Fixing thresholds in LCA PreviousNext
Mplus Discussion > Latent Variable Mixture Modeling >
Message/Author
 Diana Paksarian posted on Saturday, February 18, 2012 - 7:49 am
Hi,

I have read that one can fix thresholds in order to ensure that the largest class will be the last class, which is good for the LMR LRT. My largest class is the one with low response probability on all indicators. I have tried to fix this class as last with the following syntax (this one for an 11 class model):

Model:
%OVERALL%
%c#11%
[BF00$1-BF14$1*+2]

I ran this on 9 different models and it only seemed to work twice. Can you tell me what I've done wrong?
Thanks!
 Linda K. Muthen posted on Saturday, February 18, 2012 - 8:34 am
You should have STARTS=0; when you do this.
 Diana Paksarian posted on Saturday, February 18, 2012 - 10:13 am
Okay, thanks. I'm now trying to do a BLRT and so I've taken away the starts line and replaced it with the optseed option. Will that set the starts to 0 by default?

Thanks for your help.
 Linda K. Muthen posted on Saturday, February 18, 2012 - 11:30 am
When you use starting values to reorder the classes (note this is not fixing them), you should set STARTS=0; The OPTSEED option is not needed in this case.
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: