The Application of Fuzzy Statistics in Social Survey

在問卷設計上，一般人常用傳統勾選的方式讓受訪者作答。這種0或1的二分法方式，讓受訪者在數個選項中勾選出一個較適當之答案，但其勾選的並非絕對，而其無勾選的卻非不曾發生的，如此所得到的資料往往會扭曲事實的真相。 本論文即是利用模糊之概念來改良傳統之問卷，允許受訪者在數個選項中填寫百分比，解決上述傳統問卷之缺失，並以家庭關係問卷為例，來比較傳統及模糊問卷結果的差異性。更進一步建構一家庭關係模糊指標及定義一家庭關係警訊燈號，用來衡量一般人家庭關係之優劣。分析結果我們可以得到以下三點結論：1.傳統問卷之結果，往往容易集中在某個選項上；而模糊呈現的較為均衡。2.由於模糊問卷答題方式較為多樣化，故家庭關係模糊指標分布較為廣泛且均勻。3.整體而言，較多數人之家庭關係警訊燈號為綠燈，顯示家庭關係不錯。 In survey question design, people used the method of traditional selection to let respondent to answer. This method let respondent to select an appropriate answer in several options. However, his selections would not be absolute, and the options that he did not select would happen. Such data may skew the reality. In this paper, we make use of fuzzy concept which allow respondent to write down percentage in several options to improve traditional survey question and to resolve the shortcoming of traditional survey. Also, we take family relation survey question for instance to compare the difference of traditional and fuzzy survey question result. Furthermore, we build a family relation fuzzy index and define a family relation warning signal to judge the good and the bad of one's family relation