SEM MODELLING PROBLEM PLS HELP PreviousNext
Mplus Discussion > Structural Equation Modeling >
Message/Author
 KKO posted on Wednesday, July 07, 2004 - 12:51 am
There is an error message of my program that is
NO CONVERGENCE. SERIOUS PROBLEMS IN ITERATIONS.
ESTIMATED COVARIANCE MATRIX NON-INVERTIBLE.
CHECK YOUR STARTING VALUES.

Can u tell me how to solve t


INPUT INSTRUCTIONS

TITLE: this is the analysis of a confirmatory SEM two level modelling
DATA: FILE IS data1.txt;
VARIABLE: NAMES ARE s1-s3 x1 y1-y3 dist;
USEVARIABLES ARE s1-s3 y1-y3;
MISSING IS .;
CLUSTER is dist;

ANALYSIS: TYPE=COMPLEX;
ITERATIONS = 10000000;
CONVERGENCE = 0.00005;

MODEL:

f1 by y1 ;
f2 by y2 ;
f3 by s1-s3;
y3 on f3;
f1 ON f3;
f2 ON f1 y3;


OUTPUT: STANDARDIZED MODINDICES(0) TECH1;
 KKO posted on Wednesday, July 07, 2004 - 1:07 am
y1@0;
y2@0;
s1@0;

i put the above
into the model
but
the model still cannot be identified ?
why
 Linda K. Muthen posted on Wednesday, July 07, 2004 - 6:51 am
A good place to start if you have convergence problems is by looking in the Index of the Mplus User's Guide under Convergence Problems and following the suggestions given there. If you continue to have problems, send the full output and data to support@statmodel.com.
 Anonymous posted on Monday, January 03, 2005 - 1:41 pm
Even when following the suggestions on pp 160-62 of the Mplus Guide i continue running in convergence problems. Could it be because of the relatively small sample size (n=150) or is due to low factor variance?
 Linda K. Muthen posted on Monday, January 03, 2005 - 4:12 pm
Depdending on the model, it could be one of those things. Why don't you send your output and data to support@statmodel.com and I will take a look at it.
 Matevz Raskovic posted on Thursday, November 17, 2011 - 6:39 am
Hi, this is my first time joining the dicussion group. I am a PhD student running my first SEM in Mplus.

I have a model, which is well grounded in theory and should be ok. When running the model in Mplus, I got the following WARNING which I do not know quite how to either interpret or fix. Can you please help me with what to do?

Mplus WARNING:
"THE STANDARD ERRORS OF THE MODEL PARAMETER ESTIMATES MAY NOT BE TRUSTWORTHY FOR SOME PARAMETERS DUE TO A NON-POSITIVE DEFINITE FIRST-ORDER DERIVATIVE PRODUCT MATRIX. THIS MAY BE DUE TO THE STARTING VALUES BUT MAY ALSO BE AN INDICATION OF MODEL NONIDENTIFICATION. THE CONDITION NUMBER IS -0.544D-16. PROBLEM INVOLVING PARAMETER 52."

Thank you for all your help, Matevz
 Linda K. Muthen posted on Thursday, November 17, 2011 - 7:27 am
Please send your output and license number to supprt@statmodel.com.
 Lior Abramson posted on Thursday, August 20, 2015 - 8:54 am
Hello,
I have a data set of two variables measured at two time points. I tried to do a SEM analysis according to the example 6.13 on your guide: GROWTH MODEL FOR TWO PARALLEL PROCESSES FOR CONTINUOUS OUTCOMES WITH REGRESSIONS AMONG THE RANDOM EFFECTS.

I got the following message: THE DEGREES OF FREEDOM FOR THIS MODEL ARE NEGATIVE. THE MODEL IS NOT IDENTIFIED. NO CHI-SQUARE TEST IS AVAILABLE. CHECK YOUR MODEL.

Could the problem be that I only have two time points for each variable? How many measurements of each variable are required for this kind of analysis?

Thank you
 Linda K. Muthen posted on Thursday, August 20, 2015 - 11:47 am
Yes, this could be a problem. For a linear model, a minimum of three time points is needed. With two time points, an intercept only model can be estimated.
 Hebbah Elgindy posted on Thursday, March 03, 2016 - 11:36 pm
Hi! I'm running a longitudinal latent path analysis. I keep getting the following message.

*** WARNING
Data set contains cases with missing on all variables except
x-variables. These cases were not included in the analysis.
Number of cases with missing on all variables except x-variables: 110

This is despite having good covariance coverage.

How do I resolve this?
 Linda K. Muthen posted on Friday, March 04, 2016 - 9:00 am
There is no solution to this. Observations with all missing data on the set of dependent variables in the model cannot contribute to model estimation. Depending on your sample size, coverage may still be high.

If you think you are reading your data incorrectly, send the output, data set, and your license number to support@statmodel.com.
 Hebbah Elgindy posted on Tuesday, April 05, 2016 - 1:35 pm
Hi again!

As mentioned earlier, I am running a manifest path analysis and have run some interactions.

I would like to probe my interaction in Excel and have been looking for the output command that can provide the 'constant' value so that I can input all the correct information in my Excel sheet.

However, I have looked through the manual and could not find such an output.

So far, I use the following outputs commands:

tech1; tech4; stand; samp; mod (10)

I would be grateful for some guidance.

Best,
H
 Bengt O. Muthen posted on Tuesday, April 05, 2016 - 6:54 pm
What do you mean by "constant value"? An intercept?

For interaction plotting see also

http://www.statmodel.com/Mediation.shtml
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: