Three-path mediation, dichotomous DV PreviousNext
Mplus Discussion > Structural Equation Modeling >
Message/Author
 Rebecca Casciano posted on Thursday, November 03, 2011 - 11:15 am
I apologize in advance if this has been covered in another thread, but I couldn't find the answers I needed.

I'm running a three-path mediation analysis with bootstrapped CIs. I'd like to use logistic regression to estimate the final DV and I'm not sure how to do this.

y1=dichotomous
x1=dichotomous
m1&m2=continuous

(Abridged) code:

Usevariables are y1 x1 m1 m2 ;
Categorical are y1 x1 ;
Analysis:
Bootstrap = 10000 ;
Model:
m1 ON x1 ;
m2 ON m1 x1 ;
y1 ON m2 m1 x1 ;
MODEL INDIRECT;
y1 ind m1 x1 ;
y1 ind m2 x1 ;
y1 ind m2 m1 x1 ;
OUTPUT:
cinterval(bcbootstrap) ;


When I run this, I get estimates that are different from my estimates in Stata. I assume Mplus is giving estimates from probit regressions, but when I specify "Estimator = ML" I get an error message:

"BOOTSTRAP is not allowed with ALGORITHM=INTEGRATION."
(though I haven't specified algorithm=integration)

Any thoughts on what I'm doing wrong?

Thanks!
 Linda K. Muthen posted on Thursday, November 03, 2011 - 11:28 am
It sounds like numerical integration is required for your analysis because your mediator has missing data. The BOOTSTRAP option is not available in this case. You should remove BOOTSTRAP and CINTERVAL(BOOTSTRAP).
 Rebecca Casciano posted on Thursday, November 03, 2011 - 11:44 am
I don't have missing data on any of the variables.
 Linda K. Muthen posted on Thursday, November 03, 2011 - 1:51 pm
Then I would need to see the full output to understand the situation. Please send it and your license number to support@statmodel.com.
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: