Regression analysis is a statistical technique commonly used in the social and physical sciences to model relationships between variables. To make unbiased, consistent, and efficient inferences about real-world relationships a researcher using regression analysis relies on a set of assumptions about the process generating the data used in the analysis and the errors produced by the model. Several of these assumptions are frequently violated when the real-world process generating the data used in the regression analysis is spatially structured, which creates dependence among the observations and spatial structure in the model errors. To avoid the confounding effects of spatial dependence, spatial autoregression models include spatial structures that specify the relationships between observations and their neighbors. These structures are most commonly specified using a weights matrix that can take many forms and be applied to different components of the spatial autoregressive model. Properly specified, including these structures in the regression analysis can account for the effects of spatial dependence on the estimates of the model and allow researchers to make reliable inferences. While spatial autoregressive models are commonly used in spatial econometric applications, they have wide applicability for modeling spatially dependent data.