MULTILEVEL REGRESSION WITH MAXIMUM LIKELIHOOD AND RESTRICTED MAXIMUM LIKELIHOOD METHOD IN ANALYZING INDONESIAN READING LITERACY SCORES

Abstract

The multilevel regression model is a development of the linear regression model that can be used to analyze data that has a hierarchical structure. The problem with this data structure is that individuals in the same group tend to have the same characteristics, so the observations at lower levels are not independent. Education research often produces a hierarchical structure, one of which is PISA data, where students as level-1 nested within schools as level-2. In the PISA 2018 survey, reading literacy is the main focus. The data are sourced from the Organisation for Economic Co-operation and Development (OECD). The survey results show that the reading literacy scores of Indonesian students have decreased, thus placing Indonesia at 74th out of 79 countries. However, it is still very rare to research the reading literacy of Indonesian students' using a multilevel regression model. This study aims to apply a multilevel regression model to determine the factors influencing Indonesian reading literacy scores in PISA 2018 survey data. The results of this study indicate that the factors that influence response variable are gender, grade level, mother's education, facilities at home, age at school entry, student discipline behavior at school, and failing grade, while at the school level are the type of school and school location. The magnitude variance of student reading literacy scores can be explained by the explanatory variables the student level is 11,42% and the school level is 60,66%, while the rest is explained by another factor outside the study.