Mothers' Cognitions about Relational Aggression: Associations with Discipline Responses, Children's Normative Beliefs, and Peer Competence
Abstract
Prior research has shown that parental social cognitions are associated with child outcomes such as aggression. The goal of this study was to examine mothers’ cognitions about relational aggression, and to explore linkages between mothers’ attributions and normative beliefs about aggression and children's competence with peers. Participants included 103 mothers and children in grades 3 through 6. Results showed that mothers viewed relational aggression as more acceptable and normative than physical aggression, and they attributed less responsibility to children for using relational aggression. Maternal cognitions also predicted levels of sternness and disapproval in response to child relational aggression, and children's beliefs about the acceptability of relational aggression, which were associated with children's teacher-rated peer competence. Sex differences in the patterns of associations between maternal cognitions, discipline responses, child norms and peer competence were found. Applications of these results to parent education programs that are focused on relational aggression are discussed.
Introduction
In the last two decades, research on the social-cognitive and affective bases of parenting has significantly increased our understanding of individual differences in parenting practices such as discipline (Dix, 1991, 1993; Patterson, 1982). To the extent that parents’ appraisals of situations involving children predict their responses to children, this line of research has great potential to elucidate the processes through which parents influence children's developing social competence. The vast majority of research in this area has focused on parents’ appraisals of children's misbehavior in the home; less information is available about parents’ cognitions about and responses to children's negative behavior with peers. In addition, previous research has focused predominantly on parents’ cognitions about children's overtly disruptive behavior (e.g., stealing, physical aggression, disobedience; Dix & Lochman, 1989; Dix, Ruble, Grusec, & Nixon, 1986; Slep & O’Leary, 1998). The present study extends this work by focusing on child relational aggression. Our goals were to examine mothers’ cognitions about relational aggression as compared to physical aggression, and to explore pathways through which mothers’ social cognitions are associated with children's peer competence.
Relational aggression is a form of behavior observed in children as young as three years-old that is characterized by the removal, or the threat of removal of a relationship as the means of harm (Ostrov, Crick, & Keating, 2005). Examples of relational aggression include social exclusion, malicious gossip, and threats to withdraw friendship in order to control a peer. Recent longitudinal studies have shown that individual differences in relational aggression are moderately stable across early and middle childhood and comparable to the stability of physical aggression (Crick, 1997; Crick, Ostrov, & Werner, 2006; Ostrov, Burr, Jansen, Cullerton-Sen, & Crick, 2004). Engagement in relational aggression has also been shown to predict increases in children's social, psychological, and behavioral maladjustment over time, particularly internalizing and externalizing difficulties and peer rejection (Crick et al., 2006; Crick, 1996).
In light of these findings, the paucity of investigations into the familial origins of relational aggression is somewhat surprising. Available studies suggest that broad parenting behaviors and styles (high levels of coercive and psychological control and permissive parenting; low responsiveness) are associated with relational aggression and physical aggression (Casas, Weigel, Crick, Ostrov, Woods, & Yeh, 2006; Hart, Nelson, Robinson, Olsen, & McNeilly-Choque, 1998; Nelson & Crick, 2001). Few studies, to date, have identified unique mechanisms through which parents may socialize relationally aggressive behavior in their children.
One exception is a study by Werner, Senich, and Przepyszny (2006) who examined mothers’ responses to hypothetical situations depicting their preschool-age children engaged in relational and physical aggression. Compared to their responses to physical aggression, mothers were more likely to indicate they would do nothing in response to relational aggression. Mothers who did say they would intervene proposed strategies that were significantly lower in power assertion and in communication that the child violated a conventional or moral rule (rule violation) as compared to their responses to physical aggression. For example, mothers suggested strategies such as distracting the offending child in response to relational aggression (low power assertion), rather than imposing a negative consequence such as a time-out (moderate power assertion). Importantly, children whose mothers reported higher levels of rule violation in relationally aggressive conflicts were described by teachers as less relationally aggressive and more prosocial with peers, but they were not less physically aggressive. Although the direction of effects cannot be determined, these results suggest that the qualities of mothers’ responses to children's relationally aggressive behavior might play a role in the development of relational aggression in early childhood.
An important question remains: why do parents report distinct responses to relational and physical aggression? Werner et al. (2006) suggested that parents’ cognitions about relational aggression are likely to differ from those about physical aggression—cognitions which, in turn, influence their proposed socialization responses to children. Although they did not directly test this hypothesis, previous research supports the hypothesis that parental cognitions are sensitive to the type of misconduct displayed by children (i.e., norm violations, e.g., lying, fighting vs. failures to be altruistic; Grusec, Dix, & Mills, 1982). Moreover, several studies have found that adults view relational aggression as less hurtful, serious, and problematic, relative to physical aggression (Bauman & Del Rio, 2006; Hazler, Miller, Carney, & Green, 2001; Holt & Keyes, 2004).
The first goal of the present investigation was to directly test the hypothesis that mothers’ cognitions about relational aggression differ from their cognitions about physical aggression. We measured three types of maternal cognitions: beliefs about the acceptability of aggression (normative beliefs), beliefs about the normative nature of aggression (descriptive norms), and attributions of child responsibility for engaging in aggression (responsibility attributions).
We hypothesized that mothers would view relational aggression as more acceptable and more normative, and they would attribute less responsibility to children for using relational aggression, as compared to physical aggression.
Pathways Linking Maternal Cognitions and Children's Peer Competence
Our second goal was to examine two pathways through which maternal cognitions about relational aggression might influence children's peer competence. Previous research has found reliable associations between parental social cognitions and children's peer competence. For example, parents of physically aggressive children are likely to attribute children's misbehavior to dispositional causes and to infer that children behaved with hostile intent (Dix & Lochman, 1989; Dix, Ruble, & Zambarano, 1989; MacBrayer, Milich, & Hundley, 2003; MacKinnon-Lewis, Lamb, Arbuckle, Baradaran, & Volling, 1992; Pettit, Dodge, & Brown, 1988). Pettit et al. (1988) also reported that mothers who endorsed aggressive solutions to social problems had preschoolers who were less socially competent according to a composite of peer and teacher ratings. In a study by McDowell, Parke, and Spitzer (2002), fathers who endorsed relational-prosocial goals and strategies in social situations had children who were more socially competent as rated by their kindergarten teachers. Despite these important results, few studies have investigated the processes through which parental cognitions influence children's behavior in the peer group. At least two mechanisms might account for such linkages.
Role of Discipline Responses. According to one model, parents’ cognitions about child misbehavior predict individual differences in discipline responses, which in turn impact child behavior (see Dix, 1993, for a review). In the present study, we were interested in mothers’ proposed level of sternness and disapproval in response to relational aggression—two indicators of power assertion, according to Dix and Zambarano (2001). Parental use of power assertion is best viewed along a continuum, with harsh and coercive discipline falling on one extreme, and distracting and ignoring child misbehavior (i.e., low sternness and displays of disapproval) falling on the opposite extreme. Drawing on previous work, we hypothesized that mothers who perceive relational aggression to be less normative and less acceptable, and those who attribute greater responsibility to children for using relational aggression, will report higher levels of sternness and disapproval in response to displays of relational aggression. Furthermore, drawing on Werner, Senich, and Przepyszny's (2006) findings, we expected that the qualities of mothers’ discipline responses would be associated with children's peer competence, and that discipline responses would mediate the association of maternal cognitions and children's peer competence.
Role of Children's Normative Beliefs About Aggression. A second plausible pathway involves children's normative beliefs about aggression. Dix (1993) stated that ‘a principal hypothesis about attribution and socialization is that adults’ attributions about children influence children's internalization of values and views of themselves’ (p. 640). Prior investigations, however, have generally failed to find direct correlations between parents’ and children's intent attributions (e.g., MacKinnon-Lewis et al., 1992). One exception is a recent study by MacBrayer et al. (2003) who found that mothers’ hostile attributions were significantly correlated with daughters’, but not with sons’, attributions. In addition, McDowell et al. (2002) assessed participants’ self-generated strategies for dealing with social conflict and found that fathers’ strategies were correlated with children's.
In this investigation, we were interested in whether mothers’ attributions about children and their normative beliefs about relational aggression would predict children's normative beliefs, and whether children's beliefs would mediate the association of maternal cognitions and children's peer competence. We reasoned that mothers’ responses to situations involving relational aggression provide a unique opportunity for them to transmit values and norms specific to relational aggression. For example, mothers who themselves view relational aggression as inappropriate, hold children responsible for behaving in a relationally aggressive manner, and respond with moderate levels of sternness and disapproval, are likely to transmit the belief to children that relational aggression is unacceptable. Findings from Werner and Nixon (2005) demonstrated that children who hold beliefs approving of relational aggression, in turn, report higher levels of relationally aggressive behavior toward peers. Thus, we hypothesized that children's normative beliefs about relational aggression would either partially or fully mediate the association of mothers’ cognitions with children's peer competence in school.
Sex Differences
Our final goal was to examine sex differences in the pathways linking maternal cognitions and children's peer competence. Available information on the effects of sex (parent and child) on family processes is mixed, particularly in the area of parental discipline and child behavior. Some evidence suggests that harsh parental discipline has a greater impact on same-sex children (Chang, Schwartz, Dodge, & McBride-Chang, 2003; Deater-Deckard & Dodge, 1997), which is consistent with a social learning perspective on socialization (e.g., Bandura & Walters, 1959). Less research has focused on the effect of parent and child sex on social-cognitive processes within families. Two recent exceptions include studies by MacBrayer et al. (2003), who found that mothers’ hostile attributions were significantly correlated with daughters’ attributions, but not with sons’, and by McDowell et al. (2002), who show that fathers’, but not mothers’, social cognitions were associated with children's cognitions. On the basis of limited prior research, we hypothesized that mothers’ cognitions would be more strongly associated with daughters’ peer competence.
Summary
The goals of the current study were to: (1) determine whether mothers’ cognitions about relational aggression differ from those about physical aggression; (2) test two models linking maternal cognitions with children's peer competence (via discipline responses and/or child normative beliefs about aggression); and (3) examine sex differences in the patterns of associations studied. These research aims were explored using data collected from a sample of mothers and their third- through sixth-grade children.
Method
Participants
Participants in the current study included 103 mothers and their third- to sixth-grade children (49 females). The data for this study were taken from the second wave of data collection in a longitudinal study of family relationships and child adjustment. In the first wave of the study (Time 1, Cohort 1), participants were recruited from three local elementary schools in a small, Midwestern city. As a part of a larger, school-based investigation, 434 children in grades 3, 4, and 5 completed surveys about school climate, and homeroom teachers provided behavior ratings on target children. Letters were then sent home with all children, and parents were asked to provide contact information if they were interested in taking part in a home-based study of family relationships and child development. A total of 95 mothers and their third- through fifth-grade children completed home interviews (consent rate was 22 percent).
One year later, Cohort 1 participants were contacted and invited to take part in second round of data collection (Time 2, Cohort 1). To increase our sample size, a second cohort was also recruited using identical procedures as were used for Cohort 1. Specifically, children in the third grade, in addition to children in grades 4 and 5 who were new to the school, were sent home with letters describing the study, and parents were asked to provide contact information if interested in taking part in the research. These participants made up Cohort 2. A total of 108 mothers and children (60 Cohort 1 families and 48 Cohort 2 families) completed a home interview that year of the study. A comparison of Cohort 1 and Cohort 2 participants on demographic (i.e., maternal ethnicity, education, and marital status) and main study variables (e.g., normative beliefs about relational aggression; discipline responses) revealed no group differences after child grade was controlled (i.e., the majority of Cohort 2 families had third graders). The two cohorts were thus combined into a single sample for the purposes of this article.
