Determining Differential Item Functioning with the Mixture Item Response Theory

Abstract

Purpose: Studies in the literature have generallydemonstrated that the causes of differential itemfunctioning (DIF) are complex and not directlyrelated to defined groups. The purpose of this studyis to determine the DIF according to the mixture item response theory (MixIRT) model, based on the latent group approach, as well as the Mantel-Haenszel method, based on the observed group approach, compare the results, and determine the possiblecauses of the DIF. Research Methods: As this studyis contributing to the production of information todevelop the theory, it is considered basic research. Inaccordance with the purposive sampling method, theresearch sample consisted of 1166 fourth-grade level students from Singapore, Kuwait, andTurkey who participated in the Trends in International Mathematics and Science Studymathematics application and took the sixth booklet. During the data analysis, the model thatadapted the data according to MixIRT was determined. Then, the status of the itemsdisplaying DIF was determined according to the adaptive model. Findings: According to theMixIRT, the two latent class models fit best to the data. No significant difference by genderwas observed in either class or any country. This finding suggests that the gender variable,which is frequently used as the observed group in DIF studies, should not be dealt with alone.Implications for Research and Practice: Since it is difficult to state whether an item isadvantageous for a subgroup when DIF is determined in accordance with known groups, it isrecommended to employ the latent class approach to determine DIF.