青少年前期的間接攻擊行為：同儕估計法的建構效度

Indirect Aggression during Early Adolescence: Construct Validity of Peer Estimation Technique

本研究旨在以驗證性因素分析、社會計量地位差異，以及性別與年齡差異等三方面探討同儕估計法所測得的青少年前期間接攻擊行為之建構效度。本研究以台灣地區1111位小六與國二學生為樣本，以同儕估計法測量受試者的間接攻擊行為，並以同儕提名法測量受試者的社會計量地位。驗證性因素分析結果顯示：將間接攻擊行為視為關係操弄、社會排擠和散播謠言三個彼此獨立卻互有關聯的多向度構念獲得初步支持。社會計量地位差異分析結果顯示：爭議型孩童與拒絕型孩童所出現的間接攻擊行為最為頻繁，且兩者皆比歡迎型孩童有較多的間接攻擊行為，支持過去西方的研究結果。以性別與年齡為獨變項的分析也顯示在青少年早期，間接攻擊行為的確具有性別與年齡之交互作用。綜言之，本研究以不同的角度切入，檢驗了以同儕估計法測量間接攻擊行為之效度。

This research dealt with three issues related to the construct validity of peer-estimated indirect aggression during early adolescence. The first issue involves whether the construct of indirect aggression is a first-order factor or a second-order factor. The second issue involves the relations between sociometric status and the frequency of indirect aggressive behaviors. The third issue involves whether indirect aggression is more frequently found in girls or boys, or there is no gender difference. Therefore, the specific aims of this study were threefold. The first one was to examine the nature and factor structure of young adolescents' peer-estimated indirect aggression. The second one was to investigate how peer-estimated indirect aggression in young adolescents is related to sociometric status. The third one was to analyze gender differences of indirect aggression in different age groups. A total of 1111 sixth graders (N = 596) and eighth graders (N = 515) from 18 schools in Taiwan completed a peer nomination procedure for assessing sociometric status and a peer estimation procedure for assessing the frequency of indirect aggression. Two competing models of confirmatory factor analyses including a first order one- factor model and a second order three- factor model were proposed. Results indicated that the second order three- factor model fits the observed data better than the single factor model. This hierarchical CFA model implies that the higher- order factor, indirect aggression, has a direct causal effect on the lower- order factors, including relationship manipulation, social exclusion, and rumors spreading. Two one-way ANOVAs with sociometric status as the independent variable and indirect aggression tendency as the dependent variable yielded sociometric status differences on indirect aggression among both the 6th graders and the 8th graders. Post hoc analyses revealed that the controversial and the rejected subjects in both the 6th grade and the 8th grade are the most frequent users of indirect aggression; they showed higher frequency of indirect aggressive behaviors than those designated as the average, neglected and popular sociometric status. Another gender X age two-way ANOVA with indirect aggression tendency as the dependent variable indicated a significant main effect of gender, a significant main effect of age, and a significant interaction effect of gender and age. Specifically, the indirect aggression tendency was higher for boys than girls as well as higher for 6th graders than 8th graders. However, those two main effects were qualified by the interaction effect. That is, compared with girls in grade 6, boys in grade 6 were more likely to adopt indirect aggression. By contrast, compared with boys in grade 8, girls in grade 8 were more frequent users of indirect aggression. Additionally, compared with boys in grade 8, boys in grade 6 were more likely to adopt indirect aggression. In contrast, girls in grade 6 were not more frequent users of indirect aggression than girls in grade 8. In summary, the hierarchical model of indirect aggression proposed by French, Jansen, and Pidada (2002) and Underwood (2003) was initially supported. Using peer estimation technique, the present study not only confirmed findings from Crick and Gropeter's (1995) about the connections between indirect aggression and sociometric status but also extended the pattern from middle childhood to preadolescence. Further, the present study offered a possible explanation for why prior studies revealed conflicting gender effect of indirect aggression. Finally, the advantages of using peer estimation technique on measuring indirect aggression were discussed.