Teacher Rating of Student Adjustment

Abstract

The Teacher Rating of Student Adjustment is a 7-item instrument developed by the Fast Track Project to assess dimensions of success in adjusting to middle and high school. The first item queries how well the teacher knows the child; the remaining items target the teacher's perceptions of a student's academic performance, academic motivation, social skills, adult relationships, conduct, and personal maturity. Responses are coded on a five-point scale ranging from 1 to 5, as follows: Poor, unsatisfactory skills (1); Below average skills (2); Average skills (3); Above average skills (4); and Excellent skills (5).

The data from this measure are distinctive in that multiple teacher ratings of the target behaviors were obtained for each student in grades 6, 7, and 8. The goal was to administer the instrument to teachers in each of the student's core classes, as most middle school students move among several classrooms for core academic subjects. The decision to use multiple informants raises methodological issues that impact scaling and analysis of these data.

The level of inter-rater agreement varies considerably across behavioral domains, with low values of the Intraclass Correlation Coefficient for the "social skills" and "relationships with adults" items. The behavioral characteristics targeted with these items are likely to be the most subject to influence by the particular classroom environment. Researchers should keep in mind that the scores for these domains include considerable variation across teacher ratings for each student, a potentially important source of measurement error in analyses using these data.

When using the aggregate scores, analysts should note that the number of teachers providing ratings varies across students. In models predicting the TRSA domains, Muschkin and Malone (2003) found that, as expected, the lowest residual variance estimates correspond to the sample of students who received three or more teacher ratings. To the extent that the variation in the number of teachers rating each student is systematic, this could introduce heterogeneity of error variance (heteroscedasticity) in analyses using these data, violating an assumption of many common analytic techniques. An alternative approach is to use a single teacher's rating for each student. This sacrifices the increased precision from the multiple teacher ratings, but avoids the heteroscedasticity problem. Muschkin and Malone found no differences between using reports from a randomly selected teacher and from selection based on the item assessing how well the teacher knows the child; the former is recommended when heteroscedasticity is a concern.

Dataset Names

Raw Dataset Name: TyK
Scored Dataset Name: TSAySTc

Keywords

Academic Performance, Social Skills, Teacher Child Relationships, Behavior Problems, School Adjustment.