[R-sig-ME] Fitting a fixed effect model while modelling correlation in the residuals in 2 directions

[Bolker,Benjamin Michael]

________________________________________
From: r-sig-mixed-models-bounces at r-project.org [r-sig-mixed-models-bounces at r-project.org] On Behalf Of Adam D. I. Kramer [adik at ilovebacon.org]
Sent: Tuesday, September 15, 2009 12:22 AM
To: Thomas Bishop
Cc: r-sig-mixed-models at r-project.org
Subject: Re: [R-sig-ME] Fitting a fixed effect model while modelling correlation in the residuals in 2 directions

On Tue, 15 Sep 2009, Thomas Bishop wrote:

> slate.ar1*ar1 <- gls(Yield~as.factor(Variety),corr = corAR1(form = ~
> 1|(FieldRow+FieldColumn),data=Slate))

...data=Slate is one paren too deep. Should be ...+FieldColumn)),data=Slate)


   Does corAR1 really work for spatial problems?  I would have thought the definition would have been slightly different ...