Estimator for Two-part cross sectiona... PreviousNext
Mplus Discussion > Categorical Data Modeling >
 anne marie mauricio posted on Friday, March 02, 2007 - 12:02 am
Dera Drs. Muthen, I am running a two-part cross-sectional model. If I do not specify an estimator, the default estimator is WLSMV. However, I thought that, because the MCAR assumption is not met for two-part modeling, the estimator ML had to be used. I want to test for indirect effects in the context of the 2-part model and, thus, would like to use WLSMV as the estimator, if possible. Can I use WLSMV as the estimator w/ cross-sectional data and 2-part modeling?

Thanks so much in advance for your assistance.

 Linda K. Muthen posted on Friday, March 02, 2007 - 8:27 am
You should use maximum likelihood.
 anne marie mauricio posted on Friday, March 02, 2007 - 8:47 am
Thank you very much Linda.
 Mirim Kim posted on Thursday, November 10, 2016 - 9:59 am
Dear Dr. Muthen,

I'm running a two-part model for measurement invariance testing without using "Data Twopart" default.
I thought I might use other estimator without the default. However, there was a convergence problem, and output said that paired binary variable and continuous variable were highly correlated. Does that mean the default algorithm makes this problem subtle?

Under MLR, I found one of alpha parameters (found in tech 1 for latent class regression model part) was estimated, and that was a very small value.
Could you explain what does that mean?

I appreciate your answer in advance!



G#1 G#2
74 (Estimated) 0 (non)
 Bengt O. Muthen posted on Thursday, November 10, 2016 - 12:33 pm
We need to see your outputs to help you - please send to Support along with your license number.
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