Multilevel SEM --everything random? PreviousNext
Mplus Discussion > Multilevel Data/Complex Sample >
Message/Author
 Anonymous posted on Tuesday, June 07, 2005 - 8:30 pm
Dear Prof. Muthen

I have a question regarding fitting multilevel SEM
in mplus. I read the manual and see examples with random factor loadings (in CFA) and random path coefficients (in path analysis).
Is it possible using Mplus to fit a multilevel SEM
model with random factor loadings and
paths between latent variables to be random as well?

Thanks a lot.
 bmuthen posted on Wednesday, June 08, 2005 - 6:55 am
Yes to both. We include examples of that in our annual November Mplus course. But, the computations are heavy and very much so as soon as you have more than a few random coefficients. Random factor loadings is harder because there are many loadings, but perhaps you can hold sets of them equal. Let me know if you have an interesting application.
 Anonymous posted on Thursday, June 09, 2005 - 3:05 pm
hi Prof. Muthen

A follow-up question, I overlooked and thought the manual has examples on random factor loading but maybe I'm wrong.

I tried to use something like (from example 9.9 in manual) and try to include random factor loading (not just random intercept)
%WITHIN%
s| fw by y1-y4

but it doesn't work.
if I tricked it by:---
------------------------------------------------
TITLE: this is an example of a two-level SEM with
continuous factor indicators and a random
slope for a factor
DATA: FILE IS ex9.10.dat;
VARIABLE: NAMES ARE y1-y5 w clus;
BETWEEN = w;
CLUSTER = clus;
ANALYSIS: TYPE = TWOLEVEL RANDOM;
ALGORITHM = INTEGRATION;
INTEGRATION = 10;
MODEL:
%WITHIN%
s| y1-y5 ON fw;
%BETWEEN%
y1-y5 s ON fb w;
OUTPUT: TECH1 TECH8;
------------------------------------------------
Still doesn't work (program complained fw and fb were not defined. How can I fit this manual example to have random factor loadings from f to
y1-y5?

Thanks a lot for your help.
 bmuthen posted on Thursday, June 09, 2005 - 6:31 pm
You have to first name the factor in a BY statement even if by a dummy statement like

f by y1@0;

And then do the random slope specification:

s1 | y1 on f;

Note that you have to specify a random slope for each loading that you want to be random.
Back to top
Add Your Message Here
Post:
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Password:
Options: Enable HTML code in message
Automatically activate URLs in message
Action: