[R-sig-Geo] test spatial autocorrelation of level 2 residuals in lme()

Dexter Locke dexter.locke at gmail.com
Mon Apr 25 18:28:51 CEST 2016


Apologies for cross-posting, its unclear if this is a spatial or a mixed
models question. Maybe its both?

I am interested in extracting the residuals from level 2 of a mixed model
(created using lme()), building a spatial weights matrix, and then testing
for spatial autocorrelation using Moran's I.

While I have found methods of incorporating spatial effects into a mixed
model using corStruct, I am interested in first evaluating *if* that is an
appropriate model for a given dataset by examining the level 2 residuals'
spatial patterning - or lack thereof.

Which slot contains the residuals for level 2 and how are they ordered?

Best,
Dexter

	[[alternative HTML version deleted]]



More information about the R-sig-Geo mailing list