Commentary on the evaluation of teacher effectiveness through student test scores

Abstract

The Tennessee Value-added Assessment System is claimed to be able to estimate the impact of teachers on their students’ progress.

This has led to further claims, such as that teacher quality is paramount in improving student progress. However, TVAAS and similar schemes should not be relied upon. Explanations of TVASS in the public domain are incomplete and poorly presented. TVAAS is not a ‘test’ of anything, and other analysts have attributed the same student progress residuals as used in TVAAS to school, classroom, district, leadership, social and economic factors.

The analysis appears to be circular – effective teachers are defined by progress of students so students making progress have necessarily effective teachers. The analysis anyway cannot be sustained with the kinds of data available.

The estimated level of missing data, and of measurement and representational error in the data that is present, suggest that the estimated residuals for each student are composed largely of error terms. Sanders and colleagues make elementary statistical errors, such as using significance tests with population data.