DIMENSIONALITY ASSESSMENT OF 2023 AND 2024 WEST AFRICA SENIOR SCHOOL CERTIFICATE EXAMINATION (WASSCE) MATHEMATICS MULTIPLE-CHOICE TEST
Abstract
Evidence of dimensionality of test items is central to the validity of scores from high-stakes examinations and the appropriateness of the item response theory (IRT) models used to calibrate them. This study investigated the number of dimensions underlying the 2023 and 2024 West African Senior School Certificate Examination (WASSCE) Mathematics multiple-choice items. One research question guided the study. A cross-sectional psychometric design incorporating a descriptive survey was adopted. The study did not principally investigate student’s opinions, attitudes or perceptions; rather, it analysed their item-level responses to determine the latent dimensional structures underlying the 2023 and 2024 WASSCE Mathematics multiple-choice items. The population of the study is comprised of 22,550 SS3 students in public secondary schools in Anambra state, a sample of 1,125 SS3 was drawn using multistage sampling procedure for the study .The instrument for data collection was May/June 2023 and 2024 WASSCE Mathematics multiple-choice items. Responses were dichotomously scored, and tetrachoric correlation matrices were computed for each year because they are more appropriate for binary data than Pearson correlations. Factorability was established using the Kaiser-Meyer-Olkin (KMO) measure and Bartlett’s test of sphericity. Exploratory factor analysis (EFA) was then conducted on the matrices, with dimensionality judged by scree plots and the ratio of the first to the second eigenvalue. The data were suitable for factor analysis in both years (KMO = .91 and .93; Bartlett’s ?² = 4215.36 and 4386.56 for 2023 and 2024 respectively, both p < .001). A dominant first factor accounted for 41.2% and 38.7% of the total variance, for both years respectively and the first-to-second eigenvalue ratios were 5.56 and 6.66, exceeding the conventional threshold of 3 for essential unidimensionality. These results indicate that both examinations are essentially unidimensional, supporting the use of a single latent ability in 4PLM calibration. The findings have implications for item banking, test equating, and score reporting by examining bodies, and for further research using larger samples and multidimensional comparison models
Keywords
dimensionality, four-parameter logistic model, WASSCE, Mathematics, tetrachoric correlation, exploratory factor analysis.