خود کنترل، حمایت اجتماعی، و تجاوز در میان نوجوانان در خانواده های طلاق و دو والدینی

Abstract This study examined aggression in Israeli adolescents from divorced and two-parent families to explore self-control and social support as resources for reducing aggression, and to investigate whether the stress of divorce increases adolescents' aggression. Israeli adolescents from 127 divorced families and 308 two-parent families, completed self-report questionnaires. Major findings were: (1) Parental divorce did not correlate with increases in physical or verbal aggressive acts, but did correlate with significant increases in angry feelings and hostile thoughts (2) Higher levels of self-control and social support were found to mitigate possible adverse effects of parental divorce on adolescents' aggression. Outcomes imply that intervention designed to reduce aggression in adolescents should focus on the acquisition of self-control and the provision of social support.

1. Introduction Aggression among children and youth is a serious social problem that has sharply escalated in Israel in recent years (Benbenishty et al., 2002, Khoury-Kassabri et al., 2004 and Warman and Cohen, 2000). Aggression is usually described as an outcome of the links between hostile thoughts, angry emotions, and aggressive behaviors (Buss, 1961). Stress and distress are among the most frequent reasons the literature proposes as affecting aggression (Krahe, 2001, Pepler and Rubin, 1991 and Tremblay and Nagin, 2005). The present study focused on the incidence of aggression among adolescent children of divorced parents, compared with that of adolescents in two-parent families. It aimed to learn whether divorce as a stressful event results in increased aggression.

Results 7.1. Pearson correlations and ANOVAs As shown on Table 1, Pearson correlation coefficients were conducted for all the study variables. Significant positive intercorrelations emerged between all of the aggression components: hostility, anger, physical and verbal aggression, and overall aggression. In addition, all of the components of aggression revealed significant negative correlations with self-control and with social support, except for the correlation between social support and verbal aggression. Finally, a significant positive association emerged between self-control and social support. Table 1. Pearson correlations between study variables. Self-control Social support Overall aggression Hostility Anger Verbal aggression Physical aggression Sex Group (boarders/day attendees) Family structurea .13⁎⁎ .13⁎⁎ − .13⁎⁎ − .12⁎ − .11⁎ − .07 .07 .07 .13 Self-control – .35⁎⁎ − .37⁎⁎ − .18⁎⁎ − .33⁎⁎ − .18⁎⁎ − .31⁎⁎ − .15⁎⁎ .14⁎⁎ Social support – – − .28⁎⁎ − .35⁎⁎ − .19⁎⁎ .04 − .19⁎⁎ − .20⁎⁎ .07 Overall aggression – – – .67⁎⁎ .78⁎⁎ .58⁎⁎ .73⁎⁎ .25⁎⁎ − .19⁎⁎ Hostility – – – – .43⁎⁎ .27⁎⁎ .17⁎⁎ − .03 − .09⁎ Anger – – – – – .39⁎⁎ .40⁎⁎ − .02 − .09 Verbal aggression – – – – – – .27⁎⁎ .07 − .10⁎ Physical aggression – – – – – – – .52⁎⁎ − .22⁎⁎ a Divorced = 0; Two-parent = 1. ⁎ p < .05. ⁎⁎ p < .01. Table options As seen on Table 1, adolescents' sex correlated significantly with their overall aggression scores and their physical aggression dimension: Male adolescents reported higher aggression than females. Moreover, sex was significantly correlated with self-control and with social support: Males reported less self-control and less social support than females. Using ANOVAs, significant sex differences were found for four study variables: Compared to males, females reported higher levels of self-control (M = 13.45 vs. 6.34, respectively, p < .01), greater social support (M = 74.40 vs. 72.00 respectively, p < .001), lower overall aggression scores (M = 92.48 vs. 102.74, respectively, p < .001), and lower physical aggression scores (M = 23.45 vs. 33.76, respectively, p < .001). The ANOVAs comparing boarders and day attendees revealed significant differences for only two study variables: Compared to day attendees, boarders reported higher anger scores (M = 22.29 vs. 20.88, respectively, p < .01) and higher verbal aggression scores (M = 20.38 vs. 19.38, respectively, p < .01). In all other study variables (overall aggression, physical aggression, hostility, social support, and self-control), no differences were found; therefore, throughout the study the boarders and day attendees were treated as one group. 7.2. Hypothesis testing for total sample 1. Aggression measures and mediation. As described above, the first part of Hypothesis 1 was borne out: In the total sample of adolescents, significant positive intercorrelations emerged between all the components of aggression (physical and verbal, hostility and anger, and overall score). As seen in Table 1, hostility correlated positively with anger (r = .43), with verbal aggression (r = .27), and with physical aggression (r = .17); and anger correlated positively with verbal aggression (r = .39) and with physical aggression (r = .40; all were p < .01). Verbal and physical aggression also correlated positively (r = .27, p < .01). To examine the second hypothesis related to the role of anger in mediating the link between hostile feelings and aggressive behavior, we conducted a set of regressions predicting physical aggression. First, we proved the first three conditions set out by Baron and Kenny (1986) for the presence of a predictive variable, namely: an association between a dependent variable (hostility) and a predicted one (physical aggression); an association between a dependent variable (hostility) and a mediating variable (anger); and an association between the mediating variable (anger) and the predicted variable (physical aggression). The fourth indicator of mediation is when the association between the independent variable and the predicted one becomes lower if the mediating variable is held as a control. By way of corroboration, two regression analyses were conducted, with physical aggression as the predictive variable. In the first regression, with only hostility as a predicting variable, age and group (boarders/day attendees) were the first step, and they explained 7% of the variance in physical aggression. Next, as a second step, sex and hostility added another 19% to the explanation of the variance. When anger was not kept constant, the contribution of hostility to physical aggression was B = .19, Beta = .16, SE = .05, p < .01. In the second regression analysis predicting physical aggression, in which anger was kept constant (controlled), first the variables of age, group (boarders/day attendees), and anger were entered – which contributed 20% to the explanation of the variance – and then the variables of sex and hostility were entered, which accounted for a further 5% of the variance in physical aggression. In this analysis where anger was kept constant, the contribution of hostility to aggression declined (B = .00, Beta = .00, SE = .05, p > .05), pointing to anger as the mediating variable between hostility and physical aggression, and to the fulfillment of the fourth condition set out by Baron and Kenny (1986). This in turn confirmed the first study hypothesis: Associations emerged between all components of aggression, and anger was found to mediate between hostile thoughts and physical aggression. 2. Aggression and social support. Negative significant correlations emerged for social support with the general aggression score (r = − .28), with hostility (r = − .35), with anger (r = − .19), and with physical aggression (r = − .19; all were p < .01), but not with verbal aggression (ns). However, as seen in Table 3, social support was not found to contribute significantly to physical aggression in a stepwise regression analysis predicting physical aggression similar to the ones described above, where age and group (boarders/day attendees) were entered first, followed by sex, family structure (divorced/two-parent), hostility, anger, self-control, and social support (B = .04, Beta = .02, SE = .07, p > .05). Thus, the third hypothesis was partially supported. 3. Aggression and self-control. As described above, the results of the Pearson analyses (see Table 1) confirmed Hypothesis 4: In the total sample of adolescents, higher adolescent self-control correlated with lower aggression levels for each of the components: general aggression score (r = − .37), hostility (r = − .18), anger (r = − .33), verbal aggression (r = − .18), and physical aggression (r = − .31; all were p < .01). In addition, the stepwise regression analysis described above (see Table 3) indicated that physical aggression was partially predicted by the absence of self-control skills: In this analysis, the age and residential group variables contributed 7% to the explanation of the variance in physical aggression, and then the other variables (sex, family structure, hostility, anger, self-control, and social support) accounted for a further 31% of the variance. Self-control revealed a significant main effect in predicting physical aggression, B = − .04, Beta = − .09, SE = .02, p < .05. Thus, the fourth hypothesis was supported. A link was found between self-control and aggression, indicating that adolescents with higher self-control scores showed lower aggression scores. In sum, the first part of the study was supported. Links emerged between all the components of aggression (hostility, anger, verbal and physical aggression), and anger mediated the link between hostility and physical aggression. Also, social support and self-control skills were linked to physical aggression, indicating that adolescents with higher levels of social support or self-control reported lower rates of physical aggression. 7.3. Hypothesis testing to compare adolescents from divorced vs. two-parent families 5. Family structure and aggression. Analyses also confirmed most of Hypothesis 5. To conduct Pearson analyses, we measured divorce as 0 and two-parent family structure as 1 (see Table 1). Pearson correlations revealed that family structure (divorced/two-parent) correlated significantly correlated with overall aggression (r = − .13, p < .01), with hostility (r = − .12, p < .05), and with anger (r = − .11, p < .05), but not with verbal or physical aggression. Thus, adolescents from divorced families reported higher rates of aggression, hostility, and anger, but no such links emerged between divorced families and physical or verbal aggression. As seen on Table 2, ANOVAs conducted between divorced and two-parent families indicated that adolescents from divorced families reported significantly higher overall aggression, anger, and hostility than their counterparts from two-parent families. No significant gap was found for verbal or physical aggression while comparing the two family structures. Notably, the regression analysis described above (see Table 3) did not reveal any main effect for family structure, B = − .84, Beta = − .04, SE = .83, p > .05. Thus, most of Hypothesis 5 was supported. Table 2. ANOVAs comparing adolescents with divorced parents vs. two-parent families. Two-parent families (n = 308) Divorced parents (n = 127) Study variable n M SD n M SD F Overall aggression 299 95.29 20.60 124 101.36 20.22 7.685⁎⁎ Hostility 308 26.79 7.88 126 28.95 8.48 6.407⁎ Anger 306 21.31 7.00 126 22.94 6.69 4.918⁎ Verbal aggression 305 19.83 4.53 126 20.51 4.60 1.998 Physical aggression 304 27.59 9.77 124 29.21 9.62 2.435 Self-control 302 12.15 22.02 125 5.75 23.75 7.121⁎⁎ Social support 307 73.75 5.88 126 71.98 6.48 7.369⁎⁎ ⁎ p < .05. ⁎⁎ p < .01. Table options Table 3. Multiple regression predicting physical aggression based on age, group (boarders/day attendees), sex, family structure (divorced/two-parent), hostility, anger, self-control, and social support. Variable B SE β ∆R2 Step 1: .07 Age − 1.20 .42 − .35⁎⁎ Group − 11.00 2.44 − .54⁎⁎ Step 2: .38 Sex 9.72 .76 − .43⁎⁎ Family structure − .84 .83 − .04 Hostility .00 .05 .00 Anger .50 .06 .35⁎⁎ Self-control − .04 .02 − .09⁎ Social support .04 .07 .02 ⁎ p < .05. ⁎⁎ p < .01. Table options 6. Family structure and self-control and social support. The findings supported both parts of the last hypothesis: As shown in Table 1, family structure correlated significantly with self-control and with social support (r = .13, p < .01 for each). As seen in Table 2, results of the ANOVAs showed that adolescents with divorced parents reported lower self-control and less social support than adolescents from two-parent families. Mean self-control scores were 5.75 for adolescents from divorced families (SD = 23.75) and 12.15 for adolescents from two-parent families (SD = 22.02), F = 7.121, p < .01. Mean social support scores were 71.98 for adolescents from divorced families (SD = 6.48) and 73.75 for adolescents from two-parent families (SD = 5.88), F = 7.639, p < .01. 7.4. Additional findings To examine the contribution of the interactions with self-control and social support to predicting each of the four components of aggression: hostility, anger, physical, and verbal we conducted multiple regressions. For each component, we entered sex, group (boarders/day attendees), family structure (divorced/two-parent), self-control, and social support as Step 1, and we entered all the interactions with self-control and social support as Step 2. Results revealed no significant interactions when predicting physical aggression, verbal aggression, or anger. However, a significant interaction did emerge between self-control and sex in predicting hostility. Table 4 shows the first step of the regression model accounting for 15% of the variance in the prediction of hostility, with significant contributions by sex, self-control, and social support in predicting hostility, when the beta coefficient was negative. Step 2 added 2% to the prediction, with a significant contribution by the interaction between self-control and sex, when the beta coefficient was positive. To establish the source of the interaction (shown in Fig. 1), separate multiple regressions were conducted for male and female adolescents. These revealed that the interaction was not significant for boys but was for girls, indicating that for girls only, higher self-control levels predicted lower hostility levels (B = − 2.44, Beta = − .31, t = − 4.86, p < .001). Table 4. Multiple regression predicting hostility based on group (boarders/day attendees), sex, family structure (divorced/two-parent), self-control, and social support. Variable B SE β ∆R2 Step 1: .15 Sex − 1.83 .76 − .11⁎⁎ Group − 1.01 .80 − .06 Family structure − .41 .84 − .02 Self-control 1.39 .53 − .17⁎⁎ Social support − 2.88 .40 − .35⁎⁎ Step 2: .02 Self-control X Sex 2.04 .75 .17⁎⁎ ⁎p < .05. ⁎⁎p < .01. Table options Interaction between self-control and sex with regard to hostility. Fig. 1. Interaction between self-control and sex with regard to hostility.