Comparing the Individual Fit Indices in the Rasch Model Using the Mathematics Pass Test at Sultan Qaboos University

Abstract

This research aims to compare the effectiveness of individual fit indices (WSR, LZ, ECI4Z, and ECI2Z) using mathematics test scores at Sultan Qaboos University for the academic year 2019/2020 in light of their demographic differences. The study sample consisted of 351 students. The research used appropriate statistical methods. The study verified the assumptions of item response theory and the fit of the Rasch model for test items. The study calculated the agreement rates of the individual fit indices (WSR, LZ, ECI4Z, and ECI2Z) in detecting individuals with mismatched responses. The study results indicated that the highest agreement rate was between WSR and LZ indices, followed by ECI4Z and ECI2Z indices, while the lowest agreement rate was between LZ and ECI2Z indices. All four indices agreed with each other by over 90%. The WSR index is the most reliable among the four fit indices, with a reliability rate of 100%, followed by LZ with a reliability rate of 92.86%, ECI4z with a reliability rate of 90%, and finally ECI2z with a reliability rate of 81.82%.