唯一支持 －從2012民進黨黨內初選看策略性投票

Abstract

2012台灣民進黨總統提名選舉採「對比式投票」之全國民意調查電訪。「對比式投票」指以假設性情境問題詢問選民當民進黨推出不同的候選人作為2012年總統候選人時，其投票意向應為何，並據以選出最有希望擊敗現任總統馬英九的候選人。因此，同時支持蔡英文與蘇貞昌之選民若希望其選票能有效，必只能策略性投給其中一個候選人，藉此拉大與另一候選人之得票差距，此稱之為「唯一支持」投票。本研究使用類似民進黨初選民調之訪調資料，以分析此次初選中策略性投票行為。我們發現在同時支持蔡英文與蘇貞昌之選民中，即使在有明顯的策略性投票誘因下，仍只有18.7%同時支持蔡與蘇的選民對候選人做唯一支持。所占比例雖不大，在此次競爭激烈的初選中卻已足以影響投票結果。 我們以一個簡單的模型分析初選中影響策略性投票的因素。模型中，選民除了考慮對候選人的喜愛程度，並考慮選民主觀上對候選人在總統大選中的勝選機率。我們接著使用訪調資料測試模型的預測。我們發現，蘇蔡支持者在初選中策略性投票的行為可由模型解釋。而選民對蔡唯一支持的傾向比對蘇唯一支持的傾向強。由此可見，蔡英文陣營在初選期間對唯一支持的呼籲確有成效。我們的結果也指出，馬英九支持者並未在民進黨初選中進行策略性投票以選出較弱的民進黨候選人。另外我們以相似的模型對美國2008年民主黨總統初選投票行為進行估計，發現美國選民在初選的投票行為也同樣受到對候選人喜愛度與主觀對候選人在總統大選的勝出機率所影響。

The 2012 Taiwan Democratic Progressive Party (DPP) presidential primary election was conducted in the form of “contrast style polling,” asking respondents which candidate they would vote for in hypothetical situations in which the DPP proposed different candidates as the 2012 presidential candidates. The electoral rule stated that the candidate who beat the incumbent president, Ma, with the largest margins would be the 2012 DPP nominee. In this situation, if a Pan-Green supporter wants to make a difference in the election outcome, he/she must vote for either Su or Tsai strategically in order to widen the vote gap between the two candidates. This is called “exclusive-support” behavior in this study. We use survey data similar to the official DPP opinion polls to analyze strategic voting behavior in the DPP primary election. In spite of the explicit strategic voting incentive, we find that only 18.7 percent Pan-Green supporters demonstrated “exclusive-support” behavior. Although this is not a very large proportion, it still influenced the primary election outcome in this competitive election. We provide a simple model to analyze factors affecting strategic voting in the primary election. In our model, voters consider not only the favorability of candidates but also their perceptions of the candidate’s electability in the presidential election. We used the survey data to test the predictions of model. The results show that the strategic voting behavior of the Pan-Green supporters can be explained by the model. Voters had stronger intentions to exclusively-support for Tsai than exclusively-support for Su, suggesting that the advocation of exclusively-supporting by Tsai’s campaign was effective. We also analyze the voting behavior in the 2008 Democrat presidential primary election in the U.S. We find that American voters’ voting behavior in the primary election is also affected by their favorability of candidates and their perceptions of candidate’s electability.