多元民意調查方法的比較研究

台灣各界多以住宅電話調查方式反映民意，但手機與各種網路通訊工具普及，只使用手機的人口越來越多，以住宅電話號碼為抽樣底冊的調查涵蓋率日益縮小，加上拒訪率上升、較難接觸到年輕族群的意見等影響，使得住宅電話調查的推論結果受到嚴峻考驗。本研究之目的，即是希望透過比較不同調查方法所蒐集之民意調查結果，了解不同調查方法在涵蓋率與可觸及調查對象的模式效應，以及如何有效反應以戶籍資料加權造成的誤差問題，為達上述目的，同時使用住宅電話、手機、網路等調查工具進行資料蒐集，以四年期調查資料，證實了不同的調查方法會導致不同的樣本結構，進而得到不同的調查結果。未來民意調查應考慮不同調查方法所涵蓋母體的範圍差異問題，結合不同調查方式以提高涵蓋率，否則隨著使用手機的民眾比例逐年上升，單以住宅電話調查結果做為民調依據的偏頗將越來越大，本研究並具體提出以戶籍資料為加權依據及組合估計的建議，供未來調查參考。

關鍵字

住宅電話調查；手機訪問調查；網路調查；雙底冊調查之組合估計

並列摘要

Responsiveness is now a major concern of various normative and empirical theories of democracy. In response to citizens' need to have a more responsive and democratic civil administration, it is important for public policy makers to obtain public opinions during the policy making process. However, the rise of the Internet, various information and communication technologies (ICTs), and social media has brought fundamental changes in the way society works. Consequently, traditional public opinion surveys are facing tremendous challenges, such as low response rates and difficulty in reaching younger generations. This research collects data from landlines, cell phones, and internet surveys for four years to compare different survey methods. The results show that the telephone survey has better access to respondents who are older, the mobile phone survey has better access to male respondents, and the Internet survey has more access to respondents under 40 years old. In addition, the data have shown that the population that each survey method comes into contact with differs greatly from one method to another. For future policy-related surveys, a post-stratified weighting procedure is suggested to integrate samples from various survey methods, so that the samples contacted by the surveys conform more closely to the true populations.