If I want to use one moderating variable, do I need to keep the variable I am using as a moderator in the model as an independent variable as well (direct effect), or just the interaction (variable*modvariable).
As far as I know you need to keep both of the in the model.
I recommend you to read the Introduction to Mediation, Moderation, and Conditional Process Analysis by Andrew F. Hayes.
He provides an easy and comprehensive explanation on how to conduct moderation tests.
Hi Juliano.
In order to avoid collinearity issues related to the moderating variable you will need to create the moderating variable using the "mean center" values of the first order variables (i.e. your independent variables).
When including these variables in your regression, you need to include the original IVs and the moderation variable created with the mean centered values.
You need to verify if the explanatory capacity (R-square) increases when the interaction term (moderation variable) is included and verify what happens to the main relationship you are investigating. Does the main effect increase when you increase your moderating variable or the opposite happens? This analysis is performed usually by adding and subtracting a specific value from the moderation variable before creating the interaction term, for instance 1 std dev up and 1 std dev down or any other value that is relevant to your analysis and data. This allow to discuss a positive or negative moderation effect in your model.
If you have any questions just let me know.
- Badges
- Science topic
- Similar topics
- Psychology