Rubin, K.H. & Mills, R.S.L. (1990). Maternal beliefs about adaptive and maladaptive social behaviors in normal, aggressive and withdrawn preschoolers. Journal of Abnormal Child Psychology, 18, 419-436.
Maternal Beliefs About Adaptive and Maladaptive Social Behaviors in Normal, Aggressive, and Withdrawn Preschoolers
Kenneth H. Rubin & Rosemary S. L. Mills
University of Waterloo
Running Head: Parental Beliefs
The research described herein was supported by a grant from the Social Sciences and Humanities Research Council of Canada. Preparation of this manuscript was aided by a Killam Research Fellowship to author Rubin and by a Social Sciences and Humanities Research Council of Canada Postdoctoral Fellowship to author Mills. An earlier version of this manuscript was presented at the Biennial Meeting of the International Society for the Study of Behavioural Development, Jyvaskyla, Finland, July 1989. We would like to thank those parents, children and teachers in the Regional Municipality of Waterloo who helped make this study possible. Correspondence should be addressed to either author in the Department of Psychology, University of Waterloo, Waterloo, Ontario, N2L 3G1.
The purpose of this study was to compare mothers of normal, aggressive, and anxious-withdrawn preschoolers with regard to their beliefs about how socially competent behaviors are learned and their beliefs concerning the origins of two types of maladaptive behaviors -- aggression and withdrawal. 121 mothers of 4-year olds were questioned about how they think social skills are acquired. They were also presented with descriptions of hypothetical incidents of peer-directed aggression and social withdrawal and asked what they would do about these behaviors, how they would feel about them, and how they would explain them. Children's social behaviors were observed during free play and rated by their teachers. A norm-based multi-targetting procedure was used that resulted in the identification of 10 highly aggressive children, 6 highly withdrawn children, and 60 children who were average in social adaptation. Results indicated that mothers of withdrawn children were more likely than mothers of average children to believe that social skills should best be taught in a directive manner and that maladaptive behaviors should be responded to in a high powered, coercive fashion. These mothers were also more likely than mothers of average children to indicate that they would feel guilty and embarrassed by displays of maladaptive behavior, and they attributed these behaviors to dispositional factors. Mothers of aggressive children suggested highly directive strategies to teach social skills, but were more laissez-faire than mothers of average and withdrawn children in their reactions to maladaptive behaviors.
Maternal Beliefs About Adaptive and Maladaptive Social Behaviors in Normal, Aggressive, and Withdrawn Preschoolers
The study of parents' ideas about child development has garnered a large and growing following in recent years (e.g., Goodnow, 1988; Miller, 1988). This interest in parents' ideas about the timetabling, the causes and the plasticity of development appears to represent one of the newest derivative areas of research stemming from the cognitive and social-cognitive revolutions that swept child development research in the 1970s and 1980s. Of related significance is the assumption made by information processing theorists, that cognitions mediate behaviors; consequently, it is proposed that the ideas that parents have about development are likely to influence their choice of parenting strategies and ultimately their children's development (e.g., Rubin, Mills, & Rose-Krasnor, 1989). In support of this contention, researchers have demonstrated that parents' ideas about how children best learn particular cognitive skills are related, in meaningful ways, to the methods by which parents attempt to facilitate their children's learning of these skills (McGillicuddy-DeLisi, 1982).
From our perspective, the research extant concerning parents' ideas about child development suffers from two major shortcomings. First, most investigators have centered on parental beliefs about the growth of cognitive competencies; little attention has been paid to social and emotional growth and development. Second, researchers have been concerned with the study of typical or normative parental ideas; there has been a limited focus on individual differences in parenting beliefs or on the question of the relations between parental cognitions and child behavior (Goodnow, 1988).
Given these specific shortcomings, the purpose of the present study was to examine the associations between parents' ideas about the development of socially competent and incompetent behaviors and their children's production of competent and incompetent social behaviors. Our initial efforts to understand parents' ideas about socioemotional development began with a study of what parents of preschoolers reported that they might feel, think, and do if their children behaved consistently in socially maladaptive ways, either aggressively or in an anxious-withdrawn fashion (Mills & Rubin, in press). Although the use of moderate power was the typical reactive strategy of choice to deal with aggression, low power was most often suggested as the optimal way to react to children's withdrawal. High and moderate power, however, were more often suggested for aggression than withdrawal whereas low power, information seeking and no response were more often suggested for withdrawal. The most typical emotional response to both aggression and withdrawal was that of concern. Aggression, however, was more likely than withdrawal to elicit feelings of anger, disappointment, and embarrassment, whereas withdrawal was more likely to evoke puzzlement. Finally, parents were more likely to attribute aggression than withdrawal to transient age-related factors (e.g., a passing stage).
