Centering Test Fairness in Early Childhood Computational Thinking Assessment Practices: A Regularized MNLFA approach

• In Event: AERA Roundtable Session Wednesday 7:45 am - Gold 2
  In Roundtable Session: Computational Thinking Assessment and Learning Across K-12

Abstract

Although reliable and valid computational thinking (CT) assessments for early childhood are growing, test fairness remains underexplored. Establishing fairness is critical for making valid comparisons across groups, yet few studies have addressed item bias in CT measures. This study used a regularized Moderated Nonlinear Factor Analyses (MNLFA) to examine differential item functioning (DIF) across gender, age, and interaction in a 19-item CT assessment administered to 272 young children. Few items showed DIF, supporting fairness across groups. However, boys outperformed girls, though the gender differences narrowed over time. These findings confirm age-related changes and reveal new evidence of gender differences. This study demonstrates the utility of regularized MNLFAs in evaluating test fairness and capturing individual differences in early childhood CT assessments.

Authors

• Chungsoo Na, Utah State University

• Lingna Xu, Utah State University

• Jody E. Clarke-Midura, Utah State University

• Willa van Dijk, Utah State University