Cross-lagged Model of Bullying Victimization and Mental Health Problems in Children with Autism in Middle to Older Childhood

Abstract

Youth with autism spectrum conditions have been shown to be at an increased risk for bullying victimization. The overall aim of the present study was to identify factors associated with increased risk for bullying victimization in youth with autism during middle childhood to early adolescence (aged 5 to 12 years) and to explore the potential time-ordered associations between bullying victimization and mental health problems 12 months later. The average age of the youth with autism was 7.90 years (SD = 2.33), 86% were male, 34.6% had an intellectual disability, and 84% were white, non-Hispanic. Youth with autism who experienced bullying victimization (versus no victimization) were older in age, had more severe autism symptoms, and higher levels of internalizing and externalizing mental health problems at study onset. Using two cross-lagged structural equation models, the associations between maternal report of youth bullying victimization and teacher-report of youth mental health problems using two waves of data were tested. Internalizing and externalizing mental health problems at Time 1 did not relate to change in likelihood of being bullied one year later. In contrast, bullying victimization at Time 1 was associated with an increase in internalizing mental health problems (β = .24, p < .05).

Bullying victimization has been defined as being the recipient of repeated physical, verbal, or relational aggressive behavior by someone else (Olweus, 1993) and is estimated to occur in 30 to 60% of school-aged youth in the general population (Molcho, Craig, Due, Pickett, Harel-Fisch, & Overpeck, 2009; National Center for Educational Statistics, 2011). Longitudinal studies of typically developing youth suggest that bullying victimization is associated with increases in mental health problems such as anxiety, depression, and aggression (Arseneault, Bowes, & Shakoor, 2010; Reijntjes, Kamphuis, Prinzie, & Telch, 2010; Ttofi, Farrington, Losel, & Loeber, 2011). While bullying emerges from complex community and group dynamics (Hong & Espelage, 2012), some youth are more vulnerable to being the target of bullying than others (Ryoo, Wang, & Swearer, 2015). As a group, youth with autism spectrum conditions (hereon referred to as autism) have a heightened risk for being the victim of bullying by both peers and siblings relative to youth who are typically developing (Blake, Lund, Zhou, Kwok, & Benz, 2012; Maiano, Aime, Salvas, Morin, & Normand, 2016; Toseeb, McChesney, & Wolke, 2018; Toseeb et al., 2019) and youth with other types of developmental disabilities (Sterzing et al., 2012; van Roekel, Scholte, & Didden, 2010; Zeedyk, Rodriguez, Tipton, Baker, & Blacher, 2014). It is not clear, however, whether among youth with autism, certain subgroups have a greater risk for being bullied than others. Moreover, there is a paucity of longitudinal research examining the potential consequences of bullying victimization on the mental health of youth with autism. The goal of the current study was to identify the subgroups of youth with autism who are most at-risk for being bullied and to determine the time-ordered association between bullying victimization and mental health problems during middle and older childhood, a time when peer bullying is at its peak for typically developing youth (Arseneault et al., 2010) and those with autism (Little, 2002). This information can inform bullying prevention programs by ensuring that the most at-risk youth with autism are reflected in school-based or community-wide messaging and strategies to prevent bullying, and are educated and supported in recognizing, reporting, and safely standing up to or escaping from bullying. This information is also critical for directing research and intervention aimed at reducing negative social processes that contribute to later mental health problems in youth with autism.

Currently estimated to occur in 1 in 59 children in the U.S. (Center for Disease Control and Prevention [CDC], 2018), autism is a lifelong neurodevelopmental condition characterized by impairments in social communication and rigid and repetitive behaviors and interests and often co-occurs with intellectual disability (ID) (American Psychiatric Association, 2013). It is estimated that 70–81% of youth with autism often experience multiple comorbid mental health problems (Joshi et al., 2010; Simonoff et al., 2008) that markedly impair everyday functioning, with 42% meeting diagnostic criteria for two or more mental health conditions (Leyfer et al., 2006; Moseley, Tonge, Brereton, & Einfeld, 2011; Lecavalier et al., 2019). These mental health problems include both internalizing (e.g., anxiety and depression) and externalizing (e.g., aggressiveness, impulsivity, hyperactivity, and self-injury) difficulties (Brookman-Frazee, Stadnick, Chlebowski, et al., 2018; Skokauskas & Gallagher, 2012)

In studies on the general population and youth with other types of disabilities, there is substantial evidence that bullying victimization has a negative effect on psychosocial development, including predicting an increase in internalizing and externalizing mental health problems (Kouwenberg, Rieffe, Theunissen, & de Rooij, 2012; Rieffe, Camodeca, Pouw, Lange, & Stockmann, 2012; Sigurdson, Undheim, Wallander, Lyderen, & Sund, 2015; Turner, Exum, Brame, & Holt, 2013; van den Bedem, Dockrell, van Alphen, Kalicharan, & Rieffe, 2018). However, there is also evidence that these same mental health problems are a risk factor for bullying victimization (Arsenault, Bowes, & Shakoor, 2010; Cook et al., 2010). In other words, youth with internalizing and externalizing mental health problems may be vulnerable to being the target of bullying; and in turn, the experience of being bullied may subsequently increase these mental health problems. The majority of research on bullying victimization in youth with autism has been cross-sectional; thus, the time-ordered nature of an association between bullying victimization and mental health problems in youth with autism is unclear. In cross-sectional analyses, bullying victimization has been linked to a higher level of both internalizing and externalizing mental health problems in students with disabilities receiving special education (Blake, Kim, & Lund et al., 2016) and youth with autism (Cappadocia et al., 2012; Storch et al., 2012; Zablotsky et al., 2013). Similarly, youth with autism who were reported to have been bullied by a sibling were also reported to have a higher concurrent level of mental health problems, specifically internalizing symptoms (Toseeb et al., 2018). Other studies have found emotional functioning, specifically dysregulated anger to serve as both a predictor and consequence of bullying victimization in children and with autism (Rieffe, Camodeca, Pouw, Lange, & Stockmann, 2012). Only one longitudinal study to-date has explored the time-ordered directional associations between bullying victimization and mental health problems in youth with autism. Tipton-Fisler and colleagues (2018) examined internalizing mental health problems as both a predictor and consequence of bullying victimization in three groups: 40 adolescents with autism (without ID), 34 adolescents with ID (without autism), and 82 adolescents with typical development. Bullying victimization was measured through a composite variable, capturing mother and youth report of the severity, chronicity, and impact of having been bullied. Of those bullied (regardless of group), a higher level of internalizing mental health problems reported by parents when the youth was aged 13 years predicted a higher composite bullying victimization score when the youth was aged 15 years (Tipton-Fisler et al., 2018). In the opposite direction, bullying victimization at age 13 years did not predict change in level of internalizing mental health problems at age 15 years. Thus, during the adolescent years, youth with autism who have a high level of internalizing mental health problems appear to be vulnerable to being bullied; however, the experience of having been bullied did not subsequently increase these problems. The Tipton-Fisler et al. 2018 study did not examine externalizing mental health problems; thus, the impact of bullying victimization on externalizing mental health problems of youth with autism is not known.

Among youth with autism and other types of neurodevelopmental conditions, there is evidence that some youth are at greater risk for being bullied than others. For example, based on parent report, youth with autism (aged 5–21 years) who were the victims of bullying were younger and had more severe autism symptoms (i.e., social communication difficulties), than those who had not experienced bullying victimization (Cappadocia et al., 2012). As previously stated, a higher level of internalizing mental health problems was found to be a risk factor for later bullying victimization in youth with autism (Tipton-Fisler et al., 2018). In studies on typically developing youth, mixed findings have been reported in terms of whether males vs. females are more likely to be bullying victims (Carlyle & Steinman, 2007). Studies of students receiving special education services found no gender difference in bullying victimization (Blake, Zhou, Kwok, & Benz, 2016; Farmer et al., 2012). Whether the presence of ID in addition to autism alters the risk for being the target of bullying is not known. On one hand, the presence of ID in addition to autism may be protective and reduce opportunities for bullying victimization; youth with autism and ID are more likely to be in segregated classrooms and added supervision may mean that teachers or caregivers can intervene if peer interactions become negative (Sterzing, Shattuck, Narendorf, et al., 2012). On the other hand, the presence of ID in addition to autism may make youth more prone to being bullied, as these individuals may have a more ‘observable’ condition and/or have fewer skills to recognize, report, and/or stand up to bullying. Indeed, research has shown that youth with ID (without autism) experience a higher frequency of bullying victimization than their peers with typical development (Christensen, Fraynt, Neece, & Baker, 2012).

The present study built on previous literature by investigating youth characteristics associated with bullying victimization and the time-ordered associations between bullying victimization and internalizing and externalizing mental health problems in 187 youth with autism aged 5– 12 years. There were two study aims: 1) examine likelihood of bullying victimization based on the youth’s age, gender, autism symptom severity, ID status, and mental health problems; and 2) determine the time-ordered relation between bullying victimization and internalizing and externalizing mental health problems 12 months later in youth with autism. Based on previous findings (Cappadocia et al., 2012; Zeedyk et al., 2014), we hypothesized that youth with autism who were older (vs. younger), had a higher (vs. lower) severity of autism symptoms, and had a higher (vs. lower) level of internalizing mental health problems, would be reported to have experienced a higher frequency of bullying victimization. A priori hypotheses about the association between gender and ID status and bullying victimization were not made given the paucity of previous research to drive predictions. The experience of having been bullied was predicted to be associated with an increase in internalizing and externalizing behavior problems 12 months later, given substantial evidence of this pathway of effects from typically developing populations (Turner, Exum, Brame, & Holt, 2013), albeit this effect was not found in the only other longitudinal study on youth with autism (Tipton-Fisler et al., 2018).

Methods

Participants

Analyses are based on data from 187 mothers who had a son or daughter with autism who participated in the first and second time points of an ongoing longitudinal study examining family processes in the context of youth autism. Families were recruited through research listservs, school/childcare mailings, and fliers posted at autism clinics and in community settings (e.g., libraries). Study inclusion criteria for the larger study included families in two-parent households (parents did not have to be married and they did not have to be a biological parent of the child) who had a child aged 5–12 years who had received a medical or educational diagnosis of autism. Analyses used in the current study involved information collected during the first two waves of the larger study and focused on data collection from mothers and the child with autism’s teacher. Child autism diagnosis was verified via medical or educational records and diagnostic evaluations had to have included the Autism Diagnostic Observation Schedule (ADOS; Lord et al., 2000; Lord et al, 2012). Additionally, parents completed the Social Responsiveness Scale-Second Edition (SRS-2; Constantino & Gruber, 2012) to assess the child’s current severity of autism symptoms. Five children had a mother-reported SRS-2 Total T-score < 60. Upon case review, however, these families were included in analyses because the child had received an ADOS score in the autism range and the teacher-reported SRS-2 Total T-score was > 60.

The present study is based on two time points of data collection, spaced 12 months apart (range 12.0 to 15.2 months). Of the 187 families enrolled at Time 1 of the study, 84% of the youth with autism were white, non-Hispanic, 86% were male, 34.6% had ID, and their average age was 7.90 years (SD = 2.3). At Time 1, 75.4% of youth with autism attended an elementary school (K-5^th grade), 12.3% middle school (6^th and 7^th grade), and 12.3% another type of school (e.g., preschool or alternative education placement). The majority of youth with autism were receiving specialty services outside of home, including occupational therapy (60.3%), speech related services (72.5%), behavior support training (55.6%), and counseling (13.8%). The average age of mothers was 38.7 years (SD= 5.6), and 44% had a college degree. The average household income was $80–89K (SD = ~30K). Twenty-six (14%) of the families did not participate in data collection data at Time 2. These families did not significantly differ from those who participated at Time 2 in mother age (t (184) = .34, p =.73), youth age (t (185) = −.97, p = .33, mother years of education (t (185) = .69, p = .49), household income (t (181) = .55, p = .58), youth SRS-2 Total score (t (185) = .90, p = . 37), or mental health problems on the Child Behavior Checklist (CBCL) Total score. (t (185) = .72, p = 47).

Procedure and Measures

The Institutional Review Board of the participating university approved study measures and procedures. At Time 1 and Time 2, mothers reported on family socio-demographics as well as the youth with autism’s bullying victimization and autism symptoms. Mothers identified a teacher or staff person who had regular interaction with the youth with autism and knew him/her well. This teacher or staff person was contacted by research staff and asked to report on the youth with autism’s mental health problems. Mothers were paid $50 and teachers were given a $5 giftcard at each time point for participation.

Youth Sociodemographics

Mothers reported on the youth with autism’s gender, and this variable was coded (0) female or (1). Mothers also reported on the youth’s date of birth, which was used to calculate child age in years. Youth ID status was assessed through a medical and educational record review and coded present (1) vs. absent (0). If the youth with autism had received a medical diagnosis of ID and/or if they met criteria based on standardized scores from direct and parent-reported measures of cognitive functioning and adaptive behavior. Child gender was reported on by mothers and coded: (0) female or (1) male.

Autism Symptom Severity.

Severity of the youth’s autism symptoms was assessed via the SRS-2 (Constantino & Gruber, 2012), which is a 65-item measure that provides an overall assessment of social impairment associated with ASD. Mothers were asked to rate each item on a 4-point Likert scale (1=Not true, 2= Sometimes true, 3= Often true, and 4= Almost always true). The Total raw score for the SRS-2 ranges from 0–195, with higher scores indicating a higher severity of impairment. The internal consistency, inter-rater reliability, and test-retest reliability of the SRS-2 has been shown to be strong (Constantino & Gruber, 2012). Additionally, the SRS-2 has demonstrated criterion validity with the ADI-R, with correlations ranging .52 and .79 (Constantino & Gruber, 2012). The SRS-2 Total score had good internal consistency in the current sample (Cronbach’s alpha .87).

Bullying Victimization.

Frequency of bullying victimization was assessed via parent-report in response to the question “How often has your child been bullied in the past 4 weeks?” in relation to four types of bullying (physical, verbal, social, and/or cyber). Mothers were given five response options: never, once or twice, two or three times, once per week, and several times per week. This item has been used in previous research examining parent-reported frequency of bullying victimization of youth with ASD or ID (Cappadocia et al., 2012; Zeedyk et al., 2014). In order to examine the association of bullying victimization on later mental health problems, a dichotomous variable (no victimization = 0 and any victimization = 1) was created. Across studies, bullying victimization of youth with disabilities (e.g., Zeedyk et al., 2014) has been assessed through informant (e.g., parent and teacher) and/or self-reports. In general, moderate levels of agreement have been found between youth and parent report of bullying experiences, with parents being more congruent with youth report of bullying than were teachers (Demarary, Malecki, Secord, & Lyell, 2013). In the current study, maternal report of bullying victimization was used, as youth report was not available. In our sample, mother-reported youth bullying victimization was significantly positively correlated (r = .23 to .25; ps < .05), with individual items on the teacher-reported CBCL that tap into social difficulties with peers (“Gets teased a lot,” “Does not get along with other kids” and “Not liked by peers”). The relatively small-size of these associations may reflect differences in perspectives or awareness, but also the fact that teacher-reported CBCL items do not necessary reflect bullying, as not all social difficulties with peers surmount to bullying.

Youth Mental Health Problems.

The severity of the child’s internalizing and externalizing mental health problems were assessed through teacher report on the CBCL (TRF; Achenbach & Rescorla, 2001). Teacher-reported CBCL was used rather than maternal-reported CBCL to reduce shared variance from a single-reporter and to capture mental health problems as presented in the context of the school. Teachers rated the frequency of each problem using a 3-point scale with responses ‘Not True’ (0), ‘Somewhat True’ (1) and ‘Very True or Often True’ (2). The CBCL Internalizing and Externalizing Total T-scores were used in the present analyses. Alpha coefficients for the school-age form at Time 1 were .85 on the Total Internalizing Behavior Problems subscale; the Externalizing Behavior Problems subscale had an alpha coefficient of .90. The CBCL has been shown to have good internal consistency, strong construct validity, and has been used to assess behavior problems in ASD samples (Mazefsky, Anderson, Conner, & Minshew, 2011; Sikora, Hall, Hartley, Gerrard-Morris, & Cagle, 2008). In the current study, the Cronbach alpha for TRF Total T score was .93 to .94 across the time points. Overall, mother and teacher Internalizing and Externalizing Total T-scores were significantly associated across time points with paired samples correlations ranging from .25 to .38 (ps < .01). This moderately-sized association is consistent with ranges typically reported in studies examining parent-teacher agreement of mental health problems in youth with autism (McDonald, Lopata, & Donnelly, 2016; Stratis & Lecavalier, 2015).

Data Analysis

Descriptive data and boxplots were used to examine the distribution of study variables. Analyses were first conducted in SPSS version 24.0 to examine the association between Time 1 youth ID status, age, gender, severity of autism symptoms, and mental health problems and frequency of bullying victimization in the past 4 weeks. Chi-square analyses were used to examine differences in bullying victimization by ID status and child gender and Cramer’s V was used to determine which categories of bullying victimization levels significantly differed from one another. A one-way ANOVA and least significant differences (LSD) post hoc tests were used to examine differences in bullying victimization in remaining continuous variables (age, autism symptom severity and mental health problems). The LSD post hoc test was used to examine significant differences between categories and is recommended in the presence of unequal sample sizes (Raykov & Marcoulides, 2008).

To examine the association between Time 1 bullying victimization and Time 2 youth internalizing and externalizing mental health problems, two cross-lagged panel analyses were conducted in structural equation modeling (SEM) in MPlus software (Muthén & Muthén, 2012). In order to provide a conservative estimate of mother reported bullying experiences, the dichotomous variable of bullying victimization vs. no victimization was used. These models allowed us to examine stability and examine associations across 12 months in both directions. One model examined the association between bullying victimization and internalizing mental health problems and the other model examined association between bullying victimization and externalizing mental health problems. In both models, youth characteristics of age, gender, and ID status were included as control variables and regressed onto Time 1 bullying victimization and mental health problems. Means and standard deviations for key measures at each time point are included in Table 1.

Table 1.

Means and standard deviations of study variables by time point.

Behavior Problems (mother self-report)	Time 1 (M [SD])	Time 2 (M [SD])
CBCL Internalizing Behavior Problems	63.05 [9.59]	61.20 [10.22]
% Above clinical cut-off	51.6	44.5
CBCL Externalizing Behavior Problems	60.09 [11.14]	57.38 [11.01]
% Above clinical cut-off	36.9	27.4
Bullying victimization (% bullied)	38.5	41.3

Note. CBCL= Child Behavior Checklist; standardized T scores used for internalizing and externalizing broadband scales. Victimization analyzed as a dichotomous variable (yes/no) to indicate presence of bullying in the past month (4 weeks).

Results

Descriptive Analyses.

In order to explore potential differences in bullying victimization risk levels resulting from individual youth characteristics, frequency of victimization was categorized into three different risk groups capturing different levels of youth bullying victimization. Regardless of the type of bullying reported by mothers of youth with autism, the ‘no victimization’ category included youth of mothers who responded “never,” the ‘low victimization risk’ included youth whose mothers responded “once” to “two to three times,” and the ‘high victimization risk’ included youth of mothers who responded “once per week” or “several times per week” in the past month. This approach was utilized in a previous study of youth with autism examining parental responses to a similar item adapted from a survey of bullying experiences in school-aged children (Cappadocia, Weiss, & Pepler, 2012).

Overall, 38.5% of mothers reported that their son/daughter with autism had been bullied at Time 1 and 41.3% at Time 2. Table 2 presents results of chi-square analyses examining frequency categories of bullying victimization levels by youth characteristics. There were no significant differences in frequency of bullying victimization categories with regards to the youth with autism’s ID status or gender. However, there was a significant difference in categories in regard to severity of autism symptoms. Compared to the ‘no victimization’ category, youth with autism in the ‘some victimization’ and ‘high victimization’ categories had a higher severity of autism symptoms (F (2, 184) = 4.58, p = .01, d =.41 and d= .63 respectively). Youth with autism in the ‘some victimization’ and ‘high victimization’ categories had a higher level of Time 1 internalizing (F (2, 183) = 5.68, p =.00, d = .43 and d = .72, respectively) and externalizing mental health problems (F (2, 184) = 3.84, p =.02, d = .21 and d = .75, respectively) compared to those in the ‘no victimization’ category. There was also a significant association between mother-reported bullying victimization level and the age of the youth with autism. Youth with autism in the ‘high victimization’ group were older than those in the ‘no victimization’ and ‘some victimization’ groups (F (2, 184) = 3.87, p = .02, d =.69 and d = .54 respectively). There was also a significant difference in youth age between the ‘no bullying’ and ‘some bullying’ victimization categories, albeit this difference was of small effect size (d = .17).

Table 2.

Descriptive analyses of bullying risk levels and child characteristics at Time 1

Child Characteristics	No Victimization (n = 116)	Some Victimization (n = 56)	High Victimization (n = 15)	χ2 or F
ID status (%)	37.9	26.8	40	χ2 (2) = 2.27
Age	7.66 (2.2)a	8.04 (2.2)a	9.33 (2.6)b	F (2,184) = 3.87*
Gender	86.2	82.1	93.1	χ2 (2) = 1.30
ASD Symptoms	76.08 (11.1)a	80.34 (9.4)b	82.20 (8.3)b	F (2,184) = 4.58**
CBCL Int. Bx.	61.32 (9.1)a	65.38 (9.9)b	67.87 (9.1)b	F (2,183) = 5.68**
CBCL Ext. Bx.	58.75 (10.5)a	61.07 (11.9)a	66.73 (10.6)b	F (2, 184) = 3.84*

Note. ID= intellectual disability; CBCL Int. Bx.= internalizing behaviors T score; CBCL Ex. Bx.= externalizing behaviors T score. Means with differing subscripts within rows are significantly different at p < .05 based on Fisher’s LSD post hoc paired comparisons. Frequencies with differing subscripts within rows are significantly different at the p < .05 level based on Cramer’s V.

