How Metacognition Shapes Higher Education Adoption of LLMs: A Structural Equation Modeling Approach

Large Language Models (LLMs) are transforming education by facilitating adaptive learning experiences, yet their societal acceptance remains vital for successful integration. This study investigates the relationship between metacognition and LLM acceptance, with academic burnout as a mediating factor. Using Structural Equation Modeling (SEM), we analyzed validated psychometric data from 178 computer science students, drawn from a published dataset. Results show that metacognition significantly enhances LLM acceptance (β² = 0.220) and reduces academic burnout (β² = 0.038). While burnout positively predicts LLM acceptance (β² = 0.067), its indirect effect was negative, revealing a suppressor effect: students using LLMs metacognitively tend to experience less burnout, yet the beneficial impact on acceptance stems from proactive, strategic engagement—not from stress relief.