Pathways between Parent–Child Interactions and Peer Acceptance: The Role of Children's Social Information Processing
Abstract
This study examined the links among parents' interaction styles, their children's social information processing, and peer acceptance. Fourth-grade children (N = 159) and their parents were observed during family discussions. One year later peer acceptance and children's information processing choices (goals, strategies, and attributions) in response to social dilemmas involving their parents and peers were assessed. Fathers' interaction styles with their children predicted both girls' and boys' information processing in regard to their fathers and peers, which, in turn, were related to peer acceptance. Mothers' interactive styles with their children predicted children's social information processing in regard to parents and peers and peer acceptance in similar ways, but only for girls. This study provided evidence that parent–child interaction is linked to children's information processing concerning their relationships with parents and peers and in turn with children being liked by peers. The implications of a social information processing approach for understanding family–peer links are emphasized.
Introduction
Researchers have made considerable progress in understanding the familial correlates of children's relationships with peers, particularly in terms of the parent–child relationship. More recently, there has been much interest in understanding the mechanisms by which children's experiences in the parent–child relationship are transferred to children's adjustment in the peer context. Researchers suggest that both understanding and regulating emotions and cognitive models or representations of the meaning of social relationships are key processes that mediate the linkages between parent–child interactions and children's peer relationships (Parke et al., 1998).
In this study we examined the role of social information processing as a potential mechanism mediating links between parent–child relations and children's peer outcomes. Although several theoretical perspectives suggest the role of cognition as a bridging mechanism between family and peer contexts, including attachment theory (Bowlby, 1969/1982) and scripts theory (Nelson, 1986; Schank & Abelson, 1977), the current study was guided by the social information processing model proposed by Dodge and his colleagues (Crick & Dodge, 1994; Dodge, 1986), which suggests specific cognitive processes as mechanisms by which children's past experiences are carried forward into their current social behavioral patterns. We focus on three aspects of cognitive social information processing, namely children's goals, strategies, and attributions, because prior research conducted within the social information processing perspective suggested that those components are important processes that determine individuals' responses in peer interaction contexts (Crick & Dodge, 1994). Goals for social relationships are characterized as motivational states that induce particular behaviors in a social situation, are evoked by interpretations of relevant events, and affect the child's selection of strategies in response to a given social situation (Chung & Asher, 1996; Crick & Dodge, 1994). Children's strategies for attaining the goal consist of their ideas about how they could behave in a specified social situation and their behavioral tendencies, because strategies are generated from social behavior repertoires stored in long-term memory (Crick & Dodge, 1994). Attributions are an important part of the interpretation process and involve evaluation of why a particular social event may have occurred and may shape children's behavioral responses to peers (Crick & Dodge, 1994). For example, children who have been repeatedly rejected by peers may develop representations of the peer world as hostile toward them. As a consequence, when they encounter other peers, they may be likely to interpret the peers' behaviors as hostile or threatening and may avoid them or respond aggressively to them (Burks, Dodge, Price, & Laird, 1999). Although the role of each of these aspects of the social information processing model in the development of peer competence has been well established (Crick & Dodge, 1994; Gifford-Smith & Rabiner, 2004), less is known about the familial origins of these components. We turn to this issue next.
The Familial Origins of Social Information Processing
The social information processing model describes a specific set of cognitive components that leads a child to enact social behaviors when faced with a social situation and suggests how a child uses existing cognitive representational models of social relationships in online processing of social cues. The assumption that guides our work is that variation in children's social information processing is acquired in the context of family interactions and family relationships.
Several lines of evidence support some parts of the hypothesis that the parents' interaction styles contribute to children's social information processing concerning their peers, which, in turn, guides their behavioral styles in peer relationships. In one study focused on information processing in social dilemmas, children's social goals or attributions concerning their relationships with parents were related to their selection of goals or attributions in relationships with peers; however, this investigation did not examine children's peer relationships (Burks & Parke, 1996). In other studies it was found that children who have negative interaction patterns with their parents or who report less acceptance and support from parents had more negative beliefs about unfamiliar peers (Rabiner, Keane, & MacKinnon-Lewis, 1993), were more aggressive with their peers (MacKinnon-Lewis, Rabiner, & Starnes, 1999), and perceived peer intent as more hostile (Cassidy, Kirsh, Scolton, & Parke, 1996). In other work, parental reports of parenting practices were related to child perceptions of their peers, through the mediation of child perceptions of their parents (Domitrovich & Bierman, 2001).
In spite of the empirical support for this line of inquiry, several limitations are evident. Firstly, although goals and strategies were examined for both parents and children in several earlier studies, the links with peer acceptance were not always assessed in these investigations. Secondly, most of the earlier studies of the links between children's experiences with their parents and their peer relationships were cross-sectional, thus limiting our ability to infer direction of causality. One notable exception is the earlier longitudinal study by MacKinnon-Lewis et al. (1999) of the associations between children's interactions with mothers, their beliefs about peers, their behaviors with peers, and acceptance by peers. However, MacKinnon-Lewis et al. (1999) focused only on children's cognitions about peers and did not address the possible links between parent–child interaction and children's cognitions about their parents as well as their peers. Moreover, this prior study was restricted to mothers and sons. Building upon this earlier study, our focus is on a longitudinal examination of not only parent–child interaction but also the goals, strategies, and attributions about both parents (mothers and fathers) and peers and their links with children's peer acceptance.
Parent and Child Gender
A further aim of our study was to examine the links among both mothers' and fathers' interactions with their children and children's social information processing and peer outcomes. To date, much of the work in this area has focused on mothers, in spite of the accumulating evidence that fathers also make potentially important contributions to their children's peer relationships (Ladd & Pettit, 2002; Parke, 2002).
In this section, we review the limited prior work that has included both fathers and mothers. Recently, McDowell, Parke, and Spitzer (2002) examined the links among mothers', fathers', and children's social information processing and children's social competence. This investigation found that not only was there consistency between parents and children in goals and strategies but fathers' and children's goals and strategies were related to children's social competence as rated by peers and teachers. However, there was no measure of parent–child interaction in this study. MacKinnon-Lewis, Castellino, Brody, and Fincham (2001) have provided evidence that fathers' earlier negative behavior in interactions with their preadolescent children predicted children's subsequent attributions about their fathers. In this study, neither evaluations of peer attributions nor children's social competence were available. Crockenberg and Lourie (1996) measured all of these components, namely parent–child interaction, children's social information processing, and children's social competence. They found that paternal coercion and reasoning were related to children's conflict strategies, whereas fathers' use of coercion was negatively related to the social competence of girls as rated by both fathers and mothers. Although this study illuminates the role of fathers in children's use of social information strategies, it is limited by the use of parents as reporters of both coercion and children's social competence. Therefore, one of the goals of the present study was to examine the links among mother–child and father–child interaction and children's social information processing in relationships with mothers, fathers, and peers. By focusing on both mothers' and fathers' interactions with their children, children's social information processing, and peer outcomes, we will provide a more comprehensive examination of the role of child cognitive processing as a link between parent–child interactions and peer acceptance.
Finally, the role of mothers and fathers in relation to children's peer relationships may vary for boys and girls. For example, it has been found that parents' affect expressions were associated with children's social functioning especially in same-sex (father–son and mother–daughter) dyads (Boyum & Parke, 1995; Isley, O'Neil, Clatfelter, & Parke, 1999). Similarly, maternal hostility was a stronger predictor of victimization for girls than boys (Schwartz, Dodge, Pettit, & Bates, 1997), and parents' social coaching and interaction styles operate differentially for boys and girls (Mize & Pettit, 1997). However, in an earlier study that explored pathways linking parenting practices, children's perceptions of their parents and peers, and social adjustment, gender differences were not found (Domitrovich & Bierman, 2001). Based on the limited prior literature, we expected that there would be stronger relations between same-sex parent–child dyads than opposite-sex dyads.
The Present Study
The goal of this study was to examine the hypothesis that parents' interaction styles are related to their children's social information processing about their relationships with parents, and that children's cognitive processing about parents would generalize to their processing about relationships with peers. In turn, it is hypothesized that children's information processing concerning peer relationships is related to peer acceptance. Figure 1 outlines a model that portrays the links among these components. In our study, we selected goals, strategies, and attributions as specific components of cognitive processing about interpersonal relationships. It was hypothesized that as a result of positive experiences with parents who were helpful, warm, and responsive, children would develop non-hostile attributions, positive social strategies, and relationship-oriented goals that, in turn, would lead to higher peer acceptance. We also examined whether the paths differed for boys and girls.
Hypothesized Path Model of Family–Peer Linkages.
Method
Participants
Participants were part of a longitudinal study of children's social development in which children were randomly selected from a public school district in the Western United States in approximately equal numbers to represent five sociometric categories: neglected, popular, rejected, average, and controversial (see Parke et al., 1997;Simpkins & Parke, 2002, for detailed information on sampling and attrition). In fourth and fifth grades, 159 children (83 boys, 76 girls) participated in assessments of parent–child interaction; goals, strategies, and attributions; and peer acceptance. For 126 children both parents participated in the assessment of parent–child interaction, for three children only fathers participated, and for 11 children only mothers participated. The ethnicity of the sample was European American (50 percent), Latino (40 percent), and African-American, Asian American, or other (10 percent). Median family income was approximately $40,500 (ranging from under $10,000 to greater than $58,000), and mean educational attainment was 12.6 years for mothers and 13.0 years for fathers.
Procedures
When children were in fourth grade, they and their mothers and fathers participated in videotaped interaction tasks at a university laboratory and completed a demographic questionnaire. One year later, the children completed an interview assessing their goals, strategies, and attributions concerning mothers, fathers, and peers. Sociometric interviews were conducted in fifth-grade classrooms. Each of these procedures is described in detail below.
Measures
Parents' Interactive Styles with Their Children. Parents and their children participated in several videotaped dyadic interaction tasks (mother and child; father and child separately). Parents were instructed to discuss several issues with their children to assess their interactive styles. The order in which mothers and fathers interacted with their son or daughter was counterbalanced across families. The issues used to induce parent–child discussion were adapted from the taxonomy of problematic situations (Dodge, McClaskey, & Feldman, 1985), which asked each family member to rate how difficult it would be for the child to handle each of the situations. Parents and children were individually administered a questionnaire concerning peer and family issues. Ratings were based on a five-point scale for adults and a three-point scale for children, with higher numbers indicating more difficulty for the family. Seven issues were then chosen by the experimenter for the mother–child dyad or the father–child dyad to discuss during separate 10-minute observation sessions. At least one family member had to identify the situation as being difficult for the child to deal with in order to be included. As a result, there was some overlap in the topics that children discussed with their mothers and fathers but also some unique topics for each parent. Parents were free to discuss the topics in any order during the interaction period. There was no indication that the behavior ratings of the interactions varied by the nature of the particular topics that were used in the mother–child vs. father–child discussions or in terms of the unique vs. overlapping topics across parent–child sessions. These separate 10-minute videotaped discussion sessions (one with mother and child and one with father and child) were then rated on several dimensions based on rating schemes adapted from those developed by Conger and his colleagues (see Rueter & Conger, 1995). These included the following: encouraging independence (extent to which the parent reinforces child's initiative or confidence in child's ability to solve problems and make decisions), warmth/support (extent to which the parent offers encouragement, shows concern for child's welfare, or uses physical gestures such as touching or sitting close), positive responsiveness (extent to which the parent praises, approves, rewards, or smiles in response to appropriate behaviors or responses), inductive reasoning (extent to which the parent focuses on consequences of behaviors, avoids conflict through use of reasoning, or focuses on feelings of others), and dominance (extent to which parent attempts to influence, directs, or guides the child's opinions or behaviors, or is bossy or forceful). Each dimension was rated globally after observing the discussion of all issues, using a five-point scale (1 = ‘not at all characteristic’; 5 = ‘highly characteristic’). Five undergraduate university students were trained to code all videotapes of parent–child interaction with a goal of reaching an inter-coder reliability of .75. Twenty percent of the videotaped parent–child interaction segments were randomly selected to determine inter-rater reliability. Coders coded either mother–child or father–child interaction from an individual family. This procedure ensured that mother–child and father–child interaction from a single family were coded by independent raters. The inter-rater reliabilities assessed by calculating alphas for the raters ranged from .78 to .91 (M = .85). The predictive validity of these scales has been well established. For example, Conger and colleagues have found that these ratings of parent–child interaction were related to a variety of child outcomes including internalizing and externalizing (Conger & Elder, 1994; Scarmella & Conger, 2004). Others (McDowell & Parke, 2000; McDowell, Parke, & Wang, 2003) have found these observational ratings were linked to peer and teacher assessment of social competence with peers.
Two principal components analyses based on data from mothers and fathers were used to reduce the five dimensions to reliable and meaningful composite factors. In both cases, one factor emerged. The four positive variables and one reverse-scored dominance variable loaded on single factor, but the dominance variable was dropped due to its low loading (less than .38). The positive interaction factor consisted of the following variables: encouraging independence, warmth/support, positive responsiveness, and inductive reasoning. The positive interaction score was calculated by averaging the ratings for these variables (Cronbach's alphas = .86 and .83, for mothers and fathers, respectively).
