Moderated mediation/conditional indirect effects after multiple imputation in Stata15

[Maria Carabello]

Hi all,

I need to run a moderated mediation analysis in Stata to estimate a conditional indirect effect, similar to what is described in Model 5 outlined here: https://stats.idre.ucla.edu/stata/fa...tion-in-stata/

However, I had to use multiple imputation to deal with missing data, and now I am running into problems with the moderated mediation since the sem command is not compatible with mi estimate.

Has anyone developed a workaround for estimating conditional indirect effects after multiply imputing in Stata?

thanks,
Maria

[Clyde Schechter]

Well, if you are estimating the model with SEM, you don't need to use multiple imputation to deal with missing values, as you can use the full-information maximum likelihood estimator using the -method(mlmv)- option.