住宅電話與手機雙底冊調查的組合估計：以2016 總統選舉預測為例

Post-Stratified Estimation Procedures for the Dual Frame Telephone Survey in Taiwan: The Case of the 2016 Presidential Election

隨著資通訊技術的進步與人們生活型態的轉變，使得傳統住宅電話調查面臨愈來愈多限制。尤其在資通訊載具（如手機、網路電話及App等）漸趨多元的情況之下，單一調查工具涵蓋率不足的問題逐漸明顯，而如何確保樣本代表性和對母體推估的準確性則成為當代電話調查最嚴峻的挑戰。為補救代表性的問題，過去台灣電話民調實務界主要是使用戶籍資料作為加權的依據。但愈來愈多的研究證實，這樣的方式不僅過度簡化不同人口特徵族群的差異性，且無法解決單一調查工具涵蓋率不足的問題。爰此，本研究提出組合估計的策略，嘗試結合住宅電話與手機的調查資料，並納入不同電話使用族群的權值計算，以改善傳統單一電話調查工具在涵蓋率及樣本代表性的不足。最後，透過2016 年總統大選的電話調查資料進行檢證。結果顯示，在考量「樣本涵蓋率」的前提之下，若從整體估計差距的角度觀察，「住宅電話調查為主，唯手機族資料為輔」應是最佳的權值組合方式，而「完整的住宅電話及手機調查」與「手機調查為主，唯住宅電話族資料為輔」則是次佳的權值組合。換句話說，「最經濟又實惠」的策略就是進行住宅電話調查並搭配手機調查中的唯手機族。如此一來，所得的調查資料將可兼顧主要電話使用族群的特質，又能夠有效地降低調查的成本。

The advancement of information and communication technologies has greatly changed the lifestyle of people while using landline surveys in soliciting precise public opinion is becoming limited. As people use a variety of devices such as cellphones, internet phones and APPs in daily communication, problems of insufficient population coverage arise from relying only on landline phones to reach respondents. Therefore, a daunting task in the telephone polling industry is to ensure sample representation for obtaining precise population parameters. To achieve such an objective, a common practice by pollsters in Taiwan is to use household data as weighting statistics. Many cases, however, have shown this practice to be inappropriate. To solve the above-mentioned problem, this study proposes an estimation method based on a dual frame survey that combines landline phones and cellphones. We further use data from the 2016 presidential election to compare different estimations based on a dual frame survey. Our results demonstrate that a “landline survey supplemented by cellphoneonly” is the best combination, considering sample coverage and estimation error. The second-best alternatives are “cellphone survey supplemented by landline-only” and “use both landline and cellphone.” In other words, “the most economical and efficient” strategy of a dual frame survey is to conduct a traditional landline survey and incorporating cellphone-only respondents. The data collected in such combination not only reflect the characteristics of the population, but also cost much less than other strategies.