以入選機率調整法修正調查推估偏差的成效評估

A Study of Survey Nonresponse Bias Using Propensity Score Adjustment

10.6166/TJPS.41(151-175)

杜素豪(Su-Hao Tu)；羅婉云(Wan-Yun Lo)；洪永泰(Yung-Tai Hung)

不完整資料 ； 入選機率 ； 入選機率調整法 ； 投票行為 ； Incomplete Data ； Propensity Scores ； Propensity Score Adjustment ； Voting Behavior

政治科學論叢

41期（2009 / 09 / 01）

151 - 175

繁體中文

抽樣調查資料因樣本代表性失真而造成對母體推估偏差的補救辦法通常是採用加權處理，基本上以社會人口特徵爲依據。然而僅從樣本人口特徵的分布是否和母體相符來判斷樣本的代表性並不能保證樣本在認知、態度與行爲等主題變項在分布上的推論就不會有偏差。本文以總統選舉投票行爲的調查資料爲例，探討依據入選機率的次樣本分組(subclassification on the propensity score)所調整的電話調查結果在投票行爲推估方面的成效。首先利用二○○四年台灣地區社會變遷基本調查四期五次公民權組問卷資料，以樣本重抽法(bootstrapping)產生包含20,000案的擬母體(Pseudo-population)，再從擬母體中以簡單隨機抽樣法抽取200案爲參考樣本，另外從一個典型的電話訪問調查資料中以分層隨機法抽出800案爲試驗樣本。兩套樣本組合成爲一套1,000案的新樣本。其次依據Lee (2006)的入選機率調整法(propensity scoreadjustment, PSA)進行電訪樣本「投票行爲」估計值的調整。整個流程進行2,000次的模擬分析。評估結果確認藉由次樣本分組所產生的入選機率調整法的確有降低推估偏差的功效。

It is common to use socio-demographic variables for weighting, but this does not mean that other inferences, especially on attitudes and behavior variables, will be free of bias. This article takes voting behavior in a presidential election as an example to examine the effects of adjusting the telephone survey results using sub-classifications on the propensity score as suggested by Lee (2006). We first used the Taiwan Social Change Survey (TSCS, 2004) wave 4 module for a citizenship study to produce a pseudo-population dataset based on bootstrapping sampling. A random sample of 200 cases was drawn from this dataset. In addition, a stratified random sample of 800 cases was drawn from a telephone survey. The two samples were combined into a sample of 1,000 cases. Propensity Score Adjustment (PSA) allows us to adjust and evaluate the proper estimation of voting behavior based on the sub-classification of propensity score. The procedure was repeated 2,000 times. The results showed that the PSA method does effectively reduce telephone survey estimate bias.

