Investigation of differantial item functioning of TIMSS 2019 items in fourth grade

The purpose of this research was to determine whether the multiple-choice items in the mathematics and science subtests at the 4th grade level in the TIMSS 2019 test show Differential Item Function (DIF) by gender and having a computer/tablet at home. The study group of this research consisted of students who received the eighth and ninth booklets among the 5th grade students in the 2019 TIMSS research Turkey sample. DIF levels of the items in the TIMSS 2019 test were determined by Mantel-Haenszel (MH), Logistic Regression (LR), and SIBTEST methods. For gender variable, DIF was found in two items, one in mathematics and one in science, in the eighth booklet, and in four science items in the ninth booklet. Regarding having a computer/tablet at home variable, DIF was found in eight items, two math and six science items, in the eighth booklet, and a total of ten items, five math and five science, in the ninth booklet. Additionally, the number of items with DIF and the consistency of the methods differed by the subtests and variables. While LR method detected the most number of items with DIF, MH method detected the least number of items with DIF. Considering the compatibility of the DIF detection methods, it was found that the most compatible methods were MH and SIBTEST, while the least compatible methods were MH and LR. This research provides evidence that items with DIF in the TIMSS 2019 test provide advantages or disadvantages to some groups. This situation negatively affects the validity and reliability of the decisions taken based on the test results. In addition, the results obtained from the TIMSS application should be used with caution in comparative studies and items with DIF may result in bias in the interpretation of the results. Keywords: TIMSS 2019, eTIMSS, differential item functioning, having a computer/tablet at home, Mantel-Haenszel, Logistic Regression, SIBTEST.