In research, we most often test direct effects. That is, a direct effect is a relationship between a predictor and an outcome, such as job satisfaction predicting job performance. Sometimes we want to test, however, whether a third variable explains the relationship between two other variables, such as whether motivation explains the relationship between job satisfaction and job performance. We often call this a mediating effect or an indirect effect, which can be tested in Jamovi quite easily. This page will teach you how to do just that. If you have any questions after reading, please email me at MHoward@SouthAlabama.edu.
A mediating variable explains the relationship between two other variables. For instance, we may know that job satisfaction relates to job performance, but we may not know the exact reason. We could test to see, though, whether people’s job satisfaction leads to greater motivation, which then leads to greater job performance. If significant, we would then say that motivation is a mediator between job satisfaction and job performance. If this is the case, then we would have an indirect effect of job satisfaction on job performance via motivation. Some other questions you could ask when investigating mediation could be:
- Whether social undermining explains the negative relationship between team member Machiavellian and team performance.
- Whether time studying explains the positive relationship between conscientiousness and grades.
- Whether time partying explains the negative relationship between extraversion and grades.
And here is a visual representation of mediation:
The most basic type of mediation includes a single predictor, a single mediator, and a single outcome. This is what we will use for the current example. If you don’t have a dataset, you can download the example dataset here. In this dataset, we are investigating the direct effect of Job Satisfaction on Job Performance, and we are also testing whether Motivation mediates this relationship. In other words, we are testing whether Motivation explains the relationship between Job Satisfaction and Job Performance.
Also, this file is in .xls format, but Jamovi cannot open this format. To learn how to change this .xls file to a .csv file, which Jamovi can open, please click here. Also, the pictures below are a little small on the page. Click on the link above each picture to view a larger version of the picture in a new window.
Further, this guide uses the ModMed module in Jamovi. If you haven’t installed this module yet, you will need to do so. Please click here to discover how to install modules in Jamovi.
One last note before starting: I am not entirely certain how Jamovi or the associated medmod package calculates estimates of indirect effects. I know that it calculates bootstrapped estimates, but I could not find a direct statement regarding the exact analysis. If you know this answer, please let me know! Otherwise, I would be hesitant about using this approach for research purposes until I could get a clear answer to this question.
The dataset should look like this:
We first want to click on the medmod button, as seen below:
We then want to click on Mediation.
You should see a window like the one below.
First, we want to designate our outcome variable, which is Job Performance in this example. So, you want to click on the job performance variable, and then click on the right-facing arrow next to the Dependent Variable box.
Next, we want to designate our predictor variable, which is Job Satisfaction in this example. You want to click on the Job Satisfaction variable, and then click on the right-facing arrow next to the Predictor box.
Lastly, we want to designate our mediating variable, which is Motivation in this example. Click on the Motivation variable, and then click on the right-facing arrow next to the Mediator box.
You should get output that looks like the numbers below. If so, great! If not, take a look back and see where your procedure differed from mine.
From this window, the first line represents the indirect effect, which tells us whether our mediating variable was indeed a statistically significant mediator. In this example, we see that the p-value associated with the indirect effect was .58. Because this is greater than .05, it is not statistically significant, and we would determine that motivation was not a significant mediator of the effect between job satisfaction and job performance.
You can see the results for the direct effect on the second line, and you can see the results for the total effect on the third line. The direct effect is the relation of our predictor on our outcome when controlling for our mediator. In this example, the p-value was .94, indicating that job satisfaction did not have a significant influence on job performance when accounting for motivation. Alternatively, the total effect is the indirect effect and the direct effect put together. Again, in this example, we can see that the total effect’s p-value is .87, indicating that the overall relation of job satisfaction on job performance was not statistically significant.
That is all for Mediation with Regression in Excel via Sobel Testing! If you have any questions, be sure to contact me at MHoward@SouthAlabama.edu.