Multiple Regression with Imputed Data PreviousNext
Mplus Discussion > Missing Data Modeling >
 Larissa Gaias posted on Wednesday, November 23, 2016 - 9:50 am

I am trying to run a SEM (mediation) model using imputed data. For some reason, when I try to use my outcome variable, I get no results. An output file is created, but it only displays the input model, no errors, and no results. The model including this variable ran fine when I was using FIML, but I need to use imputation because my mediator is dichotomous. Is there something that could have happened during the imputation process that would have caused this variable to be problematic? I get no errors so it is difficult to diagnose what is going on. I tested the model using different variables as placeholders, and I get results, so I think it is definitely an issue with this variable specifically (I also have the same problem when including the same variable measured as a different time point that I have in the model as a covariate). I would love any guidance if you have seen this happen before.

Thank you for your help!

 Bengt O. Muthen posted on Wednesday, November 23, 2016 - 2:42 pm
When you say "using FIML" and needing imputation because of a binary mediator, I am inclined to think there is a misunderstanding. A common misconception is that FIML refers to continuous variables. You can use FIML also for a binary mediator. A binary mediator, however, calls for special expressions of the indirect and direct effects - see
"Counterfactual causal effects for mediation modeling" on our Mediation page

If this doesn't help, send input and data 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