The final sample for the current study consisted of 103 mothers, with complete interview data at the second year of the study, and their children (52 percent female). Children were approximately evenly distributed across grades (35 third graders, 20 fourth graders, 25 fifth graders, 23 sixth graders). The majority of mothers (79 percent) and children (73 percent) were European American; 10 percent of mothers and 9 percent of children were Asian or Asian-American; and the remaining participants made up other ethnic groups or did not provide information about ethnicity. Mothers were highly educated with 69 percent having received a bachelor's degree or higher. Seventy-two percent of mothers were married, and 22 percent of families received public assistance. These figures are consistent with the ethnic and socioeconomic composition of the region in which the data were collected (e.g., 10.9 percent of children in the school district from which participants were recruited were Asian, and 26 percent of children were eligible for the free or reduced-price meal program in May 2006). All adult participants provided written informed consent, and child participants provided written and verbal assent prior to data collection.
Measures
Parents’ Responses to Children's Behavior. The parents’ responses to children's behavior (PRCB) measure was developed for the purpose of this study to assess parental social cognitions and discipline responses to hypothetical displays of child aggression. This measure was adapted from Dix and Zambarano's discipline and appraisal inventory (Dix & Zambarano, 2001) which was designed to measure parental cognitions about child misbehavior in a general sense (specific revisions are noted later, under the descriptions of individual constructs). The PRCB consisted of eight vignettes depicting situations involving a target child behaving aggressively toward a peer. The stories depict situations involving relational aggression (four stories), physical aggression (two stories), and verbal aggression (two stories). In the current study, we focused on mothers’ cognitions about relational and physical aggression only (refer to the Appendix for story descriptions). The following story describes the target child using relational aggression:
Imagine that you happen to see an email message your child sent to one of his/her friends. In the email, your child is talking about another child from the class at school. Your child says that he/she doesn't like that classmate anymore, and that everyone should stop being friends with him/her.
Mothers were asked to read each vignette and to imagine that the target child in the story was their child, and that the events in the story recently happened. Following each story, mothers responded to 10 questions that assessed cognitions (attributions and normative beliefs) and discipline responses. Each item was rated on a seven-point Likert scale. Scores were computed for each of the constructs, to be described later, by averaging items across the stories depicting each form of aggression.
Responsibility Attributions. Three subscales on the discipline and appraisal inventory (Dix & Zambarano, 2001) assessed mothers’ responsibility attributions. Attributions of child knowledge were assessed with the following item: ‘Does your child know that he/she is acting badly or improperly?’ Mothers responded to this question on a seven-point Likert scale ranging from 1 (definitely does not know) to 7 (definitely knows). Attributions of capacity were assessed with the item: ‘Would it be reasonable to expect your child to have known that this was wrong?’ which was rated on a scale ranging from 1 (definitely no; can't expect child to have known better) to 7 (Definitely yes; child should have known better). Attributions of blame were assessed with the item: ‘How much blame does your child deserve for acting like this?’ (1 = no blame to 7 = complete blame). A factor analysis revealed that the knowledge, capacity, and blame subscales loaded on a single factor, which is consistent with Dix's conceptualization of these attributions as making up a single responsibility attribution scale. We therefore averaged participants’ scores across the three subscales and across stories depicting relational and physical aggression. Higher scores reflected attributions of greater responsibility to children for engaging in the aggressive behaviors. Cronbach's alphas were .81 for attributions about relational aggression and .70 for attributions about child physical aggression.
Descriptive Norms. Mothers’ beliefs about the prevalence of child aggression were measured with the question on the PBCA instrument: ‘In general, how typical or common is this kind of behavior among children your child's age?’ Mothers responded on a seven-point scale ranging from 1 (not at all common) to 7 (extremely common). Items were summed across stories depicting relational aggression and physical aggression; with higher scores reflecting perceptions of greater prevalence of each form of aggression. Cronbach's alphas were .72 for descriptive norms about relational aggression (after dropping one unreliable item)1 and .73 for descriptive norms about physical aggression.
Specific Normative Beliefs. Mothers’ perceptions of the appropriateness of specific incidences of child aggression were assessed with the following question on the PBCA measure: ‘Do you think your child is acting badly or improperly?’ The rating scale ranged from 1 (definitely no) to 7 (definitely yes), with higher scores reflecting beliefs that aggression is more unacceptable. After summing across stories depicting relational and physical aggression, alphas were .46 and .33, respectively. Because of their low reliability, caution must be made in interpretation of results.
Discipline Responses. Two questions from the discipline and appraisal inventory (Dix & Zambarano, 2001) were retained on the PBCA measure to assess mothers’ discipline responses to child aggression. One item assessed disapproval (‘How much disapproval would you express toward your child for doing this?’) and the second question assessed sternness (‘How much sternness would be present in your response?’). Consistent with Dix and Zambarano's conceptualization, a factor analysis conducted on the two subscales (conducted separately for relational aggression and physical aggression) revealed a single factor. We therefore created a composite score termed power assertion by summing the disapproval and sternness subscales across stories depicting relational aggression (alpha = .83) and physical aggression (alpha = .79).
General Normative Beliefs about Aggression
Mothers’ General Beliefs. Mothers’ general beliefs about the appropriateness of child aggression were assessed using an instrument adapted from the general beliefs scale of Huesmann and Guerra's (1997) normative beliefs about aggression measure. The revised instrument consisted of 14 statements about child aggression and 11 filler statements (‘It is usually OK when a child plays with someone else if his/her best friend is busy’). Parents responded to each statement on a five-point Likert scale ranging from 1 (it's really wrong) to 5 (it's perfectly OK). In the current study, we utilized two subscales: general beliefs about relational aggression (three items; alpha = .59) and general beliefs about physical aggression (2 items; alpha = .89). A sample item from the relational aggression scale is, ‘In general, it is OK when a child tells their friends not to be friends with someone they don't like’, and one item from the physical aggression scale is, ‘In general, it is OK for kids to hit other people’.
