| 摘要: | 研究人員進行抽樣調查時，常無法完全避免無反應偏差，尤其是採用郵寄問卷調查時，因為缺乏調查人員與受訪者面對面互動，更增加了提高無反應偏差機會。無反應偏差是指因抽樣設計或實際執行調查時遭遇到某些問題，致使研究人員無法從母體所選定的樣本獲得所需足夠的資訊，因而所產生之偏差。本研究旨在探討郵寄問卷調查無反應偏差改善方法效果。
本文採用Meta分析法，自最近八年在行銷領域採用郵寄問卷調查之我國企管博、碩士論文及部分學術期刊中，抽選62篇進行Meta分析，探討改善無反應偏差方法，驗證採取改善行動與未採取改善行動之無反應偏差是否有顯著差異，檢定採取不同的行動對改善無反應偏差是否有不同之效果。研究結果發現採取改善行動之無反應偏差小於未採取改善行動之無反應偏差。採取二種以上混合方法和採取單一方法的改善效果並沒有顯著差異。
It is difficult to avoid "nonresponse bias" completely in sampling surveys, particularly when the survey conducted by mailed questionnaires. Mail surveys lack face to face interaction between researchers and respondents;it is possible and often increases the possibility of “nonresponse bias".Nonresponse bias is a type of bias, researchers failure to obtain enough information needed from selected samples when some problems rise during sampling plans are designed or a survey is conducted. This paper focuses on the effects of improved methodologies of “nonresponse bias" of mail survey. In order to identify what improved methodologies are used, to test if there exists a significant difference between “nonresponse bias" whereby an improved methodology is used and in scenarios whereby such improved methodology is not incorporated, and to test the effectiveness of methodologies in reduce “nonresponse bias" of mail survey.
A meta analysis method is applied, 62 thesises are selected from marketing related area of recent 8 years' doctorate and master thesis of Business Administration and some academic journal whereby mail survey is used in Taiwan. It has been found that “nonresponse bias" is reduced when improved methodology is applied. Additionally, in reducing “nonresponse bias", there is no significant difference between combining two or more methods and one method is used. |