1. 洪永泰(2006)。台灣地區抽樣調查各種母體定義、抽樣底冊、和涵蓋率的比較。調查研究，18，9-44。
   連結：
2. 陳肇男(2001)。大型電訪之省思。調查研究，10，121-138。
   連結：
3. 黃紀、張佑宗(2003)。樣本代表性檢定與最小差異加權：以2001年台灣選舉與民主化調查爲例。選舉研究，10(2)，1-35。
   連結：
4. 劉從葦、陳光輝(2005)。Is Weighting a Routine or Something that Needs to be Justified?。選舉研究，12(2)，149-187。
   連結：
5. 羅健銘、陳素秋、賴允亮、林家瑾、陳建仁(2007)。住院癌末病患照護型態對住院醫療費用與住院天數之影響。台灣衛誌，26(4)，270-282。
   連結：
6. Just Weight! The Case for Dynamic Party Identification Weighting
7. Brehm, J.(1993).The Phantom Respondents: Opinion Surveys and Political Representation.Ann Arbor:University of Michigan Press.
8. Cochran, W. G.(1968).The Effectiveness of Adjustment by Subclassification in Removing Bias in Observational Studies.Biometrics,24(2),295-313.
9. Colasanto, Diane,Jay A. Mattlin.(1987).Evaluation of Gallup's Methodology for Predicting Likelihood of Voting.Paper prepared at the 1987 Joint Statistical Meeting,San Francisco:
10. Czajka, John L.,Sharon M. Hirabayashi,Roderick J. A. Little,Donald B. Rubin.(1992).Projecting from Advance Data Using Propensity Modeling: An Application to Income and Tax Statistics.Journal of Business & Economic Statistics,10(2),117-131.
11. D''Agostino,Jr. Ralph B.(1998).Tutorial in Biostatistics: Propensity Score Methods for Bias Reduction in the Comparison of a Treatment to a Non-randomized Control Group.Statistics in Medicine,17,2265-2281.
12. Dever, Jill A.,Ann Rafferty,Richard Valliant.(2008).Internet Surveys: Can Statistical Adjustments Eliminate Coverage Bias?.Survey Research Methods,2(2),47-62.
13. Dimock, Michael,Scott Keeter,Mark Schulman and Carolyn Miller.(2001).A Voter Validation Experiment: Screening for Likely Voters in Pre-election Surveys.Paper prepared at the 56th Annual American Association for Public Opinion Research Conference,Montreal, Quebec, Canada.:
14. Drake, Christiana.(1993).Effects of Misspecification of the Propensity Score on Estimators of Treatment Effect.Biometrics,49,1231-1236.
15. Duncan, Kristin B.,Elizabeth A. Stasny.(2001).Using Propensity Scores to Control Coverages Bias in Telephone Surveys.Survey Methodology,27(2),121-130.
16. Freedman, Paul,Ken Goldstein.(1996).Building a Probable Electorate from Pre-election Polls: A Two-State Approach..Public Opinion Quarterly,60,574-587.
17. Judkins, David R.,David Morganstein,Paul Zador,Andrea Piesse,Brandon Barrett,Pushpal Mukhopadhyay.(2007).Variable Selection and Raking in Propensity Scoring.Statistics in Medicine,26,1022-1033.
18. Kalton, Graham,Ismael Flores-Cervantes.(2003).Weighting Methods.Journal of Official Statistics,19(2),81-97.
19. Lee, Sunghee.(2006).Propensity Score Adjustment as a Weighting Scheme for Volunteer Panel Web Surveys.Journal of Official Statistics,22(2),329-349.
20. Lee, Sunghee,Richard Valliant.(2009).Estimation for Volunteer Panel Web Surveys Using Propensity Score Adjustment and Calibration Adjustment.Sociological Methods & Research,37(3),319-343.
21. Lee, Sunghee,Richard Valliant.,James M. (eds.) Lepkowske(2008).Advances in Telephone Survey Methodology.New York:John Wiley & Sons..
