|
Message/Author |
|
|
Hello, I would like to get or save the estimated covariance matrix with more than three decimals. Is there any possibility to do this. In need this for a power calculation mentioned on your side (http://www.statmodel.com/power.shtml). Currently, when I check the model in step 2, there is some missfit in the model. I think this is due to rounding errors in the estimated covariance matrix. If it is not possible to the get the estimated covariance matrix with more decimals, is it admissible that I calculate the difference between the unconstrained model (that should perfectly fit, but not does) and the constrained model to get the noncentrality parameter. Thanks in advance for the answer. Best regards Christopher |
|
|
If you increase the ficticious sample size I think you will get a closer fit. |
|
|
When you save the sample statistics using the SAVEDATA command, you obtain 8 decimal points. |
|
Back to top |
|
|