Children's General Beliefs. A revised version of Huesmann and Guerra's (1997) normative beliefs about aggression measure was used to assess children's beliefs about the acceptability of aggression. The revised instrument consisted of nine statements about relational aggression and physical aggression that were identical to those rated by mothers (general beliefs about aggression) except that they were worded for children (Werner & Nixon, 2005). Children responded to each item on a five-point Likert scale ranging from 5 (yes) to 1 (no), with higher scores indicating greater acceptability of aggression. The normative beliefs about relational aggression scale contained four items (e.g., ‘In general, it is OK to tell your friends not to be friends with someone you don't like’; alpha = .67). The normative beliefs about physical aggression scale consisted of three items (e.g., ‘In general, it is OK to hit other people’; alpha = .79).
Child Peer Competence
Homeroom teachers completed assessments of target children's peer competence using 31 items drawn from previous measures (Crick, 1996; Werthamer-Larsson, Kellam & Wheeler, 1991). Three subscales were used in the present study: relational aggression (five items), peer acceptance (six items), and prosocial behavior (five items).
Relational Aggression. Items making up the relational aggression scale were taken from the children's social behavior scale—teacher form (CSBS-T; Crick, 1996). The CSBS-T was adapted from a peer-nomination instrument (Crick & Grotpeter, 1995) that has been widely used in previous research. Teachers rated children on each item using a five-point Likert scale (1 = never true to 5 = almost always true). The relational aggression subscale of the CSBS-T has been shown to be reliable and to correlate significantly with peer reports of relational aggression (.57 for boys and .63 for girls; Crick, 1996). In the current study this scale was highly reliable (alpha = .94).
Peer Acceptance. The peer acceptance scale included two items from the CSBS-T measure (Crick, 1996; ) ‘This child is popular with same-sex peers’ and ‘This child is popular with opposite-sex peers’). Items were rated on the same five-point scale previously described. Crick (1996) reported that the peer acceptance scale was correlated with peer nominations of acceptance (.35 for boys and .25 for girls) and was negatively correlated with peer nominations of peer rejection (−.57 for boys and −.37 for girls). In the current study, the peer acceptance scale demonstrated excellent internal consistency (alpha = .86).
Prosocial Behavior. Teachers rated children's prosocial behavior on five items taken from the teacher observation of classroom behavior—revised (Corrigan, 2003; Werthamer-Larsson et al., 1991). Teachers rated each child on each behavior using a six-point Likert scale (1 = never to 6 = almost always). Sample items included ‘friendly’ and ‘is helpful to others’. This scale was highly reliable (alpha = .88).
Comparison of Participants with and without Teacher Ratings. Teacher ratings were available for 75 children who were a part of this investigation (73 percent of the total sample). Reasons for missing teacher data were twofold: (1) several parents declined to give permission for children to take part in the school-based portion of the larger investigation, and teacher ratings were one component of that study (N = 3); and (2) several teachers failed to complete ratings, despite repeated reminders (all remaining cases).
To determine whether children with teacher ratings differed from children without teacher ratings, we conducted analyses of variance (ANOVAs) and computed correlations to examine the patterns of correlations between key study variables for each group. These analyses revealed that children without teacher ratings had mothers who were less approving of relational aggression, attributed more responsibility to children for engaging in relational aggression, and reported the intention to use higher levels of power assertion, compared to those with teacher ratings. No other group differences were found (including differences in children's normative beliefs), and the correlational analyses revealed no significant differences in the patterns of relations across the two groups. These findings suggest that analyses using the subsample of children with teacher ratings should be interpreted with caution, as they might not be representative of the larger sample.
Procedures
Data for this study were collected in two different contexts. Child participants completed an assessment of normative beliefs about aggression in the context of a school-based survey, and homeroom teachers completed peer competence ratings on each child during the same academic semester. Mothers of each target child were interviewed in their homes by trained graduate and undergraduate research assistants during the same academic year. Information about mothers’ social cognitions was collected during these interviews.
Results
Descriptive Analyses
To examine main effects of child sex and grade on mothers’ social cognitions about relational aggression, we computed a series of univariate ANOVAs on the dependent variables of interest. These analyses yielded no significant main effects of sex or grade. However, a significant interaction of sex and grade for specific norms about relational aggression was found, F (3, 95) = 3.5, p < .05. Follow-up tests revealed that the effect of grade was significant for mothers of girls, F (3, 53) = 3.7, p < .05. Examination of means showed that mothers of fifth grade daughters were less approving of specific acts of relational aggression (M = 6.14, SD = .60) compared to mothers of fourth (M = 5.13, SD = .92) and sixth graders (M = 5.31, SD = .74). The effect of child sex was also significant among fifth graders, F (3, 53) = 3.7, p < .05, such that mothers of girls viewed relationally aggressive acts as less appropriate (M = 6.14, SD = .60) compared to mothers of boys (M = 5.18, SD = 1.33).
The interaction of sex and grade on mothers’ responsibility attributions about relational aggression was significant, F (3, 95) = 3.1, p < .05. Follow-up tests revealed that the effect of grade was again significant for mothers of girls, F (3, 53) = 3.8, p < .05. Mothers of fifth grade girls attributed greater responsibility to them for engaging in relational aggression (M = 5.61, SD = .60) compared to mothers of third grade girls (M = 4.66, SD = .95). The effect of child sex was also significant among third graders, F (1, 34) = 4.2, p < .05. Mothers attributed greater responsibility to their sons for engaging in relational aggression (M = 5.18, SD = .51) than did mothers of third grade girls (M = 4.66, SD = .95).
Main Analyses
The central goals of this study were to investigate: (1) the effect of aggression form (relational versus physical) on mothers’ social cognitions and discipline responses; (2) the possible mediating roles of maternal discipline responses and child normative beliefs about relational aggression in the association of maternal cognitions and children's peer competence; and (3) the effect of child sex on the pathways linking mothers’ cognitions with child peer competence. We used repeated measures, ANOVAs, and path analysis to answer the mentioned research questions.
