本研究之目的在排序學生評鑑教師教學的分數，以做為評鑑教師教學能力的參考。透過貝氏多層次隨機效果模型估計「學生評鑑教師教學量表」各個題項的隨機效果，藉以控制量表各個題項的外顯分數，以及信效度跨教師之間的測量不恆等。經由分析發現，學生評鑑教師教學的外顯分數以及信效度確實存在測量不恆等，並不適合直接將學生評鑑教師教學的分數拿來進行不同教師之間的比較，以免產生評鑑不公平的現象。後續，在控制題項隨機效果的前題下，透過貝氏估計插補出每位教師在學生評鑑教師教學的潛在能力值，以做為學生評鑑教師教學分數排序的依據。

The aim of this study is to rank teachers' teaching scores based on evaluations from students. A Bayesian multilevel random effect model is used to estimate the random effect of each item in the evaluation, controlling for explicit scores and measurement inconsistencies in reliability and validity between teachers. There is evidence, based on a random variation model that the scores of student evaluations of teacher teaching, as well as their reliability and validity, vary across teachers. To avoid unfair evaluations that could negatively impact teacher promotion, evaluation, or rewards, it is important to consider this phenomenon. The plausible values of each teacher in the student evaluation of teacher teaching are interpolated through Bayesian estimation, under the premise of controlling the random variation of the measurement items. These values are then used as the basis for the real evaluation of teachers' teaching.