[R] lme and varFunc()

[Christoph Scherber]

Dear all,

I am expecting a Poisson error distribution in my lme with 
weights=varFunc().

The "weigths= varPower (form= fitted (.))" doesn´t work due to missing 
values in the response:

Problem in lme.formula(fixed = sqrt(nrmainaxes + 0...: Maximum number of iterations reached without convergence. 
Use traceback() to see the call stack

That´s why I´ve used one of my most important explanatory variables as a variance covariate:

weigths= varPower (form=~explanatory)

With that, it worked out properly so far.

What would your suggestion in such a case be?

Regards
Christoph



Lorenz.Gygax at fat.admin.ch wrote:

>>I am currently analyzing a dataset using lme(). The model I 
>>use has the following structure:
>>
>>model<-lme(response~Covariate+TreatmentA+TreatmentB,
>>           random=~1|Block/Plot,method="ML")
>>
>>When I plot the residuals against the fitted values, I see a clear 
>>positive trend (meaning that the variance increases with the mean).
>>
>>I tried to solve this issue using weights=varPower(), but it
>>doesn´t change the residual plot at all.
>>    
>>
>
>You are aware that you need to use something like 
>
>weigths= varPower (form= fitted (.))
>
>and the plot residuals using e.g.
>
>scatter.smooth (fitted (model), resid (model, type= 'n'))
>
>Maybe the latter should also be ok with the default pearson residuals, but I
>am not sure.
>
>Possibly a look into the following would help?
>
>@Book{Pin:00a,
>  author = 	 {Pinheiro, Jose C and Bates, Douglas M},
>  title = 	 {Mixed-Effects Models in {S} and {S}-{P}{L}{U}{S}},
>  publisher = 	 {Springer},
>  year = 	 {2000},
>  address = 	 {New York}
>}
>
>  
>
>>How would you implement such a positive trend in the variance? I´ve 
>>tried glmmPQL (which works great with poisson errors), but 
>>using glmmPQL I can´t do model simplification.
>>    
>>
>
>Well, what error distribution do you have / do you expect?
>
>Regards, Lorenz
>- 
>Lorenz Gygax, Dr. sc. nat.
>Centre for proper housing of ruminants and pigs
>Swiss Federal Veterinary Office
>
>______________________________________________
>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
>
>  
>