This session will be about parallel mediation, and this is the agenda for today. We'll give you an overview of parallel mediation. What is parallel mediation? And then we'll show you a way to test for parallel mediation using SPSS, more specifically, the process macro developed by Hayes. So what is parallel mediation? So parallel mediation models, they do have multiple parallel mediators. You could have three, four, five, even more mediators than just two. In this example here, we are adopting only two mediators in order to keep things simple. And when you develop your theory, let's keep it parsimonious, okay? We need it to keep it simple. If we start adding multiple mediators your model will become too complex and too confusing to explain. So I strongly recommend you to keep it simple, two perhaps three mediators that affect the relationship between your independent variable and your dependent variable. Again, when you are running parallel mediation models or mediation models you needed to adopt model number four on SPSS, on process macro. So model number four is extremely important for you. Just to compare and contrast, there is also serial mediation models. In serial mediation models, these type of models we have one particular section on serial mediation model, so I will not spend a lot of time on serial mediation models now. But, you have a sequence of mediators influencing the relationship between your independent variable and your dependent variable. You have a distal relationship. For these models you needed to adopt model number six. Although its mediation, we needed to tell SPSS that this is a serial mediation model and not a parallel mediation model. So instead of using Model 4, you use Model 6 when conducting your analysis of serial mediation models. We'll talk more about this serial mediation model a little bit in more depth in our next session, in our session specifically about serial mediation models. Yes, we are conducting an analysis and this is a hands-on session as well. The whole workshop is a hands-on workshop in which you will have the opportunity to practice, to actually conduct parallel mediation models, serial mediation models, mediation moderation, all types of analysis. So, the model that we are testing today is this. Job meaningfulness influencing job performance via job engagement and also commitment. Your theory should inform which mediators you'll have in the model, okay? It's not random picking and choose set of variables. Your theoretical development should inform the variables that you are having in the model as mediators basically. Let's conduct this analysis. So first, I'm assuming that you're a little bit familiar with SPSS, so now we go straight to analyze regression and go back to the process macro developed by Hayes. Let's choose the variables. If you watched our session on mediation, you'll notice that the only thing different here is the addition of commitment in our invariable boxes. So now we have two mediators. Our dependent variable is performance, our independent variable is job meaningfulness, and our mediators, we have job engagement and commitment. Again, number four, model number four is extremely important to keep in mind that, and we don't have co-variants or controls. We are not looking for moderators at this point. Click on OK, and then this is what you should see. Model number four, we have M1 which is our mediator one, and M2, which is our mediator two. So when we are testing for parallel mediation models, you need to go through the steps of mediation models developed by Hayes, okay? So our step one, we look at the relationship between our independent variable and our mediator one in this case. Job meaningfulness has a positive and significant relationship with job engagement. P is less than .05. And then we look at the relationship between our independent variable and our second mediator, commitment. So here we have our job meaningfulness having a significant and positive relationship with commitment again. And finally, we look at the relationship between our two mediators now, and our dependent variable. Here we have job engagement and commitment having relationships with performance. Job engagement has a positive and significant relationship with performance. But commitment, well P is not less than .05. So, we do not have indications or evidence for the relationship between commitment and job performance. We do not have a parallel mediation model in this example. We do have one path through engagement that's significant, or at least there is preliminary evidence for the significant indirect effect. And then we look at the bootstrapping, or the composition of the effects, okay? The total effect is significant, P is less than .05. The direct effect is not significant, P is not less than .05. So when we have these two mediators in the model, the effects of job meaningfulness on job performance is not significant. And then we look at the bootstrapping procedures or bootstrapping analysis. We do find that the indirect effect via engagement is significant, because there is no zero in the confidence interval of the bootstraps. But what we find about commitment is that there is zero in this confidence interval. So the indirect effect is not significant, and that this is just reinforcing what we found in the prior slide, okay? So the relationship between commitment and job performance was not significant. So you're not supposed to find a significant relationship here. We then look at the normal theory tests and what we find here is that well, for engagement, the Sobel test is significant. But for commitment the Sobel test is not significant. This is again just for providing you more information. You don't need to report Sobal tests in your paper anymore because we know that Sobal tests, they violate this assumption that the direct effect distribution should be normal. We know that it is not. In this session, we talked about parallel mediation, and then we explained what parallel mediation models are. And also contrast and compare that with serial mediation models. We also showed you how to run a parallel mediation model on SPSS. On the macro called Process developed by Hayes.