ISSN (Online): 3108-1584
Volume 2, Issue 1 - July 2026
The psychometric quality of teacher-made multiple-choice tests in introductory university statistics courses remains under-examined in resource-constrained contexts, potentially leading to invalid ability estimates due to unaddressed dimensionality and model misfit. This study investigated the dimensionality and model-data fit of the 15-item STA 111 test administered to 3,563 first-year undergraduates at Joseph Sarwuan Tarka University, Makurdi, Nigeria, in 2022. Using Item Response Theory (IRT) with the four-parameter logistic model (4PLM), analyses were conducted via NOHARM for dimensionality and Jmetrik for fit statistics. The test was found to be three-dimensional (confirmed by sequential RMSR reductions ≥10% up to three dimensions), and all 15 items showed significant misfit (S-X² statistic, p < 0.05 for each item). These results indicate violations of unidimensionality assumptions and poor alignment with the 4PLM, likely due to content heterogeneity and local dependencies. Findings underscore the need for multidimensional IRT calibration or item refinement to enhance test validity and support fairer educational decision-making in Nigerian higher education statistics programs.
Item Response Theory, Dimensionality Assessment, Model Fit, Multiple-Choice Tests, Four-Parameter Logistic Model
A. A. Adoghe, A. D. E. Obinne, S. O. Emaikwu, "Item response theory assessment: dimensionality and model-fit in introductory statistics multiple-choice items", Cosmo Research & Science International Journal, vol. Jul-25, no. 1, pp. 392-398, 2026.
A. A. Adoghe, A. D. E. Obinne, S. O. Emaikwu (2026). Item response theory assessment: dimensionality and model-fit in introductory statistics multiple-choice items. Cosmo Research & Science International Journal, Jul-25(1), 392-398.
A. A. Adoghe, A. D. E. Obinne, S. O. Emaikwu. "Item response theory assessment: dimensionality and model-fit in introductory statistics multiple-choice items." Cosmo Research & Science International Journal, vol. Jul-25, no. 1, 2026, pp. 392-398.
@article{CRSIJ26000289,
author = {A. A. Adoghe, A. D. E. Obinne, S. O. Emaikwu},
title = {Item response theory assessment: dimensionality and model-fit in introductory statistics multiple-choice items},
journal = {Cosmo Research and Science International Journal},
year = {2025},
volume = {2},
number = {1},
pages = {392-398},
issn = {3108-1584},
url = {https://cosmorsij.com/published/CRSIJ26000289.pdf},
abstract = {The psychometric quality of teacher-made multiple-choice tests in introductory university statistics courses remains under-examined in resource-constrained contexts, potentially leading to invalid ability estimates due to unaddressed dimensionality and model misfit. This study investigated the dimensionality and model-data fit of the 15-item STA 111 test administered to 3,563 first-year undergraduates at Joseph Sarwuan Tarka University, Makurdi, Nigeria, in 2022. Using Item Response Theory (IRT) with the four-parameter logistic model (4PLM), analyses were conducted via NOHARM for dimensionality and Jmetrik for fit statistics. The test was found to be three-dimensional (confirmed by sequential RMSR reductions ≥10% up to three dimensions), and all 15 items showed significant misfit (S-X² statistic, p < 0.05 for each item). These results indicate violations of unidimensionality assumptions and poor alignment with the 4PLM, likely due to content heterogeneity and local dependencies. Findings underscore the need for multidimensional IRT calibration or item refinement to enhance test validity and support fairer educational decision-making in Nigerian higher education statistics programs.},
keywords = {Item Response Theory, Dimensionality Assessment, Model Fit, Multiple-Choice Tests, Four-Parameter Logistic Model},
month = {July}
}