Keyword: secondary data analysis

ABSTRACT: This study proposes an integrated, evidence-based model to explain the transition from artificial intelligence (AI)-related anxiety to acceptance and learning engagement in the context of Education 4.0. Drawing on a secondary data synthesis of four nationally representative South Korean datasets—KISTEP (2024), NIA (2024), KEEP (2024), and Statistics Korea (2024)—and grounded in Keller’s (1987) ARCS motivational model and Davis’s (1989) Technology Acceptance Model (TAM), the study develops a multi-layered framework linking AI anxiety, AI literacy, AI acceptance, and learning engagement. The analysis tests four hypotheses using a pseudo-SEM approach. Because the study relies on secondary data, proxy indicators are employed to operationalize key constructs, with the Digital Competency Index (DCI) used as a proxy for AI literacy. The findings show that AI anxiety is significantly and negatively associated with AI acceptance, whereas AI literacy exerts a direct positive effect and also weakens the anxiety-acceptance relationship, functioning as a cognitive buffer (moderating effect, β = .23, p < .05). In turn, AI acceptance emerges as a strong predictor of learning engagement (R² = .34). Most notably, this study reframes AI literacy not merely as a technical skill, but as a resilience mechanism that helps reduce technology-related anxiety. The theoretical contribution has direct implications for instructional design and age-sensitive policy interventions. In addition, the pseudo-SEM approach offers a replicable framework for building macro-level theory from national administrative data.