicity, χ2(3) = 0.352, p = .95, or parental income, χ2(2) = 0.84, p = .37. Students with and without missing data in Wave 3 also did not differ based on age, χ2(7) = 1.79, p = .97, gender, χ2(1) = 0.96, p = .17, or race/ethnicity, χ2(3) = 6.52, p = .09. Significant differences were found related to parental income in Wave 3, χ2(2) = 6.55, p = .04. However, Cramer’s V indicated that the strength of this association was weak (v = 0.15). To handle missing data, the full information maximum likelihood (FIML) estimation method was used in Mplus. Sampling weights from the parent survey were applied to the analyses to account for the randomized stratified structure of the dataset (Thomas, Heck, & Bauer, 2005).
Measures
The SEELS surveys were constructed after comprehensive review of measures included in other national studies. Items selected for survey inclusion functioned well in a similar format with parents and teachers and could provide comparison with the general population. The SEELS parent survey included several items taken from the Social Skills Rating System (SSRS; Gresham & Elliot, 1990), as well, for the purpose of using a norm-referenced instrument to measure social skills, problem behaviors, and academic competence (Wagner, Kutash, Duchnowski, & Epstein, 2005).
ADHD and behavioral phenotypes
ADHD diagnosis was measured directly on parent and school questionnaires through a dichotomous survey item. An item confirming either parent or school-reported ADHD diagnosis was then used to determine inclusion in analyses. Students’ ADHD subtypes were not directly measured by a specific item in the SEELS survey. Information regarding student behaviors was included, however, and a number of these aligned with the descriptions of ADHD subtypes found within Diagnostic and Statistical Manual of Mental Disorders (5th ed.; DSM-5; American Psychiatric Association, 2013) criteria and child psychopathology literature (Gaub & Carlson, 1997).
The inattention composite was comprised of six items in the teacher and parent surveys. Items from the teacher survey included “Gets easily distracted,” “Does things independently even when they are hard,” “Keeps at task until finished, even if it takes a long time,” and “Completes homework on time.” Two items were added from the parent survey: “Keeps working at something until he or she is finished” and “Is well organized.” A set of five items were combined to form the hyperactivity-impulsivity composite: “Acts impulsively,” “Fights with others,” and “Avoids situations likely to result in trouble,” from the teacher questionnaire, and “Speaks in an appropriate tone at home” and “Avoids situations that are likely to result in trouble,” from the parent questionnaire. Responses were provided on a scale of 1 to 3, ranging from “sometimes” to “almost always.” Positively worded items were reverse-coded, and then all items were averaged to create composite scores for each student. The Cronbach’s α for the inattention composite was 0.893 and 0.829 for the hyperactivity-impulsivity composite, suggesting both composites evidenced adequate reliability.
Victimization
Victimization status was assessed through three items presented on the parent questionnaire. These items asked if the student had been “bullied/picked on by other students,” “teased/called names at school,” and “physically attacked” during the most recent school year. Parents answered “yes” or “no” to these items and responses were combined to create a Victimization Composite score, ranging from 0 to 3 in value.
Parental income
Annual student household income was assessed using a single categorical item from the SEELS parent interview categorized into three income levels: low (US$25,000 and less), medium (US$25,001–US$50,000), and high (more than US$50,000).
Stimulant medication use
Medication status was measured with a dichotomous variable on the parent survey. Parents answered “yes” or “no” to whether their child was prescribed stimulant medication.
Data Analysis
The primary analysis of this study was a Longitudinal Latent Class Analysis (LLCA; Collins & Lanza, 2010), a Growth Mixture Modeling (GMM) technique that enables the grouping of individuals based on specific characteristics over time. Thus, LLCA permits differences in patterns of growth among subgroups as their variance centers on separate means. This method is commonly used to define unobservable subgroups in a sample with categorical variables, based on responses to a set of items.
