An Empirical Study on the Correlation of Undergraduate Teaching Evaluation

Abstract

Teaching evaluation serves as a fundamental approach to enhancing pedagogical quality and fostering continuous improvement in higher education. However, the current paradigms of student evaluation of teaching and supervisor evaluation have long been questioned regarding their empirical validity and reliability. Using an extensive dataset comprising five years of teaching records from a university college, this comprehensive study adopts grouped statistics, correlation analysis, and linear regression models to systematically explore the internal relationships among student academic performance, student evaluation results, and supervisor evaluation metrics. The empirical results reveal no significant linear correlation between these three distinct sets of evaluation data. Instead, only minor local differences and complex non-linear characteristics are observed within stratified samples. Additionally, course types and disciplinary attributes exert markedly heterogeneous effects on evaluation outcomes, demonstrating that correlation patterns vary substantially across different academic courses. Fundamentally, these findings suggest that different evaluators hold distinct perspectives and adopt entirely separate observation dimensions when assessing instructional quality. Consequently, the existing evaluation framework fails to align with the core requirements of Outcome-Based Education (OBE), and thus cannot truly reflect the actual effectiveness of classroom teaching. This research critically identifies structural biases within the diversified teaching evaluation system and provides robust empirical evidence for universities to optimize their evaluation mechanisms, ultimately guiding the development of more scientific, equitable, and effective systems for teaching quality assessment.