

Estimate covariance of random slope 

Message/Author 

Giselle Guan posted on Saturday, February 08, 2020  7:54 am



Dear Dr. Muthen, I'm trying to conduct multilevel path analysis using Mplus. In the model, I want to test a crosslevel interaction using random slope:  %WITHIN% s1 Y ON X_within; !the random slope between the DV and IV %Between% Y WITH X_within; Y ON X_moderator; s1 on X_moderator; ! The 2way interaction X_within with X_moderator;  Question 1: I understand we should covary all exogenous variables (in this case, X_within and X_moderator). Should I also covary s1 and the dependent variable Y? Question 2: Is s1 treated as a endogenous variable? Question 3: Should I covary s1 and X_moderator? Question 4: At the between level, instead of modeling the relationship between X_within and Y per se, I only allow them to covary. I believe in this way I can have a simpler model, which is more likely to converge. Is it correct? Many thanks! 


Because you call it X_within, I assume that you would put this covariate on the Within list. If so, you would not mention it on Between. 1. Yes, you should covary s1 and Y on Between. This is a residual covariance. 2. Yes. 3. No, you already regress s1 on X_moderator. 4. See my intro comment. 

Back to top 

