近來，民意調查大量被用以作為選舉預測的工具，其目的在於藉由民調資料預測候選人的勝算。就相當程度而言，朝野政黨參考這些民調資訊作為選舉動員的依據，參選者憑藉這些民調資料作為擬定競選策略的基礎，選民仰賴這些民調數據作為投票決定的參考。然而，2001年第五屆立法委員選舉結果揭曉之後，民意調查卻成為眾矢之的，引發諸多批評。其中，或許最受非議的焦點，在於選前公布民調結果嚴重扭曲選民投票意向，導致若干民調領先的立委候選人落選。換言之，民意調查並無法預測實際選舉結果，僅淪為候選人與政黨傳達「棄保效應」的文宣工具。本文嘗試探討批評選前民調的論點，並佐以相關資料，檢證新聞媒體選前民調資料與總體選舉結果的關係。實證數據顯示，民意調查結果確實是預測候選人得票率的有效指標。換言之，前述批評民調的理由，著實有待商榷。我們認為，民意調查作為選舉預測工具，僅為政治資訊的提供者，實不宜賦予過多是非對錯的評價。

　　　　　In the past years, opinion survey has been widely employed as the means to predict electoral outcomes. To that extent, political parties engage electoral mobilization on the basis of opinion survey information, and candidates plan campaigning strategies according to survey data, while voters rely on such information as a reference for voting decisions. However, after the Legislative Yuan election in 2001, critics charge survey research with some flaws. One of the points criticized most is that the pre-election surveys seriously distort the voters' inclination, which results in the losing of leading candidates. In other words, opinion survey cannot accurately forecast the electoral results. It at best becomes the advertising device for candidates and political parties to encourage voters to wage "strategic voting." Analyzing survey data and electoral statistics of the 2001 Legislative Yuan election, this work aims at examining the relationship between pre-election survey and electoral results. As the empirical data shown, survey research is still positively an effective indicator of predicting the vote shares of candidates. It means that these criticisms remain unconvincing. Throughout this essay and in the conclusion, we note that survey research employed as a means to predict elections is simply the provider of political information, and should not be charged too many criticisms.