درک پویایی زورگویی در میان دانش آموزان در آموزش و پرورش خاص و عام

Abstract Students in general and special education experience bullying. However, few empirical investigations have examined involvement in bullying along the bully/victim continuum (i.e., as a bully, victim, or bully–victim) among students with disabilities. A total of 816 students, ages 9 to 16, participated in the present study. From this total sample 686 were not receiving special education services (categorized as “no disability”), and 130 were receiving special education services (categorized as “observable disability,” “non-observable disability,” and “behavioral disability”). Data on students’ involvement in bullying, office referrals, and prosocial behavior were collected. Results indicated that students with behavioral disorders and those with observable disabilities reported bullying others more and being victimized more than their general education counterparts. Students with behavioral disorders also had significantly more office referrals than students in general education. Seventh graders in general education reported more bullying behavior than sixth graders and ninth grades in general education. Fifth graders in general education reported more victimization than students in all other grades in general education. However, the grade differences were not significant for students in special education. No gender differences on bullying and victimization were found. Students with disabilities reported less engagement in prosocial behaviors than their general education peers. Implications for bullying prevention and intervention across both general and special education are discussed.

1. Introduction Bullying is a major problem facing our nations’ schools and communities. Bullying behaviors consist of negative actions toward a student or a group of students perpetrated by one or more peers (Olweus, 1993a and Olweus, 1993b). Bullying can be physical (e.g., hitting, pushing, and kicking), verbal (e.g., threats, name calling, and teasing), or relational (e.g., excluding individuals from the group, spreading rumors, and saying mean things). Whether or not bullying consists of physical behaviors, verbal behaviors, or both, the common denominator is that the bullying behaviors involve aggressive acts; there is an imbalance of power between the bully and the person being bullied, and bullying behaviors are repeated over time (Espelage and Swearer, 2003, Olweus, 1993a, Olweus, 1993b and Smith et al., 2004). Bullying behaviors among school-aged youth can occur in the school building, on school grounds, going to and from school, and in cyberspace (i.e., cyber-bullying). Students may be involved in the bully/victim continuum as a bully, victim, bully-victim (i.e., both bullies others and is bullied by others), or a bystander (i.e., someone who observes bullying; Jimerson, Swearer, & Espelage, 2010). Bullying is a common experience among school-aged youth. In the most comprehensive study conducted in the United States, Nansel et al. (2001) found that 29.9% of 15,686 sixth through tenth grade students surveyed reported regular involvement in bullying. Specifically, 13% reported bullying others, 10.6% reported being bullied, and 6.3% reported involvement as a bully–victim. Other studies with large sample sizes also suggest that about 20% to 30% of adolescents are involved in bullying (Carlyle and Steinman, 2007, Dinkes et al., 2006 and National Center for Education Statistics, 2006; also see Rose, Monda-Amaya, & Espelage, 2010 for a review). Some studies with smaller sample sizes have found higher prevalence rate of bullying. For example, in one study of 250 middle and high school students, 75% reported being bullied (Hazler, Hoover, & Oliver, 1992). Of these students, 90% reported experiencing negative side effects as a result of being bullied, such as anxiety, low grades, and social rejection (Hazler et al., 1992). The wide range of estimates from different studies may be due to the different age groups studied, and various definitions of bullying, and assessment methods used. For example, some studies used a more lenient frequency cut-off point in their definition of bullying. Furthermore, some studies used students’ self-report, but other studies used teacher report or observation to measure bullying.

3. Results 3.1. Preliminary Results 3.1.1. Rates of Bullying and Victimization The prevalence rates of bullying and victimization varied depending on whether results from the single bullying/victimization item or the PRB Bullying score/ PRB Victimization score (the average of the five specific items) were analyzed. When asked about taking part in mean or negative behavior toward others (although not referred to as bullying) in the general question, 323 (39.6%) students reported they have taken part in such behavior toward others, and 493 (60.4%) students reported no such behavior. However, when provided with five specific examples of bullying (e.g., engaging in pushing, hitting or kicking or other physical ways on purpose), 311 (38.1%) students reported bullying others, and 505 (61.9%) students reported no bullying behavior. Similarly, when asked about victimization in the general question, 433 (53.1%) students reported that others had “been mean or negative” toward them at least “once or twice” and 383 (46.9%) students reported no victimization. However, when provided with specific examples, 547 (67%) reported victimization and 269 (33%) reported no victimization. In other words, when provided with specific examples, some students who reported no bullying or victimization in the general question changed their answers and reported bullying or victimization at least once or twice. Among students who reported bullying others or being victimized in their answer to the single bullying/victimization question, some of them reported that they bullied others or were victimized only by pushing, hitting, or kicking or other physical ways jokingly. These students might misinterpret the meaning of “negative actions or mean behavior” to include joking behaviors. However, it is also possible that those students did engage in bullying or victimization, but not in any of the specific forms measured by the items in our measure. 3.1.2. Validity of Groupings The t-test results indicated that there were significant difference between the bully group (M = 1.42, SD = 0.32; M = 1, SD = 0) and the victim group (M = 1, SD = 0; M = 1.55, SD = 0.50) on the PRB Bullying score, t(52) = 9.76, p < .001, and the PRB Victimization score, t(288) = − 18.78, p < .001. The bully group (M = 1.42, SD = 0.32) scored significantly higher than those not involved (M = 1, SD = 0) on the PRB Bullying scores, t(52) = 9.76, p < .001. Similarly, the victim group (M = 1.55, SD = 0.50) also scored significantly higher than those not involved (M = 1, SD = 0) on the PRB Victimization score, t(288) = 18.78, p < .001. There was also a significant difference between the bully group (M = 1, SD = 0; M = 1.42, SD = 0.32) and the bully–victim group (M = 1.69, SD = 0.51; M = 1.47, SD = 0.46) on the PRB Victimization score, t(257) = − 21.73, p < .001, but not on the PRB Bullying score, t(309) = − 0.65, p = .52. Interestingly, the bully–victim group (M = 1.69, SD = 0.51) scored significantly higher than the victim group (M = 1.55, SD = 0.50) on the PRB Victimization score, t(545) = 3.33, p = .001. These results provided additional evidence of the validity for the bully/victim grouping in the current study. Students from different special education categories were statistically compared on their ratings of all the major study variables. The behavioral disorder group and the OHI group did not significantly differ on bullying, victimization, and prosocial behavior. Students with MMH reported more bullying behaviors compared to students with SLI and HI, but the Kruskal–Wallis test found that the difference was not significant, χ2(2) = 5.56, p = .06. No other differences were found among students with MMH, SLI, and HI. These findings provided additional support for grouping students with behavior disorder and OHI together as well as grouping students with MMH, SLI, and HI together. ELL students who were not in special education (n = 80) and other students in general education (n = 601) did not differ on their involvement in bullying, t(87.23) = − 1.64, p = .11, and victimization, t(679) = 0.006, p = 1.00). Based on these results, the 80 ELL students were included in the general education group. Chi-square test indicated that students with different special education categories did not differ significantly on their bully/victim status, χ2(9) = 16.46, p = .06. Chi-square tests were used to examine race, gender, grade, and school difference across special education status. There were no race differences in special education status, χ2 (21) = 10.36, p = .97. There were no grade differences in special education status, χ2 (6) = 12.16, p = .06 (5th and 9th graders were not included in the chi-square test due to small cell number for some categories). Gender differences were significant, χ2 (3) = 26.18, p < .001. Using standardized residuals as indicators for the post hot test, results showed that boys were significantly more likely to have behavioral disorders and non-observable disorders than they were expected (standardized residual = 2.7 and 2.1). Girls were significantly less likely to have behavioral disorders and non-observable disorders than