Skip to content
Prev 199189 / 398503 Next

stepAIC(coxph) forward selection

On Nov 4, 2009, at 9:26 PM, Rupa Parvataneni wrote:

            
I am a bit puzzled by the choice of a maximal model when you are then  
specifying only "forward". Wouldn't that choice imply that you should  
be starting with;

b.cox <- coxph(my.bpfs ~ 1)
Nothing in this reply should be taken to mean that this is a valid  
approach to model selection. Stepwise methods often are an abdication  
of an analyst's responsibility to use prior scientific knowledge.