有經驗的訪員常有較高的完訪率，調查機構也珍惜有經驗的訪 員。然而，一些研究發現，經驗跟問卷資料品質並不成正比；甚至在 同一調查內，訪員的完訪經驗逐漸增多之後，其訪問或問卷資料品質 都有下降的趨勢，使得完訪次序與品質成負相關。另一方面，亦有研 究發現，訪員在同一調查內的完訪經驗愈增多，訪問所用的時間愈短， 使完訪次序與訪問速度成正相關。然而，雖然調查專家早已強調訪員 必須放慢速度，但尚未有研究真正觀察過，完訪次序與訪問速度，何 者才是真正影響品質的因素？或兩者都各有其重要性？本文以「台灣 地區社會變遷基本調查」第四期第二次的兩組問卷，利用多層線性模 式（HLM）探討這個議題。訪問品質以受訪態度及非實質回答次數為 指標。研究結果發現，當問卷內容貼近受訪者生活經驗時，訪問速度 及完訪次序對於非實質回答各有顯著獨立的非線性關係；問卷內容過 難時，速度及次序都沒有顯現對於非實質回答的影響。最後，兩問卷 的受訪態度都與次序及速度無關。文末討論研究發現對調查實務的意 義及研究限制。

Experienced interviewers often have higher response rates than novice interviewers and thus are valued by survey organizations. Nonetheless, studies found that experienced interviewers did not necessarily have better data quality. Even worse, both data quality and interview quality became worse as the interviewer gained more experience during the field period. Other research found that later interviews had a quicker pace than earlier interviews. However, even though survey experts have advised interviewers slow down their pace, no research has attempted to examine if it is the interview order or the interview pace, or both, that are related to interview quality. This paper investigates the issue by analyzing the two data files of 2002 Taiwan Social Change Survey with HLM models. I use respondent attitudes when being interviewed and counts of non-substantive answers as two indicators of interview quality. The results suggest that, when the questionnaire contents are closely related to the respondents’ life experience, both interview pace and interview order are independently related to the number of non-substantive answers. Respondent attitudes are not related to either interview order or interview pace. Limitations of the study and implications for survey practice are discussed.