This study examined the relationship among teacher, peer, and self-ratings of children's social behavior. The Pupil Evaluation Inventory was completed by 172 first-graders, 346 fourth-graders, 283 seventh-graders, and 30 teachers. Groups of deviant responders and controls were also selected from the total sample on the basis of peer-rated aggression and withdrawal scores. Interrater agreement was consistently greater between peer and teacher ratings than between self-ratings and either peer or teacher ratings. Discrepancies between raters were greatest for children with more deviant scores, with peer ratings providing the highest estimates of deviant behavior, and self-ratings yielding the lowest. Self ratings were lower than teacher or peer ratings on aggression and withdrawal, and higher on likability. Aggression produced greatest agreement between raters. Agreement was uninfluenced by the cognitive maturity of peer evaluators. The results suggest that the selection of raters should be influenced by the class of behaviors to be evaluated and the context in which they occur.

This study examined age-related changes in the organization underlying children's ratings of social deviance in their peers. Peer ratings of aggression, withdrawal, and likeability using the Pupil Evaluation Inventory (PEI) were collected from 326 first graders, 356 fourth graders, and 298 seventh graders. Measures of the . perceived similarity of all possible pairs of PEI items were derived by computing the frequency with which children at each grade level were concurrently nominated by their peers for both items comprising each pair of items. Multidimensional scaling was then employed to elucidate the structure underlying these indexes of interitem similarity. Structure ratios denoting the cohesiveness of the clusters of aggression, withdrawal, and likeability items were computed from the results of these analyses. For children of all ages aggression items and likeability items were found to comprise highly cohesive categories of behaviors that were distinct from each other. Withdrawal items, in contrast, clustered poorly at the first grade but became an increasingly cohesive category of behaviors and increasingly distinct from aggression as grade level increased.Aggression and social withdrawal represent fundamental dimensions of adult-rated deviant behavior in children (Achenbach & Edelbrock, 1978;Quay, 1979). Comparatively few studies, however, have examined children's perceptions of aggression and withdrawal in their peers. The need for research in this area is compelling because of the increasing use of peers as assessors of childhood deviance (e.g., Ledingham, 1981;Weintraub, Prinz, & Neale, 1978). Existing peer assessment instruments have largely tended to parallel adult-rating measures in their focus An earlier version of this article was presented at the annual meeting of the Canadian Psychological Association, Montreal, Quebec, June 1982. This research was supported by grants from the Quebec Ministry of Education, and Health and Welfare Canada.The authors wish to thank Claude Senneville and Denise Morin for their help in the data collection. Geoff Selig's valuable computer assistance is also greatly appreciated. We extend special thanks to la Commission des ecoles catholiques de Montreal for their continued cooperation with our research.

Previous research has revealed grade-related changes in organization underlying children's ratings of aggression and withdrawal in their peers (\ounger, Schwartzman, & Ledingham, 1985). The present investigation examined the contributions to such changes of age-related differences in the perspective of the raters (age of rater) and in the behavior of the children rated (age of children rated). Study 1 examined teacher ratings of aggression and withdrawal in first-, fourth-, and seventh-grade children in order to assess effects attributable to age of children rated. In contrast to earlier findings with peer raters, no differences were found across grade level in the organization of teacher ratings. Study 2 examined age of rater differences in the organization of first-, fourth-, and seventh-grade children's beliefs about behavior that might be displayed by hypothetical peers. Differences were found that paralleled those observed earlier in children's actual peer ratings. Study 3 examined first-and seventh-grade children's ratings of peers who were older or younger than the raters, to assess the influence of age of rater on children's ratings. Age of rater effects emerged even when children rated peers who were not their age mates. These findings suggest that differences across grade level reported in children's peer ratings largely reflect differences in the child raters' view of behavior. Implications of these findings for the use of peer evaluations are discussed.Childhood peer relations have assumed a role of central importance in the assessment and classification of childhood psychopathology (Achenbach & Edelbrock, 1978;Quay, 1979;Ross, 1980). How children relate to their peers has not only served as a means for classifying childhood behavior problems but has also been found to provide a valuable predictive index of later functioning in adolescence and adulthood (Kohlberg, LaCrosse, & Ricks, 1972;Roff, Sells, & Golden, 1972). Because children represent actual participant-observers of the social behavior of their peers and may consequently view such behavior from a unique perspective, investigators have also increasingly chosen children to serve as assessors of the social functioning of their peers. Indeed, it has been argued that the assessment of peer opinion provides the investigator with unique information

This study examined the ability of children of different ages to encode and retrieve from memory descriptions of aggressive and withdrawn behavior displayed by hypothetical peers. 90 children from the first, third, fifth, and seventh grades (aged 6-7, 8-9, 10-11, 12-13, respectively) listened to behavioral descriptions of a hypothetical aggressive and a hypothetical withdrawn boy and subsequently reported their recollections for these descriptions. Differences across grade level were found in the number of behaviors correctly recalled for the withdrawn boy but not for the aggressive boy. At the first grade, children recalled fewer descriptions of withdrawn behavior than of aggressive behavior. However, recall for withdrawal increased significantly across grade level, such that at grades 5 and 7, subjects recalled more descriptions of withdrawn than of aggressive behavior. Such differences across grade level in children's ability to encode descriptions of withdrawn behavior into memory and subsequently to retrieve the descriptions accurately were interpreted as evidence of the emerging importance of social withdrawal as a social-cognitive schema underlying children's social perceptions.

This study examined children's reasons for choosing peers for the withdrawal items on the Revised Class Play (RCP). Eighty-eight elementary-school children nominated peers they felt were best described by each RCP item. Reasons for their nominations were classified into 2 categories: passive withdrawal from and active isolation by the peer group. For 3 of the items ("Someone who would rather play alone than with others," "Someone who is very shy," and "Someone whose feelings get hurt easily"), the children's reasons were predominantly based on passive withdrawal, whereas for 3 others ("Someone who is often left out," "Someone who has trouble making friends," and "A person who cant get others to listen"), they were predominantly based on active isolation. Reasons for the remaining item ("Someone who is usually sad") were split equally between both alternatives.