My research model deals with the use of digital systems, with the following structure:
- Bild1.jpg (19.4 KiB) Viewed 1013 times
The aim is to analyse whether DV1 is explained by the IVs, and whether DV1 has an influence on DV2. The main focus is not to explain DV2 by IVs (would make little sense, as there are too many influencing factors on “job satisfaction” (DV2)).
I have no literature basis for IV--DV2, so I didn't create hypotheses for IV--DV2.
The evaluation shows: All hypotheses supported (significant, small to medium effect sizes).
Now, a reviewer demands a mediation analysis arguing that the model structure implies full mediation.
If I perform a mediation analysis according to Hair et al. (2022) “A Primer on PLS-SEM”, I include the direct effects IVs--DV2 into the model. The evaluation results in no effects (no mediation) and direct effects (no mediation), as of course including the direct relationships IV--DV2 into the model change the significance of some paths compared to the model without direct relationships.
My questions:
Q1: Do I have to perform a mediation analysis based on the model structure even if the focus is on explaining DV1?
Q2: For mediation analysis, do I have to include the direct relationships, even if I cannot derive the corresponding hypotheses? I worry that this opens up relationships that are not relevant to my research questions and rather distort my actual focus.
I would be very, very grateful for any advice as I just don't know how to deal with this!