Investigating Factors That Affect Turkish Students' Academic Success with Canonical Commonality Analysis According to PISA 2009 Results

The aim of this study was to investigate relations between the factors related to Turkish students' reading abilities and the factors related to facilities that both students and families had, according to PISA 2009 results. The main reason that researchers do not employ canonical correlation analysis to determine relationship between the variables is that interpreting the results are difficult and complex. By calculating unique and common variance associated with variables in each variable sets, canonical commonality analysis helps to determine accurately degree of multicollinearity between the variables, suppressor variable (if there is) and related importance of variables in a canonical model. Thus, it helps researcher make more accurate and reliable interpretation. In this study, predictor variable set consists of factors related to facilities that students and family had and criterion variable set consists of factors related to students' reading abilities. Relationship between variable sets were investigated with canonical commonality analysis. As a result, predictor and criterion variable sets explained 31.7 % of variance in students' academic success. In addition, according to commonality analysis, utilizing information technologies while preparing homework variable was a suppressor and there was a great multicollinearity between facilities that students had at home and socio-economic status of families variables.