ON ANALYZING OF MULTIPLE CHOICE QUESTIONS IN NIGERIA UNIVERSITIES’ COMPUTER BASED TESTS WITH ITEM RESPONSE THEORY MODELS

Abstract

Every year, university teachers face challenge of how to cope with increasing number of examination students together with the validity of test items. Due to this, multiple choice question (MCQ) test item came to address this problem but little or no intention is paid to the properties of test items. Absent of item analysis in developing these MCQ item could endanger the integrity of assessment, selection, certification, and placement. In the same way, improper use of item analysis and lopsided test items could lead to the same fate by bringing about wrong award of grade and certificate. A three-parameter logistic (3PL) model uses ability to predict the probability of a certain response as a function of student’s ability level and item properties in describing suitability of test items as well as nominal response (NR) model which estimates students’ propensity to endorse their preferred options are presented as a solution module to the problem. The estimated discrimination, difficulty, psuedoguessing, category difficult, and category discrimination parameters indices of items are presented for illustrations and guide to test developers (e-Center, NECO, WAEC, and JAMB)