|
|
 |
| Twolevel CFA problems (identification... |
 
|
| Message/Author |
|
|
|
|
Hi,
I'm doing a twolevel CFA, with teachers nested within schools, and I'm having a couple of problems.
First, I have a dichotomous variable indicating school type, middle school vs not middle school.
With variable in the model, I get the error:
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 -0.502D-18. PROBLEM INVOLVING PARAMETER 29.
(Where parameter 29 is the variance of this variable).
Even if there is nothing else in the model, I get this error.
My model looks like:
model:
%within%
%between%
midschool;
[midschool];
However, it appears that it is identified - at least, changing the starting values don't change the answers I get.
There's also a weird thing with the MIs, but I thought it would be clearer if that were in a separate post. |
|
|
|
|
The second part of that is the modification indices.
The modification indices suggest that the variable should be freed at the within level - and it has a fairly hefty MI. That's not possible, as it's a between variable
Minimum M.I. value for printing the modification index 5.000
M.I. E.P.C. Std E.P.C. StdYX E.P.C.
Within Level
Variances/Residual Variances
MIDSCHOO 1188.500 -1.000 -1.000 -1.000
Between Level
Beginning Time: 16:46:42
Ending Time: 16:46:42
Elapsed Time: 00:00:00
Is this something that should concern me?
Thanks,
Jeremy |
|
|
|
|
| Please send the relevant files and your license number to support@statmodel.com. |
|
|
| Back to top |
|
|
