Estimating The Carelessness Parameter In Large-Scale Assessments: An Item Response Theory Approach Using The 4pl Model For The 2019-2021 National Business Technical Certificate Examinations Biology Multiple Choice Test Items
Year: 2026
Type: article
Abstract: The study examined carelessness responses in the National Business and Technical Certificate Examination (NABTCE) Biology Multiple Choice Test items for the years 2019–2021 using Item Response Theory (IRT) and the Four Parameter Logistic (4PL) model. Traditional psychometric models consider parameters associated with item difficulty, discrimination, and guessing, whereas the carelessness parameter (upper asymptote) accounts for the likelihood that high ability test takers may answer items incorrectly because of carelessness or other non cognitive factors. The data comprised responses from 24,707 students across three examination years: 2019 (9,265), 2020 (9,472), and 2021 (5,970), who were selected using a multistage sampling method from all six geopolitical zones of Nigeria. The upper asymptote (d) parameters were estimated using Winsteps Rasch software version 3.75.0, and one way ANOVA was used to test for significant differences among the three years at the 0.05 level of significance. Results showed that the proportions of items with upper asymptote scores below 0.95 were 28% (n = 14) in 2019, 24% (n = 12) in 2020, and 22% (n = 11) in 2021. Statistical analysis revealed no significant differences among the three years in terms of carelessness estimates (F(2,147) = 1.23, p = .296, η² = .016). The findings emphasized the need for the use of the 4PL model in the analysis of standardized tests to provide more precise estimates of ability and enhance test validity. Recommendations included systematic item review protocols, improved test administration procedures, and the adoption of 4PL modeling in routine psychometric analyses by examination boards. Keywords: Carelessness parameter, Four Parameter Logistic Model, Item Response Theory, NABTCE
Institution: University of Benin
Language: English
Field: Decision Sciences
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7167518036