民意調查是一種對個人的認知、態度與行為的主觀測量，在政治與公共政策上的應用非常廣泛；然而，對於敏感性議題（sensitive issues）或社會道德議題（moral issue）的測量，民意調查的主觀測量是否仍具有效度一直是備受質疑；再者，媒體報導是影響民眾認知的主要因素，而民意認知與媒體報導之間的一致性迄今仍無定論；最後，政府統計資訊在賄選、治安等方面常有「黑數」，與民眾的主觀認知或媒體報導之間往往出現極大的落差，常引發是否可信的爭論。就議題屬性而言，賄選實屬於敏感性和社會道德議題，多元的資訊來源之間究竟是一致或分歧，在理論與實務上均屬值得探討的議題。因之，本研究以2010 年大臺南市三合一選舉為分析案例，以鄉鎮市區為分析單位，透過民意調查與台灣反賄選文獻之檢視，結合問卷調查、內容分析、次級資料分析等方法，以三角驗證法（triangulation）交叉檢證三合一選舉的民意調查、媒體報導與政府統計之間的關聯性，以探究民意調查與其他資訊來源之一致性程度，推論民意調查在敏感性議題的應用性，並根據研究發現在方法上與實務上進行討論。

The public poll is a subjective measurement of the individual’s perception, attitude and behavior, which has broadly been used in politics and public policy. It also has high credibility and reference value. For the measurement of sensitive issues or social moral issues, however, its validity has often been questioned. Evidence-based investigations of public opinion polls are few.
The nature of election bribery is a sensitive issue as well as a social moral issue. Additionally, Taiwan has also been strengthening the inspection of election bribery recently. Thus, it is worthwhile to examine the consistency and divergence of information on election bribery among public opinion polls, media reports and official statistics.
This paper tries to analyze the case of the three-in-one election of Greater Tainan City in 2010. This research uses the township as the analysis unit and conducts comparative analysis by the triangulation method, which includes survey, content analysis and secondary data analysis. Our results demonstrate a correlation among public opinion polls, media reports and official statistics. We discuss the methods and practice also.