|
|
 |
| Mplus error -- CONDITION NUMBER IS -... |
 
|
| Message/Author |
|
|
| Jessie Rudi posted on Tuesday, January 16, 2018 - 10:59 am
|
|
|
Hello,
I am running a path model and keep receiving an error despite attempting many fixes.
Here is the full error message:
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 -0.122D-15. PROBLEM INVOLVING THE FOLLOWING PARAMETER: Parameter 71, FCOLL
I have tried recoding variables so there are no zero values. All categorical variables are control variables, so I have not designated them as categorical variables in the input. I have also deleted data missing extensive data.
Any other ideas I could try to fix the problem?
Thank you. |
|
|
|
|
| If this parameter is a variance for a binary x variable, you can ignore the message. |
|
|
| Jessie Rudi posted on Monday, January 22, 2018 - 9:35 am
|
|
|
Thank you very much. Yes -- it is a binary x (or control) variable.
The output only gives zeros for the chi-square test of model fit, the RMSEA, a perfect 1.00 for CFI/TLI, and zero for SRMR.
Do you know what is causing this to happen? Or how I could fix it so I get true model fit information? |
|
|
|
|
| It sounds like the model is saturated/just-identified so that there are no left-out arrows. If so, the model is not testable - but can still be fine to use. |
|
|
| Back to top |
|
|
