目前國內兩個主要的政黨，皆將民意調查作為決定政黨提名的正式機制。國內政黨在提名候選人時引進民意調查，其主要的理由有二：一是試圖縮短黨意與民意的落差，以期能提名更適當的候選人而在選戰中獲勝；二是欲藉此消弭黨內初選時賄選、買票的風氣，以及「人頭黨員」、「口袋黨員」對初選所可能造成的扭曲。本文以1998年民進黨立委提名黨內初選所首度正式採行的民意調查進行個案研究。研究結果發現，民意調查並不一定能達成上述的目的，甚至初選的成績與選舉的結果之間可能還會產生重大的落差。

Two major parties in Taiwan, KMT and DPP, have implemented public opinion survey as the formal mechanism for their candidate selec-tions. The employment of public opinion survey in the process of candi-date selection is based on two hypotheses: (1) the usage of opinion survey may narrow down the gap between the "party will" and the "public opin-ion", (2) the usage of opinion survey may reduce the influence of "nominal party members". This paper examines the candidate selection process of DPP in 1998 Legislative Yuan Election, and demonstrates that none of the hypotheses above are true. Moreover, when the public opinion survey was applied for the candidate selection, the most important function of the political party will be shrunk.