從抽樣與統計方法探討網路問卷調查的可行性：比較電話訪談與網路問卷樣本的實質差異性

Abstract

本研究採同一問卷以網頁形式呈現及電話調查的方式同步進行網路人口及使用模式相關資料蒐集，然後再以兩種不同的統計方法對資料進行差異度檢測，以此方式來比較兩種樣本的異同處，並探討利用網際網路做為調查工具的可行性。研究結果發現：⑴兩種抽樣方式所得人口統計結果具有明顯差異，但與歷年來的數據比較，均有相符與不符之處；網路問卷抽樣的人口結構具有相當程度的穩定性，電話問卷的樣本變異性較高。⑵同時運用兩種統計方法來檢測單一人口變項的使用行為模式，在兩種調查模式上的抽樣結果是否有所差異，結果發現差異性並不明顯，所得的結論多半不是證據不足、無法判斷，便是差異度不大、不具實質意義。依此發現，本研究認為，研究的目的如果是強調人口統計結構的話，電話隨機抽樣是必要的調查方式。如果研究的議題著重網路使用行為模式，網路問卷具有低成本的優勢，可以是另一種調查方式的考量。

In this study, we conducted a telephone survey and a paralleling Internet-based one, using the same questionnaire. To examine how different the Internet survey sample was from the telephone survey, we implemented two statistical methods (t-test and equivalence test) simultaneously to analyze and compare the two samples. The findings indicate that the demographic configurations were different across the two sampling modes. We also found that the two samples were different for most variables concerning Internet use and media use patterns. However, the differences were in most cases only statistically significant. As far as practical significance was concerned, the samples of the two surveys did not differ that much. Implications of the findings were dis-cussed. We suggest that Internet as a survey tool has some advantages, and Internet-based survey could be complementary to traditional survey methods in the study on Internet users.