Evaluating ChatGPT Accuracy in Answering a Basketball Common Content Knowledge Test in Physical Education

• In Event: AERA Roundtable Session Thursday 7:45 am
  In Roundtable Session: Instructional Innovation and Learner Outcomes in PE/PETE

Abstract

This study aimed to (a) assess ChatGPT accuracy in answering a previously validated basketball Common Content Knowledge (CCK) test and (b) compare the performance of five versions—4o, 4.5, o3, o4-mini, and o4-mini-high. Using a comparative cross-sectional design, each version was prompted to complete the test, and accuracy was compared with Cochran’s Q test. All versions scored highly (80.6%–94.4%) with no significant differences (p > .107). These findings indicate that ChatGPT can accurately answer basketball CCK questions and may serve as a supplementary resource for PE teachers, though it should not replace professional judgment.

Authors

• Omar Albaloul, Kuwait University

• Chad Killian, University of New Hampshire