Hello scientific world,
I just simply wonder if 8 to 10 mediators that I found as important for my model is just not too much?
One idea that comes to my mind is that I could reduce this number and treat some of them as moderators...
But then I may be tempted to test moderated mediation with 5 mediators and 5 moderators....
Please state your opinion on that
Thank you
PS: Perhaps some other researchers share similar doubts.
Witold,
any such decision should be based on theory. Theory indicates the role of each variable in your model.
Also, mediation and moderation are two different things, so I would recommend thinking very carefully about which role each variable would play, instead of just trying out different model structures.
Maybe this helps:
Article Improving Our Understanding of Moderation and Mediation in S...
Finally, if you want to draw causal inferences from your model, note that you need temporal precedence as a very basic condition for causality. This implies, among other things, that mediation cannot be validly tested in cross-sectional research designs.
See:
Kline, R. B. (2015). The mediation myth. Basic and Applied Social Psychology, 37(4), 202-213.
Usually best to keep it simple (KIS) to begin with and only increase complexity if essential.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics