I run a cox frailty model(model 1) in R, by adding new co-variate to model(model 2) The ACI decreases that show the new model is better than model 1, but variance of random effect of model 2 is more than model 1, what is the reason, do I omit this co-variate from model? is it reasonable that variance increase by adding significant co-variate? How can I interpret that?
Aucun commentaire:
Enregistrer un commentaire