本研究探討政治學者與媒體記者在觀察預測選舉時的落差，並檢證媒體報導選戰時的基模 ( schema ) 與民眾所關心的焦點有無落差。一般而言，政治學者可以在選前數月用政黨認同、統獨立場、省籍、性別、及年齡等基本變項來解釋或預測選民的投票行為，並且有基本的穩定性和準確性 ( Lewis-Beck and Rice, 1992；Rosenstone, 1983 ) 。但是從媒體的報導角度來看，民意支持度的變化卻是隨著選戰策略及競選新聞事件而起伏，似乎民意是浮動且極難預測的。因此本文首先試圖從民意調查資料的模式中對上述現象提供一個可能的解釋。其次，作者想要檢驗在競選過程當中，媒體報導的選戰的角度為何，及其與選民所關心的議題之間的落差。結果顯示，媒體自有一番新聞運作邏輯，其報導選舉新聞的焦點和民眾所關心的議題之間有一定的差距。

This Study discusses the differences between journalists and political scientists while observing and predicting elections, and examines whether there are discrepancies between the media’s schema and the public’s in-terest during campaign. Overall speaking, political scientists are able to predict or explain voting behavior months before the election, using de-mographic variables such as partisanship, unification/independence stance, ethnic identity, gender, and age with considerable stability and accuracy (Lewis-Beck and Rice, 1992; Rosenstone, 1983). However, from the perspective of media report, candidates’ support ratings rise and fall along with campaign strategies and news events, public opinion seems to be variable and difficult to predict. Therefore, this article would like to first provide a plausible explanation to the above dilemma. Subsequently, the author would like to examine the media focus during campaign to see if their coverage reflects the issues people deem important. Empirical data shows that the media operates on its own logic, and there is a consider-able gap between media campaign coverage and the issues people care most about.