The estimator for Zero-Inflated NB + ... PreviousNext
Mplus Discussion > Categorical Data Modeling >
 Jae Seung Lee posted on Sunday, November 07, 2010 - 10:22 pm
I have tried to estimate a model for an count dependent variable with many zeros.
The independent variables are two latent variables with 9 indicators, as well as 17 observed variables.
In this case, what is the best estimator?
And how can I find the equation of the estimator?

Thank you for your help in advance.
 Linda K. Muthen posted on Monday, November 08, 2010 - 6:17 am
For a model with a count variable, only maximum likelihood estimation is available. The default is the MLR estimator.
 Jae Seung Lee posted on Saturday, November 20, 2010 - 7:07 pm

However, MLR or ML estimator is under normality assumption. But count data is apparently not the case.

So how is this non-normality problem dealt with in Mplus when using MLR or ML?

Thank you
 Bengt O. Muthen posted on Sunday, November 21, 2010 - 9:52 am
Maximum likelihood estimation is not limited to continuous-normal outcomes. What is required for ML is to assume a distribution for the outcomes, and then the likelihood follow. This means that you can work with many different types of outcomes and distributions, including counts assuming say Poisson or NB, or categorical assuming say logistic.
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