부산대학교 도서관

The development of assessments as part of the funded LOCUS project is described. The assessments measure students' conceptual understanding of statistics as outlined in the GAISE PreK-12 Framework. Results are reported from a large-scale administration to 3,430 students in grades 6 through 12 in the United States. Items were designed to assess levels of understanding as well as components of the statistical problem solving process as articulated in the GAISE framework. We discuss details of how the model used to develop the LOCUS assessments guided the gathering of evidence for validity and reliability arguments. Three types of validity evidence are presented: Validity Evidence via Evidence- Centered Design, Validity and Reliability Evidence via Item Analysis, and Validity Evidence via Item Calibration. [ABSTRACT FROM AUTHOR]