Analyzing PIRLS 2021 Türkiye Data Using Cognitive Diagnostic Modeling

• In Event: AERA Poster Session Wednesday 7:45am
  In Poster Session: Insights from Large-Scale Assessments: Advancing Methods, Measurement, and Learning Opportunities

Abstract

This study applies Cognitive Diagnostic Modeling (CDM) to the PIRLS 2021 Türkiye dataset to provide fine-grained insights into fourth-grade students’ reading comprehension. By applying the G-DINA model to responses from 667 fourth-grade students on the “Empty Pot” passage, the study identifies mastery of four specific attributes: Inference, Interpretation, Using Personal Knowledge and Experiences, and Examine and Evaluate. The G-DINA model demonstrated superior fit compared to alternative CDMs, enabling detailed item-level diagnostics. Results show only 5.85% of students mastered all attributes, while 10.34% failed to demonstrate any mastery. Inference was easiest; Using Personal Knowledge was most difficult. CDM provides diagnostic insights not captured by traditional scoring. The study showcases CDM’s utility in large-scale international assessments.

Authors

• Sedat Sen, Harran University

• Nihal Burcin Bircan, Turkish National Ministry of Education