I am performing ordinary least squares regression analysis on two variables: K concentration in soil as explanatory variable, and thickness of stalk of plants as the dependent. The K-concentration dataset shows significant spatial autocorrelation (p <0.01). The OLS results in ArcMap 10.4 return a very low R2, but the residuals are not spatially autocorrelated. The F-Statistic indicates that the model is not significant. Could the spatial autocorrelation (clustering) in the input variable be affecting the functionality of this model?


  • Is your K-concentration a raster layer? If so, raster layers in fact almost always show high spatial autocorrelation, i.e. for Moran´s I or Geary's C. I guess if your overall R2 is low, this is not due to the autocorrelation but due to the fact that there is very likely no autocorrelation. Spatial autocorrelation rather leads to better results which should be questioned when the residuals show a pattern caused by the autocorrelation. – Jens Jan 3 '17 at 16:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.