# Mediation: How can the Total Indirect Effect be significant when the Indirect Effects are not?

I ran a mediation analysis with two mediators (M1 and M2) using lavaan and I am somehow confused about the results:

Direct Effect: The total direct effect of X on Y is not significant. X is, however, significantly related to M1 and M1 is significantly related to Y. As for M2, no significant direct relationships are observed.

Indirect Effect: The indirect effect of X on Y through M1 is not significant. The same also applies to M2.

Total Indirect Effect: When adding the two indirect effects, a significant result is shown. How is that possible?

Total Effect: The total effect is not significant.

I am mostly confused about the Total Indirect Effect. How can it be significant when the two indirect effects are not?

It might be useful to add that I let M1 and M2 covary. Could this serve as an explanation?

• "I let M1 and M2 covary. Could this serve as an explanation?" No. Dec 28, 2020 at 20:05
• The total indirect effect is a bit like a multivariate test, where the individual effects are univariate tests. They should not be expected to agree. You picked a threshold (presumably 0.05). Pick a different threshold and this won't be true. Dec 28, 2020 at 20:06

The total indirect effect is actually the sum of all specific indirect effects, or, more simply the total effect between $$X$$ and $$Y$$ less the direct effect between $$X$$ and $$Y$$. You can obtain barely significant specific indirect effects (which I guess happened) and when you add them up, it can reach the significance threshold.