DIMENSIONALITY AND MODEL FIT ASSESSMENT OF 2022 NECO MATHEMATICS MULTIPLE-CHOICE QUESTIONS IN ANAMBRA STATE

Abstract

Mathematics multiple-choice questions (MCQs) play an indispensable role in assessing student’s knowledge level in the different aspects of mathematics. These test questions, to be valid and reliable, should have an infinitesimal variance between the scores it obtains and the student’s ability. This study evaluates the reliability and validity of 2022 NECO mathematics multiple-choice questions by assessing the dimensionality and model fit of the test items. Two research questions guided the study. The study adopted a survey design, and the population for the study was made up of all the 26,421 SSIII students of the 2022/2023 academic session who enrolled for the NECO examinations in Anambra State, out of which a sample of 1,828 was selected by the use of purposive sampling technique. Data collected was analyzed using DIMTEST statistics for research question one, and Akaike Information Criterion (AIC) and Likelihood Ratio Test (LogLik) for research question two. The findings revealed that the 2022 NECO Mathematics MCQs do not fulfill the Item Response Theory assumption of Unidimensionality, while the data was best fitted into 4 Parameter Logistic Model (4PLM). It was recommended that examination bodies, and researchers involved in the use of IRT should ascertain the conformity of the test items to the IRT assumptions in order to avoid bias in item parameter estimation.