Applying Cognitive Diagnostic Modeling to Examine Middle School Students' Argumentation Practices

• In Event: Trends and Topics in Science Education Research Methods

Abstract

This study applies a cognitive diagnostic modeling (CDM) approach to examine student performance on argumentation. We abstract three types of attributes which are deemed critical to successful argumentation practice: making claims, providing evidence, and warranting. We automatically scored responses to 19 constructed response items from 932 middle school students using machine learning algorithms, and applied cluster analysis to classify students into different groups. We then examined patterns of students’ mastery of the attributes. Results indicate that machine scoring has comparable accuracy with human expert scoring, and the items showed sufficient psychometric characteristics except for three items with negative discrimination. We found that students were clustered in seven groups, and 90% of students demonstrated seven mastery patterns.

Authors

• Xiaoming Zhai, University of Georgia

• Kevin Haudek, Michigan State University