[R-sig-Geo] advice on spatial regression for interpolated data

[Terry Griffin]

----- Original Message -----
From: "juanita choo" <juanitachoo at gmail.com>
To: r-sig-geo at r-project.org
Sent: Monday, December 27, 2010 11:19:42 AM GMT -06:00 US/Canada Central
Subject: [R-sig-Geo] advice on spatial regression for interpolated data

Hi Everyone,

I am trying figure out the best method/R package for analyzing my
data. I am testing the hypothesis that seedling density data is
correlated with both adult tree densities and seed predation
intensities. I have mapped out the locations of all adult and
seedlings in a 150x150m plot, and have estimated predation intensities
at  72 locations in the plot. All variables are spatially
autocorrelated. I was planning running a spatial regression on the
interpolated values seedling densities (dependent variable) against
adult densities, and predation intensities. But I am not sure if this
is appropriate or if there is a better method. I was looking at the
spdep package was wondering whether the lagSARlm would work in this
setting. Thanks for your advice on this.

Juanita

_______________________________________________
R-sig-Geo mailing list
R-sig-Geo at r-project.org
https://stat.ethz.ch/mailman/listinfo/r-sig-geo


Hi Juanita,

A useful discussion of using spatially interpolated variables especially as independent variables is in:

Anselin, L. 2001. Spatial Effects in Econometric Practice in Environmental and Resource
Economics, American Journal of Agricultural Economics 83, 705-710.

Sometimes interpolation methods seem attractive to remedy the dilemma of spatially disparate spatial data layers; however, Anselin 2001 describes the systematic error that is introduced into the data, causing problems in deriving inference.

Terry Griffin, Ph.D.
Assistant Professor - Economics
University of Arkansas - Division of Agriculture
501.671.2182
tgriffin at uaex.edu