|
|
I am running some Panel Regression using Procedure:
Genlin
In my data file, I have subject-ids, time of 3 levels(1,2,3) and
other covariates.
I would like to test whether there is any overall time effect controlling
for other predictors. What I did was:
Create a full model(includes time effect) and get a Goodness of Fit-Quasi Likelihood (under QIC),
and a reduced model(without time effect) and get another Goodness of Fit-Quasi
Likelihood (under QIC). Can I just simply calculate the change of
Quasi Likelihood and it's associated change of d.f. & treat it as Chisquare distribution
(similar to logistic regression)? Your help is highly appreciated.
Thank you.
Linda
|
Locked
