Interaction between latent variables PreviousNext
Mplus Discussion > Structural Equation Modeling >
Message/Author
 anonymous posted on Tuesday, January 17, 2006 - 9:26 am
I have an interaction in my model between two latent variables. I got this error message. Any suggestions to solve it? Thank you.

THERE IS NOT ENOUGH MEMORY SPACE TO RUN THE PROGRAM ON THE CURRENT INPUT FILE. THE ANALYSIS REQUIRES 7 DIMENSIONS OF INTEGRATION RESULTING IN A TOTAL OF 0.17086E+09 INTEGRATION POINTS. THIS MAY BE THE CAUSE OF THE MEMORY SHORTAGE. YOU CAN TRY TO REDUCE THE NUMBER OF DIMENSIONS OF INTEGRATION OR THE NUMBER OF INTEGRATION POINTS.
 Linda K. Muthen posted on Tuesday, January 17, 2006 - 1:44 pm
It is not realistic to estimate a model with seven dimensions of integration. We discuss this in the Mplus User's Guide. I don't know what your seven dimensions of integration come from, but I suggest rethinking your model so that you have no more than three or four dimensions of integration.
 Alex Mason posted on Thursday, January 19, 2006 - 2:44 pm
Is it accurate to say that the latent variable interaction procedure in Mplus 3.13 implements the Quasi-Maximum-Likelihood (QML) approach referred to in the Marsh et al. (2004) Psychological Methods paper and the Klein papers? Thanks!
 Linda K. Muthen posted on Thursday, January 19, 2006 - 3:20 pm
The implementation in Mplus is full-information maximum likelihood not quasi-maximum likelihood.
 Su-Ping Chen posted on Saturday, June 16, 2007 - 2:18 am
I have one question about MPlus.
QML belongs to the mew generation of methods for analyzing nonlinear effects.
Could QML approach be implemented in MPlus Version 3 or Version 4?

Thanks a lot!
 Bengt O. Muthen posted on Sunday, June 17, 2007 - 1:43 pm
If you are referring to Klein's recent QML approach for models with latent variable interactions, the answer is no.
 Sabrina Thornton posted on Friday, November 15, 2013 - 4:55 am
Hi,

I am running some latent variable level moderation effects. I do have success to run them, but I also have encountered this error message a few times:

THE MODEL ESTIMATION DID NOT TERMINATE NORMALLY DUE TO A CHANGE IN THE
LOGLIKELIHOOD DURING THE LAST E STEP.

AN INSUFFICENT NUMBER OF E STEP ITERATIONS MAY HAVE BEEN USED. INCREASE
THE NUMBER OF MITERATIONS OR INCREASE THE MCONVERGENCE VALUE. ESTIMATES
CANNOT BE TRUSTED.
SLOW CONVERGENCE DUE TO PARAMETER 53.
THE LOGLIKELIHOOD DERIVATIVE FOR THIS PARAMETER IS 0.24835653D+00.

There are times when I increase the number of iterations, the model would eventually converge, but there are times when nothing else can be done. Are there anything I can do to modify the model without changing the causa relationships that will allow me to assess the moderation effects? Thanks.
 Linda K. Muthen posted on Friday, November 15, 2013 - 11:42 am
Try increasing the number of MITERATIONS.
 Sabrina Thornton posted on Friday, November 15, 2013 - 12:17 pm
Hi Linda, just tried iterations = 1000, but it still shows the same error message. Please advise?
 Sabrina Thornton posted on Friday, November 15, 2013 - 12:17 pm
Sorry, it was miterations = 1000.
 Linda K. Muthen posted on Friday, November 15, 2013 - 1:18 pm
You can send the output and your license number to support@statmodel.com and we can take a look at it.
 Sabrina Thornton posted on Friday, November 15, 2013 - 1:21 pm
Hi Linda,

I have just tried MITERATIONS = 10000, and the model converged. What is the maximum miteration for a model to be estimated? If the same problem happen, do I just keep increasing the number of MITERATIONS? Thanks.

Sabrina
 Linda K. Muthen posted on Friday, November 15, 2013 - 3:55 pm
There is no maximum. If the error message suggests this, try it.
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: