Study sample
The study cohort sample was taken from the Special Needs Education Longitudinal Study (SNELS) database, released in 2011 by the Survey Research Data Archive of Academia Sinica (https://srda.sinica.edu.tw/index.php). The Institutional Review Board approved the data application and analysis. In the SNELS, representative samples of students with ASD were randomly selected from a list of registrants; specifically, they were identified from medical records or records held by the Special Education Needs Committees of local governments, as described in a previous study [24]. In the SNELS, data were obtained from questionnaires administered online, which both students and teachers completed. The students completed their questionnaires independently, with the teachers providing aid if necessary. Figure 1 displays the process used for participant selection. Of the 317 seventh graders with ASD involved in the SNELS in 2011, 223 were enrolled in regular classes in general education schools. Thirty-nine seventh-graders with ASD were excluded due to a lack of data regarding the studied variables. Overall, 184 seventh graders (167 boys) with ASD were enrolled in this study.
Types of victimization
The students were asked whether they had experienced social exclusion, insults or teasing, extortion, or sexual harassment over the past semester. Social exclusion was assessed by asking the students, “Have you felt left out by your classmates at this school?” Being insulted or teased was measured by asking the students, “Have you felt insulted or been teased by your classmates at this school?” Experiencing extortion was assessed by asking the students, “Have you experienced extortion for money at this school?” Being sexually harassed was measured by asking the students, “Has anyone touched your body inappropriately or made you feel uncomfortable at this school?” Students recorded their responses using a four-point Likert scale (1 = never, 2 = seldom, 3 = occasionally, and 4 = frequently) to indicate the frequency of bullying victimization. Answers of “never” were coded as 0, and all other responses were coded as 1.
Student characteristics
Student characteristics comprised of the following variables: (a) sex, reported by teachers, with male and female coded as 1 and 0, respectively; (b) BMI, collected from teacher reports, was calculated as the weight in kilograms divided by the square of the height in meters; and (c) the general tendency of the students’ learning capabilities in school was collected from teacher reports. The variable of students’ learning capabilities in school was assessed using the following ten items: (1) paying attention in class; (2) complying with instructions; (3) sitting still; (4) participating in classroom discussions; (5) asking questions; (6) answering teachers’ questions; (7) focusing on schoolwork; (8) having the motivation to learn; (9) finishing homework on time, and (10) collaborating with peers to complete tasks. These items were scored on a four-point Likert scale (1 = never, 2 = seldom, 3 = occasionally, and 4 = frequently). Cronbach’s α for the ten criteria was 0.86. (d) Psychological distress was student-reported using a six-item index adapted from the Symptom Checklist-90-Revised (SCL-6) instrument, originally developed to assess a broad range of psychological symptoms [25]. The criteria were extracted based on the most frequent symptoms experienced by students receiving special education services [26]. Each of the following six dimensions comprised one item of the SCL-6: depression (feeling down or lonely); obsessive compulsion (trouble concentrating); insomnia (restlessness or disturbed sleep); anxiety (feeling tense or uptight); psychoticism (never feeling close to another person); and anger/hostility (experiencing the urge to break or smash objects or hit, injure, or harm others). The frequency of each item was rated on the same four-point scale used to assess the frequency of victimization incidents. Cronbach’s α for the SCL-6 was 0.88.
School environment characteristics
Seven variables measured the characteristics of the school environment. The first variable was the number of friends that the students had, which was assessed using responses to “How many classmates or friends do you usually hang out with?” on a scale of 1–5 (1: no friends, 2: one friend, 3: two or three friends, 4: four or five friends, and 5: six or more friends). The second, friendship quality, was measured by the teachers’ responses to the question, “Is the student able to get along well with other students in the class?” on a scale of 1–5 (1: no interaction at all, 2: poorly, 3: not very well, 4: acceptably well, and 5: very well). The third variable, classroom climate, as measured by the teachers’ responses to the following statements, using a scale of 1–4 (1: not at all true, 2: not very true, 3: somewhat true, and 4: very true): “The majority of the students proactively participate in inter-class competitions and activities.”; “The students have positive relationships with each another”; “The students are helpful to and caring towards each another”; and “The students collaborate (i.e., discuss their homework together).”The fourth variable was whether students received assistance when difficulties occurred, which was measured by how the students responded to the question, “Are you able to get help from someone when encountering difficulties?” using a scale of 1–4 (1: almost no one, 2: very few people, 3: some people, and 4: many people). The fifth variable, the teacher’s attitude towards class integration, was assessed by the teachers’ responses to the following statements, using a scale of 1–4 (1: strongly disagree, 2: somewhat disagree, 3: somewhat agree, and 4: strongly agree): “Schools should not exclude students with disabilities from enrolment”; “Students with disabilities should be allowed to learn with their typically developing counterparts as much as possible”; “Teachers of regular classes should not neglect the needs of typically developing students in class because of the presence of students with disabilities”; “The placement of students with disabilities in the same class as typically developing students is beneficial for both groups of students”; “The teacher’s attitude is a key factor that affects the learning and adaptation of students with disabilities”; and “Teaching students with disabilities is a responsibility that is shared by teachers of both special and regular classes.” The sixth variable, parental engagement in the children’s learning, as measured by the teachers’ responses to “How would you describe the involvement of this student’s parents in their learning?” using a scale of 1–4 (1: nonexistent, 2: not very high, 3: moderately high, and 4: very high). The final variable, the amount of integration within regular classes, was measured by whether the students participated in class full-time (coded as 0) or part-time (coded as 1). Those in regular classes on a part-time basis received special education sessions at other times.
Data analysis
All analyses were performed using the IBM SPSS Statistics program for Windows, version 22 (IBM Corp., Armonk, NY, USA). First, participant demographics and differences in the school environment for the participants who had not experienced bullying were analyzed. Continuous variables were analyzed using independent sample t-tests, and categorical variables were compared using the chi-square test. Next, four separate hierarchical logistic regression analyses were conducted to examine the association between aspects in the school environment and the following types of bullying victimization: social exclusion, insults or teasing, extortion, and sexual harassment. Student characteristics were input into the models to assess their initial effects on bullying victimization. After adjusting for individual factors, the variables concerning aspects of the school environment were entered and analyzed.