[R] lme() random form

Irene Mantzouni ima at difres.dk
Sun Oct 14 05:40:09 CEST 2007


Dear all,
 
I am trying to fit a lme of the form:
m=lme(y~F*x-1, random=...,data=...)
 
F is a 2-level factor since I need the fixed part to differ for 2 groups. 
The random part depends on subgroups G. 
I would like to have a random effect only on the slope and I cannot figure out how to formulate the random part. 
If I write it like : F*x-1|G, I get random effect on both the slope and the intercept. 
If I use: F-1|G, the random effect is on the intercepts. 
If I use x-1|G I get an effect only on the slope corresponding to one of the levels of F. 
The coefs in the fixed part look like: 
F1
F2
x
F1:x
 
What kind of contrasts should I use and how can I get a random effect only on both slopes?
 
Thank you!
 
Irene



More information about the R-help mailing list