Parallel process growth models PreviousNext
Mplus Discussion > Categorical Data Modeling >
Message/Author
 Viswanathan Ramanathan posted on Wednesday, May 03, 2006 - 8:37 pm
I am trying to run a parallel process growth model with 2 categorical variables - prcat11- 19 and wdraw1-9. The unconditional growth models for both these ran without difficulties.

Here is the syntax I used to run the parallel process model -
model:
picept pslope | prcat11@0 prcat12@1 prcat13@2 prcat14@3 prcat15@4 prcat16@5
prcat17@6 prcat18@7 prcat19@8;

aicept aslope | wdraw1@0 wdraw2@1 wdraw3@2 wdraw4@3 wdraw5@4 wdraw6@5 wdraw7@6 wdraw8@7 wdraw9@8;

picept on aicept;
pslope on aicept aslope;

picept with pslope;
aicept with aslope;

I am getting this message -

THE MODEL ESTIMATION TERMINATED NORMALLY

FACTOR SCORES COULD NOT BE COMPUTED. AT LEAST ONE OF THE RESIDUAL
VARIANCES FOR THE CATEGORICAL VARIABLES IS NOT POSITIVE.

What does this mean? Are the results interpretable? Is such a warning an indication of a poor relationship between the two processes?

Should I constrain any parameters to get better results? Which ones? Can you please point me to relevant syntax?

Thanks
Visu
 patrick sturgis posted on Wednesday, May 03, 2006 - 11:03 pm
Visu

You could have a residual variance of zero which is being estimated as a very small negative value. Examine these parameters and if one is negative and the z score (est/s.e.) < 1.96, rerun the model with this paramter constrained to zero. If the negative variance is significant, you may have a problem with your model specification. In my experience, though, this usually works. Bengt/Linda may have other/better advice! Best,

Patrick
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: