投票率高誰得利？台灣總統選舉投票率與得票率的分析

Abstract

每逢選舉，坊間便盛傳投票率高低會影響選舉結果，投票率高能讓泛藍勝選的機會增加，投票率低則讓泛綠佔盡便宜。這類的認知並非空穴來風，一方面在國外行之有年的選舉研究中指出投票率會影響左翼政黨的得票率，另一方面，一般相信台灣的藍綠支持者在投票傾向上有不同的表現，泛綠支持者的凝聚力大於泛藍支持者，因此才會帶來泛綠選民一定會去投票而泛藍選民則否的印象。針對此概念，本文從藍綠兩陣營支持者的人口結構出發，探討兩陣營的人口結構是否有差異，以及投票率究竟與得票率是否有關。 為了將焦點放在台灣投票率研究中幾項未被關注的角度，本文兼採總體資料與個體資料進行分析，以全台368個鄉鎮市區為分析單位。首先利用集群分析區別藍綠鄉鎮及類型，並以交叉列表、迴歸分析，觀察各類型鄉鎮市區的人口差異。接著，再利用羅吉斯迴歸分析個體資料、以及使用迴歸分析總體資料，透過兩種方法的研究結果探討投票率是否會影響得票率。 研究發現主要有三：第一，藍綠鄉鎮的人口結構雖然有差異，但歧異程度卻不如泛藍鄉鎮在投票率高與投票率低兩類鄉鎮之間的差異。此人口結構差異突顯泛藍的支持者中有一群投票傾向較低的選民。第二，整體而言投票率高雖然對泛藍有利，但必須要視選舉時空而定，政黨得票率隨投票率起伏時，利害程度和利害轉換點都因政治背景不同而有別。第三，不同的人口結構雖然是影響投票率的因素，但若將政黨色彩納入考量，則影響因素便不全然一致。因此，可以說人口結構會影響投票率，但卻不是會使得投票率改變得票率的關鍵。

It is commonly believed turnout is a decisive factor in the outcome of elections. In Taiwan, the Pan-Blue Coalition is thought to benefit from a heavy turnout while the Pan-Green Coalition gains from lower turnout. Studies in other countries have shown that the percentage of votes won by left wing parties is affected by turnout. In Taiwan, however, it is recognized that Pan-Green support is more cohesive than Pan-Blue support, meaning that Pan-Green supporters will certainly vote and Pan-Blue supporters maybe will not. On this basis, the thesis asks whether there is a fundamental difference in the population structure of supporters of two coalitions, and attempts to show whether there is a correlation between turnout and overall electoral performance. This thesis combines individual and aggregate-level data to provide a completely new perspective on turnout in Taiwanese elections. The unit of analysis is the urban and rural township, and the sample includes 368 townships in Taiwan. First, the article employ cluster analysis to identify the prevailing partisan alignment and the turnouts in each township. Subsequently, using crosstabs and regression analysis to identify different population structures in each township. Finally, using individual data with logistic regression and aggregate data with regression analysis, we try to find out whether turnout affects the percentage of votes received. There are three main findings. First, although population structure is different in Pan-Blue towns and Pan-Green townships, the divergence is more significant in Pan-Blue townships between townships with a lower turnout and townships with a higher turnout. This difference indicates that there are Pan-Blue supporters who are reluctant to vote. Second, generally speaking a heavy turnout favors the Pan-Blue Coalition, but this can vary from election to election. How much each party benefits from different levels of turnout varies according to the prevailing political situation. Third, population structure influences turnout, but the impact on support for each party is not clear. In other words, population structure is not crucial to explaining how turnout impacts levels of support for different parties.