Scholarlink

The COVID 19 pandemic has accelerated the global shift to online learning in educational institutions. This study investigates the causal effect of online learning on student performance using propensity score matching (PSM) to mitigate confounding variables and estimate the Average Treatment Effect on the Treated (ATT). We analyzed a comprehensive dataset of 10,000 students with complete data across demographic, behavioral, and academic variables. Online learning engagement was measured through a composite score derived from online course completion and educational technology usage, with 4,780 students classified as online learners and 5,220 as traditional learners. The propensity score distributions demonstrated good overlap between groups, satisfying key assumptions for causal inference. Our findings revealed a modest positive effect of online learning, with online students showing slightly higher exam scores (raw difference = 0.25%) and pass rates (84.4% vs 83.0%). Covariate balance was significantly improved post matching, confirming the effectiveness of the PSM methodology. The directed acyclic graph (DAG) modeling approach clarified the causal pathways between teaching mode and student outcomes. These results suggest that online learning, when effectively implemented, can maintain, or slightly improve student performance compared to traditional approaches. This study contributes to the ongoing discourse on online learning effectiveness by applying a rigorous causal inference framework and emphasizing the importance of controlling for confounding variables in educational research.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

The Impact of Online Learning on Student Performance Using a Causal Inference Approach

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Ainur Alam Budi Utomo,
  • Dule Abera,
  • Preetam Kumar,
  • Sissoko Makan,
  • Tarpan Suparman,
  • April Lia Hananto,
  • Seyma Kocak,
  • Anggun Fergina

摘要

The COVID 19 pandemic has accelerated the global shift to online learning in educational institutions. This study investigates the causal effect of online learning on student performance using propensity score matching (PSM) to mitigate confounding variables and estimate the Average Treatment Effect on the Treated (ATT). We analyzed a comprehensive dataset of 10,000 students with complete data across demographic, behavioral, and academic variables. Online learning engagement was measured through a composite score derived from online course completion and educational technology usage, with 4,780 students classified as online learners and 5,220 as traditional learners. The propensity score distributions demonstrated good overlap between groups, satisfying key assumptions for causal inference. Our findings revealed a modest positive effect of online learning, with online students showing slightly higher exam scores (raw difference = 0.25%) and pass rates (84.4% vs 83.0%). Covariate balance was significantly improved post matching, confirming the effectiveness of the PSM methodology. The directed acyclic graph (DAG) modeling approach clarified the causal pathways between teaching mode and student outcomes. These results suggest that online learning, when effectively implemented, can maintain, or slightly improve student performance compared to traditional approaches. This study contributes to the ongoing discourse on online learning effectiveness by applying a rigorous causal inference framework and emphasizing the importance of controlling for confounding variables in educational research.