Increasing starting values PreviousNext
Mplus Discussion > Latent Variable Mixture Modeling >
Message/Author
 Ali posted on Sunday, June 29, 2014 - 12:33 pm
Hello,
I am doing simulations of mixture Rasch model .

I generated the data which are 2 class,and I tried to fit the data into 1 class, 2 class, 3 class with starting values 100 20.

When I fitted the 3-class solution, it always gave the message"THE BEST LOGLIKELIHOOD VALUE WAS NOT REPLICATED" or" THE BEST LOGLIKELIHOOD VALUE HAS BEEN REPLICATED.
THE MODEL ESTIMATION DID NOT TERMINATE NORMALLY DUE TO A NON-POSITIVE".

Do I need to increase staring values when fitting the 3-class solution? I increased starting values (500 125), but it took longer time. I am wondering if I should increase staring values when fitting 3 classes.
 Bengt O. Muthen posted on Sunday, June 29, 2014 - 5:21 pm
Yes, with more classes you may need more starts. Also, it sounds like you did have the best LL replicated but had a non-pos problem that should be investigated.
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: