پیش بینی خانوادگی خواهر و برادر و حل تعارض عاشقانه شریک: مقایسه نوجوانان اواخر خانوادههای سالم و طلاق

Abstract The present study examined whether predictors of romantic-partner conflict may vary as a function of family structure. Using a cross-sectional design, we tested a mediation model of conflict resolution behaviours among late adolescents from intact (n=185n=185) and divorced (n=87n=87) families. Adolescents rated conflict resolution behaviours in five dyadic relationships: interparental, mother–adolescent, father–adolescent, sibling, and romantic partner. Mother–adolescent and father–adolescent conflict resolution behaviours mediated the relationship between interparental and sibling conflict resolution. Moreover, both mother–adolescent and sibling conflict resolution behaviours mediated the relationship between interparental and romantic-partner conflict resolution behaviours. This model was supported for both positive and negative behaviours, and it applied equally well to adolescents from intact and divorced families. Results are discussed in the context of social learning theories and support conflict resolution behaviours within the family spilling over into romantic relationships for adolescents from both intact and divorced families.

Introduction Almost all close, personal relationships will experience conflict, and the successful resolution of conflict is essential to relationship satisfaction and well-being. The family is often the primary context for learning and practicing effective conflict resolution behaviours (Patterson, 1982). Research suggests that parents’ methods of resolving conflict within the family are imitated by their adolescents within sibling relationships (Reese-Weber, 2000) and romantic relationships (Martin, 1990; Reese-Weber & Bartle-Haring, 1998). Such research has typically been based on adolescents from intact families only, which does not provide information about how conflict resolution behaviours develop in divorced families. Studies have found that late adolescents from divorced families have more negative attitudes regarding romantic relationships than late adolescents from intact families (Giuliani, Iafrate, & Rosnati, 1998; Summers, Forehand, Armistead, & Tannenbaum, 1998); however, differences in predictors of family and romantic-partner conflict resolution behaviours between adolescents from divorced versus intact families have not been examined. The present study provided such a test. Social learning theory suggests that parents serve as models to their children, and that children imitate their parents’ behaviours in their own relationships (Bandura, 1989). As such, observing and practicing conflict behaviours within the family provides a number of possible learning opportunities. Adolescents can observe two parents engaging in conflict with each other (termed interparental conflict), note the conflict resolution behaviours that are effective and ineffective, and then practice conflict resolution behaviours with their parents and siblings. Indeed, research indicates that children's positive behaviours with siblings are predicted by low levels of overt interparental conflict, and negative sibling behaviours are predicted by high levels of overt interparental conflict ( Brody, Stoneman, & Burke, 1987). In addition, higher frequencies of interparental conflict are related to higher frequencies of sibling conflict concurrently and 1 year later ( Brody, Stoneman, McCoy, & Forehand, 1992). The relationship between interparental conflict and sibling conflict may be mediated by the parent–child relationship (Brody, Stoneman, & McCoy, 1994; Erel, Margolin, & John, 1998). Reese-Weber (2000) found that the relationship between interparental and sibling conflict resolution was mediated by both mother–adolescent and father–adolescent conflict resolution behaviours for both positive and negative conflict behaviours. This mediation relationship is not surprising given the direct association between parent–adolescent conflict behaviours and sibling behaviours that has been consistently found in Patterson (1982) and Patterson (1986) work. Patterson's coercive family model indicates that parent–adolescent interactions in which parents fail to stop their child/adolescent from escalating conflict and continue to engage with their adolescent may result in increasingly negative exchanges. Furthermore, the coercive exchanges between parents and adolescents can lead to similar interactions between adolescents and their sibling (Patterson, 1986). In sum, the conflict behaviours expressed within the interparental dyad are likely to spill over into the parent–adolescent dyadic relationship (Erel & Burman, 1995), and behaviours displayed in the parent–adolescent relationship are likely to be displayed in sibling conflict. Understanding the expression of conflict within the family is paramount not only because of the importance of effective family functioning but also because family behaviours have been linked to behaviours in personal relationships outside the family. One personal relationship outside the family that becomes increasingly important during adolescence is the romantic relationship (Brown, Feiring, & Furman, 1999). Martin (1990) found that, similar to research on sibling relationships, a mediation model explains the relationship between interparental, parent–adolescent, and romantic-partner conflict resolution. Late adolescents’ reports of the frequency of overt interparental conflict in their family were related to negative conflict management strategies in the mother–adolescent and father–adolescent relationships. Mother–adolescent conflict, and to a lesser extent father–adolescent conflict behaviours, were related to romantic-partner difficulties. Although Martin (1990) provided evidence that parent–adolescent conflict mediates the relationship between interparental and romantic-partner difficulties, Martin did not examine how sibling conflict fits into this complex pattern of relationships. Because the sibling relationship is the first peer-like relationship that children experience (Shantz & Hobart, 1989), the sibling relationship appears to be essential in the development of conflict behaviours (Anderson, Hetherington, Reiss, & Howe, 1994). According to social learning theory, adolescents’ conflict behaviours in the sibling relationship would be mirrored in romantic-partner conflicts, and research by Reese-Weber & Bartle-Haring (1998) supports this idea. Reese-Weber and Bartle-Haring asked late adolescents to provide their perceptions of conflict resolution behaviours typically utilized by both dyadic members in interparental, mother–adolescent, father–adolescent, sibling, and romantic relationships. Their results suggested that (1) the relationship between interparental and sibling conflict resolution strategies was mediated by both mother–adolescent and father–adolescent dyads for both positive and negative conflict resolution strategies, and (2) the relationship between interparental and romantic-partner conflict resolution strategies was mediated by mother–adolescent, father–adolescent, and sibling dyads for negative strategies and mother–adolescent and father–adolescent dyads for positive strategies. Like Martin's research, Reese-Weber and Bartle-Haring's research suggests that conflict resolution behaviours displayed within the family are likely to predict how late adolescents handle conflicts with romantic partners. Although the role of parent conflict behaviours on sibling and romantic-relationship conflict behaviours seems clear, one limitation in the research on sibling conflict behaviours (e.g. Brody et al., 1994; Erel et al., 1998; Reese-Weber, 2000) and romantic-partner conflict behaviours (e.g. Martin, 1990; Reese-Weber & Bartle-Haring, 1998) is that the samples were drawn from two-parent, intact families. That is, previous studies have not examined the sibling and romantic-partner mediation models among adolescents from divorced families. Theory and empirical data suggest that differences do exist between intact and divorced families regarding conflict resolution. For example, Gottman's (1993) balance theory of marriage suggests that couples who use more negative and less positive conflict behaviours are more likely to dissolve their marriages than couples who are more positive and less negative during interactions. Hence, late adolescents from divorced families may report more negative and less positive interparental conflict as compared to late adolescents from intact families. In addition, because divorce typically results in fathers becoming nonresidential parents (Amato & Booth, 1996), with fewer opportunities to practice conflict resolution behaviours, more negative and less positive interactions may be exhibited in the father–adolescent relationships for late adolescents from divorced families as compared to intact families. While it seems clear that mean differences may exist in the conflict resolution behaviours for various dyadic relationships, it is not known how conflict resolution within the family differentially predicts adolescent conflict resolution behaviours between adolescents from intact versus divorced families. We can speculate on how adolescents learn conflict resolution behaviours in divorced families based on a mediation model tested for both intact and divorced families regarding other child outcomes (e.g. externalizing and internalizing behaviours). Fauber, Forehand, Thomas, and Wierson (1990) found that mothers’ disruptive parenting (e.g. rejection/withdrawal, psychological control, lax control) mediated the relationship between biological parents’ interparental conflict and child outcomes in both intact and recently divorced families, but interparental conflict predicted more variance in mothers’ rejection/withdrawal from her adolescent in recently divorced families as compared to intact families. Extrapolating from Fauber et al.'s work, the mediation model for interparental, mother–adolescent, father–adolescent, sibling, and romantic-partner conflict resolution previously found with intact families (Reese-Weber & Bartle-Haring, 1998) is likely to exist in divorced families. Moreover, Fauber et al.'s (1990) data might suggest that stronger pathways exist in divorced families as compared to intact families; however, this has not been examined with conflict resolution behaviours. There are also reasons to expect that the relationships between father–adolescent and sibling conflict, as well as father–adolescent and romantic-partner conflict, might be weaker in divorced families due to less contact with fathers. Because fathers are typically the nonresidential parent, adolescents in divorced families may not have as many opportunities to observe and practice conflict resolution behaviours in the father–adolescent relationship as compared to adolescents from intact families. Hence, one could argue that a weaker relationship may be expected from father–adolescent to sibling and romantic-partner conflict resolution in divorced families because fewer learning opportunities are present. In summary, there are reasons to believe that late adolescents’ conflict resolution behaviours in romantic relationships may be predicted differently depending on whether they are in an intact family or a divorced family, but this question has not been tested empirically. The two main objectives of the present study are to (1) replicate the mediation model found for interparental, mother–adolescent, father–adolescent, sibling, and romantic-relationship conflict resolution behaviours among a sample of late adolescents from intact families and (2) to determine whether the same model applies to a sample of late adolescents from divorced families. Although the escalation of conflict may occur because of a reciprocal interaction between any two family members (Patterson (1982) and Patterson (1986)), on the basis of social learning theory (Bandura, 1989) and empirical research with intact families (Reese-Weber, 2000; Reese-Weber & Bartle-Haring, 1998), our model specified that the causal influences of conflict resolution behaviours is primarily unidirectional from parents to adolescents. Thus, we hypothesized that interparental conflict resolution behaviours directly predict mother–adolescent and father–adolescent conflict resolution behaviours, and mother–adolescent and father–adolescent conflict resolution behaviours in turn directly predict (a) sibling conflict resolution behaviours and (b) romantic-relationship conflict resolution behaviours. Sibling conflict resolution behaviours were hypothesized to be an additional predictor of romantic-relationship conflict resolution behaviours. The present study examined the above hypotheses by assessing late adolescents’ perspectives of conflict resolution behaviours utilized within their family and romantic relationships. First, late adolescents were asked to provide their perceptions of conflict resolution behaviours for each person within a dyadic relationship (e.g. their own behaviour toward a sibling and the sibling's behaviour toward the respondent). According to Patterson (1982) and Patterson (1986) coercive family model, how two family members respond to each other is more important than how one person behaves. Second, late adolescents’ perceptions of conflict resolution behaviours were used based on Grych and Fincham's (1990) cognitive-contextual framework that suggests the meaning of the conflict or the adolescents’ interpretation of the conflict behaviours is essential for understanding the impact of those conflict behaviours on them. From this framework, adolescents’ perceptions of conflict behaviours are as important for predicting their reactions, or subsequent behaviours, as observable conflict resolution behaviours or other family members’ perspectives of the conflict. Hence, late adolescents’ perceptions of conflict resolution behaviours utilized in several dyadic relationships were assessed to examine the study hypotheses.

Results Descriptive statistics Table 2 provides the means and standard deviations for each study variable. Ten t-tests were conducted to compare adolescents from intact versus divorced families. As indicated in Table 2, late adolescents from divorced families reported less positive and more negative conflict resolution behaviours between their two biological parents as compared to late adolescents from intact families. Similarly, late adolescents from divorced families reported less positive and more negative conflict resolution behaviours in their relationship with their father as compared to late adolescents from intact families. Correlation analyses Correlation coefficients are presented in Table 3 for late adolescents from intact families (below diagonal) and divorced families (above diagonal). We used a listwise deletion for the correlations, and as such the sample sizes were 171 for the intact group and 66 for the divorce group. To determine whether the 237 for whom we had complete data differed from the 35 for whom we did not have complete data in a meaningful way, we conducted chi-square analyses and t -tests on relevant demographic variables. The two groups did not differ on gender, race, biological or step sibling as closest in age, older or younger sibling as closest in age, brother or sister as closest sibling in age, seriousness of romantic relationship, who they currently live with, age, number of biological siblings, and sibling age for sibling closest in age. However, those with and without complete data did differ in family structure, such that 72% of those with complete data were from intact families whereas 40% of those with missing data were from intact families, p<.05p<.05. Because all of our analyses are focused on differences in family structure, we do not believe our missing data pose a problem for the interpretation of these analyses. Table 3. Correlations among all study variables 1 2 3 4 5 6 7 8 9 10 1. Interparental positive 1.0 .50** .34** .37** .22 −.44** −.16 .03 −.10 .02 2. Mother–adolescent positive .41** 1.0 .41** .42** .52** −.05 −.36** .12 −.00 −.10 3. Father–adolescent positive .53** .53** 1.0 .38** .26* −.16 −.24 −.59** −.18 −.26* 4. Sibling–adolescent positive .28** .49** .58** 1.0 .40** −.15 −.28* −.09 −.29* −.33** 5. Romantic partner positive .18* .53** .42** .45** 1.0 −.16 −.36** −.15 −.03 −.36** 6. Interparental negative −.67** −.28** −.34** −.20** −.04 1.0 .43** .31* .41** .31* 7. Mother–adolescent negative −.31** −.64** −.17* −.28** −.18* .49** 1.0 .19 .44** .52** 8. Father–adolescent negative −.42** −.24** −.60** −.31** −.16* .52** .44** 1.0 .41** .30* 9. Sibling–adolescent negative −.17* −.23** −.28** −.63** −.09 .35** .43** .49** 1.0 .45** 10. Romantic partner negative −.18* −.30** −.20** −.28** −.52** .22** .33** .28** .36** 1.0 *p<.05*p<.05, **p<.01**p<.01. Note : The correlations for the intact group (N=171N=171) are presented below the diagonal and the correlations for the divorced group (N=66N=66) are present above the diagonal. The bold-faced correlations indicate significant (p<.01p<.01) differences between the two groups. Table options For both groups, positive conflict resolution behaviours in one dyadic relationship were significantly positively correlated with positive conflict resolution behaviours in all other dyadic relationships. Similarly, negative conflict resolution behaviours in one dyadic relationship were significantly positively correlated with negative conflict resolution behaviours in all other dyadic relationships. For the intact-family group, late adolescents’ reports of positive conflict resolution behaviours in one dyadic relationship were negatively correlated with negative conflict resolution behaviours in other dyadic relationships (e.g. interparental positive conflict resolution was negatively correlated with father–adolescent negative conflict resolution). This same pattern was not as evident among late adolescents from divorced families. Correlation coefficients for intact and divorced groups were compared using the Fisher's z-test, which takes the differing sample sizes of the two groups into account. Because several pairs of correlations coefficients were compared, we adopted a more conservative alpha level of .01 to reduce the probability of making a Type I error. The bold-faced correlation coefficients in Table 3 indicate a significant difference between the correlation coefficients of the intact versus divorced groups. In summary, where correlation coefficients differed in strength, correlations were stronger in the intact-family sample than the divorced-family sample. Path analyses We used a multiple-groups path analysis to compare the hypothesized model for late adolescents from intact families and those from divorced families. The model was estimated separately for positive conflict resolution behaviours (i.e. compromise) and negative conflict resolution behaviours (i.e. attack). For each conflict resolution behaviour, we estimated two models. In the equal-paths model the path coefficients (i.e. beta weights) were constrained to be equal between the intact and divorced groups. For example, the beta weight between interparental conflict resolution and mother–adolescent conflict resolution was constrained to be the same in both groups. In the unequal-paths model, the beta weights were allowed to differ between groups. If the unequal-paths model provides a better fit than the equal-paths model, then the model does not apply equally well to the divorced and intact groups. We used the chi-square difference test to determine whether the unequal-paths model provided a better fit to the data. In addition, we used the Comparative Fit Index (CFI), the root mean square error of approximation (RMSEA), and the standardized root mean square residual (SRMR) to assess overall model fit. Hu and Bentler (1999) suggested that values of .95 or higher for the CFI, .06 or lower for the RMSEA, and .08 or lower for the SRMR reflect good model fit. Preliminary analyses Before conducting the hypothesis tests we wanted to ensure that our hypothesized models did not differ by gender. Accordingly, we estimated the models for (a) negative and (b) positive conflict resolution for the males and females for whom we had complete data. These multiple groups analyses indicated that neither the negative conflict model, Δχ2(8)=9.90Δχ2(8)=9.90, p>.05p>.05, nor the positive conflict model, Δχ2(8)=13.93Δχ2(8)=13.93, p>.05p>.05, differed as a function of gender. We therefore collapsed across gender in the primary analyses. Preliminary analyses also addressed whether adolescents who were in new romantic relationships (i.e. 2 months or less in duration) might not be appropriate for inclusion in these analyses because of the limited time they have had to engage in conflict resolution behaviours with their partner. We therefore conducted all hypothesis tests both with and without adolescents in new romantic relationships. Our results did not differ, so we opted to include these adolescents in our analyses as a means to increase statistical power. Finally, we wanted to ensure that the use of reciprocal scores (e.g. the sum of adolescent to mother and mother to adolescent behaviour scores) would provide the same results as the use of the adolescents’ behaviours only (e.g. adolescent to mother). Our rationale was that a strict interpretation of social learning theory might suggest that only the adolescents’ behaviours toward others would be learned from interparental conflict resolution, whereas a more process view of conflict resolution (e.g. Patterson (1982) and Patterson (1986)) would favor the use of an aggregate measure. We therefore conducted hypothesis tests (a) with the aggregated reciprocal scores and (b) with only the adolescent behaviours as variables. Results did not differ, so we report analyses based on the more reliable aggregated reciprocal scores. Negative conflict resolution behaviours For negative conflict resolution behaviours (i.e. attack), the equal-paths model provided a good fit to the data based on all indices except the SRMR, χ2(12)=10.49χ2(12)=10.49, CFI=1.00, RMSEA=.00 (90% confidence interval [CI]=.00–.06), SRMR=.09. The unequal-paths model provided a similarly good fit to the data, χ2(4)=3.58χ2(4)=3.58, CFI=1.00, RMSEA=.00 (90% CI=.00–.09), SRMR=.04. The chi-square difference test indicated that the equal-paths model provided as good of a fit to the data as the unequal-paths model, χ2(8)=6.91χ2(8)=6.91, p>.05p>.05. Thus, because it is the more parsimonious model, the equal-paths model was accepted for the negative conflict resolution behaviours. The path coefficients for the equal-paths model are presented in Fig. 1 and apply to both intact and divorced groups. The relationship between interparental and sibling negative conflict resolution behaviours was mediated by mother–adolescent and father–adolescent negative conflict (R2=.30R2=.30 for sibling negative conflict resolution behaviours). The relationship between interparental and romantic-partner negative conflict resolution behaviours was principally mediated by mother–adolescent and sibling negative conflict (R2=.26R2=.26 for romantic-partner negative conflict resolution behaviours). The only path coefficient that was not statistically significant was the relationship between father–adolescent negative conflict resolution behaviours and romantic-partner negative conflict resolution behaviours. Because this path was not significant, we estimated a trimmed model that did not include the path from father–adolescent to romantic partner. The fit of the trimmed model was very similar to the fit of the equal-paths model, χ2(13)=12.49χ2(13)=12.49, CFI=1.00, RMSEA=.00 (90% CI=.00–.06), SRMR=.10. As such, the relationship between father–adolescent negative conflict resolution behaviours and romantic-partner negative resolution conflict behaviours appears to be entirely indirect, as mediated by sibling negative conflict resolution behaviours. Path analysis of negative conflict behaviours for equal-paths model. Fig. 1. Path analysis of negative conflict behaviours for equal-paths model. Figure options Positive conflict resolution behaviours For positive conflict resolution behaviours (i.e. compromise), the equal-paths model provided a good fit to the data based on all indices except the SRMR, χ2(12)=13.47χ2(12)=13.47, CFI=1.00, RMSEA=.02 (90% CI=.00–.07), SRMR=.10. The unequal-paths model provided a similarly good fit to the data, χ2(4)=7.24χ2(4)=7.24, CFI=1.00, RMSEA=.06 (90% CI=.00–.13), SRMR=.04. As with negative conflict resolution behaviours, the chi-square difference test indicated that the equal-paths model fit as well as the unequal-paths model, χ2(8)=6.23χ2(8)=6.23, p>.05p>.05. Thus, the equal-paths model was accepted for the positive conflict resolution behaviours. The path coefficients for the equal-paths model are presented in Fig. 2 and apply to both intact-family and divorced-family groups. Again, the relationship between interparental and sibling positive conflict resolution behaviours was mediated by mother–adolescent and father–adolescent positive conflict (R2=.32R2=.32 for sibling positive conflict resolution behaviours). The relationship between interparental and romantic-partner positive conflict resolution behaviours was mediated by mother–adolescent and sibling positive conflict (R2=.32R2=.32 for romantic-partner positive conflict resolution behaviours). Similar to the negative conflict resolution model, the path from father–adolescent positive conflict resolution to romantic-partner positive conflict resolution was not significant. Hence, a trimmed model was computed that did not include the path from father–adolescent to romantic-partner conflict and had a very similar fit to the equal-paths model, χ2(13)=14.79χ2(13)=14.79, CFI=.99, RMSEA=.02 (90% CI=.00–.07), SRMR=.09. Path analysis of positive conflict behaviours for equal-paths model.4. Fig. 2. Path analysis of positive conflict behaviours for equal-paths model.4.