Differential item functioning analysis of PISA 2015 reading items: Singapore, Australia and USA

The purpose of this study is to determine whether there is DIF in the PISA 2015 reading section items from Singapore, Australia and the United States by employing the Mantel-Haenszel and the Logistic Regression methods. In order to calculate uniform DIF two research questions are asked: 1) Is there gender-related DIF and or (2) culture-related DIF in the 16 selected binary items of the PISA 2015 reading section from Singapore, Australia and the United States? The data was obtained from the PISA 2015 reading items section and the sample was divided according to the variables gender and culture for a total of 2293 students. The DIF values were determined and classified into categories A, B or C. MH findings showed that considering the variable gender, males and females were equally advantaged whereas the LR results demonstrated that females were favoured over males. Regarding the variable culture, MH showed that Australia was advantaged over Singapore, Singapore over USA and Australia over USA. However, LR analysis showed exactly the opposite results in all the cases. A total of 7 items were classified with C level and the overall levels were higher. In general, when the variable gender was studied most of the items were classified with A levels. Nevertheless, once the countries were paired to evaluate the variable culture, the DIF level increased into major occurrence of B and C levels. The overall results demonstrated that the items from the reading section of PISA 2015 functioned equally among the group members.