No abstract

No abstract

A recent study challenged the ability of evaluation teams to select technically adequate data sources. Aspects of methodology and focus of inquiry restrict the interpretation of results. A more productive line of inquiry is suggested.

Scores on three scales of the Pupil Evaluation Inventory (Aggression, Withdrawal, and Likability) were derived from a sample of 200 regular education and mainstreamed elementary students and were compared with group achievement test results and membership in referral subgroups. Results showed low to moderate correlations of the PEI measures with achievement test scores and indicated some gender differences in the magnitude of the relationships. Groups referred for special ser-vices or gifted programs showed significant differences from regular education students on all of the PEI measures. The PEI holds promise for the further study of the social and behavioral correlates of school success.

The behavior ratings of parents and teachers using the Burks Behavior Rating Scales (BBRS) were examined by means of a paradigm derived from generalizability theory. Internal consistency indices and interrater reliabilities were manipulated to determine and apportion variance into error components. The analysis suggested that large amounts of error variance were present for most of the 19 scales on the BBRS. The large amount of error or situational variance indicated that comparing scores obtained from different raters in different settings is not advisable when one is making judgments with respect to stable child characteristics. Further implications for the development and use of behavior rating scales were discussed.