Aggression Form Effects on Mothers’ Social Cognitions and Discipline Responses
Aggression form effects were initially analyzed with repeated measures analyses of variance conducted separately on mothers’ normative beliefs, responsibility attributions, and power assertion. In these analyses, child grade (third to sixth) and sex were the between subjects independent variables and aggression form (two levels: RA, PA) was the within subjects independent variable. Because the main effects of child grade and sex, and interactions of these variables, were non-significant in all analyses, we repeated the analyses using paired-samples t-tests. Means, standard deviations, and Cronbach's alphas for all dependent variables can be found in Table 1.
| Cronbach's alpha (PA/RA) | Type of aggression | ||||||
|---|---|---|---|---|---|---|---|
| Physical aggression | Relational aggression | ||||||
| Mean | SD | Range | Mean | SD | Range | ||
| Maternal cognitions and discipline responses | |||||||
| RES ATT | .70/.81 | 5.66 | .70 | 3.38–6.75 | 5.03 | .82 | 2.33–6.58 |
| GEN NORM | .89/.59 | 1.01 | .15 | 1.0–2.50 | 1.38 | .53 | 1.0–3.33 |
| DES NORM | .73/.72 | 3.40 | 1.32 | 1.0–6.50 | 4.57 | 1.26 | 1.67–7.0 |
| SPEC NORM | .33/.46 | 6.51 | .71 | 4.0–7.0 | 5.50 | .88 | 3.0–7.0 |
| POWER | .79/.83 | 18.25 | 2.22 | 9.0–21.0 | 14.43 | 2.69 | 7.5–21.0 |
| Child social cognitions | |||||||
| CHILD NORM | .73/.67 | 1.29 | .53 | 1.0–4.0 | 2.13 | .81 | 1.0–4.25 |
| Child peer competence (teacher ratings) | |||||||
| PROSOC | .88 | 4.51 | 1.0 | 2.0–6.0 | |||
| RA | .94 | 1.88 | .93 | 1.0–4.6 | |||
| ACC | .86 | 3.74 | .98 | 1.5–5.0 | |||
- Notes: RES ATT = responsibility attributions; GEN NORM = general normative beliefs; DES NORM = descriptive normative beliefs; SPEC NORM = specific normative beliefs; POWER = power assertion; CHILD NORM = child normative beliefs; PROSOC = prosocial behavior; RA = relational aggression; ACC = peer acceptance. Ns = 72–106.
The effect of aggression form was significant for all comparisons. Mothers assigned less responsibility to their child engaging in relational aggression as compared to physical aggression, t(102) = −13.36, p < .001. Mothers’ specific normative beliefs about aggression also differed by aggression form, t(102) = −12.24, p < .001, such that they viewed relational aggression as less unacceptable than they did physical aggression. The comparison of mothers’ general normative beliefs about aggression was significant, t(102) = 6.90, p < .001. Mothers evaluated relationally aggressive behaviors more positively than physically aggressive behaviors. In addition, mothers perceived relational aggression as significantly more frequent (descriptive norms) than physical aggression, t(102) = −9.52, p < .001. Finally, the analysis of discipline responses revealed that mothers proposed using significantly lower levels of power assertion (i.e., sternness and disapproval) in response to children's relationally aggressive behavior as compared to their responses to physical aggression, t(102) = −16.10, p < .001.
Pathways Linking Maternal Cognitions with Child Peer Competence
Overview. Our central analyses were designed to explore possible mechanisms through which maternal cognitions influence children's peer competence at school. Specifically, we sought to test two mediational models: one involving maternal discipline responses and the second involving children's normative beliefs about relational aggression. We first computed simple correlation coefficients between maternal social cognitive variables and teacher ratings of children's relational aggression, prosocial behavior, and peer acceptance (see Table 2). Only two significant correlations were found. Although mediational analyses were therefore not feasible, we reasoned that maternal social cognitions might still be associated with child peer competence through indirect processes involving mothers’ discipline responses and/or child normative beliefs.
| Maternal Cognitions and Discipline responses | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 |
|---|---|---|---|---|---|---|---|---|---|
| 1. RES ATT | — | −.20 | −.28 | .84** | .66** | .24 | −.18 | .08 | −.16 |
| 2. GEN NORM | .32 | — | −.29 | −.15 | −.03 | −.11 | .31 | .02 | .31 |
| 3. DES NORM | .07 | −.16 | — | −.40* | −.40* | −.06 | −.10 | −.05 | −.12 |
| 4. SPEC NORM | .49** | .46** | .14 | — | .56** | .18 | −.17 | .14 | −.11 |
| 5. POWER | .05 | −.06 | .10 | .38* | — | .13 | .00 | −.08 | .01 |
| Child social cognitions | |||||||||
| 6. CHILD NORM | .07 | .28 | .19 | −.03 | .02 | — | −.25 | .03 | −.18 |
| Child peer competence (teacher rated) | |||||||||
| 7. PROSOC | .01 | .13 | .06 | .29 | .33 | −.15 | — | −.72** | .77** |
| 8. RA | −.12 | −.30 | −.11 | −.35* | −.22 | .11 | −.73** | — | −.44** |
| 9. ACC | −.12 | −.14 | .10 | .08 | .36* | −.33 | .62** | −.42* | −.12 |
- Notes: Correlations for boys (N = 37) are above the diagonal; correlations for girls (N = 36) are below. RES ATT = responsibility attributions; GEN NORM = general normative beliefs; DES NORM = descriptive normative beliefs; SPEC NORM = specific normative beliefs; POWER = power assertion; CHILD NORM = child normative beliefs; PROSOC = prosocial behavior; RA = relational aggression; ACC = peer acceptance. * p < .05,** p < .01.
We used path analysis in AMOS 5.0 to examine the associations between maternal social cognitions, discipline responses (i.e., power assertion), child normative beliefs, and peer competence. Multigroup analyses were used to determine whether child sex moderated the relations explored in this study. Because of small cell sizes, we could not compute multigroup analyses by child grade. As is standard procedure (Jöreskog & Sörbom, 1993), we first estimated our models for sons and daughters simultaneously (unconstrained models). Next, we tested each model again with the constraint that all parameters were identical across groups (fully constrained models). If child sex moderated the relations hypothesized in the model being tested, the chi-square value of the fully constrained model would be significantly larger than the model in which all parameters are allowed to vary. When this condition was met, pair-wise parameter comparisons in the unconstrained model were examined to determine which parameters differed significantly by child sex. We report chi-square and chi square change values, comparative fit indexes (CFIs) and goodness of fit indexes (GFIs) for all models.
Pathway Involving Maternal Discipline Responses. In the first model, we hypothesized that mothers’ cognitions about relational aggression would predict their power-assertive discipline responses to relational aggression, which in turn would predict children's level of peer acceptance, relational aggression, and prosocial behavior at school. In the original model, the four social-cognitive variables (responsibility attributions, descriptive norms, specific norms, and general norms) were hypothesized to predict mothers’ power assertion scores in response to children's use of relational aggression. Power assertion, in turn, was expected to predict the three indices of child peer competence. The error terms for the social-cognitive variables were assumed to be uncorrelated with one another. This model did not fit the data well, χ2 (36) = 98.7, p < .001, GFI = .77, CFI = .66. Examination of modification indices revealed that allowing several error terms to covary would improve model fit. After making these modifications, the final model fit the data well, χ2 (28) = 31.0, p = .32, GFI = .90, CFI = .98. This model is diagrammed in Figure 1.
Mediational Model: Relations of Mothers’ Normative Beliefs and Attributions with Child Social Competence Mediated through Peer Assertion.Notes: Path coefficients are presented in standardized units. Coefficents for girls are on the right. Goodness-of-fit index = .90, χ2 (28, N = 72) = 31.0, p = .32. * p < .05, ** p < .01, *** p < .001. Coefficents in bold differ significantly at p < .05.
The multigroup analysis revealed that the fully constrained model did not fit the data significantly better than the unconstrained model, Δχ2 (22) = 45.0, p < .01, thus the latter model was accepted. Examination of the pair-wise parameter estimates in the unconstrained model revealed several sex differences (refer to Figure 1) involving both covariances and path coefficients. Of particular interest in the present study were the path coefficients. Maternal responsibility attributions about sons were positively associated with their proposed level of power assertion (β = .68, p < .01), whereas mothers’ specific normative beliefs about daughters’ relational aggression predicted their power-assertive discipline responses (β = .55, p < .001). In both cases, these sex differences in path coefficients were significant.
Examination of the pathways linking maternal power assertion with children's peer competence revealed that mothers who indicated they would respond with higher levels of sternness and disapproval to daughters’ relationally aggressive behavior had daughters who were described by teachers as significantly better accepted by peers (β = .38, p < .05) and as more prosocial (β = .35, p < .05). Although the pathway between power assertion and child relational aggression did not reach significance, it was in the proposed direction, with higher levels of power assertion associated with lower levels of girls’ relational aggression toward peers (β = −.23, p < .20). In contrast to the model for girls, the model for boys revealed no significant associations between mothers’ power assertion and teacher ratings of peer competence. These sex differences, however, were not significant according to the pair-wise comparisons.
Pathway Involving Child Normative Beliefs. The second indirect pathway investigated involved children's normative beliefs about relational aggression. We hypothesized that maternal social cognitions would impact children's peer competence via their association with children's beliefs about the acceptability of relational aggression. This model is diagrammed in Figure 2. The unconstrained model fit the data very well, χ2 (28) = 28.6, p > .10, GFI = .91, CFI = .99. The multigroup analysis revealed that the fully constrained model did not fit the data significantly better than the unconstrained model, Δχ2 (22) = 41.0, p < .01, thus the unconstrained model was accepted as the final model. The results are presented separately for daughters and sons in the next paragraphs.
Mediational Model: Relations of Mothers’ Normative beliefs and Attributions with Child Peer Competence Mediated through Child Normative Beliefs about Relational Aggression.Notes: Path coefficients are presented in standardized units. Coefficents for girls are on the right. Goodness-of-fit index = .98, χ2 (8, N = 72) = 4.85, p = .77. * p < .10, ** p < .01, *** p < .001. Coefficents in bold differ significantly at p < .05.
Mothers’ general normative beliefs about relational aggression emerged as a significant predictor of daughters’ normative beliefs (β = .42, p < .01), such that mothers who viewed relational aggression as more acceptable had daughters who held similarly positive beliefs. In addition, mothers’ descriptive norms about relational aggression were a marginally significant predictor of daughters’ normative beliefs (β = .30, p < .10), suggesting that mothers who viewed relational aggression as more typical among school-age children had daughters who viewed relational aggression as more acceptable. Interestingly, responsibility attributions and specific normative beliefs (the variables that predicted mothers’ power assertion in the previous model) were not significantly associated with daughters’ normative beliefs about relational aggression. Finally, daughters’ normative beliefs about relational aggression predicted teacher ratings of peer acceptance (β = −.34, p < .05), such that girls who viewed relational aggression as more acceptable were less accepted by peers. The pathways linking girls’ normative beliefs with relational aggression and prosocial behavior were not significant, although they were in the predicted direction.
The model for boys revealed no significant associations between maternal social cognitions and sons’ normative beliefs about relational aggression, nor between sons’ normative beliefs and teacher ratings of peer competence. The only path coefficient that differed significantly for boys and girls, however, was the pathway between mothers’ general normative beliefs about relational aggression and children's normative beliefs (refer to Figure 2).
Discussion
Our focus on maternal cognitions about relational aggression was motivated by diverse literatures including research on the social-cognitive antecedents of parenting practices, associations between parents’ social cognitions and child peer competence, and social information processing contributions to children's social adjustment (e.g., Dix, 1993; Huesmann & Guerra, 1997; MacBrayer et al., 2003). Our results demonstrate that mothers hold a distinct set of beliefs about relational aggression, and that these beliefs are associated with children's peer competence via their impact on mothers’ discipline responses and on children's normative beliefs about relational aggression. Interestingly, our data suggest that maternal social cognitions are associated with daughters’ peer competence, but not with sons’. To the extent that maternal cognitions are malleable, these results have important implications for parent education programs designed to prevent or reduce relational aggression, particularly among school-age girls.
When we compared mothers’ cognitions about different forms of aggression, we found that relationally aggressive behaviors were viewed as more normative and acceptable than were physically aggressive behaviors, and mothers attributed less responsibility to children for engaging in relational aggression. These findings contribute to a growing literature demonstrating that adults think differently about relational and physical aggression (Bauman & Del Rio, 2006; Hazler et al., 2001; Holt & Keyes, 2004; Werner et al., 2006). They are also consistent with evidence that adults’ social cognitions about children are sensitive to contextual features (Colwell, Mize, Pettit, & Laird, 2002; Dix et al., 1989). Several features of relational aggression might help explain these results. Firstly, children do engage in higher levels of relational aggression than physical aggression; thus, mothers’ descriptive norms appear to reflect accurate perceptions of the relative frequency of these behaviors. Behaviors that are more normative may be viewed as less serious and problematic (Bauman & Del Rio, 2006), a supposition that is consistent with findings reported here. In light of evidence that relational aggression peaks in early adolescence (see Crick, Werner, & Casas, O'Brien, K. M., Nelson, D. A., Grotpeter, J. K., et al., 1999), one direction for future research will be to test the hypothesis that adults become increasingly tolerant of this form of aggression as children age.
Explanations of mothers’ differing beliefs about relational and physical aggression should also consider the broader social context in which relational aggression occurs. Relationally aggressive acts often take place within friendship groups (i.e., between ‘friends’), and children are likely to play multiple roles over the course of time, such as aggressor, victim, and bystander (Craig, Pepler, & Atlas, 2000; Sutton & Smith, 1999). Moreover, the harm inflicted on targets of relational aggression is often less visible to outside observers than is the harm caused by physical aggression (Bauman & Del Rio, 2006). Taken together, these unique features of relationally aggressive conflicts may result in adults being less inclined to attribute responsibility to perpetrators, especially if they perceive relational aggression to be highly normative and to be a relatively acceptable response to conflict, as did the mothers in this study. One strategy for parent education is to use tools such as Olweus’ (1993) ‘bully circle’ to educate parents about the different roles children play in aggressive peer conflicts. This model emphasizes that all children are responsible for bullying, even children who are not ‘aggressors’, or even when a child's behavior was not intentionally hostile. This form of education should lead to changes in parents’ cognitions about relational aggression, which in turn might impact their responses to children, as discussed further.
The significance of mothers’ cognitions about relational aggression was most clearly demonstrated by our analyses investigating two pathways through which maternal attributions and normative beliefs about relational aggression were associated with children's peer competence in school (our second goal). On the one hand, we found support for the hypothesis that maternal cognitions regulate discipline responses to children. Specifically, mothers who viewed relational aggression as less acceptable reported the intention to use higher levels of sternness and disapproval (power assertion) in response to their daughters’ relationally aggressive behavior. Maternal power assertion, in turn, was associated with greater peer acceptance and prosocial behavior among daughters.
Finding that higher levels of maternal power assertion predicted greater peer competence appears to contradict previous research demonstrating the adverse effects of power assertive discipline on children (e.g., Hart et al., 1998; Patterson, Reid, & Dishion, 1992; Pettit, Bates, & Dodge, 1993). We believe that the key to understanding these results lies in the extent to which mothers’ responses facilitate children's internalization of messages about relational aggression. When a parent communicates clearly (i.e., with sternness) to a child that his/her behavior was unacceptable (i.e., the parent is disappointed), the child is more likely to internalize values that support the use of prosocial rather than aggressive behavior with peers, and positive means of conflict resolution—skills that lead to peer acceptance. If a parent ignores the situation, responds with neutral affect or with low levels of disappointment (low power assertion), the child might walk away with a very different interpretation of the parents’ message (e.g., that his/her behavior was not wrong). This line of reasoning is supported by Krevans and Gibbs (1996) who found that parental expressions of disappointment following child misbehavior predicted higher levels of child prosocial behavior. Our results suggest that parents should employ a level of power assertion that ensures children's attention to their socialization message, while avoiding arousal of anger and hostility in children (Hoffman, 1983). Mothers who do not view relational aggression as hurtful or unacceptable are unlikely to employ that ‘optimal’ level of power assertion, which might contribute to the maintenance of relationally aggressive behavior.
Further investigation into mechanisms through which mothers’ cognitions might influence children's peer competence focused on children's normative beliefs about relational aggression. Mothers who held beliefs supportive of relational aggression had daughters with similar beliefs, beliefs which in turn predicted lower levels of peer acceptance at school. These findings are notable given that few prior studies have found direct correlations between parents’ and children's social cognitions. These findings are also consistent with social-information processing models of children's social adjustment (Crick & Dodge, 1994) and with prior research showing that children who believe aggression is an acceptable response are more likely to utilize maladaptive behaviors in the peer group (Huesmann & Guerra, 1997; Werner & Nixon, 2005).
Additional research is needed to investigate the processes through which direct modeling of social cognitions occurs within families. One promising direction is to study parental advice giving, or social coaching, around relationally aggressive peer conflicts. Although most prior research on social coaching has focused on parents of young children (e.g., Mize & Pettit, 1997), recent research suggests that parents’ style of giving advice about peer issues is associated with school-age children's peer competence (McDowell, Parke, & Wang, 2003). Everyday conversations about peers provide an opportunity for parents to model benign attributions to children about others’ intentions (Nelson & Crick, 1999), to engage in problem-solving about strategies for handling peer difficulties, and to communicate specific values to children. Together with recent findings (Werner et al., 2006), the current results suggest that parents might influence the development relational aggression in direct ways not investigated in previous research.
Overall, our results suggest that mothers play a particularly important role in daughters’ peer competence during middle childhood. Huston (1983) proposed a gender role differentiation theory in which parents feel a greater responsibility for the socialization of same-sex children, and therefore use higher levels of control over same-sex children (Power & Shanks, 1989). In the present study, however, neither mothers’ social cognitions nor their discipline responses varied by child sex (although there were some grade by sex interactions). Thus, we interpret our results in terms of social learning theory, which proposes that children will identify more strongly with a same-sex role model. Daughters in the present study might have paid closer attention to mothers’ socialization messages, perceived the messages more accurately, or have been more accepting of the messages (Grusec & Goodnow, 1994), relative to sons. Support for this hypothesis is also found by Mize and Pettit's (1997) findings that mothers’ social coaching was more strongly associated with daughters’ peer competence than with sons’. Research on fathers’ cognitions is needed to further validate hypotheses about parental influences on same vs. opposite-sex children.
