DETECTING DIFFERENTIAL ITEM FUNCTIONING BY USING SEQUENTIAL LOGISTIC REGRESSION AND POLY-SIBTEST METHODS

45816

The aim of this study is to compare the differential item functioning methods used in the determination of item bias, which is an important validity problem in psychometry, and to reveal the similarities and differences in the application. For this purpose, the two preferred methods, sequential logistic regression and poly-SIBTEST, were used to analyze the DIF. PISA 2006 student questionnaire was examined to determine whether there was DIF items according to cultural and linguistic differences. Australia, New Zealand, United States and Turkey were comparatively investigated samples in this study. First, the factor structure of the student questionnaire was examined by confirmatory factor analysis. Subsequently, the sequential logistic regression and poly-SIBTEST analyzes were used to determine whether the questionnaires showed DIF between different cultures and languages. Then, the similarities of DIF determination methods were examined. According to results of the confirmatory factor analysis showed that the measurement model had the same factor structure in all cultures. Sequential logistic regression and poly - SIBTEST findings show that there are items that have different item functioning among countries, and if the linguistic and cultural differences between countries increase, the number of items that have differential item functions increases. When the compatibility of the methods was examined, it was found that both methods were compatible in determining differential item functioning, but more sensitive DIF analysis was performed with poly-SIBTEST method.

705

PISA , differential item functioning , confirmatory factor analysis , sequential logistic regression , poly , SIBTEST

Abant İzzet Baysal Üniversitesi Eğitim Fakültesi Dergisi

717

Abant İzzet Baysal Üniversitesi Eğitim Fakültesi Dergisi