Prior to conducting the LLCA, it was necessary to complete a latent profile analysis (LPA) using Mplus 8.0 (Muthén & Muthén, 2017) to divide the students into subgroups or classes based on reported victimization status across time points. LPA is a common person-centered mixture modeling technique that is used to derive classes from the means of continuous observed variables, such as the frequency or intensity of bully victimization, thus differentiating homogeneous subgroups within a larger sample (Berlin, Parra, & Williams, 2013; Berlin, Williams, & Parra, 2014). A number of fit indices were used to determine the optimal model: Bayesian information criterion (BIC; Schwartz, 1978), with lower values signifying better fit; entropy values, with values closer to 1.0 indicating higher accuracy of classification (Berlin et al., 2014); and Lo-Mendell-Rubin test (LMR; Lo, Mendell, & Rubin, 2001), with p values < .05 indicating improvement in model fit. Due to our use of sampling weights, the Bootstrap Likelihood Ratio test (BLRT; McLachlan & Peel, 2000), a commonly used model fit index could not be accessed in Mplus.
LPA results informed the LLCA, which included inattention and hyperactivity-impulsivity as covariates and potential predictors of bully victimization. The remaining disability categories were not included as covariates to avoid increasing the complexity of the analysis. The data met statistical assumptions of LPA (e.g., the use of categorical or ordinal data, independence of observations in each class). To determine whether key demographic variables were significantly associated with latent class assignment, a three-step approach compared classes based on age, gender, race/ethnicity, parental income, and stimulant medication status (Asparouhov & Muthén, 2014; Blake et al., 2016; Froehlich et al., 2007). These variables were selected due to their potential to influence trends in both victimization status and ADHD symptomatology. As none of these variables were significant in the LPA, they were excluded in the subsequent LLCA.
Results
Preliminary Analyses
A preliminary analysis of responses in the latent class analysis sample indicated that 65.1% of students in Wave 1, 48.1% of students in Wave 2, and 42.6% of students in Wave 3 experienced some form of bully victimization. Based on single item responses on the teacher survey for inattention (“Gets easily distracted”) and impulsivity (“Acts impulsively”), preliminary analyses revealed that 24.7% of students were described as “sometimes” inattentive, while 60.6% were described as “very often” inattentive in Wave 1. In contrast, 40.4% were “sometimes” impulsive and 53.6% were “very often” impulsive. Preliminary results suggested that, as expected, ADHD-related behaviors were common among children within the dataset who had a confirmed diagnosis of ADHD and received services within the Other Health Impairment category.
Behavioral Phenotype Analysis
To identify whether bully victimization status differed for different behavioral phenotypes of ADHD, an LPA was conducted of bully victimization status among students with ADHD. Students were divided into classes based on reported bully victimization across time points while controlling for key demographic variables. An LPA model with three classes was determined to provide the best fit to the data based on Akaike information criterion (AIC), BIC, entropy, and LMR values (see Table 1). Given that the three-class model provided the greatest interpretability of class delineation, it was selected for the LLCA.
Comparison of Latent Profile Analysis Models.
Note. The chosen number of latent class was bolded in the table.
The LPA revealed that a large proportion of the students (35.8%) presented with a declining high victimization profile (Class 1). However, more students fell into the low victimization profile, with 39.0% exhibiting a declining victimization trend (Class 2). Finally, 25.1% of students exhibited a stable, high victimization trend over time (Class 3). Table 2 includes the means and slopes of each profile and the trajectories of each class are depicted visually in Figure 1.
Comparison of Latent Class Characteristics.
Note. Results based on three-class Longitudinal Latent Class Analysis. Standard deviations are in parentheses for inattention and hyperactivity-impulsivity. Standard errors are in parentheses for intercept and slope.
Comparison of latent profile trajectories.
Victimization profiles and ADHD predictors
The second step of the analyses added the Inattention and Hyperactivity-Impulsivity composites as covariates in the LLCA. Results suggested that hyperactive-impulsive behaviors had a significant effect on the intercept of each victimization profile at Wave 1 (p = .03). Inattentive behaviors had only a marginally significant effect on the intercept of each victimization profile (p = .06). Hyperactivity-impulsivity was a significant predictor of latent class assignment and had a significant positive effect on the trend of the victimization profiles across Wave 2 and Wave 3 (p < .01). Inattentive behaviors had no such effect on the trend of the victimization profiles (p = .182).
