Comparing probit coefs with mediating... PreviousNext
Mplus Discussion > Structural Equation Modeling >
Message/Author
 Maarten Hermans posted on Tuesday, July 05, 2011 - 2:05 am
We have a recursive pathmodel (all observed) with a dichotomous dependent variable, the IV's are continuous or dummycoded and we have one ordinal mediating variable.

Using the WLSMV-estimator, we estimate the model for two groups, obtaining total, direct and indirect effects in the form of probit coefficients.

A reviewer requested formal tests/comparisons of the size and significance of those coeficients between the two groups.

Are there suggested procedures I can look into that are applicable in this situation? Are general procedures for comparing logit/probit coefs (e.g. Allison, 1999) applicable?
 Bengt O. Muthen posted on Tuesday, July 05, 2011 - 9:21 am
In the 2-group run you can define the indirect effects using Model Constraint and test equalities using Model Test.
 Andrew Whitehead posted on Wednesday, July 27, 2011 - 8:16 am
I have a recursive path model (all observed) with a dichotomous dependent variable, the IV's are continuous or dummy coded, and the mediating variable is dichotomous.

I use the WLSMV estimator and I obtain the total, direct, and indirect effects.

I also obtain the standardized effects for the model. I am aware that the standardized effects are not valid for binary variables in the WLSMV estimator.

Is there any way I can compare the effect sizes across the continuous and binary variables, including the binary mediating variable?
 Linda K. Muthen posted on Thursday, July 28, 2011 - 9:25 am
I don't know of any way to compare effect sizes across continuous and binary variables.
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: