The K-12 mathematics and science "frameworks" now emphasize problem solving skills and conceptual understanding. This has stimulated the development of performance tests because they are better measures of these skills than the multiple choice items relied upon in the past to assess student achievement. However, information about performance measures is sparse and fragmented. The proposed project will examine the cost, technical quality feasibility, and acceptability of alternative assessment systems (i.e., testing programs that include combinations of measures for making decisions about individuals, instructional programs, or state and national trends). The project will gather and analyze data on the performance and other types of measures used by the California Assessment Project and other such programs. The study will examine the statistical properties of these measures, the resources required to develop and use them, and the attitudes of students, teachers, and others to their inclusion in testing programs. The study also will develop several tasks from the same "shell." Data on these measures will be used to investigate the sensitivity of these tasks to the effects of instruction and practice. The project's final report will discuss the key features and implications of alternative assessment systems.
