Testing significance of level-2 variance

[ccguarnizo]

Dear All,

I’m working on multilevel logistic regression models with MCMC methods (using runmlwin). May be this is a silly question, but in my empty model I just want to test if the level-2 variance is significantly different from zero. I’ve checked the materials and do files available but couldn’t find examples of how to do this for MCMC using runmlwin. Is it just typing "test [RP2]var(cons)=0" after fitting the model?

I'd appreciate it if someone could help me with this.

With best wishes,

Carol

[GeorgeLeckie]

Hi Carol,

Using Stata's test command will perform a t-test (actually a Wald test, i.e., square of t-test) but this assumes that the sampling distribution of variance is normal which it is not. Variances tend to have +ve skewed sampling distributions

Best option is to compare DIC statistics for the two nested models

Smaller DIC stats are preferred

Best wishes

George

[ccguarnizo]

Hi George,

Thanks a lot for the helpful answer and the fast response to my question!

Best wishes,
Carol