Multilevel Model Analysis to Investigate Predictor Variables in Mathematics Achievement PISA Data

Abstract

This study aims to examine the relationship between predictor variables at the student and school levels and the interaction between variables in predicting mathematics achievement in Indonesia. Stratified analysis was implemented in Indonesia’s Programme for International Student Assessment (PISA) 2018 data. The variables of student level encompassed gender, economic, social, and cultural status (ESCS), metacognition, and learning time. This study revealed that the variables of ESCS, metacognition and learning time possessed a significant positive effect on mathematics achievement. The variables of school level are class size, school type, school size, and student-teacher ratio. This study demonstrated that only the data of class size produced a significant effect on mathematics achievement. Furthermore, the interaction between the learning time and class size also significantly affected learning achievement in mathematics. Therefore, variables increasing students’ mathematics achievement are ESCS, metacognition, learning time, class size, and interaction of learning time and class size.

Keywords

• multilevel modelling
• mathematics achievement
• pisa data
• student level
• school level