Type of Model PreviousNext
Mplus Discussion > Structural Equation Modeling >
Message/Author
 Anonymous posted on Thursday, December 16, 2004 - 6:39 am
This is a quick question with hopefully a simple solution that I am almost certain that I am overlooking. I was wondering about the type of analysis (latent means MIMIC model or Mixture model) for the following research issue:

I am trying to determine the effect of group membership (males vs. females)[x1]{discrete indicators} on a specific latent trait (crime)[f1] {continuous indicators}. However, the effect of group membership on crime is hypothesized to be mediated by another latent trait (self-esteem) [f2] {continuous indicators}. The data are cross-sectional.

With this research issue in mind, I was wondering which model will provide me with satisfactory results a modified version of a MIMIC model or a Mixture model? Also, could you point me to an example in the Mplus users guide that you think may put me on the correct track. Thank you very much.
 Linda K. Muthen posted on Thursday, December 16, 2004 - 8:24 am
It does not sound like you need mixture modeling. You would need this is you are trying to find unobserved heterogeneity in your data which it doesn't sound like you are doing. It sounds like you are interested in a model such as:

MODEL:
f1 BY ....
f2 BY ....
f2 ON x;
f1 ON f2;

Examples in Chapter 5 should guide you.
 Anonymous posted on Thursday, December 16, 2004 - 11:05 am
Would example 5.8 in the users guide be a concrete example that you are foreseeing form my research issue? Of course I would have to modify it to address my structural model needs.
 Linda K. Muthen posted on Thursday, December 16, 2004 - 12:53 pm
A modified version of this could work.
 Thomas Olino posted on Saturday, February 19, 2005 - 8:53 am
The general model that I am trying to run is: f1 BY x1-x4. However, I also want the residual of x2 to be regressed on the residual of x1, the residual of x3 to be regressed on the residual of x2, and the residual of x4 to be regressed on the residual of x3. Is this possible?

I'm having difficulty in seeing how to do this in the User's manual.

Thanks!
 bmuthen  posted on Saturday, February 19, 2005 - 9:10 am
To regress residuals onto each other, you have to define them as factors. You do that using BY statements - remember to delete the default residual by fixing its variance at zero.
 Thomas Olino posted on Saturday, February 19, 2005 - 1:18 pm
Do you treat the variance for the exogenous residual any differently?

Thanks again.
 bmuthen posted on Saturday, February 19, 2005 - 1:21 pm
No, same approach.
 Marion S. posted on Wednesday, April 27, 2005 - 7:28 pm
Is counterfactual modeling a type of SEM and if so can I use Mplus to do it?
 Linda K. Muthen posted on Thursday, April 28, 2005 - 8:26 am
If by counterfactual modeling you refer to models such as Little and Yau's Compiler-Average Causal Effect (CACE) modeling, Mplus can estimate this model. If this is not what you mean, please elaborate.
 Marion S. posted on Thursday, April 28, 2005 - 8:00 pm
Yes. That is what I have meant. Thank you.
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: