Dear All:
I have the response variable y --output at end of treatment
x ---baseline
I want to compare three models for the same data using bayesian method:
Model 1: y~f(x)
Model 2: y-x ~ f(x)
Model 3 (y-x)/x ~ f(x)
I can not use DIC for comparing them because of the different strcture of response variable.
Have you had the same experience?
Many Thanks!
Xin
|