Upgrading High-Stakes Assessment

Because of the large number of students involved, statewide assessments used throughout the United States are designed to be highly efficient with respect to administration and scoring. This constrains the formats that can be used and consequently limits testing to a subset of competencies typically associated with the education standards being assessed. The present paper outlines an alternative approach designed to help address this limitation. Performance assessments are used, but administered statewide only to samples of students. These samples provide group estimates of proficiency. Parallel performance assessments administered by teachers are then used to measure proficiency at the individual student level. Common specifications facilitate the development of comparable performance assessments at the state and classroom levels. Within classrooms, the performance assessments are used formatively with emphasis given to employing best practices when providing students feedback. The performance assessments administered by teachers are eventually used summatively, with student performance on teachers' assessments and the assessments administered to samples of students used to cross-validate each other. (Contains 1 footnote.) [This paper was produced by the Center for Advancement of Learning and Assessment (CALA) at Florida State University.]