서울시 하위지역 주거용 부동산 특성과 2012년 대선투표 결과에 관한 탐색적 공간분석

The objective of this study is to understand relationship between presidential election outcomes and residential real estate characteristics in Seoul metropolitan area. Spatial auto regression model(SAR) and spatial error model(SEM) are used to identify residential real estate characteristics that explain the 2012 Korean presidential election outcomes. By comparing those two spatial regression models to ordinary least square(OLS) model, this study concludes spatial error model(SEM) builds a better model than OLS model. This study also explores local patterns of independent variables included in spatial error model using Geographically weighted regression(GWR). Conclusions drawn from this study are as follows: First, 2012 Korean presidential election outcomes in sub-area of Seoul can be explained through residential real estate characteristics of sub-area of Seoul. Secondly, spatial methods, which controls spatial autocorrelation among spatial units, are required to analyze the election result. Finally, GWR result shows that coefficients on the average age and house size variables are globally related to a positive preference toward the conservative candidate, whereas coefficient on single household ratio variable is globally related to a negative preference to the candidate. Another global coefficient, house deterioration level, is positively related to a preference toward the candidate. GWR result also shows the local coefficient of homeownership variable in Southwestern Seoul, where preferences toward the liberal candidate is highly clustered, is positively and highly correlated to a preference toward the conservative candidate. Local coefficients of multi-family housing ratio variable and number of rooms per household member variable are positively correlated to a preference toward the conservative candidate across Gangnam, Seocho, Songpa, Gangdong Gu’s and localities near the Han River area.