[R-sig-Geo] Confidence interval for SAR model regression fit (errorsarlm, package spdep)

[Roger Bivand]

On Wed, 8 Feb 2012, Tommaso Jucker wrote:

> Dear List,
>
> I am modelling the relationship between species richness and an
> environmental variable and want to compare the fit of an OLS regression with
> that of a spatial model. I have used a SAR error model (errorsarlm()
> function in spdep) from which I extracted and plotted the fit. However, I
> would also like to plot a confidence interval for this fit. 
>
> I know how to do this for an object of class “lm” using the predict.lm()
> function which returns upper and lower confidence intervals for each
> estimate. However, the predict.sarlm() function is slightly different as it
> reports the trend (non-spatial smooth) and signal (spatial smooth), the sum
> of which is used to estimate the fit. 
>
> I was wondering if anyone knew if/how it is possible to obtain confidence
> estimates for a regression line generated from a SAR error model.

They are not provided because they are unknown in the literature so far, 
as far as I know. It might be possible to get an empirical distribution by 
simulating from the fitted model, in a similar way to the extraction of 
impacts/emanating effects for spatial lag models.

Hope this helps,

Roger

>
> Thanks in advance for any suggestions.
>
> Cheers,
>
> Tommaso Jucker
> --
> Tommaso Jucker
> Ph.D. student
> Forest Ecology and Conservation Group
> Department of Plant Sciences
> University of Cambridge
> e-mail: tommasojucker at gmail.com
>
>
> --
> View this message in context: http://r-sig-geo.2731867.n2.nabble.com/Confidence-interval-for-SAR-model-regression-fit-errorsarlm-package-spdep-tp7265503p7265503.html
> Sent from the R-sig-geo mailing list archive at Nabble.com.
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo

-- 
Roger Bivand
Department of Economics, NHH Norwegian School of Economics,
Helleveien 30, N-5045 Bergen, Norway.
voice: +47 55 95 93 55; fax +47 55 95 95 43
e-mail: Roger.Bivand at nhh.no