AI-based educational recommendation systems and their impact on student retention

Abstract

This doctoral study evaluated the impact of an Artificial Intelligence-based Recommendation System on university student retention using a mixed-methods quasi-experimental design. Results demonstrated that the experimental group using the AI system achieved significantly higher retention rates (94.2% vs 86.3%) and showed 53% greater academic engagement. Mediation analysis revealed that approximately 39% of the positive effect on retention was mediated by increased student engagement. Qualitative findings indicated high perceived usefulness, particularly in personalization and early difficulty detection. The impact was more pronounced among first-generation students and those with low initial self-efficacy, suggesting these technologies' potential as tools for educational equity. The study concludes that these systems are most effective when integrated as complements to a broader student support ecosystem, subordinating technology to pedagogical and ethical objectives.