Spatial statistics and geographic informaton systems in social sciences: From theory to application

In this thesis, various spatial statistical methods are applied to social sciences via using geographic information systems software. 2002 general election voter turnout data is used at city and district levels and same data is further classified as rural, urban and total voter turnout within each level. After mapping voter turnout percentages, spatial randomness hypothesis for each classification at each level of analysis is tested via conducting local and global spatial autocorrelation and clustering tests (Getis Ord General G, Getis Ord G, Local Indicators of Spatial Association and Global Moran's I). With the exception of city level results in Getis Ord General G test, all results showed clustering, dispersion or outliers, thus showing that the voter turnout in Turkey for 2002 general elections did not indicate spatial randomness.