04-27-2016 04:32 AM
I am actually running a Multilevel model for an auto-regressive structure. I have repeated cross-sectional data.
I made a two levels model:
- Level 1: individuals.
- Level 2: years.
My dependent variable is a dummy.
I run a null model, a random intercept, and a full random model.
i specified that the covariance structure should be auto-regressive (AR(1)). However, it is taking a loooooooong time and results never show up after 30 minuts waiting even for the null model... When I use the default option (UN) results quickly show up. I must specify the dataset is very large (>200.000 individuals, 50 years).
I will gladly recieve your advise!
Here is the code I am using for the model with predictors:
proc glimmix data=thesis.socgeg;
class acyr(ref="1964") fathereduc;
model finalresult(event="1")= time fathereduc time*fathereduc / solution dist=binary;
random intercept / subject=acyr type=AR(1);
covtest 'Ay intercept var = 0' 0;
Thank you very much for your help!