Prediction of mathematics achievement of students attending TIMSS 2015 according to student, family and school variables

This research aims to investigate how student, family and school variables predict math achievement in Turkey, Chile, Germany, Iran, Italy and Korea by using TIMSS 2015 data. Data were obtained from TIMSS 2015 database. For this purpose, the data of 28.025 students from 1,186 schools in the mentioned countries were used in the analysis. In order to answer research problems multiple hierarchical linear regression method conducted besides descriptive statistics. For this analysis, it was observed that the sample sizes were sufficient in all selected countries. Furthermore, in all selected countries, it was observed that the assumptions of normality and multiple linear correlations among independent variables were met to perform hierarchical multiple linear regression analysis. According to the research findings, variables of student, family and school in selected countries explain mathematics achievement variance as a whole between 30.2% and 53.6%. This value is 53.6% for Turkey. In selected countries, the variable group of students explains mathematics variance between 22.3% and 40%. This value is 35.9% for Turkey. In selected countries, the family variable group explains mathematics variance between 7.6% and 16%. This value is 16% for Turkey. In selected countries, the school variable group explains mathematics variance between 0.7% and 5.6%. This value is 1.8% for Turkey. Variable of student "Self-confidence in mathematics class"; variable of family "mother's education level" and the variable of school "Settlements of school" Turkey variables are variables that contributed most to the mathematics achievement. In many studies, "duration of pre-school education" is a significant predictor of academic achievement. Findings of this study reveal that, the duration of participation in pre-school education is a significant predictor in Model I (student variables) and Model II (Family variables). However, when school variables were included in the model, it was not a significant predictor. The duration of participation in preschool education is seen as a significant predictor in all three models in Italy and Korea. Keywords: TIMSS, Mathematics Education, Math Achievement, International Assessments