Association for Public Policy Analysis & Management

Both policymakers and academic analysts place heavy emphasis on test scores in evaluating schools; raw scores are frequently used in accountability measures, while researchers judge the effectiveness of schools and teachers using test-score value-added measures. However, test scores are imperfect metrics of school success, and a narrow focus on improving test scores may crowd out investments in student learning that may have more persistent effects. In this study, I develop a model of schools’ response to test score-based accountability, in which schools must choose how to allocate scarce resources between short-term (teaching to the test) and long-term (in-depth teaching of course material) investments in their students, under the constraint that their test scores must stay above a threshold that keeps them from being closed. Schools with high resource endowments are not bound by the constraint and can focus more on student learning, while poorer schools must reallocate resources to test preparation that they would otherwise prefer to put toward in-depth teaching. I test this model using student-level administrative data from high schools and colleges in Michigan, calculating each school’s test-score value added in reading and math, then determining the relationship between value added and college grades and initial placement in the same subject, across subjects, and in untested subjects.