they were expected (standardized residual = − 2.5 and − 2.0). To rule out the possible confounding effect of different school settings (across nine schools) on bullying, victimization, prosocial behavior and office referral among students, several univariate analyses of variance (ANOVA) and the Kruskal–Wallis test were conducted. There were no school differences on bullying using the single bullying item, F(8, 807) = 1.54, p = .14, partial η2 = .02, or victimization using the single victimization item, F(8, 807) = 0.80, p = .60, partial η2 = .01. However, when the PRB Bullying and Victimization scores (i.e., the mean of the five specific items) were used, the Kruskal–Wallis test showed significant differences in bullying, χ2(8) = 18.91, p = .02; victimization, χ2(8) = 16.89, p = .03; prosocial behavior, χ2(8) = 15.70, p = .047; and office referrals, χ2(8) = 19.83, p = .01. The means and standard deviations for the outcome variables by schools are listed in Table 5. Because some schools did not have students (or had only few students) in specific special education categories, we were not able to conduct a two-way (school-by-special education category) ANOVA to examine the impact of both school and special education category on students’ experience with bullying and victimization. Table 5. Means and Standard Deviations for Bullying, Victimization, Prosocial Behavior and Office Referrals across Participant Schools. School Variable A B C D E F G H I Sample size 29 15 32 21 243 166 227 48 35 PRB Bullying score 1.33 (0.41) 1.03 (0.07) 1.10 (0.20) 1.31 (0.73) 1.17 (0.31) 1.14 (0.28) 1.21 (0.43) 1.11 (0.20) 1.10 (0.17) Bullying single item 1.69 (0.81) 1.33 (0.62) 1.31 (0.54) 1.48 (0.98) 1.53 (0.75) 1.43 (0.65) 1.61 (0.82) 1.54 (0.85) 1.37 (0.49) PRB Victimization score 1.72 (0.55) 1.41 (0.27) 1.37 (0.39) 1.68 (0.91) 1.42 (0.54) 1.36 (0.44) 1.42 (0.50) 1.34 (0.37) 1.28 (0.39) Victimization single item 1.76 (1.02) 2.07 (0.96) 1.78 (0.75) 1.90 (1.09) 1.76 (0.86) 1.66 (0.84) 1.72 (0.82) 1.65 (0.81) 1.57 (0.66) Prosocial Behavior 3.01 (1.04) 3.88 (0.58) 3.70 (1.00) 3.50 (1.29) 3.57 (0.85) 3.53 (0.84) 3.51 (0.92) 3.55 (0.93) 3.77 (0.87) Office referrals 0.19 (0.47) 0.07 (0.26) 1.09 (4.62) 0.14 (0.48) 0.44 (1.34) 0.92 (2.97) 0.27 (0.83) 0.06 (0.32) 0 Note. A, B, C, and D = Elementary schools; E, F, and G = Middle schools; and H and I = High schools. Schools A and D had significantly higher bullying score (5 items) than schools B, C, F, G, H, and I. School A also had significantly higher bullying score (5 items) than school E. School G had significantly higher bullying score (5 items) than schools B and F. Schools A and D had significantly higher bullying score (5 items) than schools C, E, F, G, H, and I. Schools A had significantly lower prosocial score than schools B, C, E, F, G, H, and I. Schools C and F had significantly higher office referral than schools G, H, and I. School F also had significantly higher office referral than school E. Table options 3.2. Bullying and Victimization Statistically significant differences were found among the four groups in reported victimization, F(3, 812) = 5.47, p = .001. The partial η2 was .02, suggesting that approximately 2% of the variability in the victimization score and its associated error could be accounted for by the differences among the four groups. LSD post hoc tests indicated that students verified with a behavioral disorder reported that they were victimized more often than students who were not in special education (mean difference = 0.27, p = .001) and students who had non-observable disorders (mean difference = 0.23, p = .03). Students with observable disorders also reported that they were victimized more often than students who were not in special education (mean difference = 0.20, p = .02). There were also significant differences in students’ experience of bullying, F(3, 812) = 6.12, p < .001, partial η2 = .02. LSD post hoc tests indicated that students in the observable and behavioral disorder categories reported that they bullied others more often than students who were not in special education (mean difference = 0.12, p = .04, and mean difference = 0.20, p < .001, respectively; see Table 6). Table 6. Means and Standard Deviations for Bullying, Victimization, Prosocial, and Office Referrals across Students in Special Education and Students in General Education. Special education category Variable Non-observable Observable Behavioral General education Bullying 1.23 (0.39) 1.28 (0.45)* 1.35 (0.58) *** 1.15 (0.32) Victimization 1.42 (0.53) 1.59 (0.55)* 1.66 (0.63)*** 1.39 (0.49) Office referral 0.61 (1.58) 0.50 (1.36) 2.67 (5.31)* 0.30 (1.34) Prosocial behavior 13.38 (4.03) 13.36 (4.11) 13.28 (4.10) 14.23 (3.52) Note. Non-observable (specific learning disability; n = 51); observable (speech language impairment, hearing impaired, mild mentally handicapped; n = 36); behavioral (behavioral disorder and other health impaired; n = 43), and general education (n = 686). *p < .05, **p < .01, ***p < .001, indicating the group is significantly different from the general education group. Table options 3.3. Grade and Gender Differences ANOVA results showed that for students in general education, there was a significant difference in their bullying behavior by grade level, F(4, 681) = 3.17, p < .05, partial η2 = .02. Students in seventh grade reported that they participated in bullying more often compared with sixth graders (mean difference = 0.10, p = .002) and ninth graders (mean difference = 0.11, p = .01; see Table 7). For students in special education, the grade difference was no longer statistically significant, F(3, 123) = 0.83, p = .48, partial η2 = .02 (Ninth graders were not included in the analysis because there were only three ninth graders in special education.) For students in general education, there was a significant difference by grade level in their experience with victimization, F(4, 681) = 2.78, p = .03, partial η2 = .02. Fifth graders reported more victimization than sixth graders (mean difference = 0.15, p = .02), seventh graders (mean difference = 0.19, p = .004), eighth graders (mean difference = 0.16, p = .02), and ninth graders (mean difference = 0.23, p = .003). For students in special education, the difference by grade level was not statistically significant, F(3, 123) = 0.06, p = .98, partial η2 = .002. The t-tests indicated no gender difference in bullying or victimization for both students in special education, t(128) = − 0.16, p = .88, effect size Cohen's d = − 0.03 and t(128) = 0.59, p = .65, Cohen's d = 0.11, respectively, and students in general education, t(682) = 0.52, p = .61, Cohen's d = 0.04 and t(682) = 1.88, p = .06, Cohen's d = 0.14, respectively. Table 7. Means and Standard Deviations for PRB Bullying and Victimization Scores by Grade. Grade Variable 5th 6th 7th 8th 9th Special education Bullying 1.28 (0.44) 1.21 (0.44) 1.38 (0.62) 1.26 (0.36) 1.27 (0.12) Victimization 1.59 (0.50) 1.54 (0.54) 1.57 (0.69) 1.52 (0.53) 1.40 (0.20) Sample size 15 39 39 34 3* General education Bullying 1.19 (0.43) 1.11 (0.24) 1.21 (0.43) 1.16 (0.22) 1.10 (0.19) Victimization 1.54 (0.60) 1.39 (0.48) 1.35 (0.48) 1.37 (0.47) 1.31 (0.38) Sample size 82 216 167 141 80 Note.* This group was not included in ANOVA due to the small sample size. Table options 3.4. Office Referrals Results showed that 8 (22.2%) students who had observable disorders, 12 (23.5%) students who had non-observable disorders, 21 (48.9%) students who had behavioral disorders, and 86 (12.4%) students in general education received office referrals. The Kruskal–Wallis test indicated significant differences in office referrals among the four groups, χ2(3) = 51.76, p < .001. The Mann–Whitney test with the Bonferroni correction showed that students with behavioral disorders had significantly more office referrals than students in general education (mean rank difference = 144.54, p < .001). 3.5. Prosocial Behaviors No significant differences were found among the four groups in prosocial behavior, F (3, 812) = 2.31, p = .08, η2 = .01. A dichotomous variable (special education versus general education) was then created. The t-test result indicated that students in special education engaged in fewer prosocial behaviors compared with students in general education, t(166.3) = 2.37, p = .02. Prosocial behavior significantly and negatively correlated with students’ report on the single bullying item, r = −.14, p < .001. The correlations were also significant for students in special education, r = −.32, p < .001, and students in general education, r = −.08, p = .03. However the correlation was not statistically significant between prosocial behavior the PRB Bullying score, r = −.07, p = .06. 3.6. Mediation Using Baron and Kenny's method (1986), being in special education was entered into the regression model as one predictor for involvement in bullying in Step One and as one predictor for involvement in prosocial behavior in Step Two. Results showed that in Step One, the model fit the data well, F(1, 811) = 15.71, p < .001, R2 = .019. Being in special education was a significant predictor for involvement in bullying, β = .14, t = 3.96, p < .001, suggesting students in special education were more likely to be involved in bullying compared with students in general education. In Step Two, the model also fit the data well, F(1, 811) = 6.82, p = .01. Being in special education was a significantly negative predictor for involvement in prosocial behavior, β = −.09, t = − 2.61, p = .01, suggesting students in special education were less likely to be involved in prosocial behavior compared with students in general education. In Step Three, both being in special education and involvement in prosocial behavior were entered into the regression model to predict involvement in bullying, and the model fit the data well, F(2, 810) = 9.78, p < .001, R2 = .022. Results showed that being in special education was a significantly predictor for involvement in bullying, β = .13, t = 3.8, p < .001, but involvement in prosocial behavior did not predict students’ involvement in bullying, β = −.05, t = − 1.53, p = .13, which suggests that involvement in prosocial behavior did not serve as a mediator on the relationship between being in special education and involvement in bullying.