德菲法在政策研究領域中是一項廣爲使用的研究工具，該研究方法以專家知識爲基礎，歷經反覆的問卷回饋之後，將複雜的議題系統化以達成專家共識。過去以來均不斷對德菲法技術的修正。其中模糊德菲法即是針可能面臨的訊息遺漏、專家退出與語意模糊等問題進行改進，模糊德菲法在實施程序上亦有多種方式，較常使用的是利用問卷中獲得精確的數值，然後予以模糊化進行分析。但此法恐仍具有訊息遺漏之缺點，同時需先假設專家意見的隸屬函數型態，且計算較爲繁複。本文採用模糊德菲法由專家自行繪製隸屬函數的方式進行分析。以實證角度比較模糊德菲法與德菲法研究結果的差異，期能證明採取模糊德菲法較德菲法爲佳。並利用Mann-Whitney檢定，對相同問項的兩份問卷進行比較。結果顯示採取模糊問卷的專家共識區間較採取傳統問卷所形成的區間收斂。證明在相同議題的德菲法過程中，採取由專家繪製隸屬函數的作法會有較佳之共識程度，同時亦能兼顧語意模糊與較爲完整的訊息保留。

The aim of this research is to compare the different results from Delphi and Fuzzy Delphi method in order to suggest the better one for the following researchers of policy studies. Delphi method is the most widespread research method used in policy studies. This method systemizes complex issues from repeated feedback of questionnaire which is based on experts' knowledge. In the past different researches were conducted to revised the deficiencies of Delphi process, for example, information missing, expert quit, and fuzzy linguistic. Fuzzy Delphi is one of these researches to solve problems of Delphi method. There are different approaches of Fuzzy Delphi process. The most often used one is getting crisp value from questionnaire and analyzing these data by fuzzy algorithms. However, this approach still exists some problems such as information missing, complicate calculation steps, and presumption of the type of membership function. In this research the authors use the membership functions drawn by experts and adopt the Mann-Whitney test to examine the expert consensus both from Delphi and Fuzzy Delphi. The authors find that the convergence range of the Fuzzy Delphi is better than that of Delphi. The result also shows that Fuzzy Delphi can keep more complete information and deal with fuzzy linguistic.