[R] step method in glm()

[Prof Brian Ripley]

As the posting guide requested of you:

0) Do not send HTML mail.

1) Tell us your version of R.  Some problems in this area have been 
corrected recently: if it is not at least 2.3.1, update as asked.

2) Give a reproducible example.

On Thu, 13 Jul 2006, Birycki Konrad DomBank Warszawa wrote:

> Hello,
>  
> I estimaded two logit models via glm(). A null model (called glm00) and
> "full" model with all accessible covariates and interactions between them
> (glm1).
>  
> Then I tried to get even better model by step procedure. I tried the
> following code:
>  
> > step(glm00, scope=formula(glm1), method="both")   
>  
> and another one:
>  
> > step(glm00, scope=formula(glm1), method="forward")
>  
> In both cases step procedure terminated after several steps with a warning:
>  
> > Error in factor.scope(ffac, list(add = fadd, drop = fdrop)) : 
>             > upper scope does not include model
>  
> What does the word "model" refer to? To glm00 (the null model)?  To my
> knowledge and understanding, the initial null model should have been
> included in every step of the search. 
>  
> Am I wrong? 
>  
> Thanks for all comments :-)
>  
> Konrad Birycki
>  
> 
> 	[[alternative HTML version deleted]]
> 
> ______________________________________________
> 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
> 

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595