Learning Analytics, Academic Resilience, and Student Retention in Online Higher Education: A Comparative Study of Predictive Intervention and Human-Centered Advising Models

Abstract

The expansion of online and blended higher education has intensified institutional concern regarding student retention, academic disengagement, and educational inequality. In response, universities increasingly deploy learning analytics systems to predict student risk and guide intervention strategies. However, current scholarship remains divided concerning whether predictive analytics strengthens meaningful student support or reinforces technocratic models of educational governance. This article comparatively examines two institutional intervention models within online higher education: a predictive analytics-centered intervention system and a human-centered advising model integrating learning analytics with relational academic support. Using a comparative mixed-methods design informed by learning sciences, self-regulated learning theory, and student persistence scholarship, the study analyzes learning management system data, institutional advising records, student engagement indicators, online classroom observations, retention statistics, and policy documents collected between 2022 and 2025. The findings demonstrate that predictive systems improve identification of disengagement patterns but produce stronger educational outcomes when combined with relational advising, reflective feedback, and adaptive instructional support. Universities relying primarily on automated risk classification achieved short-term efficiency gains but experienced weaker indicators of student belonging and sustained participation. By contrast, human-centered advising environments demonstrated stronger academic resilience, retention stability, and learner self-regulation despite lower levels of automation. The study argues that learning analytics should function as pedagogical support infrastructure rather than institutional surveillance architecture. This article contributes to learning sciences scholarship by developing a comparative framework linking learning analytics, relational advising, self-regulated learning, and educational resilience in online higher education.