Investigation of the effect of missing data on differantial item functioning in mixed type tests

This research is aimed at examining the effects of Markov chain Monte Carlo (MCMC), multiple imputation (MI) and expectation maximization (EM), which are methods of coping with missing data, on the differential item functioning (DIF) in mixed type tests containing dichotomous and polytomous items. For this purpose, the study was carried out on the full data set consisting of the scores of 1160 students who took the booklet number 9 in the science test in TIMSS 2019 and answered it completely. The conditions to be examined for the effectiveness of the methods are; missing data mechanism (MCAR and MAR), DIF level (A, B and C) and missing data rate (10% and 20%). Using MCMC, MI and EM methods; data were assigned to the missing data sets created by deleting data at different rates under the missing completely at random (MCAR) and missing at random (MAR) mechanisms. DIF analysis was performed with the poly-SIBTEST method on all data sets obtained. To this end, the results obtained from the full data set were compared with the results of other data sets of reference. In terms of all conditions, EM and MCMC methods performed better for C level DIF than A and B levels. It has been observed that the most successful MI method in determining DIF in items showing DIF in 10% and 20% MCAR mechanisms. Compared with the full data set, all three methods showed similar results in the 10% MAR mechanism, while MCMC gave the closest results in the 20% MAR mechanism compared to the other methods.