قربانی زورگویی و ناسازگاری اجتماعی و عاطفی تماشاگران: تجزیه و تحلیل امتیاز

Abstract This study investigated how bystanders, who have and have not been bullied, perceive their social and emotional maladjustment depending on the form of bullying (physical or verbal) they witness. Using propensity score matching, equivalent groups of 270 victimized and 270 non-victimized bystander groups were created based on middle school students' responses on the Bully Survey-Student Version (BYS-S; Swearer, 2001). Victimized bystanders experienced higher social maladjustment than non-victimized bystanders. Path analysis results suggest that social and emotional maladjustment as a bystander is related not only to social–emotional maladjustment as victim, but to gender and the form of bullying witnessed. The way in which bystanders are influenced by their personal victimization may be a c

. Introduction As a subset of a larger aggressive social phenomenon, bullying is a distinct but pervasive social problem throughout the world (Carney and Merrell, 2001 and Cook et al., 2010). The Centers for Disease Control and Prevention define bullying as unwanted aggressive behavior(s) by another peer (individual or group) that may harm or distress the targeted youth physically, psychologically, socially, or educationally; it involves an observed or perceived power imbalance and is or is highly likely to be repeated (Gladden, Vivolo-Kantor, Hamburger, & Lumpkin, 2014). Bullying takes many forms including: physical (e.g., kicking, hitting), verbal (e.g., name calling), relational (e.g., social exclusion, gossip), and cyberbullying (e.g., hurtful messages or images through text message or email; Williams & Guerra, 2007). Bullying tends to peak in middle school (Eslea and Rees, 2001 and Nansel et al., 2001). Traditionally, research dichotomized youth who were involved in bullying into bully or victim categories without recognizing the social contexts in which bullying occurs. Bullying not only involves the youth who bully and who are bullied, but also bystanders who witness bullying as it occurs (Carney and Merrell, 2001 and Swearer, 2001). The behaviors of bystanders play an important role in initiating, mitigating, or exacerbating bullying. Although the current study focuses on all bystanders who witness bullying, it should be noted that other studies have further differentiated these roles into behaviors such as: (a) assisting or reinforcing the bully, (b) defending the victim, or (c) being an outsider (i.e., uninvolved or unaware; Oh and Hazler, 2009, Salmivalli, 1999, Salmivalli, 2001, Salmivalli et al., 1996 and Salmivalli et al., 2011). Several researchers (Goosens et al., 2006, Huitsing and Veenstra, 2012 and Sutton and Smith, 1999) have suggested that participant roles are not fixed, but static and dependent on situational factors (e.g., social status and networks). Therefore, students may be both bullies and defenders, victims and defenders (Huitsing & Veenstra, 2012), and bully-victims (Veenstra et al., 2005) depending on the social context. Bullying contributes to negative short- and long-term behavioral, psychological, social, and academic outcomes for perpetrators and victims (Demanet and Van Houtte, 2012, Glew et al., 2005, Ladd and Ladd, 2001 and Ttofi et al., 2011). Although there is a robust literature linking involvement in bullying and victimization to social and emotional maladjustment (see Jimerson, Swearer, & Espelage, 2010), far less is known about the experiences and impact of bullying on bystanders. The current study uses a social–ecological framework, which recognizes individual, familial, peer, school, and community factors (Bronfenbrenner, 1986 and Salmivalli et al., 1996), to enhance our understanding of the role of the bystander in bullying. Furthermore, we draw from Garbarino's (2001) accumulation of risk model as a framework for understanding how and when children suffer the most adverse consequences of exposure to violence. More specifically, this model suggests that although most children cope successfully with negative experiences, when risks accumulate, youth require protective factors and/or positive experiences to prevent the precipitation of harm. Using these theoretical frameworks, the current study examined the social and emotional maladjustment of bystanders who have been victimized and who have not been victimized using propensity score matching to isolate the impact of victimization on bystanders. Additionally, we explored the extent to which the forms of victimization experienced and witnessed predict social and emotional maladjustment of middle school-aged bystanders.

3. Results 3.1. Bivariate correlations Bivariate correlations between all study variables, except demographic characteristics (McCormick et al., 2013), were performed before data analyses were conducted to determine the extent to which covariates, the treatment variable, and the outcome variables were associated (see Table 2). Significant correlations between variables ranged from .08 (between “I don't like bullies” and “bullies hurt kids”) to .62 (between “bullies make kids feel bad” and “I feel sorry for kids who are bullied”). All significant variables were negatively related to both pro-bullying attitude items, except age; the relationship between the two pro-bullying attitude items; and “being bullied is no big deal” and social maladjustment as a bystander. Additionally, emotional maladjustment as a bystander was positively related to all variables except the two pro-bullying attitudes items. 3.2. Propensity score analysis Using the Match It packet in R (Ho, Imai, King, & Stuart, 2011), probabilities of being a victim or non-victim (i.e., treated or untreated, respectively) were assessed for a sample of bystanders who were victimized relative to a propensity-score-weighed sample of bystanders who were not victimized during the previous school year. Cases were matched on their total VPBS score (bystander perspective), gender, age, and ethnicity (dummy coded). Previous research has suggested that student attitudes toward bullying influence their participant roles (Swearer & Cary, 2003), making attitudes an important matching variable. Because many items on the Bullying Attitudes Scale (Swearer, 2001) were above the 10% threshold of missing data, cases were matched on individual items, rather than a total subscale score. Therefore, cases were matched on five anti-bullying scale items (Bullying is a problem for kids, I don't like bullies, Bullies hurt kids, Bullies make kids feel bad, and I feel sorry for kids who are bullied) and two pro-bullying scale items (I would be friends with a bully and Being bullied is no big deal). Cases were not matched on one item (Bullying is good for wimpy kids) because the mean difference of balance for matched data did not exceed 0.06. Furthermore, to have the most parsimonious model, we did not use polynomials or interactions terms given that the inclusion of such terms did not significantly improve our matching scores. A logistic regression method of estimation of propensity score was used. Matching and weighting were used as conditioning strategies. Matched groups were created using the greedy nearest neighbor approach so that each treatment observation was matched with the control observation with the closest propensity score (Thoemmes & Kim, 2011). Participants were matched without replacement, using one to one treatment to control units. The final matched sample (n = 540) of victimized (n = 270) and non-victimized (n = 270) cases was derived from an original sample of n = 689 bystanders. Table 5 displays the covariates used to estimate the propensity score, and the mean, standard deviation and the standardized mean difference between the treatment (victimized) group with the matched control (non-victimized) group, and the ratio of the standard deviation of the two groups. Two hundred and seventy victimized bystanders were successfully matched to two hundred and seventy non-victimized bystanders on total VPBS score (bystander perspective), gender, age, ethnicity, and seven attitudes toward bullying items. Table 5. Balance of covariate means and SDs for the propensity score matched victimized and non-victimized groups. Victimized Non-victimized Variable M SD M SD Mean difference Ratio of SD Bystander VPBS total score 1.75 0.93 1.67 0.85 0.07 0.92 Gender 0.62 0.49 0.63 0.48 − 0.01 0.99 Age 12.32 0.91 12.35 1.02 − 0.03 1.13 White 0.73 0.44 0.74 0.44 − 0.01 1.00 Latino/Hispanic 0.01 0.11 0.01 0.11 0.00 1.00 Native American 0.03 0.17 0.03 0.17 0.00 1.00 Eastern European 0.03 0.17 0.02 0.15 0.01 0.88 Black/African American 0.06 0.24 0.04 0.21 0.02 0.88 Asian American 0.01 0.10 0.01 0.12 0.00 1.20 Middle Eastern 0.01 0.10 0.01 0.12 0.00 1.20 Asian 0.02 0.14 0.02 0.15 0.00 1.07 Biracial 0.10 0.30 3.34 0.30 − 0.09 0.97 Bullying is a problem for kids 4.36 1.25 4.42 1.18 − 0.06 0.94 I don't like bullies 3.44 1.75 3.44 1.69 0.00 0.97 Bullies hurt kids 3.29 1.83 3.26 1.75 0.03 0.96 Bullies make kids feel bad 3.41 1.80 3.39 1.71 0.02 0.95 I feel sorry for kids who are bullied 1.71 1.11 1.68 1.09 0.03 0.98 I would be friends who a bully 1.51 1.07 1.55 1.11 − 0.04 1.04 Note. Mean differences are not exact due to rounding. Table options 3.3. Social and emotional maladjustment for victimized vs. non-victimized bystanders Differences in Bystander Social Maladjustment and Bystander Emotional Maladjustment by Victimization Status (victim or nonvictim) were examined using a Multivariate Analysis of Covariance (MANCOVA) with Victimization Status as the independent variable, the mean score of Bystander Social Maladjustment and Bystander Emotional Maladjustment of the SeMS as the dependent variables, and School as a covariate. Using Wilks' criterion, the combined dependent variables and covariate differed significantly based on Victimization Status, F (2, 537) = 3.36, p = .04, partial η2 = .01. The Bystander Social Maladjustment of victims was higher (M = 0.22, SD = 0.55) than nonvictims (M = 0.13, SD = 0.41), F (1, 537) = 5.84, p = .02, but the effect size was small (partial η2 = .01). There were no differences in the Bystander Emotional Maladjustment of victims (M = 1.33, SD = 1.05) compared to those who were not victims (M = 1.18, SD = 0.94), F (1, 537) = 3.17; p = . 08. 3.4. Form of victimization witnessed as a predictor of social and emotional maladjustment A path analysis was conducted to examine the extent to which Gender, Ethnicity, mean Victim Social–emotional Maladjustment, Victim Physical Bullying, Victim Verbal Bullying, Bystander Verbal Bullying, and Bystander Physical Bullying predicted Bystander Social Maladjustment and Bystander Emotional Maladjustment. Table 6 presents correlation matrices between study variables. Due to the high correlation, r = .96 between the mean Victim Verbal Bullying and mean Bystander Verbal Bullying, mean Victim Verbal Bullying was removed from analysis. Correlations between variables used in this analysis ranged from .00 (between Ethnicity and Bystander Verbal Bullying) to .62 (between mean Victim Social–emotional Maladjustment and Bystander Emotional Maladjustment). Except for the association between Gender and Victim Physical Bullying (males experiencing more physical bullying than females), all correlations were positive. Table 6. Bivariate correlations between study variables for subsample of victimized bystanders (n = 270). Study variables M SD 2 3 4 5 6 7 8 9 Predictor variables 1. Sex – – 0.02 − .17** .22** .16** − 0.02 .15* .20** 0.05 2. Ethnicity – – – 0.02 − 0.03 − 0.05 0.09 − 0.004 − 0.06 0.005 3. Victim physical bullying 0.79 0.86 – – .26** .49** .49** .25** .34** .31** 4. Victim verbal bullying 2.15 1.01 – – – .39** .53** .96** .39** .17** 5. Victim social and emotional maladjustment 1.09 0.92 – – – – .25** .34** .62** .42** 6. Bystander physical bullying 1.15 1.13 – – – – – .55** .29** .25** 7. Bystander verbal bullying 2.2 1.00 – – – – – – .34** .15* Outcome variables 8. Bystander emotional maladjustment 1.33 1.05 – – – – – – – .40** 9. Bystander social maladjustment 0.22 0.55 – – – – – – – – Note. *p < .05. **p < .01. Bystander social maladjustment and bystander emotional maladjustment are mean scores. Victim social–emotional maladjustment, victim physical bullying, victim verbal bullying, bystander verbal bullying, and bystander physical bullying are mean scores. Table options The saturated path analysis shows path coefficients for mean Victim Social–emotional Maladjustment was significant for both Bystander Emotional Maladjustment (b = 0.60, SE = 0.06, p < .001, b* = 0.52) and Bystander Social Maladjustment (b = 0.22, SE = 0.04, p < .001, b* = 0.37). While Gender (b = 0.25, SE = 0.11, p = .02, b* = 0.12) was significant for Bystander Emotional Maladjustment, Victim Physical Bullying (b = 0.08, SE = 0.04, p = .02, b* = 0.17) was significant for Bystander Social Maladjustment. In addition, the variance explained was 42% and 21% for Bystander Emotional Maladjustment and Bystander Social Maladjustment, respectively. Standardized and unstandardized parameter estimates, standard errors, and z values are presented in Table 7. Table 7. Path analysis parameter estimates for subsample of victimized bystanders (n = 270). Parameters Unstandardized Standardized value Value SE 95% C.I. Z-value Bystander emotional maladjustment Gender 0.25* 0.12 0.01–0.49 2.37 0.12* Ethnicity − 0.12 0.11 − 0.34–0.10 − 1.04 − 0.05 Victim physical bullying 0.07 0.07 − 0.07–0.21 0.97 0.06 Victim social and emotional maladjustment 0.60** 0.06 0.12–0.72 9.33 0.52** Bystander physical bullying 0.09 0.06 − 0.03–0.21 1.60 0.10 Bystander verbal bullying 0.08 0.06 − 0.04–0.20 1.34 0.08 Bystander social maladjustment Gender 0.03 0.07 − 0.11–0.17 0.39 0.02 Ethnicity 0.01 0.07 − 0.13–0.15 0.10 0.01 Victim physical bullying 0.06 0.05 − 0.04–0.16 1.26 0.09 Victim social and emotional maladjustment 0.22** 0.04 0.14–0.30 5.63 0.37** Bystander physical bullying 0.08* 0.04 0.00–0.16 2.32 0.17* Bystander verbal bullying − 0.05 0.04 − 0.13–0.03 − 1.32 − 0.09 Parameter estimates (fit) Bystander emotional maladjustment and bystander social maladjustment 0.07* 0.02 0.11–0.03 2.91 0.18 Note. Bystander emotional maladjustment and bystander social maladjustment for the subsample of victims were correlated moderately (r = .40). C.I. = confidence interval. *p < .05. **p < .01. Bystander social maladjustment and bystander emotional maladjustment are mean scores. Victim social–emotional maladjustment, victim physical bullying, victim verbal bullying, bystander verbal bullying, and bystander physical bullying are mean scores.