In this session we will talk about serial mediation. So first we'll talk about, we'll explain what is serial mediation and then we'll walk through a way to test for serial mediation models. So what is serial mediation models? Initially we'll compare that a little bit to the parallel mediation model and we covered this in our prior session. So again parallel mediation model we have multiple mechanisms connecting our independent variable to our dependent variable. Our job meaningfulness to our job performance via job engagement and commitment. But now we are talking about serial mediation models. We have our independent variable influencing our dependent variable via a series of mediations. So there is one distant relationship between our independent variable and our dependent variable. The example that I will be using here is very similar to the example that I've been using so far and we have our job meaningfulness influencing our job performance. But first the relationship that we will look at is job meaningfulness and the job engagement and we are theorizing that the job engagement influences job commitment, and then job commitment influencing performance. The distinction between parallel mediation and serial mediation should be done in your theory, in your theoretical development. You could have similar variables running the models are not- I mean is not a very complex or complicated thing to do, but you have to think about, think through your model in your theory development session to make the distinction between parallel and serial mediation models. An important distinction from an analytical standpoint from a tools perspective is that you needed to use a model six instead of model four to test for serial mediation models. Remember I've emphasized that for mediation we needed to use model four. But for serial mediation models, we needed to use model six off the process method developed by Hayes. So this is the model that we will be testing. So we have meaningfulness influencing job engagement and then job engagement influencing job commitment and then job commitment influencing performance. In a sequel order in a sequence. So again there is this distant relationship between meaningfulness and performance and that relationship is mediated by engagement and commitment. There is no limitation on the number of mediators that you can have in between. OK, look at here we have only two. We could have many mediators but again for parsimonious models we want to keep it simple. Let's keep it simple. Usually we look at two perhaps three mediators not more than that otherwise, it will become too complex, too difficult to explain in your theoretical development session. So now let's run this analysis. How do we run mediation models or serial mediation models I should say? Again, go to analyze and you should be familiar with that now if you watched all the prior sessions we explained how to get to these screen here. But again go to analyze and then all the way down to regression. And when you click on a regression there is a series of options that you can choose from. We are using the process macro developed by Hayes. So click on the process macro. That's the next screen that you'll see and you'll have to move your variables to the right position here to the right variable name. So for our outcome variables we have performance, for our independent variable we have meaningfulness, and for our mediator variables we have job engagement and commitment. This is exactly the same screen that you have in our parallel mediation session. What changes though is again the model number. Now you needed to choose model number six, not model number four. Once you do that click on OK and this is a screen that you should see as your outcome file. Model number six, always double check your model number, performance, meaningfulness, job engagement, job commitment and our sample size. You'll also notice that we have M1 for mediator one and M2 for a mediator two. That's exactly the same information that we get in our parallel mediation model testing. The only difference again is the selection off of the model. Now it is model number six. The way that we test for and the way that we look in our output in screen here on output file is very similar to our parallel mediation model. But first what we are looking at is a positive and significant relationship between job meaningfulness, our independent variable to our first mediator job engagement. We do find a significant positive in this case, relationship between meaningfulness and job engagement. So this is the first step. The second step we needed to find a significant relationship between our mediator one and our mediator two, in this case job engagement and commitment. And we do find that job engagement has a significant relationship with commitment here. P is less than 0.05. This is a step two. And then we have step three. In our step three we look for the relationship between our second mediator commitment and our digital outcome variable performance. In this case we look at commitment and performance. What do you see here? This is not significant. P is not less than 0.05. So we do not have evidence for a serial mediation model. We did not because our third step here is not significant. We don't find a significant relationship between commitment and performance. So let's look at the composition of the facts. We have total direct, and indirect effects. The total effect is significant, the direct effect is not significant. And here what's interesting now is that we have multiple indirect effects. We have indirect effect number one, which is job meaningfulness, influencing performance via engagement and from other one, we do have a significant indirect effect. Why? Because we don't have zero in the confidence interval of our bootstrapping procedures. Our model two here is the zero mediation model that we initially wanted to test. So meaningfulness, influencing engagement, influencing commitment, influencing performance. And we do find that zero is in the confidence interval of our bootstrapping procedures, so that is not a significant serial mediation model. And then we have this indirect three, which is the third possible indirect effect that we had in the model, which is the effects of job meaningfulness on performance via commitment. And that is not significant as well because we do have, oh, actually that is significant because zero is not in the confidence interval. Almost got you. OK, so number three, indirect three, the indirect effect is significant because zero is not in the confidence interval. We have our bootstrapping procedures with 1,000 of repetitions and our confidence interval is at 95 percent. So in this session we talked about serial mediation models. We made a distinction between serial mediation and parallel mediation models. We emphasized that we needed to adopt a model number six on the process macro developed by Hayes, and then we walked through a test of a serial mediation model. We found that, that particular model; job meaningfulness influencing job performance via engagement first and commitment second was not significant because zero was in the confidence interval of this serial mediation model.