Assessing the Dimensionality and Model Fit of the 2015 NECOMathematics Multiple-Choice Items in Nigeria Using the Four-Parameter Logistic Model for Achieving SDG 4

Abstract

Evidence on the psychometric quality of large-scale examinations remains essential for ensuring valid educational decisions and monitoring learning outcomes. This study examined the dimensionality and model fit of the 2015 National Examinations Council (NECO) Mathematics items in Nigeria using Four-Parameter Logistic (4PL) item response theory model. An ex-post-facto research design was adopted. The population comprised 24,718 candidates who wrote the 2015 NECO Senior School Certificate Mathematics III Examination in Osun State, Nigeria. Using a multistage sampling procedure, responses from 3,000 examinees were selected for analysis. Data were analysed using the MIRT package in R, with item parameters estimated through marginal maximum likelihood estimation and the expectation-maximisation algorithm. Dimensionality was assessed using factor analytic procedures, while model-fit was evaluated using Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), Sample-Size Adjusted Bayesian Information Criterion (SABIC), and Hannan-Quinn Information Criterion (HQ). Results indicated that the unidimensional model provided a superior fit to the data compared with multidimensional alternatives, yielding lower information criterion values (AIC=1,142,691; SABIC=1,144,092; HQ=1,145,043; BIC=1,145,043). Item parameter estimates further indicated that most items demonstrated acceptable discrimination, moderate difficulty, low pseudo-guessing parameters, and minimal upper-asymptote effects. These results indicate that the examination items functioned satisfactorily within the 4PL framework and met the unidimensionality assumption. The study provides empirical evidence for the psychometric soundness of the 2015 NECO Mathematics examination and underscores the importance of rigorous measurement practices to enhance the validity, fairness, and accountability of large-scale educational assessments in support of Sustainable Development Goal 4 (SDG 4).