臺灣選民的真誠投票之推估：模型建構與實證分析

Abstract

隨著臺灣政治的民主化發展與傳播資訊管道的多元化，使得選民在選舉前就能快速獲得豐富的資訊，正因如此，也使部分選民的投票行為可能受到民調變化或其他人的集體意志而左右，最後作出的選擇並非原先心中所屬意的選項，即所謂非真誠投票行為。探究非真誠投票類型中，策略投票議題一直是政治學者關注及研究的焦點，文獻中，策略投票主要指涉選民考量對選舉結果的期待，在知道其所支持的候選人當選機率渺茫，為避免浪費選票且阻止最討厭的政黨候選人當選，只好將選票轉投給其他政黨中次偏好的選項。從過去國外研究證實，不論單一選區、甚至比例代表制或複數選區制都可觀察到選民曾出現策略投票。至於國內學界，近年來也紛紛投入研究國內各種選舉可能存在策略投票的情況，以及提出相關概念如分裂或一致投票的定義與測量，且許多研究經常關注選民手中不同選票投出分裂投票的結果，從而推測可能係來自策略投票行為。實際上，策略投票實際動機可能非常多元，但是真誠投票的定義則相對明確無爭議。本研究以2012年總統大選合併立委選舉為例，嘗試使用臺灣選舉民主化調查（Taiwan`s Election and Democratization Study, TEDS）所提供的選後追蹤調查為主，加上綜合考量選前、選後調查數據，以及比較選舉總體資料。作者們另闢蹊徑，引進美國政治學界研究方法之前沿，參採反事實推理（Counterfactual reasoning）以及隨機效用模型（Random Utility Model, RUM）等相關政治學文獻，應用於修正原先根據民意調查所得的數據，重新校準臺灣地區選民實際投票情境下來自「真誠投票」的合理比例，並歸納不同投票對象對應其真誠選民的特質，以及論述真誠、策略和分裂投票存在各種統計差異的意義。

Along with the democratic development of Taiwanese politics and the diversification of information channels, voters now have access to abundant information prior to elections. Owing to this, the final decision of some voters might be swayed by changes in public opinion polls or by the collective will of groups of people. The actual vote of these citizens may not be what they originally preferred, which cannot be characterized as sincere voting behavior. In the investigation of different types of non-sincere voting behavior, strategic voting is undoubtedly a major research issue among scholars of election studies delving into voter psychology. Strategic voting primarily refers to voters who decide to cast their votes for candidates with better prospects of winning so as to avoid ＂wasting＂ their vote. Past overseas literature has confirmed that whether in single-member districts or in proportional representation or multi-member districts, strategic voting has been observed among voters. As for academia in Taiwan, increasing effort has been made in recent years to study strategic voting that may take place in domestic elections, and the definition and measurement of related concepts, such as the effect of split-ticket and party voting. Most studies, however, are confined to observing the results of split-ticket voting, from which they surmise the possibility of strategic voting. In fact, the actual motivation for strategic voting may be very diverse, but the definition of sincere voting is relatively clear and uncontroversial. Instead, this study attempts to base itself mainly on post-election panel records provided by Taiwan`s Election and Democratization Study (TEDS), together with an integrated consideration of a pre- and post- election survey and a comparison of election outcomes. With Taiwan`s 2012 presidential-cum-parliamentary elections as the source of empirical evidence, this study adopts counterfactual reasoning and literature on the random utility model, applying them to revise the survey results of the original poll data so as to estimate a reasonable proportion of actual sincere voting. Furthermore, it sums up important characteristics of sincere voters who had different vote choices and demonstrated the subtle differences between split-ticket voting, sincere voting and strategic voting. Finally, the study discusses the various statistical differences between these three voting behaviors.