Investigation of Differantial Item Functioning of National Antalya Mathematics Olympic items

In this study, it was investigated whether the items in the 26th National Antalya Mathematics Olympiad 2nd Stage exam held in 2022 showed Differantial Item Function (DIF) in terms of gender and the residential area where the students lived. In DIF analysis, Mantel Haenszel (MH), Logistic Regression (LR) and SIBTEST methods were used and then the results obtained from these three methods were compared. The data of gender and the region where the student lives variables used in this study were obtained from the student information questionnaire. The study carried out with data from 262 students who answered 20 common question in the 26th National Antalya Mathematics Olympiad Stage 2 exam and participated from 4 geographical regions. In order to obtain information of the group destribution before DIF analysis, descriptive statistics were applied according to the variables of the entire group, gender and the region where the student lives. As a result of the DIF analysis conducted according to the gender variable, it was determined that one and the same item showed DIF in the MH, LR and SIBTEST methods only compared to the MH and LR methods. The item found to show DIF was found to word in favor of boys. As a result of the DIF analysis conducted according to the variable of the region where the student lives, four items showed DIF in the LR method, while one item showed DIF in the MH and SIBTEST methods. According to the MH method, the same substance works in favor of the Marmara Region in different comparisons; According to the LR method, different items worked in favor of all other regions except the Ege Region. In the SIBTEST method, the same item worked in favor of the Marmara and Ege Regions. When MH, LR and SIBTEST methods were examined in terms of the number of items showing DIF, it was seen that different results were obtained in all three methods and the LR method determined a higher number of items. The method with the least number of DIFs was SIBTEST.