Construction and Calibration of an Item Bank in the Educational Measurement and Evaluation Course Based on the One-Parameter Item Response Model for Students of the Faculty of Education – Sana'a University

Abstract

The study aimed to construct and calibrate an item bank for the Educational Measurement and Evaluation course based on the one-parameter item response theory model (1PL). The study sample consisted of 119 third-year students from the Faculty of Education – Sana’a University, representing 37% of the total study population. The research instrument was an item bank covering the first three units of the textbook assigned to third-year students. Scientific procedures were followed in building the item bank, which comprised 112 multiple-choice items distributed across three test forms (A, B, and C). Each test form included 44 items, with 10 common anchor items shared among the three forms. The Bilog-MG3 software was used to verify the fit of the item bank data to the one-parameter model, calibrate the items based on their difficulty levels, and estimate the parameters of item difficulty, individual abilities, test information function, empirical reliability coefficients, and the reliability index. Additionally, SPSS software was used to test the assumption of unidimensionality. Among the most prominent findings: after verifying the assumptions of unidimensionality, local independence, and speededness, it was found that no items were excluded by the Bilog-MG3 program under the one-parameter model. However, based on the chi-square goodness-of-fit test, 16 items were excluded for not fitting the model. No participants were excluded from the three test forms. The item bank was successfully calibrated according to the one-parameter model, with a mean difficulty parameter of 0.65, a standard deviation of 0.907, and item difficulties ranging from -1.93 to 2.27, corresponding to ability estimates ranging from -2.20 to 3.37.