The Child's Social Information Processing Regarding Parents. Children's cognitive processing was assessed by using measures of the child's goals, strategies, and attributions. Children were individually presented with two hypothetical ambiguous provocation situations involving a parent as the target character and were interviewed with close-ended questions concerning their goals for the situations, their response strategies, and their interpretation of the parent's intent. Each ambiguous provocation vignette described a situation in which a parent's behavior has brought harm to the child, but the parent's intention was not clear (see Appendix for descriptions of these vignettes). Children were asked to respond on a five-point Likert scale (1 = ‘definitely would/should not do’ to 5 = ‘definitely would/should do’) to each of alternatives for questions concerning their goals and strategies (‘What would your goal be?’ and ‘What would you do?’). The goals included an aggressive goal (e.g., try to get back at parent), two relational goals (e.g., try to work things out peacefully; try to still get along with parent), and an avoidant goal (e.g., try to stay away from parent). The strategy responses included two aggressive responses (e.g., spill coffee on something of the parent's; not let the parent touch their things again), a relational response (e.g., ask parent why he/she spilled coffee on child's project), and an avoidant response (e.g., avoid being near parent when working on a project in the future). Children were also asked to select from two options (either ‘on purpose’ or ‘by accident’) for the attribution of parent's intent. The alternative responses were randomly ordered for each story presented. Children were interviewed twice: once with mother as the story character and once with father in the role of the story character. The order in which children were interviewed about mothers and fathers was counterbalanced.
Eight goal items and eight strategy items from two story vignettes with each parent were combined into three goals (relational, aggressive, and avoidant) and three strategies (relational, aggressive, and avoidant), and the related items were averaged for each of three goals or three strategies. The coefficient alphas of three goals and three strategies with mother and father ranged from .65 to .83 (M = .75). The attribution variable score was the mean of the ratings across stories (a hostile interpretation was coded as 1 and a benign interpretation was coded as 2, so that the higher the score, the more positive the attribution). As a result of preliminary analyses, attributions about parents were excluded in analyses due to its low variance for fathers' intent (most children endorsed the statement that father behaved ‘accidentally’) and zero variance for mothers' intent (all children endorsed ‘accidentally’). Two child cognitive processing variables regarding each parent were generated by principal component analyses using three goal and three strategy scores: positive cognitive processing (a single item loaded: relational strategy) and negative cognitive processing (five items loaded: relational goals [reverse-coded], aggressive goals, aggressive strategies, avoidant goals, and avoidant strategies; Cronbach's alphas = .73 and .75 for children's cognitive views about their fathers and mothers, respectively. The items were standardized and averaged for the composite variables).
The Child's Social Information Processing Regarding Peers. Children's cognitive processing with peers was also assessed with the child's goals, strategies, and attributions. Children were individually presented with two hypothetical ambiguous provocation vignettes except that a peer, instead of a parent, was the target character. The ambiguous provocation stories and the response goal, attribution types, and strategy categories were identical to those used for parents (except that juice was substituted for coffee). The children were asked questions in order to identify their goal and strategy (‘What would your goal be?’ and ‘What would you do?’) as well as their attributions (‘accidentally’ or ‘on purpose’). The approach of using similar stories allowed for a comparison of children's responses to father, mother, and peers in analogous situations.
The goals, strategies, and attribution regarding peers were generated using the same guidelines that were used for the parent-based vignettes (see Appendix for descriptions of these vignettes). Eight goal items and eight strategy items from the two vignettes were combined to generate three goals and three strategies. In light of an examination of the reliabilities of the variables, relational strategy was dropped because of its low alpha (less than .65). The alphas of three goals (relational, aggressive, and avoidant) and two strategies (aggressive and avoidant) ranged from .65 to .78 (M = .72). The attribution variable score was also generated using the same method as the attribution about parents' intent. Three goals, two strategies, and attribution scores were reduced to one component by principal component analysis: negative cognitive processing. Reversed score of relational goal, aggressive goal, avoidant goal, aggressive strategy, avoidant strategy, and reversed score of attribution were standardized and averaged (alpha = .85).
Children's Peer Acceptance. Sociometric interviews (Asher, Singleton, Tinsley, & Hymel, 1979) were conducted in fifth-grade classrooms in order to assess the degree to which each child was liked by peers and disliked by peers, by asking students to nominate three classmates with whom they liked to play or spend time and three classmates with whom they disliked to play or spend time. Brief interviews were conducted with all children whose parents granted permission to participate. These nominations were tallied for each child and standardized within each classroom.
Results
Path analyses using AMOS software was employed to examine the hypothesized model that children's cognitive processing with parents and peers would be related to both parents' interaction and children's peer acceptance. Little's test was used to determine whether the missing data were randomly distributed throughout the dataset rather than revealing a systematic pattern. The nonsignificant test confirmed the assumption that the missing values were randomly distributed, χ2(50) = 48.88 p > .05.1 To address this issue of missing data, we used full information maximum likelihood procedures to estimate our path models as implemented in the AMOS program (Arbuckle, 2003).
As a preliminary step, bivariate correlations were examined in order to determine whether relations among variables in the path model conformed to expectations. Table 1 presents the bivariate correlations, means, and standard deviations for the variables included in the model for the sample with corrections for missing data. Correlations revealed that (1) children who experienced more positive interactions with their mothers and fathers tended to have less negative cognitive processing (goals and strategies) with each parent and with peers one year later, but there was no significant relation between mothers' and fathers' interactions and peer acceptance scores; (2) children who endorsed more negative cognitive processing in regard to mothers and fathers were likely to endorse more negative cognitive processing in regard to peers. Children who endorsed more negative cognitive processing in regard to fathers were less liked by peers, but there was no significant relation between cognitive processing in relation to mothers and peer acceptance; (3) children who held more negative cognitive processing in regard to peers were less liked by peers.
| Variable | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 |
|---|---|---|---|---|---|---|---|---|---|
| Parents' interactions | |||||||||
| 1. Fathers' positive interactions | — | ||||||||
| 2. Mothers' positive interactions | .12 | — | |||||||
| Cognitive processing with fathers | |||||||||
| 3. Positive cognitive processing | −.14 | .02 | — | ||||||
| 4. Negative cognitive processinga | −.20* | −.13 | .17+ | — | |||||
| Cognitive processing with mothers | |||||||||
| 5. Positive cognitive processing | −.10 | −.09 | .63*** | .07 | — | ||||
| 6. Negative cognitive processinga | −.10 | −.18+ | .14 | .95*** | .07 | — | |||
| Cognitive processing with peers | |||||||||
| 7. Negative cognitive processinga | −.27** | −.19* | .04 | .45*** | .12 | .44*** | — | ||
| Peer acceptance | |||||||||
| 8. Peer like | −.01 | .05 | .01 | −.19+ | .03 | −.14 | −.35** | — | |
| 9. Peer dislike | −.07 | −.11 | .10 | .05 | .02 | .08 | −.04 | −.24** | — |
| M | 2.87 | 3.07 | 4.26 | 1.60 | 4.40 | 1.53 | −.00 | .18 | −.16 |
| SD | .76 | .71 | .98 | .73 | 1.00 | .67 | .76 | .92 | .78 |
- a These variables were a composite of standardized measures.
- + p < .10,
- * p < .05,
- ** p < .01,
- *** p < .001.
Links among Parental Interactions, Children's Social Information Processing with Parents and Peers, and Children's Peer Acceptance
Hypothesized Path Model. In our path model, we tested whether the links between parental interactions (in fourth grade) and peer acceptance (in fifth grade) were mediated through children's cognitive processing concerning parents and peers (in fifth grade). The hypothesized links were tested separately for models including each parent using AMOS and then we estimated the fit of each model. The standardized path coefficients for the path model for fathers and mothers are presented are presented in Figure 2. The fit of the model was assessed with commonly used indices including the χ2-test statistic and associated level of significance, the comparative fit index (CFI; Bentler, 1990), the Tucker–Lewis index (TLI) (or non-normed fit index [NNFI]; Tucker & Lewis, 1973), and the root mean square error of approximation (RMSEA; Browne & Cudeck, 1993). We considered models acceptable if the CFI value and the TLI value were greater than .95 (Hu & Bentler, 1999), and the RMSEA value was around .05 (but less than .08) (Kaplan, 2000).
Hypothesized Path Model for Fathers and Mothers (in Parenthesis). Standardized path coefficients are provided.Note: +p < .10, *p < .05, **p < .01.
The overall hypothesized model for fathers' interaction and children's cognitive processing with fathers was found to have an acceptable fit to the data, χ2(9) = 15.86, p = .07, TLI(NNFI) = .98, RMSEA = .07, CFI = .99. The hypothesized model for mothers' interaction and children's cognitive processing with mothers was found to have a strong fit to the data, χ2(9) = 10.87, p = .29, TLI(NNFI) = 1.00, RMSEA = .04, CFI = 1.00. As expected, based on earlier correlation analyses, fathers' and mothers' interactions did not predict children's positive cognitive processing with fathers and mothers and, further, children's negative cognitive processing with peers was not significantly associated with their dislike ratings by peers.
Reduced Model. Although the hypothesized models were judged to fit the data well, further modifications of the hypothesized model were made to improve fit and parsimony by dropping non-significant pathways from the model. The paths from parents' interaction to children's positive cognitive processing regarding parents and the paths from children's negative cognitive processing regarding peers to peer dislike were eliminated. The standardized and unstandardized path coefficients for these resulting reduced models for fathers and mothers are presented in Figure 3.
Reduced Model for Fathers and Mothers (in Parenthesis). Unstandardized path coefficients and standardized path coefficients (bold typed) are provided.Note: +p < .10,*p < .05, ***p < .001.
The reduced model including fathers' interaction and children's cognitive processing regarding fathers provided a strong fit to the data, χ2(3) = 4.62, p = .20, TLI(NNFI) = .99, RMSEA = .06, CFI = 1.00. Estimated path coefficients were consistent with our predictions. The path estimates of the association between fathers' interaction and peer liking through children's cognitive processing in regard to parents and peers were all significant. These results suggest that children who had more positive interaction with their fathers in fourth grade were less likely to suggest negative goals and strategies in regard to solving interpersonal problems with their fathers in fifth grade and that they were also less likely to endorse negative goals and strategies in solving social dilemmas with peers and finally to be liked by peers. Given that the path estimates for the hypothesized linking processes were significant, estimates of indirect effects were obtained. Fathers' positive interactions were indirectly associated with children's negative cognitions regarding peers (−.09) and being liked by peers (.04). The indirect paths of the children's negative cognitions in regard to their fathers through children's negative cognitions about peers to being liked by peers were estimated at −.20.
The reduced model including mothers' interaction and children's cognitive processing with mothers also demonstrated a strong fit to the data, χ2(3) = 1.65, p = .65, TLI(NNFI) = 1.00, RMSEA = .00, CFI = 1.00. Mothers' positive interaction with their child in fourth grade was marginally negatively related to children's negative goals and strategies in relation to their mothers in fifth grade, but the paths from children's negative goals and strategies regarding mothers and peers to peer liking were significant. Although there was a marginally significant association, estimates of indirect effects for pathways of the model were obtained. Mother's positive interaction was indirectly associated with children's negative goals and strategies regarding interaction with peers (−.09) and peer acceptance (.04). The indirect effect of the children's negative goals/strategies with regard to mothers through children's negative goals/strategies with regard to peers to peer acceptance was −.22.
Moderation by Gender of Child. We next examined whether the child gender moderated the associations found in the reduced model. Multigroup analyses using AMOS were conducted dividing the sample into families with girls (N = 79) and with boys (N = 83) to determine if the model was equivalent across child gender. We imposed a cross-group equality constraint on all path estimates, and then assessed whether the goodness-of-fit for the constrained models was as good as for the unconstrained models. If the fit of the constrained model was worse than that for the corresponding unconstrained model, then it was concluded that model effects differed by group.
For the multigroup analyses of the path model including fathers' interactions and child's goals/strategies with fathers, the unconstrained multigroup model fit the data well, χ2(6) = 7.29, p = .30, TLI(NNFI) = .993, RMSEA = .037, CFI = .998, and the constrained multigroup model still fit well, χ2(9) = 8.05, p = .53, TLI(NNFI) = 1.003, RMSEA = .000, CFI = 1.000. Because this constrained model still fit the data, this indicates that the hypothesized path model was not moderated by gender and held equally for boys and girls.
For the multigroup analyses of the path model including mothers' interactions and child's goals/strategies with mothers, the unconstrained multigroup model fit very well, χ2(6) = 3.60, p = .73, TLI(NNFI) = 1.011, RMSEA = .000, CFI = 1.000, but the constrained multigroup model, χ2(9) = 11.73, p = .23, TLI(NNFI) = .991, RMSEA = .044, CFI = .996, fit significantly worse than the unconstrained model (Δχ2(3) = 8.13, p < .05). Because this result indicated that the model was moderated by gender, we explored the source of gender differences by freeing constraints on the basis of the LaGrange multiplier test. The equality constraint for each of two path estimates that showed a distinct difference in the magnitude and significance of unstandardized coefficients across gender was dropped. This revealed that the path from mothers' positive interaction to negative goals/strategies in regard to mothers (B = −.44, p < .01, B = .01, p > .05, for girls and boys, respectively) and the path from negative goals/strategies with mothers to negative goals/strategies with peers (B = .61, p < .001, B = .27, p > .05, for girls and boys, respectively) were significant for girls but not boys. With the constraint of the path from mothers' positive interaction to negative goals/strategies with mothers freed, the model fit the data well, χ2(8) = 5.85, p = .67, TLI(NNFI) = 1.008, RMSEA = .000, CFI = 1.000, and made a significant improvement over the constrained model (Δχ2(1) = 5.89, p < 05). With the constraint of the path from negative cognitive processing (goals/strategies) with mothers to negative cognitive processing (goals/strategies) with peers freed, the model fit the data well (χ2(8) = 9.49, p = .30, TLI(NNFI) = .995, RMSEA = .034, CFI = .998), but it was not significantly different from the constrained model (Δχ2(1) = 2.24, p > 05). These results suggest that gender moderation was evident for the path from mothers' positive interactions to children's negative goals/strategies with mothers, but not for the path from children's negative goals/strategies with mothers to children's negative goals/strategies with peers. Although the overall model fit for both girls and boys, the hypothesized path from mothers' positive interactions to negative goals/strategies with mothers was significant only for girls. Moreover, the indirect effect of mothers' positive interaction through negative goals/strategies with mothers and negative goals/strategies with peers to peer liking was significant only for girls. For girls, the path coefficient from mother's positive interaction to negative cognitive processing with mothers was estimated at −.44, the coefficient from negative cognitive processing with mothers to negative cognitive processing with peers at .61, and the coefficient from negative cognitive processing with peers to peer liking was estimated at −.40. Therefore, higher mother's positive interaction with girls was associated with girls' higher peer liking indirectly through their lower levels of negative goals/strategies with peers.
Discussion
The goal of this study was to extend prior work regarding selected aspects of children's social information processing regarding social relationships with parents and peers as mediating links between family and peer relationships. Guided by social information processing theory, we tested the hypothesis that parents' interaction styles are related to their children's information processing about relationships with parents, and, in turn, children's information processing regarding parents generalizes to their cognitive processing about relationships with peers. In turn, their information processing about peers is related to peer acceptance.
Links among Parent–Child Interaction, Children's Social Information Processing with Parents and Peers, and Peer Acceptance
Although there were no significant direct associations between parents' prior interaction and children's peer acceptance one year later, evidence was found for indirect associations among the variables in the model. Children who had more positive interactions with their parents endorsed fewer negative goals and strategies in regard to their parents and fewer negative goals, strategies, and attributions in regard to peers one year later. In turn, children's cognitive processing in regard to peers was associated with peer acceptance. Our findings extend the results of earlier work that used children's perceptions of parenting as an indicator of their representation of their relationships with parents (Domitrovich & Bierman, 2001) by more closely specifying and evaluating clear dimensions of social information processing by children. These results are consistent with prior research by MacKinnon-Lewis et al. (1999), who reported that mothers' interactional negativity was directly related to lower peer acceptance concurrently but indirectly related to peer acceptance one year later via prior peer acceptance and current beliefs about peers. Although children's social information processing regarding peer relationships as expressed by their choice of goals and strategies might be influenced by prior experience with parents, their cognitive processing regarding peers may be influenced by prior peer interaction experiences as well. This inference was based on the expectation that children's experiences with peers should gradually exert a growing influence relative to parents throughout middle childhood. Further studies guided by a longitudinal transactional model are needed to examine how the relative contribution of peers and parents to children's peer acceptance varies over time.
Secondly, our results suggest that there are unique as well as similar relations between parents' interactive styles and children's cognitive processing about parents for mothers and fathers. Although the plausibility of children's cognitive processing as a link between fathers' interaction and children's peer acceptance was confirmed by the fit of the overall path model for both boys and girls, the model including mothers was supported only for girls. Boys' cognitive processing with mothers was not predicted by mothers' previous interactive styles, indicating that fathers may have greater influence for boys in middle childhood than do mothers. This partial support for our expectation that the links among the predictors would be stronger for same-sex parent–child dyads is consistent with earlier work (Isley et al., 1999; Schwartz et al., 1997). Our findings underscore the importance not only of including fathers in our studies of parent–peer relationships, but examining mother and father influences in relation to the gender of the child as well.
Children's cognitive processing (i.e., selection of goals and strategies) regarding parents, which is related to their prior interactive experiences with parents, was associated with children's selection of goals and strategies in regard to their peers and in turn, to the quality of their peer relationships. Children who suggested more aggressive and avoidant goals and strategies and fewer relational goals in regard to their parents suggested more negative goals and strategies as well as more hostile attributions in addressing disagreements with peers as well. In addition, children who made more hostile attributions and selected more negative goals and strategies in relation to peers were less accepted by peers. As has been argued by other researchers (Cassidy et al., 1996), these findings suggest that cognitive processing regarding interpersonal relationships is represented within the individual and can be applied in multiple social contexts.
The finding of significant positive relations between children's social information processing regarding parents and peers replicates earlier work (Burks & Parke, 1996). The linkage between children's cognitive processing with parents and with peers provides potential evidence that the information processing about parents are carried over to the information processing about relationships with peers. It assumes that representational models of self, others, and relationships established in early family relationships influence expectations concerning responsiveness and attitudes of others and, in turn, recreate similar pattern of relationships with peers (Shulman & Collins, 1995; Sroufe & Fleeson, 1986). Although it is more reasonable to posit such a direction of influence between the two than the reverse direction, it is possible that peer experiences may alter children's cognitive processing about parents as well (Carlson, Sroufe, & Egeland, 2004). This direction of effects issue cannot be settled by the current study.
What are the mechanisms that account for these similarities in social information processing about family and peer social relationships? The pathways in the family context by which children's cognitive processing (i.e., goals, strategies, attributions) about others are learned might be through their observations of parent's moment-to-moment interactions with either children themselves or their siblings, by the advice parents offer their children in conflict situations with parents, siblings, or peers, or by witnessing or being told about resolution strategies parents use in settling disputes with their own adult peers. This reasoning is supported, in part, by prior findings that parent's social information processing in regard to their adult peers and in regard to their children were related to their children's social information processing in regard to peers (McDowell et al., 2002). Finally, parental marital conflict resolution strategies that are observed by children may be a further source of children's social goals, strategies, and attributions (Grych & Cardoza-Fernandes, 2001), which may, in turn, be linked with variations in the quality of peer relationships (Gottman & Katz, 1989; Parke et al., 2001).
The findings concerning the relations between children's social information processing about peers and peer acceptance are consistent with a body of earlier literature indicating that children's information processing is linked with peer acceptance (Crick & Dodge, 1994; Gifford-Smith & Rabiner, 2004). In our models, peer liking was found to be associated with children's cognitive orientations toward peers but, surprisingly, being disliked by peers was not related with cognitive processing. Perhaps being disliked by peers is more strongly affected by children's negative behaviors than by negative cognitions. Negative cognitions may be linked with negative peer rating most often when these cognitions are also manifest in observed behavior toward peers. Future studies need to include cognitive processing measures as well as observations of behavioral interactions among peers as well as assessment strategies to increase the variability in dislike ratings. Assessment contexts may be important to consider as well. Whereas we used an ambiguous intent context, other studies using conflict situations have found clear links between cognitions and negative peer outcomes. For example, prior work showed that revenge goals in a conflict situation were associated with friendship maladjustment (Rose & Asher, 1999). Nevertheless, these results provide a consistent picture of negative cognitive processing as a strong negative predictor of peer acceptance, a finding that supports our theoretical model.
Conclusions, Limitations, and Future Directions
This study provided empirical support for notions derived from both attachment theory and social information processing theory, specifically that cognitive processing about parents is developed from experience with parents and, in turn, this processing contributes to peer acceptance. However, we note several limitations. The first limitation concerns our study design, which involved measurement of parent–child interaction in one year and assessment of cognitive processing one year later. A design in which both interaction and social cognitive processing were measured at both time points would have permitted clearer inferences about the direction of influence between parent–child interaction patterns and children's social information processing choices. A final caveat concerns our use of similar vignettes for assessment of cognitive processing about mothers, fathers, and peers. Although the context (home vs. school) and the object (juice vs. coffee) that was the focus of the provocation varied across parents and peers, the structure of the story vignettes was similar. Therefore, our associations between parent and peer cognitive processing may be inflated due to the similarity of story structure. However, several methodological features mitigate against this argument. A variety of procedures (e.g., questionnaires, behavioral assessment of display rules) intervened between the children's social information processing assessments and, as noted above, the order of mother and father assessments was counterbalanced. Nevertheless, in future studies, different stories should be used for parents and peers in order to provide a more stringent test of our model. Despite these limitations, our study extends understanding about the nature of the social cognitive processes involved in the linkages between family and peer systems. By focusing on the issue of the relations between children's interactive experiences with parents and children's social information processing we have demonstrated that children's cognitive processing is one important pathway between families and children's social relationships with peers.
Appendix
Story vignettes for child goals, strategies, and attribution with parents (in parenthesis) and peers: Two hypothetical ambiguous provocation situations
- 1
Imagine that you have finished a project for school. You've worked on it a long time and you're really proud of it. Another child (your dad/mom), holding a glass of juice (a cup of coffee), comes over to look at your project. You turn away for a minute and when you look back, child (dad/mom) has broken your brand new toy/game.
- 2
Imagine that you brought a new toy/game to school (home) today. You saved up your allowance to buy the toy/game and you want to show it to the other kids (dad/mom) at school. You let another child (dad/mom) play with it for a few minutes while you go get a drink of water. When you get back you realize that the child (dad/mom) has broken your brand new toy/game.
References
Acknowledgments
This research was supported by NSF Grant BNS 8919391 and NICHD Grant HT 32391 to Ross D. Parke. We would like to thank the Fontana and Jurupa Unified School Districts for their generous and continued support of the University of California Riverside Social Development Project. We are especially grateful to all of the families whose ongoing involvements make this project possible.
