Measuring the Impacts of Teachers I: Evaluating Bias in Teacher Value-Added Estimates,” Working Paper 19423, National Bureau of Economic Research

Are teachersimpacts on studentstest scores (value-added) a good measure of their quality? One reason this question has sparked debate is disagreement about whether value-added (VA) measures provide unbiased estimates of teacherscausal impacts on student achievement. We test for bias in VA using previously unobserved parent char-acteristics and a quasi-experimental design based on changes in teaching sta¤. Using school district and tax records for more than one million children, we \u85nd that VA models which control for a students prior test scores exhibit little bias in forecasting teachersimpacts on student achievement. We are indebted to Gary Chamberlain, Michal Kolesar, and Jesse Rothstein for many valuable discussions. W