An investigation into the measurement invariance of Akdeniz University International Student Exam (Akdeniz YÖS - 2018)

During this research, measurement invariance of candidates'scores from Akdeniz YÖS-2018 General Aptitude Test and Maths Test was examined depending on the variables of gender, exam language and country. The research was carried out on a data set of 4.310 students'responses for the test items. At first, information about the data set was given using some item statistics, then the convenience of the data set before advanced analyses was investigated via descriptive statistics, normality tests and reliability coefficients. Confirmatory factor analysis (CFA) was employed for each group of variables, then the fit indices gathered were discussed and the findings about the factorial structure were reported. When χ2 values and fit indices were calculated via multi-group confirmatory factor analysis (MGCFA), nested model comparisons (Brown, 2006) were made and measurement (non-)invariance findings were detailed. Regarding the research findings of the General Aptitude Test, the proof of measurement invariance was observed in the gender groups, however measurement invariance findings were not obtained for the other two groups of variables (exam languages and candidates'countries). With respect to the latter two variables, it was understood that the best model fits were the ones of strong factorial invariance, which is also known as Model C. Regarding the Maths Test, measurement invariance was reported for the gender groups and candidates'countries, yet no invariance was observed in the groups of exam languages. In relation to the language groups, it was understood that the best model fits belonged to the weak factorial invariance, which is also known as Model B.