Teacher reports and student achievement: How sensitive are results to who is surveyed?

Abstract

This paper examines how within-school variation in teacher survey responses affects estimated relationships between teacher-reported measures and student achievement. Teacher questionnaires are widely used in large-scale education assessments and are frequently linked to student achievement data to study the role of teaching practices, working conditions, and school environments in shaping learning outcomes. In many such assessments, however, only one or two teachers per school are surveyed, implicitly assuming that these responses represent the broader instructional environment experienced by students. Using a dataset that includes questionnaire responses from multiple teachers within the same schools combined with student achievement data, this paper examines the extent of within-school variation in teacher survey responses and assesses how this variation affects estimated relationships between teacher-reported measures and student achievement. The analysis first documents the dispersion of teachers’ responses across a range of commonly used survey constructs. It then shows how estimated teacher–achievement relationships change when different teachers are randomly selected to represent the same school. The results highlight an underappreciated source of uncertainty in studies that rely on teacher questionnaire data and suggest that estimates linking teacher survey constructs to student achievement may be sensitive to the particular teachers surveyed within each school. This version is a working paper.