I would be very grateful about any advices or comments on an issue I have recently come across. We tested a simple mediation analysis, hypotezing idirect effect (X->M->Y), which has been found to be significant. The direct effect was also significant, however in the opposing way (thats: while X had an decreasing effect on Y through M, X had also increasing direct effect on Y). This situation indeed resulted in a non-singificant total effect. From the conceptual standpoint its very interesting that the studied effect is thus differentiated (and it has some sense), however:

Is it possible to interpret the indirect/direct effects even though the total effect is not significant? And the second one - would it be possible to interpret just indirect effect if both total and direct effects were insignificant?

Thank you!