Confirmatory GGMM? PreviousNext
Mplus Discussion > Growth Modeling of Longitudinal Data >
Message/Author
 Jessica Saunders posted on Wednesday, January 14, 2009 - 11:21 am
Is there a way to perform a confirmatory GGMM?

If my theory makes specific hypotheses about trajectory groups, can I put in the model specifications apriori and then test them against an unrestricted model? I tried this and failed miserably, and I'm not sure why - if my model specifications were written incorrectly, if what I tried to do is impossible, or if the model really didn't fit.

I was trying to replicate someone else's findings with my data. Guo et al (2002) found a 4 group model for marijuana use trajectories: (1) early high, (2) escalator, (3) late onsetter, (4) nonuser. I tried to replicate these group patterns by using model constraints.

First I labeled all the parameters I wanted to constrain:

%c#1%
[i] (a);
[s] (b);
[q] (c);
%c#2%
[i] (d);
[s] (e);
[q] (f);
%c#3%
[i] (g);
[s] (h);
[q] (i);
%c#4%
[i] (j);
[s] (k);
[q] (l);

Model Constraints:
b >0;
c < 0;
f >0;
0 < i < f;
a=0;
b=0;
c=0;

It came up with a computing error.

I also tried estimating it without any of the constraints and then put in constraints that matched what the unconstrained model found and it didn't work either. What am I doing wrong?

Thanks!!!
 Linda K. Muthen posted on Wednesday, January 14, 2009 - 4:18 pm
I think the error is likely from 0 < i < f; Confirmatory with inequalities may be difficult to do. I would just do an exploratory analysis and see if I find the same classes.
 Jessica Saunders posted on Thursday, January 15, 2009 - 11:16 am
Thanks!!
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: