在教育数字化转型的背景下，提升在线学习质量已成为高等教育的重要命题。本研究聚焦于在线学习投入这一预测学业成功的关键变量，构建了整合“前因诊断—客观验证—机制解释”的双阶段实证体系。研究1通过大样本问卷调查（N=593），揭示了高校学生在线学习投入中行为投入、认知投入、情感投入和交互投入特征，发现个体自身内部因素和同伴互动是驱动学习投入的关键要素。研究2依托高校混合式课程开展纵向追踪（N=32），证实学习者认知投入与平台学习日志高度吻合，在预测学业成绩方面，认知投入具有最强的解释力（R²=0.272），认知投入在在线的学习时长与学业成绩之间起着完全中介作用。研究揭示评价在线效果的误区，指出在线学习行为必须转化为深层认知加工方能真正提升学业成效。研究结果为多模态学习分析、促进学习者的高阶认知的课程设计及优化协同交互机制提供了理论支撑与实践指导。

Against the backdrop of digital transformation in education, enhancing the quality of online learning has become a significant proposition in higher education. This study focuses on online learning engagement, a key predictor of academic success, and constructs a two-stage empirical framework integrating "antecedent diagnosis—objective verification—mechanism explanation." Study 1, based on a large-sample questionnaire survey (N=593), revealed the characteristics of behavioral engagement, cognitive engagement, emotional engagement, and interactive engagement in university students' online learning, finding that individual internal factors and peer interaction are key drivers of learning engagement. Study 2, conducted through longitudinal tracking (N=32) in a blended university course, confirmed that learners' cognitive engagement is highly consistent with platform learning logs. In predicting academic performance, cognitive engagement demonstrated the strongest explanatory power (R²=0.272), and it played a fully mediating role between online learning duration and academic achievement. The study reveals the misconception in evaluating online effectiveness, pointing out that online learning behaviors must be transformed into deep cognitive processing to truly enhance academic outcomes. The findings provide theoretical support and practical guidance for multimodal learning analytics, course design that promotes learners' higher-order cognition, and the optimization of collaborative interaction mechanisms.