Analysis of predictor variables of PISA 2015 science literacy of students in Turkey by regions

The aim of this study is to test the path model consisting of affective variables that affect the students' science literacy in the evaluation of PISA 2015 and examine the invariance of the model on regions. In addition, science literacy of Affective variables used in the model is the determination of differences by examining the sample of Turkey and the regions on which it is significantly tended. In this research conducted in the relational screenning model, analyses were made on the Turkey sample data set consisting of 5895 people who participated in PISA 2015. Interest in broad science topics, enjoyment of science, science self-efficacy, epistemological beliefs about science and students' science activities, which are variables of investigating various sensorial features related to science literacy in PISA 2015 student questionnaire, were included the model. In the first stage, the models consisting of these variables were separately validated by the confirmatory factor analysis (CFA). In the second stage, path analysis was used to determine the path model, which is defined science self-efficacy, interest in broad science topicsand students' science activitiesas an external variable, epistemological beliefs about scienceas mediator and enjoyment of scienceas internal variable, were validated on the Turkey sample. As a result of the analysis, it was determined that the path model, which is the tool of the epistemological beliefs about science variable, was confirmed on the Turkey sample. All the pathways established between the variables were found to be statistically significant. The epistemological belief variable was semi-mediated in each path that appeared in the model. Multi Group Path Analysis was conducted to examine the invariance of the path model over regions. As a result of the analysis, it was determined that the path model is invariant between regions. In other words, it is possible to say that the relationships and mediating effect between variables in the model on regions are the same. Multiple Linear Regression analysis was used in order to determine the significant predictor of science success in Turkey sample and regions of interest in broad science topics, science self-efficacy, epistemological beliefs about science and students' science activitiesaffective variables in the path model. As a result of the analysis conducted on Region 2 and Region 3 of Turkey sample, it was determined that the variable of interest in broad science topics did not significantly predict science literacy. As a result of the analysis conducted on Region 1, it was found that all variables included in the analysis predicted science literacy. Among science literacy predictor variables, epistemological belief variable was found to be the most important variable on Turkey sample and regions.