In this thesis, we provide a survey of specification and estimation of spatial autoregressive models, which consist of a spatial lag of the dependent variable as a regressor or a disturbance term that is spatially autoregressive. The spatially lagged dependent variable is typically correlated with the disturbance term, and hence the ordinary least squares estimator is typically not consistent in such a situation. One important goal of this thesis is to explain spatial autocorrelation concept. Spatial autocorrelation, measures the degree of dependency among observations in a spatial sample. Classic spatial autocorrelation includes Moran's I and Geary's C. Definition of these indices requires a spatial weights matrix that reflects the intensity of relationship between observations in a neighborhood. In this thesis, we describe specification tests for specifying spatial autocorrelation and then introduce consistent methods for estimating spatial autoregressive models. We also introduce statistical softwares (R and GeoDa) for analyzing spatial autocorrelation.