LL difference testing between two models PreviousNext
Mplus Discussion > Growth Modeling of Longitudinal Data >
Message/Author
 Johnny Wu posted on Tuesday, November 21, 2006 - 7:55 am
Dr. Muthen,

The "Difference Testing Using the Loglikelihood" strategy applies for testing a more restricted model vs. a more freed model.

However, what if I want to test two models that have the same variables, same parameters, but different restrictions. Because p0-p1 = 0, I cannot use the difference testing formula.

For example, I have an i & s, and two covariates x1 & x2.

The models that I wish to compare are:

(1) i ON x1 x2@0;
s On x1 x2@0;

vs.

(2) i on x1@0 x2;
s on x1@0 x2;

How would I determine which is a better fit?

Thanks, johnnywu
 Bengt O. Muthen posted on Tuesday, November 21, 2006 - 8:09 am
You can compare the loglikelihoods - I know of no test. You can also use Model test and impose the restrictions in (1) and in (2) and compare those p values.
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: