Investigation of measurement invariance in groups created according to the variables predicting PISA 2018 reading skills

The results of PISA, one of the international education researches, are used to make comparisons in different subgroups. In order for the comparisons to be meaningful, it must be proven that the applied measurement tool measures the same structure in the relevant subgroups. The purpose of this research is to examine the measurement invariance in the subgroups created for three variables (reliability assessment, summarization, economic-social-cultural situation) that explain the success of reading skills and are important for PISA in the PISA 2018 cycle. The sample of the study consists of the responses of 6890 students from Turkey who participated in the PISA 2018 cycle to the student survey and 5 units in the reading skills basic test. After it was determined that the data met the assumptions, a multi-group confirmatory factor analysis (MGCFA) was performed through the R software and measurement invariance was tested step by step. In addition to evaluating the scaled fit coefficients obtained at each stage, the comparison of the stages was made using the difference in scaled CFI and scaled TLI fit indices. It was concluded that out of 15 analysis results obtained with 5 units and 3 variables, 4 of them remained at the structural invariance stage, 10 of them remained at the metric invariance stage, and the fit coefficients could not be obtained because the model did not converge in 1 unit.