An investigation of the PISA 2015 in terms of differantial item functioning based on culture

The aim of this study is to research whether or not the attitude items towards sciences in PISA 2015 student questionnaire including differential item functioning (DIF) in terms of language and culture. Generalized Mantel Haenszel (GMH), ordinal logistic regression (OLR) and poly-SIBTEST methods are applied for DIF analyzes. The GMHDIF program was used for the GMH method and the R software was used for the poly-SIBTEST and OLR methods. The study works with PISA 2015 dataset. This dataset encompasses 5089 students in Turkey, 5157 student in Ireland, 5000 students in USA and 4417 students in England. As a result of the analysis carried out on 9 items, DIF is determined in six items by using GMH and OLR methods and four items by using poly-SIBTEST method in the England-Ireland data. In the England-USA data, DIF is determined in eight items by GMH method, nine items by OLR method and three items by poly-SIBTEST method. In the England-Turkey data, DIF is determined eight items in the GMH method, nine items by OLR method and five items by poly-SIBTEST method. Based on three methods, five items in comparison with England-Ireland and England-USA, and also seven items in comparison England-Turkey are detected attitude items including DIF. As a result, it is found that there was accordance the number of DIF items determined by OLR, GMH and poly-SIBTEST methods. On the other hand there are differences in the levels of DIF according to OLR and poly-SIBTEST methods. Furthermore, when the items containing DIF are examined by OLR and poly-SIBTEST methods, it is observed that there was accordance between the groups with the advantages.