Guessing Behaviour of Multiple Choice Test Items: An Item Response Theory Analysis Using the 3pl-Model for 2019–2021 National Business And Technical Certificate Biology Examinations in Nigeria

Abstract Guessing remains one of the most persistent psychometric threats to the validity of multiple choice test scores in large scale standardised examinations. This paper examined the guessing parameter (c-parameter) of the 2019, 2020 and 2021 National Business and Technical Certificate Examination (NABTCE) Biology Multiple Choice Test (MCT) items in the Three-parameter Logistic (3PL) Item Response Theory (IRT) model. The sampling method was a multistage probability sampling whereby it was used to obtain 24,707 candidates out of a target population of 79,809 students across the six geopolitical zones of Nigeria. Winsteps 3.75.0 software was used in calibration of data. It was revealed that 98 percent, 94 percent and 100 percent of items in 2019, 2020 and 2021, respectively, met the acceptable guessing threshold of 0.25 on four-option Multiple Choice Tests. One-way Analysis of Variance (ANOVA) showed the significant difference in the estimates of parameter guessing in three examination years (F = 3.228, p = 0.004) as significant. These findings demonstrated commendable psychometric quality of the NABTCE Biology Multiple Choice Test items. This paper suggested compulsory pre-administration IRT calibration, reconstruction of flagged items with specificity, and incorporation of guessing parameter analysis in the policy of national assessment in order to ensure integrity of the outcomes of Biology education in Nigeria. Keywords: Guessing parameter, Item response theory, 3PL model, Multiple choice items, NABTCE, Biology assessment