It is important to note, however, that normative feelings and beliefs are not necessarily conducive to the development of competent social behavior (Barrett & Boggiano, 1988). As such, it remains to be determined whether the normative parental affects and socialization strategies uncovered in our recent study are associated with social competence in early childhood, and whether deviations from the norm are associated with the development of less competent forms of social behavior such as social withdrawal or aggression. Do socially competent children tend to have parents who choose power assertive strategies that differ in kind from those of aggressive or anxious-withdrawn children when either attempting to proactively teach social skills or when reacting to children's aggression or withdrawal? Are their emotional reactions to or causal attributions about their children's displays of aggression and/or withdrawal different from those of parents of aggressive or anxious-withdrawn children?
In the existing literature, parents of socially competent children have been portrayed as moderately directive and flexible in their use of power vis-à-vis both reactive and proactive socialization (Maccoby & Martin, 1983), whereas parents of aggressive children have been found to be inconsistent, sometimes overreacting with highly power assertive strategies and sometimes underreacting with no response whatsoever (Baumrind, 1967; Hetherington & Martin, 1986; Patterson, 1982). It has also been reported that unresponsive or undercontrolling parenting vis-à-vis the proactive socialization of social skills is significantly associated with childhood aggression (Pettit & Bates, 1989). Thus, one might speculate that the mothers of highly aggressive youngsters would report a mixed pattern of over- and underreaction to their children's maladaptive behaviors and a low degree of proactive involvement in the socialization of social skills relative to the norm. What feelings and attributions they have about their children's maladaptive behaviors are, as yet, unknown and thus have been examined herein.
Given the present lack of data concerning the parent-child relations of anxious-withdrawn children, patterns of reaction or proaction for parents of these children is even more subject to speculation. It has been posited that children who are withdrawn or who have other difficulties of an internalizing nature have parents who are overcontrolling or anxiously overinvolved (Hetherington & Martin, 1986; Olweus, 1980). Interestingly, Pettit and Bates (1989) found that unresponsive proactive socialization of social skills was associated not only with aggression but also with withdrawal. Generally, however, little is known about the parents of anxious-withdrawn children.
In summary, our objective was to compare mothers of normal, aggressive, and anxious-withdrawn children with regard to their proactive beliefs about how socially competent behaviors are learned and their reactive beliefs concerning the origins of two types of maladaptive behaviors -- aggression and withdrawal. Given our reading of the current literature, we predicted that because mothers of anxious-withdrawn children may be overcontrolling, they would suggest the use of more power assertive socialization techniques than mothers of normal, average children. We predicted also that mothers of aggressive children would be more power assertive than mothers of average children, at least in response to their children's displays of hostile behavior (Patterson, 1982). In keeping with suggestions that parents of aggressive children are inconsistent in their use of power assertion, we predicted that mothers of aggressive preschoolers would be less controlling and responsive in their proactive socialization of social skills than mothers of average children (Pettit & Bates, 1989).
The sample consisted of 121 four-year-olds (51 girls, 70 boys) and their mothers. The participants were recruited from 18 preschools or day care centres in the Regional Municipality of Waterloo, a community in Southwestern Ontario with a population of 225,000. The mean age of the mothers was 31.8 years (SD = 5.3). All but 5.9% had completed high school; a large proportion were college- or university-educated (45.4%).
Two methods were used to evaluate children's social adaptation: observations of free play and teacher ratings of social behaviors. Information concerning mothers' beliefs about the most effective means by which to proactively teach social skills was obtained by means of a questionnaire. Data concerning mothers' beliefs about the origins of maladaptive social behavior, their expected affective reactions to their child's display of aggression or withdrawal, and their reported strategies in reaction to these displays were gathered by presenting them with descriptions of hypothetical situations and questioning them about their reactions.
Play Observation Scale. Observations of free play were made using Rubin's (1982, 1986) Play Observation Scale. Each child was observed during free play for thirty, non-consecutive 10-sec. time samples each day over a period of 5 non-consecutive days. Behaviors were coded on a checklist that included the social participation categories of solitary, parallel, and group activities. Nested within these social categories were the cognitive play categories of functional-sensorimotor, constructive and dramatic play and games-with-rules. Other observational categories included unoccupied, onlooker and transitional behavior, reading, rough-and-tumble play, aggression, exploration, and conversations with peers and teachers. Finally, the affect associated with each interaction (group play, conversation, rough-and-tumble play, aggression) was coded as positive (e.g. cooperation, affection, smiling, protection), negative (e.g., noncompliance, teasing, blaming, physical aggression), or neutral .
The frequencies of these observed behaviors yielded scores indexing the amount of isolate behavior (ISOPLAY, defined as the total frequency of solitary play + unoccupied behavior + onlooker behavior) and negative behavior (PNEGINT, defined as the proportion of all social behavior coded as negative. Social behavior included the sum of total group play + conversations with peers + rough-and-tumble play + aggression).
During training, reliability was assessed by pairing each observer with the trainer for several hours of observation time, at the end of which reliability estimates were obtained for a 15- to 20-minute observation period for each observer. The percentage of agreement (ie., number of agreements/agreements plus disagreements) exceeded 85% for each behavioral category at this time. A second 15-20 minute reliability check was also conducted immediately prior to onset of actual data collection for each observer and, again, reliability exceeded 85% for each behavioral category. Reliability spot checks were conducted blind in the latter quarter of data collection, constituting approximately 8 to 10 percent of all observational data. Reliability, as assessed during these spot checks, exceeded 85% for all behavioral categories.1
Preschool Behavior Questionnaire (PBQ). Teacher ratings were obtained using the Preschool Behavior Questionnaire (Behar & Stringfield, 1974). The PBQ is a 30-item scale that best yields two factors for preschoolers and elementary-school children (Moller & Rubin, 1988): an internalizing factor (TRINT) consisting of items describing fearfulness, anxiety and solitude, and an externalizing factor (TREXT) consisting of items describing hostile-aggressive and impulsive-distractible behaviors. Moller and Rubin (1988) have reported significant relations between TRINT and observations of social solitude and between TREXT and observed aggression. As such, the use of the PBQ, in concert with the observational data, was based on previously determined significant relations between these particular indices.
Each child received two factor scores (TRINT and TREXT). Higher factor scores were indicative of greater internalizing and externalizing difficulties. Most of the children had two teachers, both of whom provided ratings. Correlations computed to determine the degree of consistency between the two teachers on the TRINT and TREXT factor scores indicated that they were highly consistent on TREXT, r (94) = .83, p< .001, and moderately consistent on TRINT, r (94) = .49, p< .001. The scores used in the analyses reported below were computed from the averaged ratings.
Maternal proactive beliefs. Mothers' beliefs about the most effective proactive strategies for teaching social skills were assessed by means of a questionnaire drawn from the work of Elias and Ubriaco (1987). This consisted of short descriptions of four social skills (getting acquainted with someone new; resolving peer conflicts; getting accepted into an ongoing play group of unfamiliar peers; persuading others to do what she/he wants), each of which was followed by a list of eight strategies that are potentially influential in the development of these skills: (a) being rewarded for appropriate behavior and punished for inappropriate behavior; (b) observing what other children do; (c) being told exactly how to act; (d) experiencing interactions with others; (e) being taught and encouraged at school; (f) observing what adults do; (g) being told why one should act in a certain way; and (h) experiencing the feelings that arise when being with someone. Mothers were asked to choose the three most likely strategies, among the eight presented to them, and to rank them in order of importance. Scores of 3, 2, and 1 were assigned to ranks of 1, 2, and 3 respectively, so that higher scores would represent greater importance, and a score of 0 was assigned to each strategy not ranked among the top three. Scores were then summed across pairs of similar strategies to create summary scores reflecting the importance of four general proactive strategies: (1) the child's personal experiences in the social milieu (personal experiences = strategies [d] and [h] as described above); (2) imitation of parents and peers (observational learning = strategies [b] and [f]); (3) receipt of explanations offered by adult socializing agents (explanations = strategies [e] and [g]); and (4) parental commands and use of reinforcement contingencies (directive teaching = strategies [a] and [c]). These summary scores could range from 0 to 5.
Maternal reactions: Attributions, emotions, and strategies. Mothers' causal attributions, emotional reactions, and behavioral responses to peer-directed aggression and withdrawal were assessed by reading them, or asking them to read, descriptions of hypothetical incidents of peer-directed aggression and withdrawal. There were four stories, two describing aggressive acts with peers occurring either in an activity group or at home, and two describing social isolation at preschool or at a birthday party. Following each story, mothers were first asked how they would feel seeing their own child act this way several times in a row. The phrase "several times in a row" was intended to make the situation intermediate with respect to the presence of temporal consistency, in order to pull for individual differences in mothers' feelings about and interpretations of aggression and withdrawal. Following the question, "How do you feel when you see your child act this way several times in a row?", mothers were asked to rate each of nine emotions on a 3-point scale ranging from "not at all" (1) to "extremely" (3). The nine emotions were: angry, embarrassed, amused, disappointed, concerned, pleased, surprised, puzzled, and guilty. Mothers were then asked two open-ended questions: "Why do you think your child has been acting this way?" and "What, if anything, would you do about your child's behavior?" Because it was considered essential to assess mothers' immediate affective reactions, they were asked about feelings first, causal beliefs second, and anticipated socialization strategies third.
Coding of Causal Attributions and Socialization Strategies
Mothers' causal beliefs were coded both in terms of locus and type of attribution. Distinctions were drawn between attributions made to internal stable factors, internal unstable factors, and external factors. Internal stable factors were defined as traits or dispositions having the quality of consistency across situations and over time. Internal unstable factors referred to temporary or changeable conditions, including age or age-related factors (e.g., a passing stage or a skill not yet learned), transient states (e.g., a mood, or fatigue), and acquired habits. External factors referred to situationalinfluences.
Mothers' anticipated socialization strategies were coded as either high, moderate, or low in power assertion, or as indirect. Strategies involving strong force or coercion, such as punishment, strong commands, and threats, were labelled high in power assertion (e.g., "I would tell him to apologize") ; strategies such as reasoning, modeling, and gentle directions were designated moderate in degree of power assertion (e.g., "I'd tell her she should try to play with others, they're not always going to be doing what she wants to do"); and nondirective strategies (e.g., asking the child for information, redirecting the child's behavior) were considered low power. Other responses were considered to be quite indirect because they did not involve direct interaction with the child; these included information-seeking (e.g., "I'd talk to the teacher and find out why it happened") and planful strategies (deciding to create opportunities for the child to play with others; deciding to provide more love and attention). The latter appeared to involve long-term goals, such as creating opportunities for the child to gain experience. A final category was labeled no response.
Assessment of reliability. Training in the use of the coding system continued until the level of agreement between each of two coders (both unfamiliar with the hypotheses of the study) and one of the investigators reached 80% for a small subset of the transcripts. Two reliability checks were done, each distributed evenly across sex of child and based on approximately one-quarter of the mothers' responses. One check was done mid-way through the coding phase of the study and the other was done at the end. Reliabilty was calculated as the number of agreements divided by the number of agreements plus the number of disagreements. For causal attributions, reliability was 90.3% and for suggested socialization strategies reliabilty was 90%.
Identification of Target Groups
Norm-based extreme-group targeting procedures were used to identify children who were aggressive-externalizing, withdrawn-internalizing, or socially "average." Two methods of identification were used: observational and teacher assessment. Aggressive-Externalizing (A-E) children were identified as those who were at least three-quarters of a standard deviation above the mean for children of the same sex on both observed negative interaction (PNEGINT) and teacher-rated externalizing difficulties (TREXT). Because negative interaction rarely occurred, particularly in girls, the criterion for PNEGINT was lowered to one-half of one standard deviation above the mean for girls only. Similarly, Withdrawn-Internalizing (W-I) children were identified as those who were at least three-quarters of a standard deviation above the mean for children of the same sex on both observed social isolation (ISOPLAY) and teacher-rated internalizing difficulties (TRINT). Average (AVG) children were those whose scores on all four criteria (i.e., observations and teacher ratings of both isolation-internalizing and aggression-externalizing behaviours) fell within the normal range for same-sex children, i.e., whose scores were below the cut-off points just described. These criteria resulted in the identification of 6 W-I children (5% of the sample, comprised of 3 girls and 3 boys), 10 A-E children (8% of the sample, comprised of 4 girls and 6 boys), and 60 AVG children (50% of the sample, comprised of 25 girls and 35 boys). Thus, a total of 13% of the sample could be construed as 'at risk,' a figure consistent with that broadly accepted as appropriate. Interestingly, four of the six children in the W-I group had target scores that exceeded one standard deviation on each of the two criterion measures; seven of the 10 A-E children had scores exceeding one standard deviation on both criterion measures.
There remained 45 children who could not be classified using this procedure (37% of the sample, comprised of 19 girls and 26 boys). By and large, this group was comprised of children whose scores exceeded the criteria on only one of the two measures.
Maternal Proactive and Reactive Beliefs
To determine whether there were any differences between mothers of W-I, A-E, and AVG children in their beliefs about proactive teaching of social skills and in their reactive feelings, attributions, and strategies for modifying their children's unskilled social behaviors, a series of analyses of variance was computed with Target Group (Withdrawn-Internalizing, Aggressive-Externalizing, Average) as the between-subjects factor. In the analyses reported below, only significant effects involving Target Group will be reported.
Mothers' beliefs about proactive teaching strategies. For each of the four social skills -- getting acquainted with someone new, getting accepted into an ongoing play group of unfamiliar peers, resolving conflicts with peers, and persuading other children to do what one wants -- scores reflecting the importance of each of the four general proactive strategies (directive teaching, observational learning, explanation, and personal experience) were computed by summing across pairs of specific strategies, as described in the Method section. The resulting scores could range from 0 to 5. Means and standard deviations for these variables are given in Table 1.
Insert Table 1 about here
To determine whether there were differences in the importance mothers placed on the four general strategies as a function of target group and type of skill, a series of mixed-model analyses of variance was computed on the summary scores, with Target Group as the between-subjects factor and Type of Skill (getting acquainted with someone new, getting accepted into an ongoing play group of unfamiliar peers, resolving conflicts with peers, persuading other children to do what one wants) as the within-subjects factor. Interdependence among the scores for the different types of proactive strategies necessitated a separate analysis for each strategy.
A significant effect of Target Group was found for only one of the four proactive strategies: there was a significant main effect of Target Group on the importance assigned to directive teaching, F (2, 69) = 3.23, p < .05. Comparisons among the three target groups revealed that, for all four social skills combined, the mothers of W-I children placed more importance on directive teaching (M = 3.00) than did the mothers of A-E children (M = 2.00), t (69) = 1.82, p < .05 one-tailed. Both these groups of mothers considered directive teaching to be significantly more important than did the mothers of AVG children (M = .97), t (69) = 4.54, p < .001 one-tailed, and t (69) = 2.76, p < .01 one-tailed, for the W-I and A-E groups, respectively. No significant effect of Target Group was found for any of the other three proactive strategies.
Mothers' reactive strategies for withdrawal and aggression. To control for variability in the number of reactive socialization strategies mothers suggested for each type of behavior, the extent to which each mother reported a given strategy was indexed as a proportion of the total number of strategies she anticipated using in response to the two incidents describing each type of behavior. The tendency for a mother to choose highly coercive strategies was defined normatively with reference to the degree of power assertion that was predominant for the sample as a whole. On the basis of earlier analyses which revealed that moderate-power strategies were the predominant choice for aggression and low-power strategies were the predominant choice for social withdrawal (Mills & Rubin, in press), "high coercion" was defined as the proportion of high power strategies reported for dealing with aggressive behaviors and as the proportion of high plus moderate power strategies reported for dealing with socially withdrawn behaviors ; "low coercion" was defined as the proportion of low- plus moderate-power strategies reported for dealing with aggression and as the proportion of low-power strategies reported for dealing with social withdrawal . Mean proportions reflecting the extent to which mothers reported high and low coercion for dealing with aggression and social withdrawal, and the extent to which they reported indirect or no-response strategies (those which did not involve an immediate reaction to the child's behaviour, such as seeking information, arranging opportunities for peer experience, or simply choosing not to respond), as well as the standard deviations for these variables, are shown in Table 2.
Insert Table 2 about here
To determine whether there were differences in the extent to which mothers reported high coercion, low coercion, and indirect-no response strategies as a function of target group and the type of behavior to which they were reacting, a series of mixed-model analyses of variance was computed on the proportions, with Target Group as the between-subjects factor and Type of Behavior (aggression, social withdrawal) as the within-subjects factor. Interdependence among the proportions of the different types of strategies necessitated a separate analysis for each type.
Significant or near significant main effects of Target Group were found for the extent to which mothers reported high coercion, low coercion, and indirect-no response strategies, F's (2, 73) = 8.17, p < .001, 2.97, p < .06, and 3.04, p < .05, respectively. Following from our predictions, planned comparisons among the three target groups revealed that, for both social withdrawal and aggression, mothers of W-I children were more likely to report highly coercive strategies (M = .51) than the mothers of either the A-E (M = .21) or the AVG children (M = .22), t's (73) = 2.35 p < .03 one-tailed, and 2.77 p < .01 one-tailed, respectively. Conversely, the mothers of W-I children were less likely to report indirect-no response strategies (M = .24) than were the mothers of either the A-E (M = .65) or the AVG children (M = .51), t (73) = 3.60, p < .001 one-tailed, and t (73) = 2.77, p < .01 one-tailed, respectively. Mothers of A-E children did not differ from mothers of AVG children in the extent to which they reported highly coercive strategies, t (73) = .08, but they were more likely to report indirect-no response strategies than mothers of AVG children, t(73) = 1.84, p < .05 one-tailed. Mothers of AVG children were more likely to report the use of low coercion (M = 1.06) than the mothers of either W-I (M= .75) or A-E children (M= .93), t's (73) = 3.06, p < .001 and 1.67, p < .05 one-tailed, respectively.
Emotional reactions to withdrawal and aggression. Ratings of each of the nine emotional responses were summed across the two stories depicting each type of behavior (aggression and withdrawal). The resulting scores ranged from 2 to 6. The means of these scores for the two stories combined, and the standard deviations for these variables, are shown in Table 3.
Insert Table 3 about here
To determine whether mothers' emotional reactions varied as a function of target group and the type of behavior to which they were reacting, a mixed-model analysis of variance was computed on the scores for eight of the nine emotions (there was no variance among scores for "pleased" about displays of aggression), with Target Group as the between-subjects factor and Type of Behavior (aggression, social withdrawal) and Type of Emotion (concern, anger, etc.) as the within-subjects factors.
This analysis revealed a significant interaction between target group and type of emotion, F (14, 511) = 1.95, p < .02. Subsequent unplanned two-tailed tests of simple effects indicated that mothers of AVG children were more puzzled about displays of withdrawal and aggression (M = 8.32) than the mothers of A-E (M = 7.60) and W-I (M = 7.17) children, t's (511) = 2.19, p < .05, and 2.81, p < .01, respectively. Mothers of both A-E and W-I children were more angry about these behaviors (M's = 7.50 and 7.33, respectively) than mothers of AVG children (M = 6.30), t's (511) = 3.56, p < .001, and 2.52, p < .02, respectively. Further, mothers of W-I children felt more disappointed (M = 9.17), embarrassed (M = 6.50), and guilty (M = 6.50) than mothers of A-E children, M's = 8.10, 5.70, and 4.90, and t's (511) = 2.16, p < .05, 1.62 n.s., and 3.23, p < .002, respectively, and mothers of AVG children, M's = 7.75, 5.47, and 4.83, and t's (511) = 3.46, p < .001, 2.52, p < .02, and 4.07, p < .001, respectively, while there were no significant differences between the two latter groups.
Causal attributions about withdrawal and aggression. To control for variability in the amount of causal reasoning mothers offered when asked why their child acted in the manner described in the story, the extent to which mothers made each of five types of attribution was indexed by computing proportions of the total number of attributions made. Mean proportions reflecting the extent to which mothers made each type of attribution, and the standard deviations for these variables, are given in Table 4.
Insert Table 4 about here
To determine whether there were differences in the causes to which mothers attributed displays of unskilled social behaviors as a function of target group and the type of behavior, a series of mixed-model analyses of variance was computed on the proportions, with Target Group as the between-subjects factor and Type of Behavior (aggression, social withdrawal) as the within-subjects factor. Interdependence among the proportions of the different types of attributions necessitated a separate analysis for each type.
A significant effect of Target Group was found for two of the five types of attribution: attributions to a trait in the child and to age-related factors. A significant main effect of Target Group was found for the extent to which mothers attributed aggression and social withdrawal to a trait in the child, F (2, 73) = 3.59, p < .03. Two-tailed, unplanned comparisons among the three target groups revealed that there was a tendency for mothers of W-I children to attribute both types of unskilled behavior to a trait in the child (M = .26) to a greater extent than mothers of AVG children (M = .10), t (73) = 1.84, p < .10. Mothers of A-E children did not differ from either of the other two target groups (M = .14), t's (73) = 1.12 and .63 for comparisons with mothers of W-I and AVG children, respectively.
For attributions to age-related factors, a significant interaction was found between target group and type of behavior, F(2, 73) = 3.72, p < .03. Subsequent unplanned two-tailed tests of the simple effects revealed that mothers of A-E children were more likely to attribute aggression to age-related factors (M = .41) than mothers of either the W-I (M = .06) or the AVG children (M = .21), t's(73) = 3.05, p < .005, and 2.68, p < .01, respectively; the two latter groups did not differ significantly, t(73) = 1.50. There were no differences between any of the three target groups in the extent to which mothers attributed social withdrawal to age-related factors (M's = .08, .08, and .14 for A-E, W-I, and AVG children, respectively).
The primary purpose of the present study was to examine possible differences in the ways that mothers of aggressive, anxious-withdrawn, and average children think about adaptive and maladaptive social development . In general, despite the relatively small number of subjects in each of the two extreme groups identified on the basis of multiple criteria, we found a coherent pattern of significant differences between the mothers of Aggressive-Externalizing, Withdrawn-Internalizing, and Average children.
Mothers of Withdrawn-Internalizing children placed greater importance than did mothers of Average children on the directive teaching of social skills. If we take this together with the finding that mothers of Withdrawn-Internalizing children were more likely to choose high power strategies for dealing with unskilled behaviors than the other mothers, one conclusion we can draw is that mothers of Withdrawn-Internalizing children believe strongly in a directive approach to the proactive teaching of social skills, and in their reactive strategies they may be described as overcontrolling. This is a conclusion others have come to with regard to socialization factors associated with the development of anxious-withdrawal in children (e.g., Hetherington & Martin, 1986), but to our knowledge it has not previously been reported as an empirical finding.
Our findings with respect to mothers' emotional reactions to and causal attributions about social withdrawal and aggression suggest some hypotheses about why it is that mothers of anxious-withdrawn children may be overcontrolling. Mothers of Withdrawn-Internalizing children were significantly more likely than other mothers to blame themselves when their children displayed unskilled social behaviors. They felt not only more angry and disappointed in their children than mothers of Average children but also significantly more guilty and embarrassed. The internalizing responses of guilt and embarrassment suggest that these mothers may model an internalizing style of emotional regulation for their children. We cannot be certain that this actually contributes to their children's development of an internalizing style, but it would undoubtedly help to perpetuate any tendency these children may already have to internalize their emotions. Interestingly, these mothers were also more disappointed than the other mothers and more inclined to blame unskilled behaviors on a trait in their child. The impression one gleans from these findings is that anxious-withdrawn children are exposed to a rather complex mix of conflicting emotions and attributions in their mothers, an impression which tends to support the suggestion that mothers of internalizing children may feel overidentified with and somewhat ambivalent toward their child (Levy, 1943; Parker, 1983).
