Assessment of Item Parameters of 2017 National Examinations Council’s English Language Multiple Choice Test Using Item Response Theory

Abstract

The study aimed to evaluate the adequacy of the NECO 2017 English Language examination items in measuring their intended objectives among secondary school students in Osun State. It assessed the dimensionality, local independence, item difficulty, and discrimination indices of the examination. A non-experimental design of descriptive research type was employed, involving 26,127 senior secondary students who sat for the exam in June/July 2017. Using a stratified random sampling technique, the students were divided based on characteristics such as gender, school type, and academic performance. Data from the NECO headquarters, including responses and scores, were analyzed using the two-parameter logistic (2PL) IRT model. Three research questions were addressed, revealing that the test was unidimensional with a maximum DETECT value of -0.1331, ASSI of -0.211, and RATIO of 0.142. However, 30 item pairs were locally dependent. Additionally, 19 out of 100 items were found to be of poor quality based on difficulty parameters outside the acceptable range (-3 to 3), and 29 items had poor discrimination indices (less than zero). The study concluded that while the exam's multiple-choice items were largely unidimensional and independent, suggesting reliable scores. It was recommended that there is a need to identify specific skills influencing student performance and develop targeted interventions to improve overall test performance.