| Relational aggression is a pivotal concept associated with the trend toward increasing attention toward aggression in girls. Specifically, acts of relationship manipulation are hypothesized to be especially salient for girls, who are expected to covet social acceptance and inclusion more than boys. Past studies have supported a distinction between relational and physical forms of aggression and identified unique correlates of relational aggression. However, characteristics of relational aggression have been largely examined in isolation, thereby providing only indirect information about the nomological network of relational aggression.;The present study adopted a construct validation approach to replicating and expanding upon existing research. Two hundred twenty-four children, aged 11 to 14, completed measures of relational, reactive, and proactive aggression as well as various other behaviors. One parent and one teacher per child completed parallel measures.;At the manifest variable level, hierarchical regression analyses, which were conducted separately for child, parent, and teacher reports, varied across informants. Overall, these analyses suggested that relational aggression is both uniquely reactive and uniquely proactive in nature. Consistent with previous research, relational aggression was associated with such harmful correlates as social and internalizing problems. In many cases, such associations were mediated by either reactive or proactive aggression. Moreover, although there were generally no gender differences in mean levels of relational aggression, there was some suggestion that the implications of relational aggression may differ for boys and girls.;At the latent variable level, a multitrait-multimethod approach was used to model associations between relevant constructs. Overall, these analyses provided consistent support for the convergent validity of all relevant constructs. However, there was less evidence for the discriminant validity of relational, reactive, and proactive aggression, suggesting that the distinguishing features of these hypothesized aggression subtypes may become obscured when combining multiple informants' perspectives. Method effects were identified for parent and child reports.;Differences between results at the manifest and latent variable levels highlighted potential limitations of combining multiple reports. As direct tests of fundamental assumptions of a multi-informant approach were unavailable, it is suggested that the unique manifest variable results may offer more promising directions for future research. |
