Recent years have seen a push for assessments to better support teachers as they make instructional decisions and place less of burden on students. One approach that can meet both needs is the use of diagnostic classification models (DCMs). DCMs are psychometric models that provide fine-grained results on specific skills that a student has mastered. Additionally, because DCMs provide categorical results (e.g., mastery vs. nonmastery), we can achieve highly reliable results with very short assessments that can be embedded into normal instructional cycles. In this talk, we will describe the psychometric properties of DCMs, including how DCMs can be used to evaluate proposed relationships among the measured skills (e.g., learning progressions). We then provide examples of how student mastery results and learning progressions can be combined for reporting to students, parents, and teachers to inform instructional decisions. Finally, we’ll share applied examples of DCMs being used innovative through-year assessment models such as the Pathways for Instructionally Embedded Assessment (PIE) project, a proof-of-concept 5th grade mathematics assessment, and the Dynamic Learning Maps (DLM) alternate assessment system, which is used to as an accountability assessment for students with the most significant cognitive disabilities in over 20 states in the United States. These programs demonstrate how DCMs can be used to refocus the purpose of assessment as being in the service of student learning.

Authors:
Jake Thompson, University of Kansas, United States

About the Presenter(s)
W. Jake Thompson, Ph.D., is the Assistant Director of Psychometrics for ATLAS at the University of Kansas, USA. He is the lead psychometrician for the Dynamic Learning Maps assessment and creator of the R package measr for diagnostic modeling.

Connect on Linkedin
https://www.linkedin.com/in/wjakethompson/

See this presentation on the full schedule – Monday Schedule

Bookmark