Mediation analysis

[psychstud]

Hello ! For an undergraduate project we have to do a mediation analysis. Here is the experimental design :

• We have a first variable V1 (mediator) that we will measure a first time.
• Then, one week later, we manipulate an independant variable V2 (which has 2 groups) and then the V1 again and other dependent variables.

This should enable us to see if V2 increase or decrease V1 so if V2 has an effect on V2 and other dependent variables. We also think that the effect of V2 on the dependent variables is mediated by V1 (hope it is clear).

My question is how to estimate the total sample size needed in G*Power ? I chose [ANOVA : Repeated measures, within-between interaction] but how to estimate the sample size when we add such a mediation hypothesis.

I also would like to know, in an experimental design with a mediation hypothesis, should the mediator always be between the independant and the dependent variable. Or for example, can the mediator be measured after the dependant variables.

Last question : For the mediation analysis to be verified, the complete indirect effect should be significant and the direct effet should not be significant. But can we still verify the hypothesis if the direct effect is still significant but reduced ?

THANKS FOR THE HELP !!