ABSTRACT
This chapter concerns the effects of geography on our models. First, the chapter extensively explores what spatial correlation is and how to detect it. It also covers why it violates the assumptions of our modeling techniques from last chapter.
With that, four modeling methods are covered: lagged variable regression, expansion method, geographically weighted regression, and spatially-lagged expansion method. Their strengths and weaknesses are covered as each technique is applied to the 2011 constitutional referendum in Egypt in search of evidence of unfairness. While the search is in vain, the maps do raise several questions about how the election was conducted.
