0
$\begingroup$

I found a very old question here - Mediator reverses direction of causal variable. How to interpret?, but it isn't answered. I am having almost the same issue as @jsd. I've run mediation following Baron and Kenny and confirmed it with the Sobel test. Here's what I get when I run the regressions (all variables are continuous):

  • X significantly negatively predicts Y
  • X significantly negatively predicts M
  • M significantly positively predicts Y
  • X + M significantly positively predict Y
  • In the last model with both X and M, M no longer significantly predicts Y.

I checked moderation since the relationship reversed. There is no moderation. X and M do not significantly interact.

This post (Mediation (b path sign changes from the correlation)) suggests it is mediation, but that it is suppression. I looked at suppression, but if I'm understanding it right, X and M are not supposed to have a relationship in suppression. So, do I have mediation? Or suppression? Or something else?

Thank you in advance for any help!

$\endgroup$

1 Answer 1

0
$\begingroup$

I think I just figured it out. I have inconsistent mediation - Equivalence of the mediation, confounding and suppression effect MacKinnon et al., 2000 found here: http://europepmc.org/article/PMC/2819361

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.