How Do Multiple-Choice Assessments Reflect Student Performance? An Educational Data Mining Analysis Using Machine Learning Classifiers
ICCK Transactions on Educational Data Mining · 2026
DOI: 10.62762/tedm.2026.107680 ↗Kurz gefasst
Fachartikel in der Zeitschrift ICCK Transactions on Educational Data Mining, 2026. Verfasst von Niusha Shafiabady und 4 weiteren Autor:innen.
Abstract (Originalsprache)
Multiple-choice questions (MCQs) are widely used in higher education because they efficiently assess knowledge and a broad range of cognitive skills across disciplines. This study examines how MCQ-based assessments contribute to predicting final student grades, using data from 2,390 students (comprising 26,698 non-missing attribute values) enrolled in sixteen offerings of three undergraduate and postgraduate economics units at Charles Darwin University (CDU) between 2016 and 2019. Six machine learning classifiers were trained on assessment marks, assessment weightings, final examination performance, and the proportion of MCQs in the examination. Feature importance analysis was then used to quantify the contribution of each assessment component. The study compares MCQ and task-based examination questions, online MCQ and task-based assessments, and online and written MCQ components. Support Vector Machines achieved the highest accuracy (96.07%), followed by XGBoost (93.76%). Across the higher-performing classifiers, final examination performance and the task-based assessment were the strongest predictors of final grades. Online and written MCQ components contributed less strongly, although MCQs remain useful indicators of foundational knowledge. These findings support balanced assessment designs that combine MCQs with task-based activities, and they demonstrate how explainable AI can inform assessment analytics within Educational Data Mining.
Zitieren
APA
Shafiabady, N., Pfitzner, D., Nawaz, T., Dianati, S., & Brady, M. (2026). How Do Multiple-Choice Assessments Reflect Student Performance? An Educational Data Mining Analysis Using Machine Learning Classifiers. ICCK Transactions on Educational Data Mining. https://doi.org/10.62762/tedm.2026.107680
Vancouver
Shafiabady N, Pfitzner D, Nawaz T, Dianati S, Brady M. How Do Multiple-Choice Assessments Reflect Student Performance? An Educational Data Mining Analysis Using Machine Learning Classifiers. ICCK Transactions on Educational Data Mining. 2026. doi:10.62762/tedm.2026.107680
BibTeX
@article{Shafiabady2026,
author = {Shafiabady, Niusha and Pfitzner, Darius and Nawaz, Talha and Dianati, Seb and Brady, Mark},
title = {How Do Multiple-Choice Assessments Reflect Student Performance? An Educational Data Mining Analysis Using Machine Learning Classifiers},
journal = {ICCK Transactions on Educational Data Mining},
year = {2026},
doi = {10.62762/tedm.2026.107680}
}
EUVIMED