Types of Data Used in Spatial Models
Step 3 Estimate a GLS model based on the results from Step 2. The results you get will be unbiased due to the inﬂuence of spatial autocorrelation. The results from this estimation yielded an estimate for spatial autocorrelation of 0.476, thus removing any impact of this possible source of false positive ﬁndings. For example, the impact of percent young male, found to be signiﬁcant in the initial model, was found to be statistically insigniﬁcant, going from positive to negative in sign. In this case, spatial autocorrelation was not fully accounted for by the independent variables, and spatial autocorrelation had caused the estimate for young males to be inﬂated.