

Correlations between IVs in multileve... 

Message/Author 


Is is correct to say that Mplus always estimates the correlations between the independent variables (or predictors) although they are not shown in the model output? With regard to multilevel models, in specific, it seems not possible to estimate a correlation between a Level 1 and a Level 2 predictor in Mplus using a WITH statement. So, can I then just assume that this 'crosslevel correlation' is modeled but that I can't see the value in Mplus (and if I wanted to see it, I should calculate them in a program like SPSS)? 


Answers to your two questions: Yes. You can correlate the cluster mean of a variable observed for the individual with a clusterlevel variable. The number of observations is the number of clusters. 

Back to top 

