Abstract
This study aimed to assess the properties of MCQs using various courses in a novel holistic approach employing quantitative-comparative design. The data were collected from two university campuses in Saudi Arabia offering a Bachelor of Science in Nursing (BSN) program between 2018 and 2020. The multiple-choice questions on 'item difficulty' (34.36 +/- 17.42), 'item discrimination' (0.32 +/- 0.20), and 'option affinity' (0.32 +/- 0.16) were reasonably good. The 'lexical density' was "Very Complex" (85.08 +/- 13.37) with the Basic Adult Care Nursing final examination (91.54 +/- 9.42). The 'readability index' was low (7.65 +/- 3.08) in the Fundamentals of Nursing I Theory final examination (6.18 +/- 3.13), but "High" in information entropy (4.23 +/- 0.10) with the Fundamentals of Nursing I Theory examination (4.25 +/- 0.11). Statistical differences (<.001) were noted in option affinity,' 'lexical density,' and 'readability index. Further, the multilayer perceptron (MLP) prediction model shows 'lexical density' (100%) as the most important. Findings indicate the need for quality assurance measures in the form of faculty training.