I'm working on a moderation mediation model where the independent variable has four categories, the moderator is binary, and the mediator and the outcomes are continuous. I followed Hayes et al. (2013) and created three dummy codings to first conduct a mediation model and then chose a multigroup approach (because the moderator is binary and there exist two groups) to detect possible moderating effects. It is clear how to make inferences for relative indirect effects as "Evidence that at least one relative indirect effect is different from zero supports the conclusion that M mediates the effect of X on Y." (Hayes et al., 2013, p. 462). However, I'm not quite sure how to make inferences about the moderating effects on the paths between the three coding variables and the mediator. Is it similar that the moderator moderates at least of the three paths between the three coding variables and the mediation is enough to conclude that there is a moderated mediating effect (I attached an image to show the conceptual model for clarity)? Any suggestions and help would be appreciated!
Reference:
Hayes, A. F., & Preacher, K. J. (2014). Statistical mediation analysis with a multicategorical independent variable. British Journal of Mathematical and Statistical Psychology, 67(3), 451-470. Article Statistical mediation analysis with a multicategorical indep...
Hello Bo,
If any of the dummy IV X moderator interaction effects is non-zero (with mediator as the downstream/target variable), then you have evidence for moderation. If you want to know the collective explanatory power of the moderation, then R-squared of the full model (using mediator as DV) - R-squared of a reduced model, omitting the three interaction terms, would give you this value.
Good luck with your work.
- Badges
- Science topic
- Similar topics
- Computer Science
- Computer Communications (Networks)
- Internet