UCLA's Institute for Democracy, Education, and Access

Russell examines the extent to which California’s test based policies promote and/or inhibit California students’ access to basic learning resources and conditions. Russell reviews the history of California’s assessment systems and describes the current Academic Performance Index (API), analyzes recent survey data on the impact of testing, considers the experiences of other states, and simulates results under an alternative design for accountability in California. He finds that California’s accountability system has numerous technical shortcomings that prevent it from being a valid and useful indicator for either holding schools accountable for student learning or determining which schools are “low-performing” and warrant state intervention. The design of the system sets a target for schools that far exceeds national averages in student performance. The standardized test on which the API rankings are based is not aligned to state standards; in addition, problems in calculating accountability rankings from results on the test make the former poor expressions of student achievement. However, he also concludes that even if the technical shortcomings were fixed and/or prior decisions were altered to make expectations for most schools more reasonable, California’s single-minded focus on student outcomes as measured by standardized tests fails to adequately prevent, detect or deter gross disparities in education. A system that focuses solely on student learning outcomes, no matter how broadly defined, cannot provide schools and their constituents with information that allows them to identify why students succeed or fail to succeed. Furthermore, reliance on standardized testing tends to result in a narrowing of curriculum and instructional strategies. Evidence from other states shows that more successful accountability systems are possible. Russell concludes that an improved accountability system in California should include an emphasis on educational inputs, the use of better and more varied output measures, and a sensitivity to local contexts.