An investigation of the changes in PISA reading-mathematics-science literacy scores by using multivariate-multilevel modeling

The aim of this research is to determine the variables at school and student level that affect together the maths-reading-science performances of Turkish students that participated in PISA 2009-2012-2015 practices. Within the scope of this aim, gender, programme type, mother's and father's educational status, cultural opportunities and educational resources at home, opportunities at home, economical-social-cultural state index of the student level variables were used. With respect to school level, size of the school, computer usability index, teacher-student ratio were included in the model. Data on the PISA 2009-2012-2015 Turkish students were analysed by using Multivariate-Multilevel Regression Model. Findings, differences in maths-reading-science scores of the students were observed among schools. The student level variables that predict the maths-reading-science scores of the students in the PISA 2009-2012 Turkey practice together meaningfully were ascertained to be father's educational status, educational resources at home respectively, and mother's educational status, economical-social-cultural state in the PISA 2015 practice. In all three practices, the student level variable that predicts maths and reading performances meaningfully was revealed to be gender. The school level variable that predicts performance of the students in the PISA 2009 Turkey practice together meaningfully was found to be teacher-student ratio. It appeared that for PISA 2012 and PISA 2015, the school level variables that predict meaningfully were computer usability index. Research findings, reducing the said differences in the student success among schools might be proposed. Effort may be put to organise domestic educational environment of students.