社會科學院: 國家發展研究所指導教授: 鄧志松蘇郁涵Su, Yu-HanYu-HanSu2017-03-032018-06-282017-03-032018-06-282016http://ntur.lib.ntu.edu.tw//handle/246246/274616　　2014年臺北市長選舉無黨籍候選人柯文哲擊敗眾多競爭對手，橫掃全臺北市贏得選戰，外界稱為「柯文哲現象」。本研究之目的在於探討2014年臺北市長選舉中柯文哲的選舉表現，假設柯文哲乃奠基於民進黨支持者的基礎上擴展票源，以柯文哲得票率減去2010年民進黨候選人蘇貞昌得票率所得之「得票率變化」來解釋所謂的柯文哲現象。 　　本研究利用空間分析方法，以臺北市456個里為分析單位，從人口結構、經社背景以及柯文哲在選前所提出的相關競選政見、空間因素等面向，分析這些因素對得票率變化的影響程度，並透過傳統與空間迴歸模型的幫助，希望能更加完整的解釋本次選舉結果。 　　研究發現民進黨得票率對得票率變化的影響程度甚深，因此將民進黨得票率設為控制變數，以期準確地觀察各項變數與得票率變化之關係。透過迴歸分析，更得知人口結構和經社背景、相關競選政見變數對得票率變化皆有影響。此外，亦發現鄰近效應也影響著選舉結果的呈現。　　As an independent candidate of Taipei City mayoral election in 2014, Ko Wen-Je defeated several rivals to win the election. The abnormal phenomenon called “Ko Wen-Je phenomenon” was worth noticing. The purpose of this study was to analyze the percentage of votes Ko Wen-Je obtained. This study assumed that the votes Ko won were from Democratic Progressive Party (DPP) traditional stronghold and he newly obtained. 　　This study used spatial analysis and set 456 villages in Taipei City as analysis units, analyzing different aspects like population structure, socio-economic background, the relevant platform and spatial factors. Besides, through the traditional and spatial regression models, the election can be described more completely. The study found that Ko’s percentage of votes was deeply influenced by the DPP’s. Thus, it was essential to set it as a control variable. Furthermore, by using spatial regression analysis, this study proved that population structure, socio-economic background, and relevant platform all affected the percentage of votes. Moreover, the neighborhood effect also had an influence on the election result.論文使用權限: 不同意授權柯文哲臺北市選舉空間分析空間自相關Ko Wen-JeTaipei Cityelectionspatial analysisspatial autocorrelationthesis10.6342/NTU201602787