EFried posted on Wednesday, June 06, 2012 - 7:07 pm
Dear Linda & Bengt.
We would like to use MPLUS to rerun a previous R analysis for publication. Mixed effects models in R use listwise deletion, and we would prefer using FIML.
We are using a multivariate mixed-effects model in R, exploring the differential effects of covariates x1 through x10 on 9 response variables Y1 to Y9. 5 measurement points.
Y ~ Y_index * ( time + x1 + x2 + x3) + (time|subject)
Y_index denominates the 9 response variables, Y the outcome on all response variables. The random effects "time" and "subject" are correlated. The data are in a long-long format (every subject has time*Y_index lines=45 lines).
The data are usually considered "multilevel" because of the 5 measurement points, but there are no additional grouping factors.
For such models, the MPLUS manual refers the reader to chapter 6 (within chapter 9), but if we understand correctly a parallel growth model is fundamentally different from a multivariate mixed effects model with random effects.
Chapter 9 and Chapter 10 are not what we are looking for, either (our data are not complex survey and we're not looking for mixture modeling).
It sounds like you want 9 parallel processes expanding on Example 6.13.This example uses wide format. If you want long format, expand Example 9.16. I don't know what you mean by a multivariate mixed effects model with random effects.