兩輪投票制對選民政治態度左右極化之影響－2016年奧地利聯邦總統選舉之跨層次分析

Abstract

奧地利為「準內閣制」（quasi-parliamentarism）國家，聯邦總統具任命總理及部長之權，其政黨立場亦會對歐盟政策造成影響，故重要性不亞於國民議會（Nationalrat）。而總統選舉係採「兩輪絕對多數決選制」（die absolute Mehrheitswahl mit einer Stichwah），若第一輪投票無候選人得票率過半，則由得票率最高兩人進行第二輪決選，故第二輪選舉常造成左右立場選票集中，甚至意識形態兩端極化。在2016年4月24日第一輪投票中，各黨均未得票均未過半，故由奧地利自由黨（FPÖ）Norbert Gerwald Hofer，及綠黨Alexander van der Bellen進入第二、三輪決選，並由Alexander van der Bellen得票53.79 %當選。此次選舉顯示，兩輪投票制及總統選舉特性，促使不同政黨之選民，將選票轉向當選機率最高之前二名候選人。然僅從選舉結果不能得知選民於兩輪投票中的轉變，必須運用集體資料，進行「跨層次推論」（cross-level inference）。本文將以「階層貝式模型」（hierarchical Bayesian model），在多列聯表架構下，分析各黨選民集中左右兩端之轉變趨勢。

Austria is a semi-parliamentary(quasi-parliamentary) country, in which the federal president has the power to appoint prime ministers and ministers. The president`s political party position also has impacts on European Union (EU) policies, playing a role of importance no less than the National Council (Nationalrat). The presidential election adopts ＂the absolute majority runoff voting＂ system (die absolute Mehrheitswahl mit einer Stichwah), in which the election proceeds to a second round when no candidate has received a simple majority (more than 50%) of votes cast in the first round, and then the two candidates who received the most votes would be candidates in the second round. In this context, the polarization of ballots and ideologies can often be observed in the second round voting. Take the election on April 24, 2016, as an example. No candidate received a simple majority (more than 50%) of votes cast in the first round, and thus Norbert Gerwald Hofer of the Freedom Party of Austria ( Freiheitliche Partei Österreichs, FPÖ) and Alexander van der Bellen of the Green Party became the candidates in the second and third rounds, and eventually Alexander van der Bellen won the election with 53.79 % of the vote. The election showed that the characteristics of the second ballot system and the presidential election had encouraged voters with different political party preferences to shift their votes to the two candidates with the highest election chance. However, the change of the structure of voters in between the two rounds could not be observed from the election results, which can only be obtained through cross-level inference by the utilization of aggregate data.This study uses the ＂Hierarchical Bayesian Model＂ to infer the trends of changes in the polarization of the voters of different parties under a structure of multi-way contingency tables.