Temporal Stability of Item Response Theory Parameters in National Business and Technical Certificate Examination (NBTCE) 2018–2023 May/June Economics Multiple-Choice Items

Le résumé fourni par la source

This study investigated the temporal stability of the Item Response Theory (IRT) parameters of the National Business and Technical Certificate Examination (NBTCE) May/June Economics multiple-choice items administered between 2018 and 2023. The study adopted an ex post facto research design using candidates' response data obtained from the National Business and Technical Examinations Board (NABTEB). The population comprised 15,521 candidates' response records from the South-South geopolitical zone of Nigeria, while a sample of 10,509 response records was selected through simple random sampling. The study utilized the 50-item NABTEB Economics multiple-choice examinations administered between 2018 and 2023. Item calibration was performed using the four-parameter logistic model in jMetrik IRT software. Frequency counts and percentages were used to determine the temporal stability of the difficulty, discrimination, guessing, and carelessness parameters. The findings revealed that the difficulty parameter exhibited relatively low temporal stability, while the discrimination parameter showed moderate stability across the six examination years. Conversely, the guessing and carelessness parameters demonstrated high temporal stability, with the carelessness parameter recording the highest level of consistency across administrations. Among the examination years investigated, the 2023 examination exhibited the greatest overall temporal stability. The study concluded that although the psychometric quality of the NABTEB Economics multiple-choice examination was generally satisfactory. It was recommended that NABTEB institutionalize routine IRT-based item calibration, regularly review unstable items, and adopt the four-parameter logistic model as part of its quality assurance procedures.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.