The Practical Implications of Test Timing and Summer Learning Loss for Value-Added Estimation

• In Event: Issues in Value-Added Models for Teacher Accountability

Abstract

It is agreed that teachers should not be held accountable for the determinants of achievement that fall outside of their control. However, value-added models (VAMs) that measure gains in student achievement from one school year to the next necessarily conflate students’ summer learning with learning that occurs during the school year. We use nationally representative survey data in which students took both fall and spring tests to examine the sensitivity of VAM-based rankings of classroom effects generated by “spring-to-spring” test-score gains relative to analogous rankings generated by “fall-to-spring” gains. Preliminary results suggest that the unequal distribution of students’ summer learning rates across classrooms leads to nontrivial differences between classroom-effect rankings generated by “spring-to-spring” and “fall-to-spring” test-score gains.

Authors

• Seth Gershenson, American University

• Michael S. Hayes, American University