Multigroup comparison with latent int... PreviousNext
Mplus Discussion > Multilevel Data/Complex Sample >
Message/Author
 Nancy Hill posted on Tuesday, September 27, 2016 - 8:49 pm
Hi,

I would like to test the fit of a model with a latent interaction term across 2 groups. Here's the original model:

Analysis:
TYPE = random;
Algorithm = integration;

MODEL:
FAMILY by sefsw1 pilew1 ppeaw1 pipsw1;
SCHOOL by scpsw1 rhsbw1;
JOBMKT by woiw1 wosw1;
ENGAGE by mssrw1 cecsw1;

JOBMKT ON FAMILY;
JOBMKT ON SCHOOL;
ENGAGE ON FAMILY;
ENGAGE ON SCHOOL;
ENGAGE ON JOBMKT;
FAMxJMKT | FAMILY xwith JOBMKT;
engage on famxjmkt;

I know I must use knownclass command because of the latent interaction term.
I tried this:

CLASS = c(2);
KNOWNCLASS is c (PREDBAW1 = 1 PREDBAW1 = 2);

MODEL:
%OVERALL%
FAMILY by sefsw1 pilew1 ppeaw1 pipsw1;
SCHOOL by scpsw1 rhsbw1;
JOBMKT by woiw1 wosw1;
ENGAGE by mssrw1 cecsw1;

JOBMKT ON FAMILY;
JOBMKT ON SCHOOL;
ENGAGE ON FAMILY;
ENGAGE ON SCHOOL;
ENGAGE ON JOBMKT;
FAMxJMKT | FAMILY xwith JOBMKT;
engage on famxjmkt;

Analysis:
TYPE = mixture random;
Algorithm = integration;

I get output for both classes but how do I compare fit?
 Bengt O. Muthen posted on Wednesday, September 28, 2016 - 1:01 pm
You don't get the usual fit statistics with XWITH because the sufficient statistics are not means, vars, and covars but raw data.

You can, however, test if parameters are different in the two groups.
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: