Research on educational effectiveness most often uses student assessments of classroom instruction for measuring aspects of teaching quality. Given that crucial inferences on the success of education are based on these assessments, it is essential to ensure that they provide valid indicators. In this study, we illustrate the application of an innovative application of a multilevel bifactor structural equation model (ML-BFSEM) to examine the validity of student assessments. Analyzing a large-scale data set of 12,077 fourth-grade students in three countries (Finland, Norway, and Sweden), we find that (i) three aspects of teaching quality and subject domain factors can be established; (ii) metric and scalar invariance could be established for the ML-BFSEM approach across countries; and (iii) significant relations between students’ assessments of how easy the teacher is to understand and achievement in all subjects exist. In support of substantive research, we demonstrate a methodological approach for representing the complex nature of student assessments of teaching quality. We finally encourage substantive and methodological researchers to advance the ML-BFSEM.
This item's license is: Attribution 4.0 International