forum.smartpls.com

Hi,
I have used a single model to test mediation effects in SmartPLS 3. The model includes the direct path from the independent variable to dependent, and indirect paths from the independent variable to mediator, and mediator to dependent. Results show that the direct effect is negative and significant (-.19), the indirect effect is positive and significant (.06), and the total effect is negative and insignificant (-.13). How to interpret the results. Is it negative confounding or inconsistent. And is competitive mediation same as suppression. MacKinnon, Krull, & Lockwood (2000) say that suppression includes negative confounding and inconsistent mediation. Nitzl et al (2016) say competitive mediation includes negative confounding and inconsistent mediation. Please help.

Based on the framework by Zhao et al. it is competitive mediation. Suppression is more a statistical term than a theoretical phenomenon. The more important question is: Does it make sense? You obviously have a negative direct and total effect, but there are aspects that also work positive and thereby reduce the negative direct effect to a less severe negative total effect. I have seen many example where such a motal makes totally sense and there is no problem with it. If it makes sense, I would not complicate the model with unnecessary statistical terms.