More than one categorical latent in LCA PreviousNext
Mplus Discussion > Latent Variable Mixture Modeling >
 Dipali Rinker posted on Sunday, September 08, 2013 - 4:15 pm
Hello, do you have any resources for interpreting the Mplus output for more than one categorical latent? I have already looked at example 7.14 in the manual. Thank you.
 Bengt O. Muthen posted on Sunday, September 08, 2013 - 8:10 pm
Look under Papers, Latent Class Analysis on our website. For instance, I think the Feingold paper has this flavor. Then there is of course the Latent Transition Analysis literature.
 Georg Datler posted on Tuesday, November 12, 2013 - 9:48 am
Can Mplus also perform "EXPLORATORY latent class factor analysis" (well, the term used by Magidson & Vermunt 2001)?

If yes, how would Ex. 7.14 have to be altered?

Many thanks!
 Linda K. Muthen posted on Tuesday, November 12, 2013 - 12:12 pm
I think this is our TYPE = MISTURE EFA. See Example 4.4.
 Georg Datler posted on Wednesday, November 13, 2013 - 4:09 am
Thank you!
I would like to explore how many categorical latent variables ("factors") are necessary to grasp the pattern in a set of dichotomous observed variables.
(.i.e. there would be no continous latent variable as in Ex 4.4.)

Magidson & Vermunt call it the "latent class factor approach"
Vermunt, J. K., and Magidson, J. (2001), Latent class factor and cluster models, bi-plots and related graphical displays, Sociological Methodology, 31, 223264.
 Bengt O. Muthen posted on Thursday, November 14, 2013 - 8:26 am
I think that means using several latent class variables. The UG has examples of LCA with several latent class variables. So with 2 such variables you have e.g.

Model c1:

[different observed-variable means/thresholds for different c1 classes];

Model c2:

[different observed-variable means/thresholds for different c2 classes];

I think the special feature here is that for both Model c1 and Model c2 the observed-variable means/thresholds are given for all the observed variables to make the analysis exploratory.

Depending on your theory, c1 and c2 can be uncorrelated (default) or specified to correlate.
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