學生認知歷程與背景變數對於學生評鑑教師的影響：潛在類別偏差校正與混合迴歸分析

Influences of Different Background Variables on Student Ratings of Instruction: Bias-Adjusted Three-Step Mixture Regression Analysis

在潛在類別模型中加入共變項時，如果沒有經過偏差校正，共變項與潛在類別之間的估計參數將產生偏誤。基於此，本研究在探討學生層次變項對學生評鑑教師教學影響時，除了加入與學生學習有關的認知歷程變項外，並進行潛在類別偏差校正與混合迴歸分析。研究對象為東部某大學大學部學生，總樣本數為6,111人。研究發現，學生學習的認知歷程改變可以分為五個潛在類別群組，當學生在該科目所保留或遷移的認知歷程最多時，給教師的分數最高。此外，當學生認知歷程存在潛在異質差異時，在學生層次不同背景變項上對學生評鑑教師教學的影響不同。針對上述結果，本研究對學生評鑑教師教學的議題發展及模型建構提出相關建議。

This study used a bias-adjusted three-step mixture regression model to evaluate the influences of students’ cognitive process on their ratings of instruction. Data were collected from 6,111 students enrolled at a university in Taiwan. The results indicated that students’ gender, year in the university, course, department, and learning interest had a significant impact on student ratings, and students’ cognitive process demonstrated a moderating effect. Furthermore, the implications of these findings for student ratings policies and theirs effects on university faculty and students are discussed.