07-17-2017 11:50 AM
I am trying to predict poor vs. high self rated health in a multi-national survey with complex sampling. I have two time points and I am using proc glimmix with cluster and sample weights. All of my covariates are recorded at baseline, and my outcome is at follow up. When I add baseline self rated health at time 1, the model does not converge. What could be the reason for this?
proc glimmix data=final;
class multim(ref="6") marpartdumw5 educ femalew5 csmokerw5 country;
model poorfairsrhw6 = poorfairsrhw5 multim marpartdumw5 educ femalew5 csmokerw5 BMIw5 agew5 hnetww5 / dist=binomial link=logit
random intercept / subject = country;