Family and child characteristics in reading achievement milestones using machine-learning-based survival analysis

Abstract

Objective: This study aimed to identify early reading achievers and uncover family and child factors that mitigate reading skill disparities. Background: Literacy standards guide educational policy to prevent literacy issues in at-risk children. Many studies lack accurate methods to measure reading milestones, relying on static approaches that miss dynamic longitudinal processes. Method: This study used machine-learning-based survival analysis on Early Childhood Longitudinal Study, Kindergarten Class of 2010–2011 (ECLS-K: 2011) data to analyze children's time to reach reading milestones, examining how family structure, socioeconomic status, gender, and behavioral problems relate to reading achievements. Results: Being female, from a higher-income family, and not exhibiting behavioral problems increased the likelihood of surpassing reading milestones. Higher socioeconomic status had a stronger positive relation with reading achievement in two-parent families. Externalizing behaviors had a stronger negative relation with reading achievement in girls than boys. The survival tree analysis showed children from two-parent families with incomes at or above 200% of the poverty threshold reached reading milestones earlier. Among these children, those with lower externalizing behaviors achieved them the earliest. Conclusion: This study supports the family systems theory and the bioecological model, indicating family and child factors, and their interplay, relate to children's reading achievement. Implications: Machine-learning-based survival analysis enhances the assessment of reading milestones, facilitating early diagnosis, targeted interventions, and effective family policies.