摘要：Every-grade, every-year testing plays a prominent part in U.S. education policy and research, but the rise of standardized testing has been met with frustration and opposition. In response, policymakers have proposed legislation designed to curb the amount of standardized testing. There is little empirical evidence, however, about the potential impact of these alternate approaches on current evaluation systems. Using data from a large, urban school district, we compare value-added (VA) estimates from every-year, every-grade testing to those from two reduced-testing scenarios. We find marginal changes in the value-added estimates in both approaches relative to more traditional VA estimates. Estimates from annual testing in alternating subjects are slightly less precise but have lower associations with prior student achievement than biennial testing in both subjects. Further, there is significant decrease in the number of teachers for whom scores can be estimated in both approaches, exacerbating long-standing concerns with VA methodology.
关键词：high; stakes testing; evaluation; policy; accountability; regression analysis; value added
