The Spatial Factors in Electoral Studies: The Example of Presidential Elections in Taiwan

本文嘗試以空間分析的方法，探討台灣三次總統選舉呈現的空間因素。作者首先介紹空間分析的意義與特質，接著介紹空間分析常用的方法，對選舉地理學重要文獻作簡要回顧，最後以三次總統選舉為例，探討選舉結果呈現的空間因素。使用的方法包括視覺化呈現、空間探索分析、空間迴歸建模三個層面。研究發現第一次總統選舉，因李登輝效應影響，選票的空間結構比較異常。其後選舉，藍、綠選票加總之後，縱使得票率互有消長，仍呈現相當穩定的空間結構。空間迴歸方面，作者針對民進黨總統選舉獲票的增加率進行分析，發展誤差的空間自相關現象同時來自於空間異質與鄰近效應。The author tries to apply the methods of spatial analysis into the studies of election campaign with the purpose of identifying the influence of spatial factors. After making a short introduction of the meaning of spatial analysis and the techniques of this research approach, the author takes the three presidential elections held in Taiwan from 1996 to 2004 as example. The methods used in this study include geographical visualization, exploratory spatial data analysis and spatial modeling. The author finds that the voting results show a high degree of spatial stability, although the party supports in each election was changing. In the spatial regression analysis, the author finds there are both spatial heterogeneity and autocorrelation existed which makes error violates the assumption of OLS regression