Session 5 | Moderation and Mediation

Question 1

What is the difference between Moderation and Mediation?

Answer 1

Moderation implies that there is a third factor that strengthens the relationship between two variables.

Mediation implies that there is an underlying mechanism we are attempting to assess. E.g., hours studied do not improve test performance individually, but it is the fact that it impacts test anxiety that improves performance.

Question 2

How is the Multiple Regression structured in the case of Moderation?

Answer 2

Y= I + aX + bM + cXM
which means
X has a fixed effect on Y
M has a fixed effect on Y
The interaction between XM has an effect on Y

Question 3

In the context of moderation, which factors should I look into to verify if it is truly taking place?

Answer 3

p-value is very small = statistically significant
R^2 is HIGHER = there is a better fit
F-statistic is HIGHER = there is a better fit
ivmod is NOT too small

Question 4

How do I know if the moderator has a positive or negative effect on the dependent variable based on it relationship with the independent variable?

Answer 4

The relationship moderator-independent variable has a positive effect is the ivmod is positive and significant eg. 0.25 and negative if ivmod is negative eg. -0.34

Question 5

How do I mean center manually in R?

Answer 5

I create a new column where I calculate the difference between the current variable A and the general mean of all variables within the column.

Eg. TrainingHours_mc<-TrainingHours - mean(TrainingHours)

Question 6

How do I manually on paper calculate the slope of a moderation? What is the formula?

Answer 6

coff. of independent variable + coff of interation*SD moderador

Question 7

How do I calculate the Moderator SD+1 in R? What is the formula?

Answer 7

Mean Centered Moderator - sd Moderator (or opposite sign if I want the SD-1)

eg. df$mod_high<-df$mod_mc-sd_mod

Question 8

How do I run a slope regression analysis in conceptual terms of a Moderator?

Answer 8

sum up: Independent variable + moderator + I(Independent variable * moderator high)

You have to do the same for every high or low moderator

Question 9

Which conditions must hold for a mediation to take place?

Answer 9

1) The Independent Variable must affect the mediator
2) independent variable must affect the dependent variable
3) The Mediator must affect the dependent variable

Question 10

In Mediation, what is the difference between c and c’?

Answer 10

c = effect with the presence of the Mediator
c’= effect from independent variable to dependent variable

Question 11

Which one has to be bigger in mediation, c or c’, for us to indicate that a mediation exists?

Answer 11

c (effect with mediator) has to be bigger than c’ (effect without mediator).

Question 12

When does Perfect Mediation take place?

Answer 12

Perfect Mediation takes place when we include M and because of this the effect that seemed to originate from the independent variable drops. This means that when we have a perfect mediation the independent variable loses statistical significance.

Perfect Mediation = M explains fully the relationship between independent and dependent variables.

Question 13

What does Mediation analysis look at?

Answer 13

Mediation analysis verifies if an effect produced by the Independent Variable is carried by the Mediator.