Swiss Journal of Psychology 65 (2), 2006, 81–91Original CommunicationAggression, Self-Understanding,and Social Competence inSwiss Elementary-School ChildrenThis study examined the relation between aggression, self-understanding, and social competence in a sample of 93 Swiss elementary-schoolchildren. Aggression was rated by the parents using the aggression subscale of the Child Behavior Checklist (CBCL/4-18). Self-understandingwas assessed with a short version of Damon and Hart’s (1988) self-understanding interview. The social competence of the children was observed in a quasi-experimental, cooperative play situation. The results revealed that aggression was related to domain-specific content aspectsof self-understanding. Aggression was, however, negatively associated with social competence. Moreover, nonaggressive children with highlevels of self-understanding showed more social competence than aggressive children with both high and low levels of self-understanding.Keywords: aggression, self-understanding, social competence, elementary-school childrenAggression is one of the most stable problem behaviors inchildhood (Loeber & Stouthamer-Loeber, 1998). Out of themany different explanations for the genesis and persistenceof aggression, some theories focus on the role of self-understanding and social competence. According to identitytheory, the quality of coping with natural crises of identityand the construction of a developmentally adequate selfview determine general psychosocial adjustment and theaccomplishment of developmental tasks (Damon & Hart,1992; Erikson, 1968; Kegan, 1982). Correspondingly, theorists assumed that aggression is related to a distorted selfview (Baumeister, Smart, & Boden, 1996; Edelstein, 2005).From the cognitive-developmental perspective, social interactions with peers foster the development of social-cognitive competences and social skills, and thereby influencethe level of psychosocial adjustment (Piaget, 1932/1965;Youniss, 1994). From this viewpoint, aggression may relate to a lack of opportunities for reciprocal interaction anddiscourse with peers, opportunities that would normallysupport the development of differentiated self-understanding and decentration in thinking. This lack may augmentpeer relationship problems. Rejection by peers may also result in the reinforcement of aggression, and, in the worstcase, establish vicious circles that impede further development of social competence. In accordance with thesetheoretical perspectives, previous studies have shown thataggression is related to specific problems in self-underDOI 10.1024/1421-0185.65.2.81standing and to behavioral deficiencies (e.g., Baumeister etal., 1996; Willner, 1991). However, there are both empirical inconsistencies and theoretically unresolved issues (Sutton, Smith, & Swettenham, 1999a, 1999b). Thus, the purpose of the present study was to examine further the questionof how aggression and cognitive aspects of self-understanding interrelate and contribute to the social competenceof elementary-school children.Aggression and Self-UnderstandingSelf-understanding is synonymous with the term self-concept (Damon & Hart, 1988). It is conceptualized as a multidimensional construct that defines a child’s view of his orher self or personal identity and comprises cognitive andaffective aspects (Damon, 1989; Greve, 2000; Marsh,1990). Investigations on the relations between aggressionand self-understanding in childhood have so far focused primarily on the affective aspect of self-understanding, theglobal self-worth, which is considered to be central for psychosocial adjustment (Leary & Baumeister, 2000). However, the results of the corresponding studies are contradictory: While some studies found that a generally lowself-esteem was related to aggression and/or low psychosocial adjustment in childhood (e.g., Barry, Frick, & Killian, 2003), other studies either found no consistent relaSwiss J Psychol 65 (2), © 2006 by Verlag Hans Huber, Hogrefe AG, Bern 82T. Malti: Aggression, Self-Concept, and Social Competencetions (e.g., Bushman & Baumeister, 1998), or positive associations (e.g., Hughes, Cavell, & Grossman, 1997;Hymel, Bowker, & Woody, 1993). The present study focuses on the cognitive aspects of self-understanding, because of our interest in how the structural complexity of selfdescriptions (stages) and the conceptualization of the self(content themes) relate to aggression. In the following, wefirst outline the theoretical model of self-understanding(Damon & Hart, 1988), and then elaborate possible reasonsfor an association between aggression and cognitive aspectsof self-understanding. According to Damon and Hart(1988), cognitive self-understanding can be subdivided into a structural and a content component. The structural component means that the ability to reflect on self as an objectunderlies structural developmental