So now let's run a model. How can we do this? How can we run this model on SPSS? And the model that we'll be testing today is this indirect effect of meaningfulness on job performance via engagement. So the mediator here is job engagement, our independent variable is meaningfulness, and our dependent variable is performance. Again, there is this language that we use, so the mechanism through which meaningfulness influences performance could be engagement. And that's our mediator. How do we run this analysis? The first step, is to click on Analyze, and you'll see a list of potential analysis that you could run. The one that we are interested today for the mediation models is Regression. So, we needed to click on Regression, and then, youll have another tab here with a number of options. You can run linear regressions, you can run binarial logistic models, original models, probit models, but for mediation models, we are adopting the PROCESS macro, developed by Hayes. So click on that particular option. Once you do that, this is a screen that you see. So on your left side, here, you have the list of the variables you find in your data set. If you chose to use our on data set, you will have this list of variables. If you are adopting your own data set, you may have a different set of variables, there. But, again, the magic number is eight, eight characters per variable. If you have more than eight characters in the name of the variable, well, you will get an error message. So our next step then is to select the correct number. With mediation models, we are adopting model number four, and that this is critical. If you don't select the right model number, you will get error messages or output that is not applicable to your model, so you may be finding different results. Or results that actually don't test for your hypothesis. Hayes and his colleagues put together a series of models that are more than 70 models, 7, 0 models. So we also added in the description of this video, a link in which you can get access to all these models. There are more than 70 models. So, yeah if you want to learn more about the other models, just download that file you'll have an idea about what other models you can learn. But for mediation, again, model number four. Let's bring the variables to the right place. So performance is our outcome variable, or Y. Some researchers we will use this terminology, Y for the dependent variable, X for independent variable. In the end, they are the dependent variable or independent variable. And then, we have our meaningfulness as our independent variable and engagement as our mediator. You'll notice that there is place here to add your covariates, or controls. In this analysis, to keep it simple, we do not have controls. But you can, you can have, for example, I don't know, age, or in this case, commitment could be a control. What's important to know about controls is that, you have to have a theoretical reason, a theoretical justification for adding controls to your model. What controls do is, basically, to partial out various related to those particular variables. So it's important to have a theoretical justification to add controls to your model. And then, what we also see, here, is this section in which we have the moderators. So we have moderator W, Z, V and Q. We are not adopting moderators at this point. But you should know that it's extremely important for the process macro when you're conducting moderating equations, or moderating models, to enter the variable in the right moderator. But we'll talk more about that in one of our future sessions. So, if you want to know more about how to use those places here, how to put the variables here, yeah, watch the future sessions in this workshop, and the next workshops. So, the next step is to click on Option, and select total effects. There are many different options here for mediation models. We needed to click on total effect and also the Sobel test. Remember, we should not report Sobel tests when writing about the results, or reporting the results. But for completeness, we will show you the Sobel tests for the mediation models as well. Now, just click on OK, and there you go. We have the output for our analysis. This is the first screen that you should see. And if you scroll down, you'll get to this point. Always double check the model number. Remember, for mediation, is model number four. And then, we check for the variables. We have performance, meaningfulness, and engagement. They are the dependent variable, independent variable, and mediator. Our sample in our data set, we have 1,000 observations, that's how we created the data set, so the sample size is 1,000. Everything looks good now. So let's take a look at the output, the coefficients now. So we have two steps. In the first step, remember, when we are running the current task for mediation, the first thing that we needed to do is to see if the independent variable predicts our mediator. What's the relationship between meaningfulness and engagement? We do find that meaningfulness has a significant relationship with engagement. P, here, is less than 0.05. And then, we move to our second step. And in our second step, we are looking for the relationship between our mediator and our independent variable. Job engagement, and performance. And in this case we have, also, a significant relationship between engagement and performance. P is less than 0.05. And you'll notice, we are controlling for our independent variable, which becomes non significant. So that relationship with performance is not significant at this point. If you keep scrolling down your screen, or this file, you'll see the effects composition. And here, we have that the total effect is significant. So the total effect is the indirect effect plus the direct effect. If we look at the direct effect, the direct effect is not significant. So when we have the mediator in the model, the relationship between our independent variable, meaningfulness, and our dependent variable, performance, is not significant. The next step, is to look at the indirect effect and, here, we look at the bootstrapping confidence interval. We'll see here that 0 is not in the 95% confidence interval, which indicates to us that this indirect effect is significant. So now, we have evidence for the indirect effect. Yes, meaningfulness influences performance via engagement. Just to keep this complete for the sake of completeness, we also conduct the Sobel test. And you'll notice that the terminology that we have in the output file is not the same of the terminology that we click it when we were asking SPSS to run this model. Instead of giving us a Sobel test coefficient, it gives us a normal theory test for the indirect effects, which is the Sobel test. So in this particular case, our P value for the Sobel test is less than 0.05, so it is significant as well. But remember, we don't report Sobel tests when writing about the results or reporting the results anymore. Well, in this session, we talked about SPSS. How to get your data in SPSS, so now you can run mediation models. When we got to that place, that position, we then ran the mediation model.