Message-ID: <1352716577698-4649300.post@n4.nabble.com>
Date: 2012-11-12T10:36:17Z
From: tfarkas
Subject: Adding Spatial Correlation Strucutre to Logistic Regression / Contingency Analysis
>From what I can tell by reading forum posts etc., this is not a trivial
issue. An answer in 2008 indicated some directions, but I'm curious whether
any developments have been made since then.
In my data set I have 182 observations with a binomial response, a 2-level
explanatory factor, and x and y coordinates. By visually inspecting the
spatial distribution of standardized residual error and multi-directional
variograms I must conclude that adding spatial correlation structure to my
logistic regression model (currently implemented using glm) would be
prudent.
Can anyone please offer advice as to how I should go about incorporating
this spatial auto-correlation to my model?
Is it perhaps easier in my case, given that I have only a single 2-level
factor as an explanatory variable? If so, I'm actually interested in how to
do this for logistic regression (with continuous variables) too, so a method
toward this end would be interesting as well.
--
View this message in context: http://r.789695.n4.nabble.com/Adding-Spatial-Correlation-Strucutre-to-Logistic-Regression-Contingency-Analysis-tp4649300.html
Sent from the R help mailing list archive at Nabble.com.