2014年4月

手機使用對臺灣地區電話調查涵蓋率之影響評估
An Evaluation of the Effect of Cellphone-only Users on Telephone Survey Coverage Rates in Taiwan

研究論文 Article

洪永泰, 洪百薰, 林宇璇, 呂孟穎, 許勝懋, 吳淑惠, 卓仲彥, 徐書儀
Yung-Tai Hung, Baai-Shyun Hurng, Yu-Hsuan Lin, Ming-Ing Lu, Shen-Mao Hsu, Shu-Hui Wu, Chung-Yen Cho, Shu-Yi Hsu

7-30

臺灣地區的電話調查一向有嚴重的涵蓋率問題，不僅如此，隨著民眾使用手機日愈普及，傳統以有線電話號碼為抽樣底冊的電話調查涵蓋率也日愈縮小。這不但在面臨眾多困境的電話調查問題上雪上加霜，也使得各種推論工作的成效受到嚴重考驗。本研究以2009-10年國民健康調查、2009 年健康危害行為監測電話調查及2009年成人吸菸行為電話調查三項大致上在同時間完成的調查資料進行比較分析，藉由面訪受訪者的四種電話使用分類方式（有有線電話有手機，有有線電話無手機，無有線電話有手機，無有線電話無手機）來推估臺灣地區電話調查因涵蓋率不足而可能造成的推論偏差。分析後發現「唯手機族（cell phone only）」佔了臺灣地區18 歲以上民眾的9.3％。這對一般電話調查的涵蓋率確實有影響，但是還沒有到嚴重的地步。透過吸菸率調查結果的比對，本研究證實了加權處理會改善傳統電話調查的推估偏差，不加權一定更糟。國民健康調查面訪案在對全臺灣地區進行推估時除了依照0縣市人口比例併檔外並不需要加權，至於其他的面訪案則需要投入更多的後續研究、納入多種面訪案後才能得到較穩定的評估成果。

There have been serious coverage problems in telephone surveys in Taiwan. The situation is getting worse because of the increasing usage of cellphones, hence raising public doubt about the inferences made from traditional telephone surveys. This study uses three largescale survey databases to investigate the issues involved, namely, the2009–10 National Health Interview Survey (NHIS), the 2009 Behavior Risk Factor Surveillance System Telephone Survey, and the 2009 Adult Smoking Behavior Telephone Survey. Four types of telephone usage, i.e., line phone with cellphone, line phone without cellphone, cellphone-only, and neither line phone nor cellphone, assessed from the NHIS face-to-face interviews, are used for the estimations of inference bias caused by the coverage deficiency of traditional telephone interviews. The results show that among adults aged 18 and above, 9.3% of the sample interviewed belonged to the cellphone-only group. This does show some effect on traditional telephone survey coverage, but the problem is not serious. The study uses adult smoking rate as an indicator to demonstrate the fact that a weighting scheme is a necessity to correct the bias caused by coverage and nonresponse problems for traditional landline telephone surveys. Further studies from large-scale face-to-face surveys are needed in order to reach more concrete research conclusions.

電話調查, 涵蓋率, 手機使用者, 健康行為指標
telephone survey, coverage rate, cellphone-only, health behavior indicators

社會學
Sociology

https://doi.org/10.7014/SRMA.2014040001

全文下載