TEACHER PERCEPTION, PEER PERCEPTION, SELF-PERCEPTION, CLASSROOM VARIABLES, AND SOLITARY BEHAVIOR (MINNESOTA).

Various assessment measures have been used to identify socially isolated children. A model relating constructs associated with social behavior was developed to examine causal relationships among various types of assessment measures. The subjects were 267 third and fourth grade children from public school classrooms in Minnesota. Teacher perception, peer perception, self-discomfort, assertiveness, positive solitary behavior and negative solitary behavior were conceptualized as dependent latent variables. The effects of academic achievement and sex on the dependent latent variables also were examined. Observable indicators of the latent variables include peer sociometric measures; three teacher ratings of social interaction; self-report measures of assertiveness, anxiety, and loneliness; classroom observations of solitary behavior; and the Iowa Test of Basic Skills. Covariance structure analysis procedures (LISREL) were used to link observed measures of social behavior to their respective latent variables through confirmatory factor analysis (CFA) and to examine hypothesized causal relationships among the latent variables. CFA of the dependent variables indicated that the initial model did not provide an acceptable fit with the data. T-values indicated that the observed variables loaded on the hypothesized latent constructs except for two self-report measures. Subsequent model testing indicated that some observed variables loaded on more than one latent variable. The initial covariance structure model was revised based on the CFA. The LISREL analysis indicated that neither the initial nor the revised models of social behavior provided an adequate fit with the data. The significance of individual parameters was examined. The observed self-report measures of assertiveness, loneliness, and anxiety demonstrated low reliability. Hypotheses supported include that academic achievement was causally related to peer perception and that teacher perception, peer perception and academic achievement were causally related to the level of negative solitary behavior. The results support the assessment of perceptions of significant others and academic achievement as screening measures to identify socially isolated children. Additionally, the validity and reliability of using self-report measures with elementary students needs further investigation