AI 전환기 대학교육에서 교수 역량이 학업성취에 미치는 영향: LMS 로그데이터를 활용한 다층모형 분석

초록

This study aims to empirically examine the effects of faculty competencies on students' academic achievement amid the structural changes in higher education driven by the proliferation of generative AI. In particular, considering the hierarchical data structure in which students are nested within faculty members, this study investigated how self-reported faculty competencies are structurally linked to actual student academic achievement using hierarchical linear modeling (HLM). The participants included 312 faculty members and 7,416 undergraduate students enrolled in their courses at a four-year university during the second semester of the 2025 academic year. Faculty competencies were measured across five sub-dimensions, and both LMS log data-based teaching behavior indicators and students' grade point averages (GPA) were utilized. Descriptive statistics and correlation analyses were conducted for the key variables, followed by the application of HLM distinguishing between student-level (Level 1) and faculty-level (Level 2) variables to examine the predictive power of faculty competencies. The results indicated that approximately 24% of the total variance in students' academic achievement was attributable to faculty-level differences in the null model (ICC=.240), confirming the appropriateness of applying multilevel modeling. As faculty competency variables and cross-level interaction terms were sequentially introduced, the faculty-level variance decreased and the ICC dropped to .147 in the final model, demonstrating that faculty competency variables significantly explained between-faculty differences in student achievement. Specifically, curriculum design and assessment competency, AI-integrated teaching competency, and instructional management and adaptability significantly predicted students' academic achievement. This study is significant in that it integrates self-reported faculty competencies with LMS log data-based behavioral indicators and applies multilevel modeling reflecting the hierarchical data structure to rigorously examine the effects of faculty competencies. By empirically demonstrating that faculty competencies are structurally linked to actual learning outcomes in the context of the AI transition, this study aims to provide foundational data for faculty development and evidence-based educational policy in higher education.

키워드