臺灣地區的電話調查一向有嚴重的涵蓋率問題，而隨著民眾使用 手機日益普及，傳統以有線電話號碼為抽樣底冊的電話調查涵蓋率也 日益縮小。本研究以2012 年及2013 年臺灣選舉與民主化調查進行比 較分析， 運用「近用性」（accessibility） 為定義的「唯手機族 （cell-phone-only, CPO）」當作基礎，分析後發現「唯手機族」佔臺 灣地區20 歲以上民眾的6.2%。「唯手機族」帶來的估計偏差約為0.93 個百分點，正負方向則要視變數性質與人口特徵與政治態度的關聯而 定。本研究透過具有「黃金標準」的「選舉投票率」與調查結果進行 比對，在考量選後定群追蹤調查及樣本流失的情況下，證實加權處理 會改善傳統電話調查的推估偏差。相較於美國，臺灣應該準備啟動手 機調查的探索性研究，雖然代價偏高、成本效益偏低，但在研究方法 上已有大致共識，未來應針對相關議題擬定合理的解決模式。

The proportion of adults with cell-phone-only (CPO) is growing rapidly. The traditional telephone survey sampling frame has serious coverage bias. Using 2012 and 2013 data from the Taiwan Election and Democratization Survey (TEDS), this article is among the first to reveal that non-coverage can result in non-negligible bias for traditional landline telephone surveys that do not cover cell phones. We selected TEDS responses obtained from cell phone calls, and found that 6.2% of the total population older than 20 years old is the cellphone- only (CPO) group. The estimation bias varies between plus and minus 0.93%, depending on the characteristics of the variables and the population, and their political propensity. This study uses voter turnout rate as the reference for comparison. The results show that weighted estimation improves the bias in the traditional telephone survey. With the results of this paper, we would like to suggest that Taiwan should be ready to launch an exploratory study investigating surveying by cell phone. Although cell phone surveys are still very expensive and not cost-effective, there is consensus among researchers that sampling and weighting methods for cell phone surveys should be further developed.