The main purpose of this research is to find the influence of the ethnic distribution in the pan-Green camp percentage of votes by spatial analysis. This research compare electoral spatial distribution patterns by Geoda software. We use the ordinary least square (OLS) model、spatial lag(SLM) model and spatial error(SEM) model to find: whether the influence of the ethnic distribution in the pan-Green camp percentage of votes has difference with the different level elections or time changing. This research found that the electoral spatial distribution patterns has space association , spatial effects and spatial heterogeneous. The influence of the ethnic distribution is higher in Central level elections than place level elections and it’s lower in the 2008’s president election than 2004’s.