^*p < .05;

^**p < .01;

^***p < .001.

Cross-Lagged Panel Models.

Figure 1 and Figure 2 show the path coefficients for the significant associations between parent-reported bullying victimization and teacher-reported mental health problems in the cross-lagged panel models. Table 3 presents the unstandardized path coefficients associations for internalizing mental health problems (Model 1) and followed by the externalizing mental health problems (Model 2). Global measures of fit using the Chi-Square statistic and root mean squared error of approximation (RMSEA), and incremental fit indices, including the Tucker-Lewis index (TLI) and the comparative fit index (CFI), were examined. The model examining associations between bullying victimization and internalizing mental health problems (CFI = 1.00, TLI = 1.00, and RMSEA = 0.00) indicated a good fit (Hu & Bentler, 1999). Model fit indices for the model examining associations between bullying victimization and externalizing behavior problems similarly indicated a good fit (CFI= 1.00, TLI= 1.00, and RMSEA= 0.00).

Figure 1.

Cross-lag panel design examining relationships between teacher reported child internalizing mental health problems and maternal report of bullying victimization

Note. Unstandardized estimates and standard errors represent associations between child internalizing mental health problems and bullying victimization. Significant effects are marked in solid lines and insignificant effects are marked in dashed lines. *p ≤ 0.05; **p ≤ .01; *** p ≤ .001.

Figure 2.

Cross-lag panel design examining relationships between teacher reported child externalizing mental health problems and maternal report of bullying victimization

Note. Unstandardized estimates and standard errors represent associations between child externalizing mental health problems and bullying victimization. Significant effects are marked in solid lines and insignificant effects are marked in dashed lines. *p ≤ 0.05; **p ≤ .01; *** p ≤ .001.

Table 3.

Standardized path coefficients for two-wave cross-lagged models for child behavior problems and bullying victimization

Model Path	Internalizing Mental Health Problems (1) β (SE)	Externalizing Mental Health Problems (2) β (SE)
Stability Effects
T1 MH to T2 MH	.43***	.67***
T1 Bullying to T2 Bullying	.38***	.39***
Cross-lagged effects
T1 MH to T2 Bullying	.04	.15+
T1 Bullying to T2 MH	.24*	.11
Control Variables
ID to T2 MH	−.20*	.15+
ID to T2 Bullying	−.19*	−.19**
Age to T2 MH	.08	−.10
Age to T2 Bullying	.05	.05
SRS-2 to T2 MH	.09	.04
SRS-2 to T2 Bullying	−.01	−.04

Note. Values reported in this table are standardized. MH= Mental health problems. Bullying= Bullying victimization risk.

⁺p < .10;

^*p < 0.05;

^**p < .01;

^***p < .001.

Table 3 displays path coefficients for cross-lagged effects between youth bullying victimization and internalizing mental health problems (Model 1). Cross-lagged panel analysis demonstrated stability effects across the two time points for both internalizing mental health problems (T1-T2: β = .43, p < .001) and bullying victimization (T1-T2: β = .39, p < .001). In Model 1, we found a significant cross-lagged effect of bullying victimization on later internalizing mental health problems (T1-T2: β = .24, p < .05); Time 1 mother-reported experience of bullying victimization was associated with higher level of internalizing mental health problems at Time 2. In contrast, Time 1 internalizing mental health problems was not significantly related to subsequent bullying victimization at Time 2 (T1-T2: β = .04, p > .05). In Model 2, youth externalizing mental health problems (T1-T2: β = .67, p < .001) and bullying victimization (T1-T2: β = .39, p < .001), both fairly stable across the two waves of data collection. There was not a significant cross-lagged effect of Time 1 bullying victimization on Time 2 externalizing mental health problems (T1-T2: β = .11, p = .22). There was also not a significant cross-lagged effect of Time 1 externalizing mental health problems and Time 2 bullying victimization at Time 2 (T1-T2: β = .15, p = .10).

