Mediator variable turn insignificant in the c' relationship

Question

Dear all, I am currently running a mediation analysis and I'm running into some trouble interpretting the insignificance of the variables. - When running a regression for path A, this is significant at 0.017 - When running a regression for path B (not controlling for IV), it is marginally significant at 0.068 - When running a regression for path C (not controlling for mediator), it is significant at 0.004

When I include both the Mediator and the IV predicting the DV, then the signifance level of the IV is 0.009 but the Mediator is insignificant at 0.194.

My question is now, how do I interpret these results? Because in order for a mediation effect to be present, the value of the mediator has to be significant right?

Thank you in advance for your help!

Answer 1

In order to claim a mediation effect, the product of the A and B paths should be different from zero. So far, you haven't tested that. Failing to find a significant B path when controlling for the IV indicates that you likely will not find mediation, though. This doesn't mean mediation is absent; you just haven't found evidence for it (e.g., because your sample is too small or the signal is too weak).

So, you would interpret these results as follows: There is a significant total association effect of IV on DV. There is not evidence of mediation by the proposed mediator.

There isn't much more to say. The near significant relationship between the mediator and DV when not controlling for IV cannot be meaningfully interpreted because the relationship is clearly confounded by the IV.