changes and developsover stages from more concrete attributes to increasinglyabstract levels of self-description (Damon, 1989).The content component of self-understanding impliesthat self-descriptions can be differentiated into specific content themes, which classify on the one hand the self-descriptions of the ‘Me’ (‘self-as-object’) into physical, active, social, and psychological self attributes, reflecting‘objective’knowledge of one’s own characteristics (Damon& Hart, 1992; James, 1890; Mead, 1934). These contentthemes are assigned to different content domains of selfunderstanding (e.g., a child’s evaluation of self). On the other hand, the self-descriptions of the ‘I’ (‘self-as-subject’)can be classified into the procedural categories of self-continuity, distinctness, and agency, reflecting the subjectively experienced sense of identity.Based on the general theoretical assumption that thequality of social behavior is related to differentiation andextension of social-cognitive concepts (e.g., Youniss, 1980),it is possible that aggression may also relate to a less differentiated self-understanding and the use of specific content themes to describe the self. More specifically, aggressive children may use physical and/or active attributes ratherthan social or psychological attributes to describe themselves, because in their daily lives they have fewer opportunities for social interaction, which would help create asense of social identity. There appear to be surprisingly fewresearch studies on the relation between these cognitive aspects of self-understanding and aggression or social behavior in childhood. A study by Pior (1998) found an association between the self-concept of social integration andthe continuity of solitary play or cooperative play in children. This result supports the assumption that aggressionrelates to fewer social attributes in self-descriptions. Research studies based on Damon and Hart’s self-understanding-model have shown that self-understanding differsbetween children with developmental problems and normally developed children (e.g., Lee & Hobson, 1998).Further, a study by Melcher (1986) revealed that adolescents diagnosed as having a conduct disorder showed aless differentiated level of self-understanding and mentioned different content themes when describing themselvescompared to adolescents without this diagnosis. Taken toSwiss J Psychol 65 (2), © 2006 by Verlag Hans Huber, Hogrefe AG, Berngether, these findings and case studies (e.g., Selman &Schultz, 1988) provide some evidence that children withaggressive behavior may have delayed self-understandingand use specific content themes to describe themselves. Insum, systematic research on the relation between the cognitive aspects of self-understanding corresponding to Damon & Harts’ model and aggression in middle childhood islacking. It seems to be rather unclear up to now how thesecognitive aspects of self-understanding relate to aggression.Due to the fact that self-understanding explains the distinctiveness of a person, whereas most other social-cognitive concepts focus on interpersonal relatedness (e.g., roletaking; Damon & Hart, 1988), the content themes of selfunderstanding may be especially interesting in relation toaggression. It is possible that aggressive children have difficulty in adequately distinguishing themselves from others, due to their history of negative experiences in socialrelations; thus, they cannot regulate the dimension distancecloseness suitably and tend to be too distinct, which maybe reflected in asocial self-descriptions.Aggression, Self-Understanding, andSocial CompetenceAccording to cognitive-developmental theory, childrenlearn through social interactions with peers, co-constructing reality, and shaping their social-cognitive developmentand the quality of their social behavior (Piaget, 1932/1965;Youniss, 1994). The quality of social interactions promotesin particular decentration in thinking as well as differentiation and coordination of perspective-taking (Keller &Edelstein, 1991).Correspondingly, uncooperative and aggressive behavior may lead to rejection by peers, which in turn corroborates the hostile attribution-biased cognitions of such children and thereby maintains or even reinforces theiraggressive behavior, impeding socially competent reactions(Crick & Dodge, 1996; Orobio de Castro, Veerman, Koops,Bosch, & Monshouwer, 2002; Warman & Cohen, 2000).Previous research has shown that children who were rated as aggressive showed less social competence in peer interactions (Dodge, Pettit, McClaskey, & Brown, 1986;Hughes, White, Sharpen, & Dunn, 2000; Stormshak & Webster-Stratton, 1999). Also, it has been confirmed in researchthat aggression is related to a lack of cooperation in socialplay. For example, in a behavioral observation study byWillner (1991), it was found that aggressive boys showedmore solitary