Improving Student Performance Through the Sequencing of Test Items: Recommendations for AI-Driven Assessment

• In Event: AERA Roundtable Session Wednesday 12:40 pm Four Seasons Ballroom 2-3
  In Roundtable Session: Artificial Intelligence in Education: Innovations in Assessment, Literacy, and Performance Analysis (Table 5)

Abstract

The study examines the effects of item sequencing on student performance in Mathematics and Science multiple-choice tests (MCTs) and provides recommendations for AI-driven testing and assessment. Standardized MCT items from the West African Examination Council, were organized into easy-to-hard, hard-to-easy, and random, and administered to 370 high school students in Ghana.
The results indicate significantly higher student performance in the easy-to-hard sequence in both subjects. These findings suggest that AI models may also enhance students’ performance when the training data is structured from simpler to complex. The study also highlights the need for fairness, transparency, and accountability in educational and AI-oriented assessments, which can contribute to more effective and equitable evaluation environments.

Authors

• John Baffoe, University of Wisconsin - Milwaukee

• Daniel Asamoah, Universiti Brunei Darussalam

• Auwal Halabi Kabara, Universiti Brunei Darussalam