[Mediation-information] question about moderated mediation

[Dora, J. (Jonas)]

Dear list,


In my next experiment, I would like to test a (mixed-model using lme4) moderated mediation using the mediation package.
All variables in the mediation are nested within participants (the mediator is continuous, the outcome is binary [0/1]).
The moderator is at the participant level and is also binary (0/1).
In your paper (Tingley et al., 2014), you describe two ways moderated mediation is implemented in your package.
I tried them both with simulated data and it appears to me that the test.modmed() function cannot handle lme4 models – however I could not find a definitive confirmation of this anywhere.
The second approach, where I run the mediation at two levels of the moderator (in this case 0 and 1) works, however I am having trouble understanding how to reach a conclusion with regards to my hypothesis.
Is there evidence for the moderated mediation if one of the indirect effects is significant and the other is not, or if their confidence intervals don’t overlap, or is there another way to interpret the results?
I would appreciate your help very much!

Cheers,
Jonas

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/mediation-information/attachments/20180327/93b9bcdf/attachment.html>