The Impacts of Student-Level and School-Level Factors on Students’ Reading Achievement: An Application of Hierarchical Linear Modeling

Abstract

Reading achievement of students is one of the most significant predictors of their academic performance and competitiveness in society. Researchers have been investigating the reading achievement related factors for decades from different aspects. This study aimed to examine relevant factors that are associated with reading achievement from both the student and school levels. The student-level factors included English language proficiency (ELL), students’ reading motivation, and students’ home resources. The school-level factors included school SES, teacher’s characteristics, school literacy readiness, and grade-level reading proficiency. A large-scale data set of 3,001 fourth-grade students from 133 elementary schools in the United States was included in the current study. The dataset was part of the Progress in International Reading Literacy Study (PIRLS) 2016 international public database. Considering the nested structure of the dataset, Hierarchical Linear Modeling (HLM) was utilized to analyze the impacts of student- and school-level predictors on the fourth-grade students’ reading achievement simultaneously. The results indicated that the factors at student level were all significantly predicting reading achievement. The inclusion of student-level predictors reduced 8.3% of the total variance in reading achievement. The results also showed that school SES and grade-level reading proficiency were significant predictors of reading achievement at school level, whereas the later predictor demonstrated extremely weak predictive capacity in the prediction. The variations in the intercepts of different schools were explained by school-level factors and the vast majority of the 13% of the total variance in reading achievement was accounted for by school SES. In addition, the full model with both student- and school-level predictors made significant improvement compared to the unconditional model and student-level model by providing the best model-fit. The educational implications for improving reading achievement and limitations were discussed as well.