|
|
Moderated mediation with count DV |
|
Message/Author |
|
w posted on Tuesday, November 27, 2018 - 5:19 pm
|
|
|
Hi, I'm trying to test for moderated mediation in the model: X -> M -> Y with a 2nd stage moderator W. X is binary, M and W are continuous, and Y is a count. To get counterfactually defined conditional indirect effects I can use the MODEL INDIRECT command MODEL INDIRECT: y MOD m w(low high increment) mw x; This gives separate indirect effects for every level of w. But is there a way to test for moderated mediation by comparing the indirect effects at just 2 levels of w? Perhaps by bootstrapping the resulting estimates? |
|
|
You can do it by expressing the 2 indirect effects and their difference using Model Constraint. The approach to a moderated indirect effect for a count Y is discussed in Section 8.2.1 in our RMA book. It is given for when the moderator W is M itself but can be modified for your case where W is not M. |
|
Back to top |
|
|