One problem with describing election vote shares using ordinary least squares (OLS) is that it ignores the possible presence of spatial error correlation, whereby the errors are correlated in a systematic manner over space. This omission can bias OLS standard errors. We examine the 2004...