訪員如何影響訪問績效？一個互動式的觀點

Discovering the Characteristics of a Good Interviewer: Analyzing the Performance of Interviewers from Their Interaction with Respondents and the Survey Environment

訪問品質一直是調查研究者所關心的重要問題，如何維護訪問品質，降低拒訪率與降低問卷無反應率都是研究者努力的方向。以面對面訪問所進行的調查研究來說，最重要的三個元素莫過於訪員、受訪者與情境，因此這三者的特徵對於調查研究的影響正是本研究所關心的焦點之一。本研究結合上述兩類行為者特徵，再加上訪問區的情境因素，並且從互動式的觀點，探討這三者間的互動如何影響訪問的成功與否以及訪問內容的完整性。透過多層模型的建立與分析， 本研究發現相對於沒有做過任何訪問的訪員而言，有做過TEDS（Taiwan’s Election and Democratization Study, TEDS）的訪員比較能取得受訪者的同意受訪，尤其是都市化程度愈高的訪區差異愈大。另外，有做過TEDS訪問的訪員比沒有做過訪問的訪員，在低都市化地區與面對低教育程度受訪者時，問卷項目無反應率也較低。訪員的性別因素也是影響訪問績效的重要因素。相對於女性訪員，男性訪員顯然更容易被拒訪，尤其是在面對女受訪者時。但是，訪員的性別差異並沒有反映在他們的問卷完整性上。在其他的因素方面，本研究也發現受訪者的年齡愈高愈可能拒訪，且問卷無反應率也越高。男性受訪者比女性受訪者更傾向接受訪問，在問卷無反應率方面也比女性受訪者來得低。本研究所關注的訪區因素方面，都市化程度愈低的訪區，受訪者比較傾向接受訪問，但是他們提供有效答案的比例也顯著較都市化程度較高地區的受訪者低。另一個訪區因素，即訪區的犯罪率則沒有造成任何的顯著影響。本研究認為，倘若研究者愈能夠掌握訪員、受訪者與訪問區域的元素，且愈了解這些元素之間彼此的互動，則愈有利於調查研究上人力的安排。

The quality of survey research is the researcher’s main concern and he always tries to reduce the refusal rate and item nonresponse rate. The interviewer, respondent and survey environment are the most important factors affecting the result of a face-to-face interview. This article analyzes the characteristics of these factors and their interaction in deciding the result of a face-to-face interview. Using the multilevel model, the analysis reveals that the interviewer who had TEDS interview experience was significantly more successful in persuading the respondent to accept the interview, compared to those who had no experience at all. The difference is more significant when the respondent is coming from a highly-urbanized area. The experienced interviewer has a lower item nonresponse rate than one without experience, when the respondent has less education or the respondent is from an area with a low degree of urbanization. Compared to the female interviewers, the male interviewers are more likely to be refused, especially when they meet female respondents. Besides, male respondents have both a lower unit nonresponse rate and item nonresponse rate. A respondent from a highly urbanized area is more likely to refuse the interview, but a respondent from a less urbanized area has a higher item nonresponse rate. Finally, this paper suggests that those researchers who engage in survey research should obtain as much information about the interviewer, respondent and environment as they can.