Declining high victimization profile (Class 1)
Students assigned to the first latent class (35.8%, n = 104) presented with an average intercept of 1.63 (p < .01) and an average slope of −1.20 (p < .01). They exhibited a high level of victimization initially that steeply declined across Waves 2 and 3. In relation to ADHD symptoms, students in Class 1 showed the highest level of inattention of all classes (M = 1.76, SD = 0.32) and a high level of hyperactivity-impulsivity (M = 2.15, SD = 0.35). To examine the magnitude of inattention separately in each group, an analysis of variance (ANOVA) with post hoc tests using Bonferroni adjusted alpha levels of .017 per test (0.05/3) were employed. A significant group effect was found, F(2, 287) = 9.82, p < .01. Analysis using the Scheffé post hoc criterion for significance indicated that students in Class 1 (M = 1.76, SD = 0.32) exhibited a significantly higher level of inattention in comparison to Class 2 (M = 1.56, SD = 0.34), p < .01, but not Class 3 (M = 1.65, SD = 0.34), p = .09.
Declining low victimization profile (Class 2)
Students in the second latent class (39.0%, n = 114) presented with an average intercept of 0.06 (p = .75) and an average slope of −0.71 (p = .02). They initially displayed a low level of victimization and the level of victimization slightly declined across Waves 2 and 3. In relation to ADHD symptoms, students in Class 2 showed the lowest level of inattention (M = 1.56, SD = 0.34) and the lowest level of hyperactivity-impulsivity (M = 2.05, SD = 0.36). To examine the magnitude of hyperactivity-impulsivity separately in each group, an ANOVA with post hoc tests using Bonferroni adjusted alpha levels of .017 per test (0.05/3) were employed. Results indicated that the effect of group was significant, F(2, 287)= 5.55, p < .01. Post hoc analysis using the Scheffé post hoc criterion for significance indicated that students in Class 2 exhibited the lowest level of inattention (p < .01) and hyperactivity-impulsivity (p < .01) in comparison to other classes.
Stable high victimization profile (Class 3)
Students assigned to the third latent class (25.1%, n = 73) showed a stable, high victimization profile across time. They presented with an average intercept of 2.90 (p < .01) and slope of 0.39 (p = .89). Students in this class reported a very high level of victimization at Wave 1 and continued reporting high victimization over time. In relation to ADHD symptoms, students in Class 3 showed a high level of inattention (M = 1.65, SD = 0.33) and the highest level of hyperactivity-impulsivity (M = 2.21, SD = 0.32). Furthermore, the ANOVA of group effect on hyperactivity-impulsivity with post hoc analysis using the Scheffé post hoc criterion suggested that students in Class 3 exhibited the highest level of hyperactivity-impulsivity in comparison to other classes, F(2, 287) = 5.55, p < .01.
Discussion
Children with ADHD are at higher risk for bullying victimization in comparison to those without the disorder and those with other types of special needs (Blake et al., 2016; Twyman et al., 2010). Given the lack of study into whether behavioral phenotypes of ADHD influence bully victimization trends, the purpose of this study was twofold to examine whether students presenting with a high degree of hyperactive/impulsive behaviors versus inattentive behaviors would be at greater risk for victimization and to explore the effect of changes in ADHD symptom presentation on victimization risk. We hypothesized that students with ADHD exhibiting a high level of hyperactive/impulsive behaviors would be more likely to experience bully victimization given their status as provocative victims (Olweus, 1993). We also hypothesized that the trajectory of bully victimization would shift over time for students with ADHD given the tendency of hyperactive and impulsive behaviors to lessen in frequency and intensity as children mature (Willcutt et al., 2012).
Consistent with our hypotheses, we found that hyperactive/impulsive behaviors were indeed predictive of both students’ initial victimization experiences and the trajectory of victimization across time. More specifically, students exhibiting the highest level of hyperactive-impulsive symptoms fell into the third victimization class that was initially the highest at Wave 1 and remained high across Waves 2 and 3. Contrary to our second hypothesis, however, students who were highest in hyperactivity/impulsivity (Class 3) showed a stable trajectory for high victimization experience, rather than increasing or decreasing over time. They presented with a significantly greater level of hyperactivity-impulsivity in comparison with those in the first and second classes, who initially reported a lower level of victimization. Thus, we found that hyperactive/impulsive behaviors were more common among students reporting a higher level of victimization, to the extent that hyperactivity/impulsivity significantly predicted their trajectory.
With regard to our second hypothesis, inattention was not a significant predictor of students’ victimization trajectories. The profile with the highest level of inattention (Class 1) exhibited a decreased victimization trajectory over time. However, inattention was a pervasive symptom, exhibited to a moderate or high degree by the majority of students in the sample. Students presented with a higher level of inattention than hyperactivity-impulsivity within all three victimization profiles. Overall, our findings suggest that most participants’ behaviors coincided best with a combined presentation of ADHD.
Limitations
A number of limitations must be considered when interpreting the findings of this study. First, although parent and teachers ratings of ADHD-related behaviors may be more accurate than self-reported ratings by students, parent-reported victimization may not yield as accurate ratings as peer, teacher, or even self-reported victimization (for discussion, see Blake et al., 2016). This is particularly true for students with ADHD, due to parents’ reduced ability to observe interactions with peers and reliance on the reports of their children, who may have a less accurate perception of others’ behaviors toward them (e.g., desire to harm, intentionality of actions). In addition, parents were not provided with a definition of bullying to guide their responses, which may have resulted in an overestimation of bullying incidence (Lessne & Yanez, 2018). Future research should incorporate a definition of bullying into the survey and utilize multiple sources of information to better account for students with ADHD victimization experiences.
Second, the scope of this study was limited by the variables collected within the SEELS dataset. These variables reflected only students’ victimization experiences and did not take into account bullying perpetration. In addition, they did not incorporate key aspects of the definition of bullying (such as repetition, imbalance of power), and the data did not contain adequate information related to the type victimization experienced or how students responded. Finally, the dataset did not inquire about ADHD subtypes when parents confirmed diagnosis of the disorder, which was a key aspect of the present study.
With regard to ADHD subtypes, using composites to measure behavioral phenotypes could also be viewed as a limitation. The composites were based on items related to functional impairment that did not reflect all DSM-5 diagnostic criteria for the hyperactivity/impulsivity and inattention subtypes; however, they did evidence strong psychometric properties suggesting that they assessed core features of these symptoms. Students were not separated by predominant subtype presentation in this study, but all students’ inattention and hyperactivity-impulsivity were considered together for latent class analyses and then divided naturally based on their victimization experiences. This approach may be advantageous because many children with ADHD are diagnosed with a combined presentation type, having both inattentive and hyperactive-impulsive symptoms of differing severity. However, division by subtype based on an accurate preexisting diagnosis would provide results that are more easily applicable to school and community settings.
Implications and Future Directions
The major finding of this study suggests that among students with ADHD in special education, hyperactive-impulsive behaviors are predictive of bully victimization. These results are not only consistent with Olweus’ conceptualization of the “provocative victim,” but provide clarification as to which provocative behaviors place students at risk for being bullied. Students who tend to exhibit inappropriate movements or verbalizations, have difficulty waiting their turn, interrupt frequently, talk or play loudly, lack regard for others’ personal space, and/or grow highly excitable are more likely to bother peers and elicit bullying behaviors. Furthermore, students exhibiting a high degree of such behaviors are more likely to continue experiencing victimization over time, which may eventually lead to peer rejection and subsequent negative outcomes (Holmberg & Hjern, 2008; Wiener & Mak, 2009). While students with ADHD exhibiting fewer hyperactive/impulsive symptoms may struggle with peer relationships and certainly experience bully victimization, those with a profile characterized by elevated hyperactive/impulsive symptoms are likely to need a higher level of intervention.
Previous study findings highlight that students with ADHD are at risk of victimization across age and grade levels (Taylor et al., 2010; Timmermanis & Wiener, 2011). This, combined with the current study, helps increase understanding of how researchers can modify prevention and intervention strategies for the specific subpopulation. First, it is essential for individuals working with students with ADHD, such as parents, teachers, paraprofessionals, and other school staff, be observant of their peer relationships. There may be opportunities to assist these students in interpreting social situations and helping them build friendships with other students. If parents or school staff observe peer rejection or bully victimization, it is important to consult with mental health professionals and administrators who can intervene and assist the victim in the situation using appropriate preventive measures. The key role of student bystanders in helping victims gain assistance and raising awareness of bullying also cannot be understated, as student-initiated actions may be more powerful than those taken by adults (Denny et al., 2015).
Research examining prevention and intervention tools to help reduce victimization among children with disabilities is scarce (Rose et al., 2011), but is slowly gaining momentum. While universal anti-bullying programs are now common in schools, most of these programs have not been extensively evaluated to determine whether they can effectively reduce bullying among special needs populations (Hartley et al., 2017). To date, there are two interventions that meet criteria for evidence-based practice (Houchins, Oakes, & Johnson, 2016). Ross and Horner (2014) assessed the effectiveness of integrating simple bullying prevention strategies into existing School-Wide Positive Behavior Interventions and Supports for elementary students. These strategies involved using a “stop” signal when disrespectful behavior is experienced or witnessed, walking away to remove attention from the behavior, and telling an adult if the behavior continues. Espelage, Rose, and Polanin’s (2015) randomized controlled trial of the Second Step: Student Success through Prevention Program (Second Step) with a sample of middle school students with disabilities is rated highly as meeting standards for evidenced-base practice. The program provides explicit instruction in social-emotional learning, focusing on key areas such as emotional awareness, communication skills, problem-solving, and education about school violence, making it ideal for students with ADHD who might exhibit difficulties with impulse control. Whereas Second Step resulted in a significant reduction in bullying perpetration for students with disabilities, these effects were not found for victimization, suggesting that other areas of intervention must be examined to assist students with disabilities targeted as victims.
Given our findings that ADHD-related symptoms in and of themselves are predictive of bully victimization, research on effective treatments for ADHD may provide needed guidance and future direction for this subgroup. In addition to developing greater understanding of social situations and coping skills for victims, tools focused on promoting increased behavioral inhibition, self-control, and emotional regulation might be beneficial for students with ADHD who experience bullying (Rahill & Teglasi, 2003; Vessey & O’Neill, 2011). Integrating these topics into existing support groups and curricula may help reduce the frequency of victimization. Although medication may affect these aspects of ADHD presentation as well, it is not guaranteed to generalize to peer interactions or improve social skill deficits. The most effective treatment for ADHD symptoms resulting in reduced functional impairment across settings combines medication with behavioral parent management training (Chronis, Jones, & Raggi, 2006). No such training programs are designed for teachers at this time. However, many teachers engage in consultation with school mental health professionals who recommend behaviorally based classroom intervention strategies (e.g., differential attention, behavior-specific praise, token/point systems, and daily report cards; Pelham, Wheeler, & Chronis, 1998). Utilizing appropriate behavior management techniques may have a positive impact on both student–teacher relationships and peer relationships, as disruptive behaviors become less frequent and teachers’ model positive interactions with students who have ADHD.
A unique aspect and strength of this study was its longitudinal investigation of students in special education. Participants represented are among those with the highest level of impairment and intervention needs. Consistent with prior research, preliminary results confirmed that bully victimization is a significant problem for students with disabilities, regardless of disability category, and that ADHD-related behaviors were a common concern. Although not all students with ADHD qualify for and are ultimately placed in special education, many students are actively receiving services. The bullying literature would benefit from a continued study of students in special education to identify better methods of prevention for this population. This research will hopefully result in greater awareness of ADHD as a disability with significant functional impairment, particularly among school professionals, and inspire the development of more inclusive classrooms to help students with ADHD succeed.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.