play, reacted more often in a hostile-aggressive way, displayed more egocentric and less prosocial behavior. Although there have been many studies on the relation between aggression and social competence (Dodge et.al., 1986, Hawley, Malti, & Keller, 1998; Keller & Malti,1999), research results are not consistent. Contrary to theresearch results above, some studies found that aggressionpartially relates positively to social competence (e.g., Hawley, 2003), and to self-understanding (Edens, Cavell, & T. Malti: Aggression, Self-Concept, and Social CompetenceHughes, 1999). Correspondingly, the assumption that aggressive children are incompetent has been critized. Thus afew researchers have argued that some aggressive childrenmay be quite competent and use their social skills to manipulate other children, in order to reach personal goals(e.g., Sutton et al., 1999a). To further investigate these diverging standpoints, it is necessary to differentiate the social-cognitive concepts used in previous studies and/or toemploy new ones (Crick & Dodge, 1999). In the presentstudy, we use cognitive self-understanding in order to determine whether or not aggression is necessarily related tolower social-cognitive capability.Additionally, we wanted to investigate whether aggression is related to specific self-descriptions. We are, therefore, interested in how aggression and self-understandingdetermine a child’s social competence, because the inconsistent results in previous research suggest that there maybe subtypes of aggressive children: some with a high levelof self-understanding who show social competence, butothers with a low level of self-understanding who are socially not competent. In sum, we explored two researchquestions:The first question addressed the issue of whether aggression is determined by a less differentiated view of self(structural level of self-description) and the use of specificcontent themes. Based on the assumption that fewer opportunities to interact and cooperate aggravate decentrationin thinking, we assumed that aggression is related to a lower level of self-understanding. This view is supported byprior research in the moral domain (e.g., Gibbs, 1991). Further, we hypothesized that aggression is related to the useof physical and/or active attributes and to the absence orless frequent use of social attributes while describing theself, because this may reflect the daily experience of children with less social interactions. The use of physical and/oractive attributes may also help to perpetuate a consistentself-view and/or reduce cognitive dissonance; aggressivechildren may recognize that they have had negative experiences with other children and that they are not able to buildup a positive sense of identity through their social relations,but rather by physical or active attributes. The self-definitions of aggressive children may therefore rely on creatingdistance between themselves and others, reflected in asocial self-descriptions. Thus, an aggressive child may referto his or her physical strength, while a nonaggressive childmay mention positive relations with other children whiledescribing the self.However, it is possible that aggression is only in specific domains (e.g., self-definition) related to these specificcontent aspects, because the use of the content aspects variesbetween the domains of self-understanding (Damon & Hart,1988). We therefore wanted to explore whether the relationbetween aggression and these content domains depends onthe domain of self-understanding, or whether it is general.Second, we wanted to investigate whether the negativerelationship between social competence and aggression isdependent on the level of self-understanding or not. Based83on previous research, which has shown that not all aggressive children lack social-cognitive and social skills (Suttonet al., 1999a, 1999b), we wanted to find out whether, in addition to the socially less competent children with aggressive behavior and low levels of self-understanding, there aresocially competent children who employ aggressive behavior but have high levels of self-understanding.Aggression and social competence belong to the samedevelopmental facet (the behavior of the child) and are conceptually overlapping developmental dimensions (definitions of aggression contain partially the absence of socialcompetence, e.g., cooperation), whereas the levels of selfunderstanding refer to complex internal processes. Wetherefore assumed that these developmental levels of selfunderstanding are less strongly related to social competencethan aggression at this age. Thus, we expected nonaggressive children to show more social competence than aggressive children, independent of the level of self-understanding, but also that aggressive children with different levelsof self-understanding possibly vary as to social competence.We hypothesized that high levels of aggression in combination with low levels of self-understanding predict the lowest social competence, followed by high levels of aggression and high levels of self-understanding, followed bynonaggressive children with low or high levels of self-understanding. In addition, age, gender and socioeconomicstatus effects were investigated.MethodParticipantsThis study was part of a larger project on social-cognitivedevelopment and aggression in middle childhood (Malti,2003). A total of twelve classes from three different elementary schools in two German-speaking cantons inSwitzerland were chosen for participation. Requests for participation were sent to 198 parents, and 153 parents (77%)gave consent for their child’s participation. School boardpermission was obtained as well. Of the participating children 56% belonged to the first and second grade, and 44%to the third and fourth grade. All 153 children took part inthe self-understanding interview. Due to school absences,140 children participated in the behavioral observation. Sixty-one percent of the parents returned the questionnaire,some after a reminder was sent by mail, and thus we derived parental aggression ratings for 94 children. From theseremaining 94, one child was excluded from data analysesby virtue of being too old. Thus, the final sample included93 children (6.9 – 11.1 years, M = 8.4, SD = 1.3; 45 boys,48 girls) and their parents. The socioeconomic status of thesample was calculated by a revised version of the Hollingshead (1979) four-factor index (Dodge, Pettit, & Bates,1994). The sample represented basically middle-class families (M = 12.3; SD = 4.0; range = 3-20). The socioeconomicdistribution of the sample was comparable to the socioecoSwiss J Psychol 65 (2), © 2006 by Verlag Hans Huber, Hogrefe AG, Bern 84T. Malti: Aggression, Self-Concept, and Social Competencenomic distribution of Zurich’s population as reported in theStatistical Yearbook 1999 (Malti, 2003), and we thereforeassume that there was no systematical sample bias due tothe relatively high drop-out rate of the parents.is with other people (e.g., ‘when I am with babies, I am gentle’). A complete description of the interview questions hasbeen given elsewhere (Malti, 2003).Coding of the Self-Understanding InterviewProceduresThe study consisted of the following parts: In the first session, children were individually interviewed about theirself-understanding in a separate, quiet room in their school.The interviewers were trained undergraduate psychology students. The interviews were taped and later transcribed.In the second session, the children’s social behavior duringplay was observed in a quasi-experimental, dyadic play situation in a separate, spacious room located within theschool. The testers were other undergraduate psychologystudents, who received intensive training in the observationmethod. All sessions were videotaped for five minutes. Parallel to the child’s sessions, parents received a questionnaireon the social development of their child by mail; they thencompleted it and returned it to the school.MeasuresAggressionThe extent of aggressive behavior of the children was rated by the parents using the German version of the aggression subscale from the Child Behavior Checklist (CBCL4/18; Achenbach & Edelbrock, 1983; Döpfner et al., 1998).Cronbach’s α for the scale was α = .90. The extent of aggression was low-to-moderate in this sample (M = 7.2, SD= 5.9, range = 0-31).The coding scheme and scoring criteria from Damon & Hart(1988) were employed to classify the children’s answers in(a) one of four developmental levels of self-understanding(stage levels) and (b) one of four content aspects of self-understanding (content categories). The four developmentallevels reflect the range in complexity of self-description.Lower levels contain self descriptions in concrete, observable terms, whereas higher levels comprise complex anddeeper concepts of self-description. The four content aspects of self-understanding represent different contentthemes of the ‘self-as-object’: Physical aspects, activities,social aspects, and psychological aspects of self description(e.g., a physical self-description refers to physical characteristics or material possessions of the self). Variables weresummarized and the following scores derived: (a) sixteendomain-specific sum scores for the content categories(physical, active, social and psychological) in the domainsself-definition, self-interest, self-evaluation, and self-in-relation and b) a global sum score for the level and four global scores for the four content categories. Cronbach’s α forthe scores ranged from α = .48 to α = .74 with a mean of α= .58. In order to determine the interrater-agreement, 15%of the interviews were coded by two independent raters. Theinterrater-agreement was calculated according to the procedure suggested by Damon & Hart (1988, p. 88).The percentage agreement between the two raters was93% for the levels, and 96% for the categories of self-understanding. Disagreements were discussed, and a consensus was found.Self-UnderstandingSocial BehaviorThe self-understanding of the children was assessed by ashort version of Damon and Hart’s semi-structured interview of self-understanding (1988) for children and adolescents. It contained seven main questions regarding the ‘selfas-object’ (e.g., ‘can you tell me what you are like?’). The‘self-as-subject’ was not assessed, because we were mainly interested in the set of beliefs children use when defining themselves (‘self-as-object’). The questions in the interview were assigned to four different content domains ofthe self-understanding: Self-definition, self-interest, selfevaluation, and self-in-relation.Self-definition responses refer to the way a child describes the self (e.g., ‘I am tall’), self-interest responses reflect a child’s hopes, wishes, or desires (e.g., ‘I wish to havea lot of money, because then I could buy me a big car’), selfevaluation responses refer to the child’s evaluation of self(e.g., ‘I am proud that I am so big, because then I can playfootball’), and self-in-relation responses reflect how the selfSwiss J Psychol 65 (2), © 2006 by Verlag Hans Huber, Hogrefe AG, BernThe children were asked to play a game with a classmateof the same-age. The children were randomly assigned toeither same-sex or opposite-sex dyads. Almost half of thedyads were same-sex, and the other half opposite-sex. Twodifferent games were used in counterbalanced order(‘Stapelmanniken’ and ‘CheeseWiz’). After entering theroom, children were instructed by the tester how to play thegame and how to reach the goal of the game (e.g., build hightowers). The tester motivated the children to cooperate, butthe decision to play the game either cooperatively or alonewas left to the children (e.g., build one big tower together,or build two towers, i.e., one each). The tester remainedseated in the observation room during the observation period and pretended to read. Play was recorded by a camera,located at the opposite end of the room. T. Malti: Aggression, Self-Concept, and Social CompetenceCoding of Social BehaviorThe five-minute play sequences were divided into 30 tensecond intervals. The total amount of solitary versus interactive play during the five minutes was protocolled. Playwas coded as solitary if a child played with him- or herselfor was passive. Play was coded as interactive if the child interacted with the playmate in a positive way (e.g., cooperatively) or in a destructive way (e.g., aggressively). A revised coding system was employed in order to code sociallycompetent behaviors (Dodge et al., 1986; Malti, 2003). Thecategories for social competence were: social initiation(e.g., a child initiates a social interaction), helping/sharing(e.g., a child shares its play materials), praise/reinforcement(e.g., a child praises the other child), request (e.g., a childrequests something), play suggestion (e.g., a child expresses a constructive play suggestion), agreeableness (e.g., achild agrees on a play suggestion), and setting up rules (e.g.,a child sets up constructive play rules).The occurrence/non-occurrence of the behavioral categories in each ten-second interval was coded, and dummyvariables for each behavioral category were created (0 =non-occurrence, 1 = occurrence). Each category wassummed up over the 30 intervals, and the proportionalamount of occurrence was divided by the total number ofintervals in order to derive standardized variables. An overall score for ‘social competence’ was built by aggregatingthe behavioral categories. Cronbach’s α for the social competence score was α = .61. The mean score of social competence in the sample was M = 0.76 (SD = 0.87). To determine the interobserver agreement, 20% of the dyads werecoded by two independent observers. The interobserveragreement was calculated by the number of agreements divided by the number of disagreements plus the number ofagreements. The average agreement for the behavioral categories was 94%, with a range from 88% to 100%. Nonagreements were discussed and a consensus was found.85sion sum score and of the global level of self-understanding score. Next, two dummy variables were created (0 = lowaggression, 1 = high aggression; 0 = low level of self-understanding, 1 = high level of self-understanding). Childrenin the low aggression group had a mean aggression scoreof M = 3.31 (SD = 1.90), and in the high group, the meanaggression score was M = 12.05 (SD = 5.64). Children inthe low level of self-understanding group had a mean of M= 1.72 (SD = 0.25) while for the high group the mean wasM = 2.30 (SD = 0.19). Based on these two dummy variables,four subgroups were created as follows: Children with highaggression scores and low levels of self-underst…