The mothers of Aggressive-Externalizing children showed a different pattern of beliefs and feelings. Although these mothers believed more strongly than mothers of Average children in taking a directive approach to the teaching of social skills, and although their children were highly aggressive and they indicated that this kind of behavior certainly made them feel angry, they were nevertheless more inclined than the other groups of mothers to choose very indirect strategies or no strategies at all to deal with both aggression and withdrawal. Thus, in these mothers there seems to be some disparity between their proactive beliefs and their choice of reactive strategies. As to why this is so, we wonder if it has something to do with their attribution of aggressive behaviors to age-related factors. One set of dynamics that suggests itself to us is that these mothers' attempts to directively teach social skills may be unsuccessful and they may be somewhat intimidated by their child's aggressiveness. Perhaps they attribute their child's behavior and their own lack of success to short-lived age-related factors and choose less direct strategies in order to lessen their anxiety, avoid confrontation, and wait out what they hope is a passing phase. It seems to us, then, that these mothers are attempting to normalize their children's behavior despite the fact that it makes them feel quite angry. Such conflicting emotions and attributions could perpetuate a high level of aggression in the child in precisely the way Patterson and his colleagues have found, i.e., by engendering an erratic pattern of behavior in which undesirable behavior is sometimes rewarded or ignored as the parent attempts to avoid confrontation, and desirable behavior is sometimes, out of frustration, dealt with in an authoritarian fashion (Patterson, 1982; 1986).
This is, of course, but one possible scenario, and a highly speculative one at that. Since our data were correlational in nature, it is impossible to determine the direction of causality for the relations found. We cannot tell whether mothers' beliefs evolved in response to their children's behaviors or whether children's behaviors developed as a result of maternal beliefs. Most probably, both processes contributed. An implication of the reciprocal association between maternal cognitions and child behavior may be that it makes the pattern of parent-child interaction a complementary one. The present findings help to call attention to the kind of information that is needed to further our understanding of the complementary patterns of parent-child interaction associated with the development of socioemotional difficulties in childhood.
In summary, we have demonstrated that mothers of anxious-withdrawn, hostile-aggressive, and average preschool age children differ significantly from each other concerning how they would go about socializing social skills, and how they would react to (i.e., feel about, explain, and attempt to influence) their children's consistent display of aggressive or withdrawn behaviors. On the other hand, our data tell us little about mothers' actual (i.e., behavioral) proactive and reactive socialization strategies, nor do they inform us about possible developmental differences in the beliefs and behaviors of mothers of aggressive and anxiously withdrawn children. Furthermore, our data are correlational in nature and cannot help us determine causes of children's appropriate and inappropriate behavior patterns. Nevertheless, they do suggest directions in which to pursue underlying causes. In particular, it appears to be important to investigate the emotions and causal attributions that occur during parent-child interaction. Our data also provide us with a good beginning to the study the correlations and causal relations between parental beliefs and patterns of children's social behaviors.
1 Although Cohen's (1960) kappa would have been the preferred method of obtaining reliability estimates, this procedure was not performed on the observational data. In a more recent unpublished study, we used the same observational procedures for a sample of preschool and grade 2 children. Kappas were computed on 20% of the 180 10-sec observational intervals for each of 90 children. For the social participation categories of relevance herein, kappa = .84.
Baumrind, D. (1967). Child care practices anteceding three patterns of preschool behavior. Genetic Psychology Monographs, 75, 43-48.
Behar, L., & Stringfield, S. A. (1974). A behaviour rating scale for the preschool child. Developmental Psychology, 10, 601-610.
Elias, M. & Ubriaco, M. (1987). Linking parental beliefs to children's social competence: Toward a cognitive-behavioral assessment model. In R. Ashmore & D. Brodzinsky (Eds), Thinking about the family: Views of parents and children. Hillsdale, N.J.: Lawrence Erlbaum Associates.
Goodnow, J. J. (1988). Parents' ideas, actions, and feelings: Models and methods from developmental and social psychology. Child Development, 59, 286-320.
Hetherington, E. M., & Martin, B. (1986). Family factors and psychopathology in children. In H. C. Quay & J. S. Werry (Eds.), Psychopathological disorders of childhood (pp. 332 - 390). New York: Wiley. 3rd ed.
Levy, D.M. (1943). Maternal overprotection. New York: Columbia University Press.
Maccoby, E.E. & Martin, J.A. (1983). Socialization in the context of the family: Parent-child interaction. In E.M. Hetherington (Ed.), Handbook of child psychology, Vol.4, Socialization, personality, and social development (pp.1-102). New York: John Wiley.
Miller, S. A. (1988). Parents' beliefs about children's cognitive development. Child Development, 59, 259-285.
Mills, R.S.L. & Rubin, K.H. (in press). Parental beliefs about problematic social behaviors in early childhood. Child Development.
Moller, L. & Rubin, K.H. (1988). A psychometric assessment of a two factor solution for the Preschool Behavior Questionnaire in mid-childhood. Journal of Applied Developmental Psychology, 9, 167-180.
Olweus, D. (1980). Familial and temperamental determinants of aggressive behavior in adolescent boys: A causal analysis. Developmental Psychology, 16, 644-666.
Parker, G. (1983). Parental overprotection: A risk factor in psychosocial development. New York: Grune & Stratton.
Patterson, G.R. (1982). Coercive family processes. Eugene, OR: Castilia Press.
Patterson, G. R. (1986). Maternal rejection: Determinant or product for deviant child behavior? In W. Hartup & Z. Rubin (Eds.), Relationships and development. Hillsdale, N.J.: Lawrence Erlbaum Associates.
Rubin, K. H. (1982). Non-social play in preschoolers: Necessarily evil? Child Development, 53, 651-657.
Rubin, K.H. (1986). Play, peers and social development. In A. Gottfried & C. Caldwell Brown (Eds.), Play interactions: The contribution of play materials and parental involvement to child development. Lexington, MA: Heath (pp. 163-174).
Rubin, K.H., Mills, R.S.L. & Rose-Krasnor, L. (1989). Maternal beliefs and children's social competence. In B. Schneider, G. Attili, J. Nadel-Brulfert, & R. Weissberg (Eds.), Social competence in developmental perspective. Holland: Kluwer (pp. 313-331).
Mean Importance of Proactive Strategies for Teaching Social Skills, as a Function of Target Group
MSDMSDMSD Directive Teaching
Getting Acquainted .19 .64 .44 .88 1.00 1.55
Getting Accepted .05 .29 .22 .44 .17 .41
Resolving Conflicts .58 1.10 1.00 1.66 1.00 2.00
Persuading Peers .14 .44 .33 .71 .83 1.33
Getting Acquainted 2.47 1.70 2.11 1.54 1.17 1.17
Getting Accepted 2.02 1.04 2.56 1.13 2.83 .75
Resolving Conflicts 1.23 1.41 1.11 1.97 .83 1.33
Persuading Peers 2.46 1.50 2.56 1.67 3.00 1.41
Getting Acquainted .56 .91 1.00 1.23 1.33 .82
Getting Accepted .88 .97 .67 1.00 .17 .41
Resolving Conflicts 1.63 1.36 1.89 1.62 2.17 1.47
Persuading Peers .40 .80 .56 1.67 1.17 1.17
Getting Acquainted 2.72 1.66 2.33 1.32 2.50 2.17
Getting Accepted 2.95 1.23 2.56 1.42 2.83 .75
Resolving Conflicts 2.54 1.62 2.00 1.66 2.00 1.90
Persuading Peers 2.86 1.43 2.56 1.51 1.00 .89
n 57a 9a 6
a Sample sizes lower than those reported in the "Identification of Target Groups" section reflect missing data.
Mean Proportion of High and Low Coercion and Indirect-No Response Strategies for Reacting to Aggression and Social Withdrawal, as a Function of Target Group
M SD M SD M SD
High Coercion .19 .22 .23 .22 .47 .40
Low Coercion .68 .21 .57 .25 .53 .40
Indirect-No Response .13 .16 .20 .25 .00 .00
High Coercion .24 .25 .19 .23 .54 .27
Low Coercion .37 .23 .36 .25 .22 .25
Indirect-No Response .39 .28 .46 .17 .24 .20
n 60 10 6
Mean Intensity of Emotional Reactions to Aggression and Social Withdrawal, as a Function of Target Group
Concerned 4.80 .90 5.10 .99 4.50 1.05
Angry 4.27 1.07 5.10 .74 4.67 1.03
Disappointed 4.23 .96 4.60 .52 4.50 .84
Surprised 3.97 1.12 3.40 1.35 4.00 .63
Puzzled 3.82 1.00 3.70 1.16 3.17 .75
Embarrassed 3.37 1.09 3.50 1.27 3.67 .82
Guilty 2.48 .73 2.50 .71 3.50 1.23
Amused 2.05 .22 2.10 .32 2.00 .00
Pleased 2.00 .00 2.00 .00 2.00 .00
Concerned 4.58 1.20 4.40 1.08 4.83 1.33
Angry 2.03 .18 2.40 .97 2.67 1.63
Disappointed 3.52 1.30 3.50 1.27 4.67 1.21
Surprised 4.12 1.22 4.10 1.10 4.17 1.60
Puzzled 4.50 1.14 3.90 1.10 4.00 1.41
Embarassed 2.10 .35 2.20 .42 2.83 1.33
Guilty 2.35 .88 2.40 .70 3.00 .89
Amused 2.05 .29 2.20 .63 2.83 1.33
Pleased 2.15 .48 2.50 .53 2.33 .82
n 60 10 6
Mean Proportion of Each Type of Attribution About Aggression and Social Withdrawn, as a Function of Target Group