An investigation of measurement invariance for TIMSS 2015 fourth grade mathematics test according to countries

In this study, it is aimed to examine the measurement invariance of the scores obtained from fourth grade mathematics achievement test of TIMSS 2015 among OECD countries. OECD countries included in the research are Germany, America, Australia, Belgium, Czech Republic, Finland, the Netherlands, England, Ireland, Spain, Sweden, Italy, Japan, Canada, Korea, Hungary, Norway, Poland, Portugal, Slovakia, Slovenia, Chile, Turkey and New Zealand. Total number of students from 24 countries participated in the TIMSS 2015 application is 132,226. This research, which was conducted in relational survey model, was conducted on the data of 9,641 students who received the Booklet 7. Within the scope of the research, the main assumptions that the data should have before the measurement invariance analysis were examined and then the confirmatory factor analysis was performed to determine whether the measurement models of the countries were verified. It was concluded that the single factor measurement model of each group was validated on the basis of the fit indices. Then, multi-group confirmatory factor analysis and measurement invariance analyzes were performed. The findings of multi-group confirmatory factor analysis were evaluated by comparing hierarchical models; structural invariance model, weak factorial invariance model, strong factorial invariance model and strict factorial invariance model. The comparison of these four models was performed using nested models. According to the findings when the models are compared, the best working model was the strong factorial invariance model. Based on this finding, it has been concluded that TIMSS 2015 fourth grade math achievement test scores do not have measurement invariance among OECD member countries and therefore; the given countries cannot be compared in terms of students' performances.