LCA with one known/zero class PreviousNext
Mplus Discussion > Latent Variable Mixture Modeling >
 kgb posted on Sunday, January 18, 2015 - 6:18 pm
Excuse me if this is a silly qn.
I am proposing a study where I want to use LCA to identify sub-groups of problem gamblers based on 5 continuous latent class indicators. However, I want to specify a 'non-problem gambler' class (a zero class) a priori.
Is this possible, and if so, can you direct me to where I can find more information on how to do it. Im thinking it may be possible using the training variable command, but I cant find any information it.

I have done it before in a different analysis with LCGA but not by using the training command (instead I specified a zero class with I s q @ 0 and then defining high thresholds so that the probabilities of exceeding these high thresholds are zero)
 Bengt O. Muthen posted on Monday, January 19, 2015 - 9:11 am
It is hard to specify a zero class with continuous indicators because it implies zero variance for them, which results in singularity. Perhaps the continuous variables have a lower censoring point at zero in which case maybe specifying them as censored or two-part opens up a way to do it.
Back to top
Add Your Message Here
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Options: Enable HTML code in message
Automatically activate URLs in message