Testing the item bias of PISA 2006 student questionnaire by using sequential logistic regression and Poly-SIBTEST methods

This study investigated whether student questionnaire of PISA 2006 show differential item functioning with respect to different cultural and linguistic characteristics (English- Turkish) with the samples of Australia, New Zealand, United States of America and Turkey on a comparative basis.The sample of this study consisted of 4 countries selected among 57 countries participating in the PISA implementation and students selected among approximately 400000 students, namely 14170 students from Australia, 4823 students from New Zealand, 5605 students from United States of America, and 4942 students from Turkey. In this study, due to cultural and linguistic differences adressed as the possible causes of differential item functioning, the data of the 4 countries having similar and different cultural-linguistic characteristics was studied. Australia has been taken for reference and the countries were compared with the same language "similiar culture (Australia- New Zelland ), the same language "different culture ( Australia-America ) and different language-different culture ( Australia- Turkey ) .In the first phase of the study, the factor structure of the student survey was examined by using a confirmatory factor analysis (CFA). In the second phase of the study, it was investigated whether questionnaire items show differential item functioning between different kültürse ant languraysa bay usan sequential logistic regression and poly-SIBTEST.The results of the confirmatory factor analysis showed that the measurement model had the same factor structure in all cultures. The findings of sequential logistic regression and poly-SIBTEST revealed that there were items showing differential item functioning between the countries. While these methods showed consistency in determining differential item functioning, they differed in the levels of differential item functioning.As a result of the analyses, it was observed that when linguistic and cultural differences increased among countries, the number of items showing differential item functioning increased, as well. In conclusion, the items may show differential item functioning in terms of cultural and linguistic differences. Thus, this fact should be taken in consideration when using these tests on different populations from different culteres and languages.