Limitations and Future Directions
We were surprised at our failure to find significant associations between maternal and child predictor variables and children's use of relational aggression. Teacher ratings of the three indices of peer competence were moderately to highly intercorrelated; thus, it is possible that the effects of maternal discipline responses and child normative beliefs on relational aggression were mediated through prosocial behavior or peer acceptance. Alternatively, teachers might underestimate children's use of relational aggression, thus reducing the ability to detect significant effects on this less visible form of social behavior. Future studies will benefit from using multiple informants to assess relational aggression.
Participants in this study were highly educated; thus, they may have engaged in socialization practices that are generally more effective compared to less educated parents. Importantly, though, we found that mothers utilized the full range of response options on most scales assessing cognitions and socialization responses. Moreover, mothers’ education level was not significantly associated with any of the social cognitive and discipline variables. Nonetheless, additional research is needed to replicate these findings with a larger, more diverse sample of parents. Another limitation pertains to the group of children with missing teacher ratings. Mothers of these children differed in several ways from mothers of children with complete data. This introduces additional questions about the generalizabilty of our findings and points to the need for replication before firm conclusions can be drawn.
The cross-sectional nature of this study prohibits us from making inferences about the direction of effects. Although children's history of social behavior might influence parents’ attributions, Dix (1993) argued that, ‘available evidence suggests that parent variables may predict dispositional attributions about children more strongly and more consistently than measures of children's actual dispositions and behaviors’ (p. 634). Support for parent-to-child effects is bolstered in our study by findings that teacher ratings of children's peer competence gathered one year earlier on a subsample of participants were uncorrelated with maternal cognitive measures at Time 2. In addition, the use of different informants to measure maternal and child variables reduced problems with shared method variance.
Another limitation concerns the use of a newly developed measure of mothers’ cognitions about relational aggression. Although the subscales of the PRCB taken from Dix and Zambarano's (2001) measure achieved acceptable reliability, the three measures of mothers’ normative beliefs about relational aggression were less reliable. Examination of mothers’ responses to scenarios depicting relational aggression revealed that they made more fine-grained distinctions than expected, which reduced the internal consistency of our measures. For example, mothers perceived rumor spreading as less appropriate than giving the ‘silent treatment’. Mothers might also distinguish between reactive and proactive acts of relational aggression (Werner et al., 2006). Future studies should examine parents’ sensitivity to different features of relationally aggressive episodes and how those features influence cognitions and responses to children.
Finally, observational studies are needed to determine whether parents’ cognitions about relational aggression predict actual behavior with children. Laboratory-based studies of parents’ social coaching around relationally aggressive episodes represent an important direction for future research. Observing parents engaged in ordinary conversation about actual peer conflicts children have experienced, or using video-based stimuli as discussion prompts (Mize & Pettit, 1997), might offer a ‘window’ through which to view parents’ social cognitions and the process through which cognitions influence children's developing peer competence.
Implications for Parent Education
This investigation extends our understanding of parental influences on relational aggression in middle childhood by demonstrating that mothers hold a different set of beliefs about relational as compared to physical aggression, and that specific patterns of cognitions about relational aggression are associated with daughters’ peer competence via their impact on maternal discipline and on daughters’ normative beliefs about relational aggression. This information has important implications for family-focused prevention programs for relational aggression. Parents of school-age children need to be educated about the negative impact of relational aggression on aggressors and victims in order to modify their beliefs about relational aggression. However, such programs should go beyond this and should teach parents how to take advantage of everyday opportunities to discuss peer relationships and to transmit important messages to children about the hurtful nature of relational aggression and the alternatives to aggression. Although the few existing programs focused on relational aggression have yet to be evaluated, programs that include both family skills training and a school-based intervention (e.g., Espelage & Swearer, 2004) hold the greatest promise.
Appendix
Parental Responses to Child Behavior Measure (PCRB)
| Aggression type | Story description |
|---|---|
| Relational | Imagine that you overhear your child on the phone one afternoon with a friend. Your child is telling an embarrassing story about another child he/she sometimes hangs around with. You know about the situation and also know that your child promised not to tell anyone about it. When you confront your child about the promise, he/she says that the child has been spreading rumors about them in school. |
| Relational | Imagine that you happen to see an email message your child sent to one of his/her friends. In the email, your child is talking about another child from the class at school. Your child says that he/she doesn't like that classmate anymore, and that everyone should stop being friends with him/her. |
| Relational | Imagine that your child's birthday is coming up and you are planning a party together. Your child is writing a guest list but leaves off of the list the name of a child who just moved in next door and is in your child's class at school. You’ve heard from the mother of this child that he/she is making a lot of friends at school. When you ask your child why he/she doesn't want to invite them, he/she says that they just don't want that kid to be in his/her group of friends. |
| Relational | Imagine that your child comes home from school one day very upset because his/her best friend made fun of the new shirt they wore to school. Your child tells you that, even though the friend tried to apologize, they refused to talk to them all day at school. |
| Physical | Imagine that you just received a call from the parent of a child in your child's class at school. The parent says that your child has been pushing their child up against the lockers at school lately. |
| Physical | Imagine that you are chaperoning a field trip with your child's class. During a lunch break, you overhear a child say something mean to your child. The next thing you know, your child has punched the other child in the arm. |
References
Acknowledgments
Special thanks to the co-principal investigators on the ISOP project, Drs. Laura Griner Hill and Matt Bumpus, and to the numerous graduate and undergraduate students who worked on this study over the years. We would also like to thank the principals, teachers, and students at Lincoln Middle School and Franklin, Sunnyside Elementary and Jefferson Elementary Schools for their support of this project.
This study is based in part on the M.A. thesis project of the second author.
