[R] model selection, lme

[kMan]

Dear Kay,

It is not an issue of whether it is "allowed". The issue is understanding
what comparisons are taking place between the models. Your advisor should
help with that.

Perhaps you had a more focused question related to R?

Sincerely,
KeithC.

-----Original Message-----
From: Kay Cichini [mailto:Kay.Cichini at uibk.ac.at] 
Sent: Wednesday, April 28, 2010 3:56 AM
To: r-help at r-project.org
Subject: [R] model selection, lme


hello,

i'd like to evaluate the 2 factors (f1 with 4, f2 with 2 levels) and its
interactions in my model for writing up a summary table with
likelihood-ratio tests for publication purpose.
   
now i'm unsure about which terms of my model may be dropped from the full
model and which not.

mfull<-lme(resp~f1*f2,random=...)
m1<-lme(resp~f1+f2,..)
m2<-lme(resp~f1,..)
m3<-lme(resp~f2,..)

i compared each model with dropped terms with the full model, but is this
allowed for m2 and m3?
or would i rather have to sequentially compare mfull/m1,m1/m2,m1/m3.

i'd very much appreciate help with this.
it may be a trivial problem - but though i could not figure out.

thanks in advance,
kay




-----
------------------------
Kay Cichini
Postgraduate student
Institute of Botany
Univ. of Innsbruck
------------------------

--
View this message in context:
http://r.789695.n4.nabble.com/model-selection-lme-tp2068889p2068889.html
Sent from the R help mailing list archive at Nabble.com.