本文先從一個比較宏觀的角度，討論民意調查的科學基礎、民意匯集功能與限制，爾後再循序導入我國政府首長施政滿意度調查的個案分析，進而從方法學和民意調查的政治功能兩個面向，來反思當前政治民意調查或施政滿意度調查所引發的爭議問題，本研究發現目前的政府首長施政滿意度調查．仍舊面臨許多技術面的難題，例如施政滿意度的概念製作與問卷設計、電話簿的編製架構與行政區不一致、電話調查系統的抽樣架構涵蓋率偏低及其衍生的問題等，這些因素相當程度限制了政冶民意調查的科學描述、解釋及預測能力。更有甚者，在調查技術尚待精益求精、尋求本土化突破之際，國內對於民意調查的操作與解讀經常偏向政治上的考量，充斥著結果論、簡化論、信賴敘述推論和至上論等迷思，本文最後提出重建政治民意調查的公信力，首要在於還給施政滿意度調查一個獨立自主的運作空間，避免政治力量或商業利益的不當侵擾，並將其制度，化為政府例行業務的一部份，定期實施各種民意調查來昭信於社會大眾。

　This paper is first to explore the scientific foundation of public opinion polls, their functions and constraints with the articulation of public opinion. Based on a macro perspective, the paper then evaluates several cases of polling on the satisfaction with governor's performance in Taiwan. Finally, I make acouple of treatises on the current polls of the satisfaction with governor's performance, with an emphasis on the niethodologicalissues and their political functions. The findings indicate that current polls on the satisfyyactionwith governor's performance encounter many difficulties in bothtechnical and political aspects. Specifically, the technical issuesare pertinent to those problems such as, conceptualization of the satisfaction with governor's performance, questionnaire design, low coverage and inconsistency of the sampling framework.These problems have detrimental effects on the abilities of thepolls to describe, explain and predict public opinion. Moreover ,the utilization of the polls tends to be very political-orientation ,over emphasizing on the outcomes, simplification, ability to forecasting, and paramountcy in decision-making. Toreconstruct public confidence in the polls, I suggest that pollingon the satisfaction with governor's performance should beinstitutionalized to be part of governance work.