I am running a cross level interaction (random intercept and random slope) in Mplus. Is there a way to get the R square for it? I know standardized values are not plausible.
R^2 has a few meanings, which come to the same answer in the OLS case, but not generally. A good page on measuring model predictive value when there are random variables is: http://www.bristol.ac.uk/cmm/software/support/support-faqs/pval.html . This also includes several useful references.
One simple measure is to calculate the correlation squared between the observed response and the predicted response , I am talking about the normal-theory model. You can choose what to include in the predictions including random slopes, but of course the latter are really latent and are unknown. You would not include level 1 random terms.