Latent Class Errors PreviousNext
Mplus Discussion > Categorical Data Modeling >
Message/Author
 Itziar Familiar posted on Tuesday, October 12, 2010 - 12:18 am
Hi,

I did an LCA using 38 categorical indicators and 337 observations to fit a 3 class model.
I wanted to use the likelihood ratio chi2 to explore adequacy of model fit but I keep getting the following message:
THE CHI-SQUARE TEST IS NOT COMPUTED BECAUSE THE FREQUENCY TABLE FOR THE
LATENT CLASS INDICATOR MODEL PART IS TOO LARGE.

Are there any other options?

tks,
 Linda K. Muthen posted on Tuesday, October 12, 2010 - 1:32 pm
The chi-square test referred to is for the multiway frequency table for the categorical variables. This test works well for 8 or fewer variables. With 38, the table is too large. You can look at TECH10 to assess model misfit.
 Itziar Familiar posted on Tuesday, October 12, 2010 - 9:24 pm
Thanks for the recommendation.
However,when I added "tech10" I found the tech10 output was empty (using MPLUS 5.2)
 Linda K. Muthen posted on Tuesday, October 12, 2010 - 9:44 pm
TECH10 was added in Version 5.2. It sounds like there may be a problem for your situation.
 Itziar Familiar posted on Friday, October 15, 2010 - 2:25 pm
Hi,
I solved the problem and got the tech10 output.
I'm not sure which column I should be looking at in the bivariate model fit results (residuals between pairs of indicators). Is it the last one with the standardized residuals (z-score)? If so, is the -/+ 1.96 an appropriate cut-off point for an alpha =0.05?

tks
 Linda K. Muthen posted on Friday, October 15, 2010 - 3:13 pm
Yes, you should look at the standardized value which is a z-score and use appropriate z-score cutoffs.
 Itziar Familiar posted on Friday, October 15, 2010 - 4:18 pm
Great thanks!
In relation to tech11 and tech14 outputs; can the likelihood values for the Vuong-Lo-Mendell-Rubin and bootstrapped ratio tests be exactly the same?
 Bengt O. Muthen posted on Friday, October 15, 2010 - 4:24 pm
They typically are not.
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: