[R] linear hypothesis test in gls model

John Fox jfox at mcmaster.ca
Sat Jun 16 19:21:55 CEST 2007


Dear Tine,

linear.hypothesis() currently has no method specifically for gls objects,
and so this usage invokes the default method. I'm not sure off-hand what's
appropriate for an F-test in this context (and indeed why the default test
is inappropriate). Can you describe the correct test or supply a reference?
I suspect that it shouldn't be hard to write a linear.hypothesis method for
gls objects that fixes up the result returned by linear.hypothesis.default. 

You might take a look at car:::linear.hypothesis.default to see that it does
-- the computations are pretty straightforward.

I hope this helps,
 John 

--------------------------------
John Fox, Professor
Department of Sociology
McMaster University
Hamilton, Ontario
Canada L8S 4M4
905-525-9140x23604
http://socserv.mcmaster.ca/jfox 
-------------------------------- 

> -----Original Message-----
> From: r-help-bounces at stat.math.ethz.ch 
> [mailto:r-help-bounces at stat.math.ethz.ch] On Behalf Of Tine Huyghe
> Sent: Saturday, June 16, 2007 4:19 AM
> To: r-help at stat.math.ethz.ch
> Subject: [R] linear hypothesis test in gls model
> 
> Dear all,
> 
> For analysis of a longitudinal data set with fixed 
> measurement in time I built a gls model (nlme). For testing 
> hypotheses in this model I used the linear.hypothesis 
> function from the car package. A check with the results 
> obtained in SAS proc MIXED with a repeated statement revealed 
> an inconsistency in the results. The problem can be that the 
> linear.hypothesis function (1) only gives the asymptotic chi 
> square test and/or (2) only uses the residual error. Is there 
> another solution to testing linear hypotheses in a gls model?
> 
> Thanks in advance
> 
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide 
> http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>



More information about the R-help mailing list