Discussion

A growing number of studies have shown that as a group, youth with autism experience an elevated rate of bullying victimization than their typically developing peers and peers with other types of neurodevelopmental conditions (Forrest, Kroeger, & Stroope, 2019; Maiano, Normand, Salvas, Moullec, & Aimé, 2016). In the current sample, 39% of youth with autism were reported to have been the victim of bullying in the past month based on maternal report, with rates increasing to 41% the following year. This is within the range reported by previous studies examining bullying victimization in youth with ASD using a similar time frame (e.g., within past month) ranging from 17% to 77%, (e.g., Blake et al., 2012; Schroeder, Cappadocia, Bebko, Pepler, & Weiss, 2014; Sterzing et al., 2012). Bullying victimization across the two time points in the current study was fairly persistent. In other words, 85% (N = 66) of the 71 youth with autism who were in the “some” or “high” bullying victimization categories at Time 1 continued to experience “some” or “high” bullying victimization at Time 2. This finding highlights that bullying victimization is often a chronic experience for youth with autism.

Consistent with prior research (Cappadocia et al., 2012; Zablotsky et al., 2013; Zeedyk et al., 2014), our findings suggest that some youth with autism are more vulnerable to being the target of bulling than others. Specifically, in concurrent analyses, youth with autism who had a higher severity of autism symptoms and a higher level of internalizing and externalizing mental health symptoms at study onset were reported by mothers to have experienced a higher frequency of bullying victimization than those with a lower severity autism symptoms and level of mental health problem. Within our sample, older youth with autism also had a higher frequency of bullying victimization in the last month than those who were younger, suggesting that older childhood more so than middle childhood is a particularly vulnerable time period for peer bullying experiences in youth with autism.

In our sample, gender and ID status did not relate to differences between our three bullying victimization status groups, when other characteristics (i.e., autistic symptoms, age, behavior problems) are also taken into account. However, in regards to ID status, it should be noted that the majority (80%) of youth with autism in our sample were in inclusive mainstream classrooms. It is possible that the presence of ID in addition to autism alters the risk for bullying victimization when considering youth with comorbid ID placed in restricted or non-inclusive classrooms. Indeed, previous studies have found that educational placement in restricted or non-inclusive settings is linked to reduced bullying victimization (Hebron & Humphrey, 2014; Sterzing et al., 2012), as it possible that these settings provide more monitored supervision, smaller class sizes, and reduced perceptions of difference due to shared commonalities among students in special education placements.

Our study also adds to an understanding of the direction of associations between bullying victimization and mental health problems in youth with autism. In our cross-lagged models, bullying victimization at Time 1 was related to a higher level of internalizing mental health problems (i.e., anxiety and withdrawn behavior) one year later. In contrast, initial level of internalizing mental health problems was not associated with subsequent risk for bullying victimization. This finding suggests that that the experience of bullying victimization may contribute to internalizing mental health problems in youth with autism during middle childhood to early adolescence, but not vice versa. This time-ordered pattern is in opposition to a previous study (Tipton-Fisler et al., 2018) of youth with autism (without ID) in their mid-adolescence. It is possible that the time-ordered pathways between bullying victimization and internalizing mental health problems differ at older vs. younger developmental stages. Our findings are consistent with longitudinal studies from non-autism samples of youth in middle childhood to early adolescence which have similarly found evidence of adverse effects of bullying victimization on later internalizing mental health problems (Arseneault et al., 2010; Reijntjes et al., 2010). It may be that with increasing age, the cumulative effects of social isolation and peer victimization also serve to predict future victimization due to limited social networks in high school and unpopularity with peers. In the current study, bullying victimization was not found to be associated with changes in externalizing mental health problems (i.e., aggression, hyperactivity, and impulsivity) across the two time points, albeit concurrent associations between externalizing mental health problems and bullying victimization were found. Findings regarding the temporal nature of bullying victimization and externalizing mental health problems in non-autism samples have been mixed (Cook et al., 2010). Thus, in both autism and non-autism populations, bullying victimization may have greater implications for internalizing than externalizing mental health problems.

Summary and Implications for Practice

Overall, we found that more than one-third of youth with autism in middle childhood to early adolescence were reported by mothers to have been bullied in the past month. Bullying victimization was largely a chronic experience, with the majority of youth reported to have experienced bullying victimization in the past month at study onset also reporting this experience one year later. Bullying victimization was higher in youth with autism in older childhood than for those in middle childhood, and higher for youth with more severe autism symptoms and mental health problems in concurrent analyses. When examining time-ordered associations, mother report of bullying victimization was associated with an increase in teacher-reported internalizing mental health problems one year later, but not vice versa.

