I am working on a mediation analysis Using Sobel's test, I am studying the mediating role of B in the relationship between A & C. So far, I have these regression results:
(A --> B): significant
(B --> C): significant,
(A --> C): not significant.
Upon using the Sobel test equation, the z-score was also significant suggesting a mediating role of B in the relationship between A&C.
the question is: do you think Bootstrapping is necessary in such a case?
Hello Haitham,
Sobel's test dates back a bit, but all the pieces you've presented match up with a pattern consistent with a mediation effect.
While it's always possible that an asymptotic method and a bootstrap estimate will differ to some degree, I wouldn't bother re-analyzing your data via bootstrap. Just be clear on your chosen methodology in your write-up.
Good luck with your work.
- Badges
- Science topic
- Similar topics
- Mathematics