有鑑於媒體或機構通常採用的固定權重大學排名方式難以反映各大學的辦學特色,本研究提出一種結合了保證區域與層級資料包絡分析之模式,對臺灣各大學的教學績效進行評估,研究發現,保證區域資料包絡分析法所獲得之排名結果,能使一些資源有限但致力發展辦學特色的大學,獲得較佳的排名,同時也改善了傳統資料包絡分析法可能產生之排序鑑別力下降及權數分配極端的情形。此外,為了同時解決效率點估計值間差異顯著性及分群判定上之困難,本研究在保證區域資料包絡分析法架構上,利用了層級資料包絡法之分群方式,將教學績效相當的大學透過效率邊界理論進行績效分層。藉用本研究提出之新模式,除了能解決前述傳統資料包絡分析法之困難外,透過分層也能幫助績效較差的大學制定更為合理的績效改善策略。
Due to the limitation of the conventional fixed-weighted university ranking method commonly adopted by medias, which often fails to capture the unique focus of each university, this study proposes a new ranking method, which combines Assurance Region Data Envelopment Analysis (AR/DEA) with Tiered Data Envelopment Analysis (TDEA), called AR/TDEA model. We apply this proposed method to evaluate the teaching performance of universities in Taiwan with student population exceeding 3,000 and staff size exceeding 150. The research findings reveal that the rankings obtained through the proposed AR/TDEA allow some universities to achieve better rankings due to their dedication to developing unique and distinctive academic features, while avoiding problems of extreme weight allocation and low discrimination power in conventional DEA. The proposed model also overcomes the difficulties in statistically identifying efficiency difference and grouping among DMUs. Using the proposed AR/TDEA, this study can stratify a group of universities with similar teaching performance into the same efficiency group. This efficiency group stratification process could also aid lower-performing universities to make a more reasonable plan to improve their performance step by step.