Currently, bullying prevention efforts have focused on school-wide messaging, policies, and strategies aimed at reducing bullying behaviors and creating a positive school climate (Rose, 2010). These programs have shown to reduce bullying perpetration and victimization in the general population samples (Ttofi & Farrington, 2011). However, across studies, effects are greatest for studies conducted outside the United States and those using homogenous samples of students. There is also evidence that attitudes about bullying may be reduced more than actual bullying behaviors (Merrell et al., 2008). Given the high rate of bullying victimization in youth with ASD, it has also been argued current school-wide approaches may not be sufficient for addressing the needs of specific at-risk groups, and thus programs specific to youth with autism are emerging (e.g., Laugeson et al., 2012; Rose & Monda-Amaya, 2012). Information from the current study suggests that increased internalizing problems follow bullying victimization in youth with autism, and thus, highlights the critical need for this work. Moreover, our findings suggest that these bullying programs may need to direct particular attention to subgroups of youth with autism who may be most at-risk for bullying victimization such as those with more autism symptoms and those who were bullied in the past (given the chronic nature of bullying). These youth may benefit from specialized training and support around identifying, reporting, and standing up to or escaping bullying.

The sustained nature of bullying at older ages for autistic youth, combined with increasing mental health concerns in adolescence, may both contribute to greater social difficulties. From a socio-ecological perspective, it is important to utilize multilevel strategies that target the mechanisms and relational nature of bullying experiences across different perspectives of bullies, victims, bystanders, teachers, and families (Espelage & Swearer, 2010). However, our study, along with previous studies, highlights important information that may be helpful to adapt approaches to intervention at the individual level. School-wide approaches are not sufficient for addressing bullying in schools, especially in youth with autism that may require more tailored approaches of intervention focused on skill development in the areas of social skills, adaptive emotional and behavioral regulation, assertive social communication, and social problem-solving. There are a few interventions that target specific subgroups of youth who are most vulnerable and at increased risk of bullying victimization including those with ASD or other types of disabilities (e.g., Laugeson et al., 2012; Rose & Monda-Amaya, 2012). Our findings highlight the importance of examining longitudinal trends in bullying and autism and their effect on youth indicators of psychological well-being.

Strengths, Limitations, and Future Directions

The current study had several strengths. The study expanded on previous cross-sectional studies by including two time points of data collection in a relatively large sample of youth with autism. Problems of single reporter shared variance were avoided by using mother-report of youth bullying victimization and teacher-report of youth mental health problems. The use of informant reports also allowed us to include a broader range of youth with autism in the study, including those with ID.

There are several limitations to the current study. Although parent-report of bullying has been found to have moderate agreement with self-report of bullying victimization (Sawyer, Mishna, Pepler, & Wiener, 2011; Zeedyk et al., 2014), mother report may underestimate bullying victimization as mothers may not always be aware of these experiences. Future research that includes youth self-report as well as teacher-report of bullying experiences is important. Similarly, future studies should include self-report of mental health problems, given evidence that teacher-report may similarly be an underestimate of mental health problems, especially internalizing mental health problems (Cunningham & Suldo, 2014; Dyer, Nicholson, & Battistutta, 2006). Reliance on teacher-report also only captures mental health problems expressed within a school-setting. The current study is also limited in that it only assessed frequency of bullying victimization. More extensive and nuanced measures of bullying victimization (e.g., severity and type) may be better able to capture the full range and consequence of these experiences on the mental health of youth with autism. The current sample is also limited in terms of largely representing families of non-Hispanic White racial/ethnic backgrounds and of middle socioeconomic status, and involving two-parent families. It is possible that the frequency of bullying victimization and/or its association with mental health problems differ in families of minority racial/ethnic backgrounds, those of lower socioeconomic status, and single-parent households, who reflect groups at-risk for experiencing additional daily stressors related to family functioning (Tucker, Finkelhor, Turner, & Shattuck, 2014; Wolke, Tippett, & Dantchev, 2015).

The current study is also limited in that it assessed the association between bullying victimization and mental health problems across 12 months. It is possible that the negative effects of bullying victimization unfold across a shorter or longer time frame. In addition, the current study did not examine how school or classroom characteristics (e.g., school climate, classroom placement, monitoring, discipline practices, or parent and teacher involvement) alter bullying victimization and/or its consequences for increasing internalizing mental health problems. This is an important topic for future research given that bullying is shaped by complex group and community dynamics (Espelage, 2014; Hong & Espelage, 2012) as well as emotional processes that may contribute to relational aspects of bullying involvement (e.g., Menesini & Camodeca, 2008). Finally, future research should examine the link between bullying victimization and clinically significant mental health disorders such as depressive disorders. Youth with autism have an increased risk of depressive disorders (Gotham, Unruh, & Lord, 2015), and it is not yet known to what extent negative peer experiences such as bullying victimization contribute to this risk.