A Network Perspective On Factors Affecting Reading Achievement

Abstract

Objective: This study applies network analysis in Progress in International Reading Literacy Study (PIRLS) 2016 data to 1. explore the strongest explanatory variables for the overall reading score among variables from student, family and teacher domain; 2. explore the most impactful nodes in the network; 3. explore the pattern of their interconnections. Method: A total of 310,537 students from 49 countries or regions participated in PIRLS 2016 are included in a regularized partial correlation network analysis using the graphical LASSO algorithm to explore the inter-relations among the variables. The direct association between pairs of variables are revealed via controlling for the influence from all other variables Centrality indices are computed to reveal those variables with highest impact. Results: The strongest explanatory variables for the overall reading score includes 1. Student domain: student’s reading efficacy; 2. Parent domain: children’s book amount at home; 3. Teacher domain: students with reading difficulties in classroom. The variables with highest centrality impact includes students like reading, parent’s literacy activities with children before primary school, teacher’s major in reading-related training. Separate networks of respective domain are further investigated to explore the network pattern. Conclusions: Network analysis not only reveals the most explanatory variables for a single outcome variable, but also offers insight into the most impactful variables among the network, and their potential mediation pathways for future causal hypotheses investigation and inform intervention.