SEM Structural Model: Should Moderators and Mediators Be Tested Together or Separately?

When conducting SEM analysis, the approach to testing mediators and moderators can significantly impact your results. Here’s a step-by-step note:

1. Test All Mediators Together: Begin by including all mediators in the model. If the results align with your hypotheses, keep the model as is.
2. Add Moderators: Next, include all moderators alongside the mediators. If the results are favorable—such as moderated mediation effects appearing in the expected direction—retain the combined model.
3. Isolate Moderators if Needed: If the results are not satisfactory, separate each moderator while still moderating all mediators simultaneously. Check if this adjustment improves the model’s performance.
4. Test Moderation Lines Individually: If the above step still doesn’t yield clear results, analyze each moderation line separately to identify potential issues or nuances.