Is it possible to have significant direct and indirect effects but the total effect is not significant?
Also, is it possible to have significant direct effect but the total effect is not significant?
What about having the total effect the only significant one (direct and indirect effect are not significant), is that also possible?
If any one of these questions is answered yes, what does it mean? any example to simplify it?
Thanks
Hello Mona,
assumed that the model is correctly specified anyway (i.e., no omitted confounding and no reverse effects), yes, everything is possible:
1) If the direct and indirect effect differ in their sign, the total effect (i.e, their sum) may be non-significant, as both add to the total effect and may eliminate each other.
2) If the direct and indirect effects are of equal sign but miss significance, their addition (=total effect) may be significant.
HTH
Holger
I completely agree with Holger Steinmetz .
2) sometimes it happens within mediation models through full transmittal approach (Memon et al., 2018). Direct path (c) is significant and positive, a path is significant and positive, but b is significant and negative. Thus, the total effect on the outcome variable is the sum of a positive direct effect (c) and a negative mediated effect (b). Therefore, it loses significance.
Memon, M. A., Cheah, J.-H., Ramayah, T., Ting, H., & Chuah, F. (2018). Mediation analysis: Issues and recommendations. Journal of Applied Structural Equation Modeling, 2(1), i–ix. https://doi.org/10.47263/JASEM.2(1)01
If the total effect direct effect is insignificant there is no need to report mediation.
- Badges
- Science topic