22. Loosveldt, Geert,Nathalie Sonck.(2008).An Evaluation of the Weighting Procedures for an Online Access Panel Sucvey.Survey Research Methods,2(2),93-105.
23. Rosenbaum, P. R.,Brian S. Everitt (eds.) David C. Howell. (eds.)(2005).Volumn 3 Encyclopedia of Statistics in Behavioral Science.New York:John Wiley & Sons..
24. Rosenbaum, P. R.,D. B. Rubin.(1983).The Central Role of the Propensity Score in Observational Studies for Causal Effects.Biometrika,70(1),41-55.
25. Rosenbaum, P. R.,D. B. Rubin.(1984).Reducing Bias in Observational Studies Using Subclassification on the Propensity Score.Journal of the American Statistical Assocation,79(387),516-524.
26. Schonlau, Matthias,Arthur van Soest,Arie Kapteyn.(2007).Are 'Webographic' or Attitudinal Questions Useful for Adjusting Estimates from Web Surveys Using Propensity Scoring.Survey Research Methods,1(3),155-163.
27. Schonlau, Matthias,Kinga Zapert,Lisa Payne Simon,Katherine Haynes Sanstad,Sue M. Marcus,John Adams,Hongjun Kan,Rachel Turner,Sandra H. Berry.(2004).A Comparison between Responses from a Propensity-Weighted Web Survey and Identical RDD Survey.Social Science Computer Review,22,128-138.
28. Strümer, Til,Sebastian Schneeweiss,M. Alan Brookhart,Kenneth J. Rothman,Jerry Avorn,Robert J. Glynn.(2005).Analytic Strategies to Adjust Confounding Using Exposure Propensity Scores and Disease Risk Scores: Nonsteroidal Antiinflamrnatory Drugs and Short-term Mortality in the Elderly.American Journal of Epidemiology,161(9),891-898.
29. Traugott, Michael,Clyde Tucker.(1984).Strategies for Predicting Whether a Citizen Will Vote and Estimation of Electoral Outcomes.Public Opinion Quarterly,48,330-343.
30. Zanutto, Elaine L.(2006).A Comparison of Propensity Score and Linear Regression Analysis of Complex Survey Data.Journal of Data Science,4,67-91.
31. 內政統計年報－人口年齡分配、十五歲以上人口教育程度
32. 吳統雄(1984)。電話調查：理論與方法。台北:聯經。
33. 吳齊殷(1996)。戶中抽樣與否對樣本代表性的影響：以大台北地區電話訪問爲例。調查研究，1，39-65。
34. 呂金河(1999)。依號碼尾數修正的戶中選樣法。中國統計學報，37(3)，257-278。
35. 李隆安、韓成業(2008)。The Study of Weights by the Raking Method。「2008年度中國統計學社學術研討會」論文，台北:
36. 林佳瑩、陳信木(1996)。各種電話號碼抽樣方式之比較分析。調查研究，2，111-141。
37. 林彩玉、洪永泰、鄭宇庭(2004)。調查研究中『無反應』選項問題之分析。調查研究，15，31-60。
38. 洪永泰(2006)。台灣地區電話訪問調查資料加權處理之探討。「第六屆調查研究方法與應用國際學術研討會」論文，台北:
39. 洪永泰(2007)。台灣地區選前電訪民意調查推估偏差的補救。「政治思想、政黨與選舉行爲－紀念謝延庚教授學術研討會」論文，台北:
40. 洪永泰(1996)。抽樣調查中樣本代表性的問題。調查研究，1，7-37。
41. 洪永泰(1996)。戶中選樣之研究。台北:時英。
42. 洪永泰(1989)。抽樣調查中訪問失敗問題的處理。社會科學論叢，37，33-52。
43. 洪永泰(2000)。調查時間、抽樣方法與樣本代表性的實務探討。「國內調查研究之抽樣設計：理論與實務研討會」論文，台北:
44. 隋杜卿(1986)。問卷調查中的樣本代表性研究。思與言，23(6)，72-86。
45. 黃毅志(1997)。抽樣調查中訪問失敗的問題之探討：以台灣地區社會變遷調查爲例作說明。調查研究，4，113-129。
46. 戴立安(1997)。對國內電話調查之多段隨機撥號抽樣方式的初析。調查研究，4，131-198。
47. 蘇建州(1998)。穩健電話調查程序之初探。調查研究，6，119-143。
48. 蘇蘅(1986)。傳播研究調查法。台北:三民。

1. 陳鴻嘉,柯秀憓(2019)。眾裡尋「輕」千百度：電訪戶中選Young之調整。調查研究-方法與應用，42，7-81。
2. 黃東益、張鐙文、洪永泰(2017)。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。選舉研究，24(2)，65-96。
3. 涂志揚、俞振華(2017)。探討以電訪資料及「入選機率調整法」修正網路調查偏誤的可行性。政治科學論叢，73，81-126。
4. (2015)。比較入選機率分組與其他加權方法對電話調查樣本的調整：上網率的推估。臺灣社會學刊，56，115-150。