An investigation of math subtests item bias in 2014 free boarding and scholarship examination in terms of gender variable

The aim of the research is to determine whether or not 2014 free boarding and scholarship examination (FBSE) includes item bias in terms of gender variable. The descriptive model of quantitative research is used for the study. Data for 5th, 6th, and 7th grade secondary education students attending from seven regions of Turkey was obtained by the General Directorate of Measurement, Evaluation and Examination Services in Turkey. 20% students were selected randomly within 192075 students. Therefore, the research group consists of 38415 students. 2014 FBSE includes Turkish, Math, Science and Technology, and Social Sciences. In the study, It is examined through 38415 students' answers to math subtest consisting of 25 items that whether bias exists or not in terms of gender. Rasch Model, one of the determining methods of Different Item Function (DIF), in respect to gender variable. Before passing on to DMF analysis, it has been examined whether the data satisfy the Rasch model's qualities. In this direction, the primacy is to test the assumption of unidimensionality. Confirmatory factor analysis (CFA) was applied to twenty-five math questions. After that, NFI, TLI, CFI, IFI, RFI, GFI, RMSEA, and SRMR were examined for validity and values of goodness of fit. Modification index values of confirmatory factor analysis were examined and then some of the items were omitted because these items gave error in terms of model conformity and conceptualism. That's why,5th grade math subtest is reduced 22 questions, 6th grade math subtest is reduced 19 questions and 7th grade math subtest is reduced 22 questions. The tests are considered to be unidimensionality. When 5th grade math subtest was examined, values were between -0.21 and +0.20. Values in 6th grade math subtest range from -0.21 to +0.30. When 7th grade math test was examined, values were -0.30 and +0.20. Analysis shows that 5th,6th,7th math subtests do not include item bias bases on gender. In the analysis for he subtests, no contrast value, that is suitable for the DIF, could be found. In other words, it couldn't be found any item that shows any different function in terms of gender in subtests.. This result indicates that there is no item that works for the behalf of or to the detriment of students from different gender groups. Key Words: Free Boardıng And Scholarshıp Examınatıon, İtem Bias, Diffirential İtem Function, Winsteps, Rasch Model.