1.2 Install and run medmod

In jamovi, you can perform mediation /moderation analysis (will be introduced in the next section) with the module medmod.  The medmod is a module to do simple mediation and moderation analysis. To use medmod, you will have to install medmod in jamovi. 

To install medmod, please follow the steps: 

• Install the latest version of jamovi and run it. 

• Go to Modules menu on the upper right corner, and go to the jamovi library. 

• Go to the Available tab, look for medmod.

• Install medmod. 

You can now click the medmod icon and select an analysis to your mediation / moderation analysis.

12. Simple Mediation

2.1 Basic idea

• Again, mediation is about how an effect occurs. A basic form of mediation form is the simple mediation model- there is only one mediator to help explain the effect. 

• This model is firstly proposed by Baron & Kenny (1986) and modified by Preacher & Hayes (2008). They specify a conceptual model with several directional pathways- 

  • Path c (Total Effect): when a third variable is absent, the effect of X on Y

  • Path a:  directional effect from the independent variable (X) to the mediator (M)

  • Path b: directional effect  from the mediator (M) to the dependent variable (Y)

  • Indirect pathway of ab (Indirect Effect): the production of a and b, indicating the effect of X on Y through the effect of M

  • Path c' (Direct Effect): the effect of X on Y after considering the effect of M

• After introducing M, we can rewrite the effect of X on Y as:

Total Effect (c) = Direct Effect (c') + Indirect Effect (ab)

2.3 Example 1 - Mediation

A recent article suggested that a possible association between self-control and the duration of sleep. The less capable a person to regulate and inhibit his / her behaviours or bad habits, the more likely he / she will sacrifice sleep for doing them, for example, addiction to social media.

We now want to know how Basic Self-Control, Sleep and Social Media are related to each other, and whether their relationships depend on one another. Therefore, we need to test 1) the direct relationship between Basic Self-Control and Sleep; and 2) the indirect relationship of Social Media between Basic Self-Control and Sleep.

• Q: How do we test the direct and indirect relationships between BSC and Sleep, through SNS?

• A: We use “Mediation” under “medmod” to perform our mediation analysis.

Results Interpretation 

• The Sobel z-test showed that the indirect effect of SNS in the link between Sleep and BSC was significant, indirect effect = 0.215, SE = 0.041, z = 5.203, p < .001, indicating that there was a significant mediation effect between the hours of sleep and the basic self-control via the hours of spending on social network sites. 

If the standard error is bootstrapped, 

• A mediation model with (number of bootstrapping) bootstraps indicated that the indirect path was significant, estimate = (the effect), SE = (the bootstrapped SE), 95% CI [(lower and the upper limit of the estimate)].

2.4 Example 2 - Mediation 

Many people are wondering what are the qualities that make a person more experienced in romantic relationships. Some insist on the physical qualities, that is the attractiveness of a person, while others are convinced by personality, for example, extraversion. In fact, we know that both physical qualities and personality traits play a role in romantic relationships, and sometimes influencing each other.

Given that how a person perceived himself / herself as attractive, how extroverted and outgoing he / she is, and the number of romantic relationships is associated, we now want to know how are they actually related to each other correlational and causally. Therefore, we need to test 1) the direct relationship between one’s Perceived Attractiveness and Number of Relationships; and 2) the indirect relationship of Extraversion between Perceived Attractiveness and Number of Relationships.

• Q: How do we test the direct and indirect relationships between PerAttract and NumRel, through BFM_E?

• A: We use “Mediation” under “medmod” to perform our mediation analysis.

Results Interpretation 

• The Sobel z-test showed that the indirect effect of extraversion in the link between perceived attractiveness and numbers of relationships was significant, indirect effect =  0.0008, SE = 0.0004, z = 2.547, p = .011, indicating that there was a significant mediation effect between perceived attractiveness and numbers of relationships via extraversion. 

3. Moderation

3.1 Basic idea

• As described in the introduction part, moderation is about when an effect occurs, like under what circumstances the effect between two variables, X and Y, will be stronger or weaker or even disappear.

• In statistical terms,  we are testing the interaction effect by calculating an interaction term, X*M (the product between the independent variable and the moderator)

• When the interaction terms significantly predict Y (dependent variable), it means that there's an interaction (moderation) effect

• Once we obtain a significant interaction effect, we can use simple slope analysis to make the interaction effect readily to be understood

3.2 Example 1 - Moderation : BFM_E * LTS_AC -> PerAttract

The way we perceived attractiveness depends on many factors, personality in particular. Among all personality traits, extraversion seems to be one of the very popular measurements for attractiveness. Extroverts tend to be more likable and popular in a group of people, hence a better perception of attractiveness. Yet, extraversion cannot tell or even predict one’s lying (or cheating) tendency, especially in a relationship. How likely a person tell lies can indeed affect one’s perception of attractiveness.

Given that Perceived Attractiveness is positively correlated with extraversion, we now want to know whether this perception of attractiveness could be indirectly affected by one’s lying tendency. We will need to test the interacting effect of Extraversion and Lying Tendency on Perceived Attractiveness.

• Q: How do we test the moderated effect of LTS_AC to BFM_E on PerAttract?

• A: We use “Moderation” under “medmod” to perform our moderation analysis.

Results Interpretation 

• The interaction effect of extraversion and lying tendency on perceived attractiveness is significant, Estimate = -2.098, SE = 0.831, p = .012, indicating that the higher lying tendency, the weaker the positive relationship between extraversion and perceived attractiveness. 

3.3 Example 2 - Moderation: SNS* IPS -> GPA

One of the biggest challenges most people are facing nowadays is to not check on social media frequently, so do many college students. There have been some debates on whether the use of social media affects students’ academic performance. Some people argue that the more frequent use of social media does bring a negative impact on academic performance, whereas some people raise the concerns of the procrastination tendency on the internet, which is manifested into the behaviour of frequent use of social media.

We now want to further study the relationship between Use of Social Media, Procrastination, and Academic Performance, whether Procrastination on top of the Use of Social Media will affect students’ Academic Performance. Therefore, we need to test the interacting effect of Use of Social Media and Procrastination on GPA.

• Q: How do we test the moderating effect of IPS to SNS on GPA?  

• A: We use “Moderation” under “medmod” to perform our moderation analysis.

Results Interpretation 

• The interaction effect of procrastination and the use of social media was not significant, Estimate = -0.012, SE = 0.011, p = .263, indicating that the level of procrastination did not influence the negative relationship between the use of social media and students' academic performance. 

Module Exercise (4% of total course assessment)

Complete the exercise!

• • Now, if you think you're ready for the exercise, you can check your email for the link.

  • Remember to submit your answers before the deadline in order to earn the credits!