Dear all.
I would like to use NLMIXED,for the following data.
I have a data dependent variable Y =success,failure, T categorical
variable (1,2,3), X categorical variable (1,2) and Z categorical variable
(1,2) , the variables are measured for student in different school the
variable school is R with 5 categories
I write my syntax as follows
proc nlmixed data=library.table qpoints =10;
param beta0=-1 beta1=1 beta2=1 beta3=1 s2u=2;
eta=beta0+beta1*t+beta2*X+beta*Z+u;
expeta=exp(eta);
p= expeta/(1+expeta);
model Y ~ binary (p)
random u ~ normal(0,s2u)
subject = R out =eb;
run;
The above syntax suppose the independents variables continuous, I would
like to ask if this model will give a s2u for each school (which
is considered as a level 2 variable in eb table) and estimate parameter for the
parameters.
If I want to consider the discrete characteristic of my variables I
have to create a set of dummy variables for t,X and Z like in the logistic
model? and to include them in the equation of eta
Thank you for any help
Adel
---------------------------------
ALL-NEW Yahoo! Messenger - all new features - even more fun!
|