Moderators and mediators

OK, this just fried my brain, but I think I have it straight now.  In creating a predictive model, two relationships can be seen if a direct relationship between the predictor (x) and the outcome (z) cannot be observed.  In the case of mediation, a quasi-causal relationship can be established between predictor and mediator variable, and a quasi-causal relationship can be established between the mediator and outcome variable.  This is depicted in Figure 1 on the right.  In the case of moderation, the impact of the predictor variable on the outcome variable is directly influenced by the value of the the moderator variable, and there is no causal relationship between the predictor variable and the outcome variable.

Figure 1: Mediator ModelFigure 2: Moderator Model