In this session, we'll be covering dual-stage moderated mediation. So the agenda for today is, again, pretty simple. Let's think about how we theorize and then how we test for these dual-stage models. I must tell you, since the beginning, that we don't many dual stage models, in our journals. This is also an opportunity to start looking at dual stage models, but also our liability, if you will, because models can get too complex and very difficult to explain. So I would rather, if you would, err on the parsimonious side and have stage 1 or stage 2 models. But if you have a strong theoretical argument for having dual stage models, go for it. We can definitely learn from these models. So, dual-stage models, we have two moderators basically. We have a moderator that is in the first stage, and a moderator that's in the second stage of this interact effect. So, the first stage, do you remember, is in this first path. So we are looking for the interactive term of the independent variable and moderator one on the mediator. And the second stage is the second path. The path between the mediator and the dependent variable. And we'll be looking at the interaction term between the mediator and the second moderator on the dependent variable. So you can find a few papers that describe these dual-stage models. One, for example, is Liu, Liao & Loi in 2012. You can go back to that paper and take a look on how they described these models. The first step when conducting these dual-stage models is to theorize about the indirect effect. Again, the starting point for moderated mediation models is the indirect effect. That's different from mediated moderation models in which the first step is the moderation, okay? So here we are looking at the indirect effect. So theorize about the effects of independent variable, only the dependent variable via a mediator. And then in step 2, we theorize about the relationship between our independent variable and our mediator being a function of the moderator one. Okay, so we are looking at the interaction term and the effects of that interaction term on the mediator. And for our third step and this is different, we are adding another moderation. We are looking at the indirect, or the effect, if the effects of our mediator on the dependent variable varies based on levels of our moderator two. So in this case we are looking at an interaction term between mediator and moderator two on the dependent variable. Finally we theorize about this overall more complex model, if you will. And we are theorizing that the effects of the independent variable on the dependent variable via the mediator can change, based on different levels of our first moderator or our second moderator, I should say. How do we test for that? Pretty similar to what we've been doing so far. We first do the random model, a mediation model. And we adopt Hayes' PROCESS macro for that. And at step 2, we run a traditional test of moderation and we are looking at the relationship, the interaction term of the independent variable and moderator on the mediator. Step 3 is a similar moderation model but now we are looking at the interaction term between the mediator and our second moderator in that effect on our dependent variable. And then we adopt process. Now we are looking at model number 29. It's important to chose the correct model again. In one of our prior sessions, I mentioned that there are more than 70 models. And it's critical to select the correct model, otherwise you may have error terms or just bogus output or outcomes that don't test for your theoretical model. Finally you look at the bootstrapping analysis and the output of that.