An analysis of the differential item function for the items available in the PISA 2009 mathematics literacy sub-test through Mantel-Haenszel, SIBTEST and logistic regression methods

This study performs a Differential Item Function (DIF) analysis in terms of gender and culture on the items available in the PISA 2009 mathematics literacy sub-test. The DIF analyses were done through the Mantel Haenszel, Logistic Regression and the SIBTEST methods. The data for the gender variable were collected from the responses given by 332 students to the items in the mathematics literacy sub-test during the administration of the 5th booklet in the PISA 2009 application whereas the data for the culture variable were collected through the application of the 5th booklet in Turkey, Germany, Finland and the United States in the PISA 2009 application. Besides, the descriptive statistics and the reliability analyses were conducted for the mathematics literacy sub-test, and general information concerning the group distribution prior to the DIF analysis was obtained. Following the DIF analysis, which had been performed for the gender variable, 2 DIFs were specified in the MH technique while 3 were found in the LR technique and 4 were found in the SIBTEST technique. It may be said that 4 of the 5 items with DIF were in favour of male students, and only one was in favour of female students. Based on the MH analysis results which had been performed in relation to the data obtained from the students in Turkey and Germany, the DIF was found in 10 items; and 5 of them were to the advantage of students in Germany and the other items were to the advantage of those in Turkey. With the LR analysis the DIF was found in 10 items, and 7 of them were to the advantage of German students and 3 were to the advantage of Turkish students. In consequence of the SIBTEST, the DIF was found in 12 items; 7 of them were found to be to the advantage of German students and 5 were found to be to the advantage of Turkish students. According to the results of the MH analyses conducted for the data obtained from students in Turkey and Finland, the DIF was found in 12 items; and 6 of the items were to the advantage of Finnish students whereas the other 6 were to the advantage of Turkish students. In consequence of the LR analysis, the DIF was found in 10 items; and all of the items were to the advantage of Finnish students. The SIBTEST results, on the other hand, demonstrated that the DIF was available in 13 items; and that 7 of them were to the advantage of Finnish students while 6 were to the advantage of Turkish students. Based on the MH analysis results for the data obtained from the students in Turkey and in the United States, the DIF was found for 11 items; and 5 of them were to the advantage of American Students whereas 6 were to the advantage of Turkish students. According to the LR analysis results, the DIF was found in 9 items; and 4 of them were to the advantage of American students while 5 were to the advantage of Turkish students. According to the results of the SIBTEST, however, the DIF was found in 13 items; and 8 of them were to the advantage of American students while 5 were to the advantage of Turkish students.