I have only seen it when the moderator variable has only 2 levels/groups/categories (dichotomous). What if my moderator variable has 3 levels? Should I just use a two-way ANOVA then since it also tests interaction effects? Or can I do moderated multiple regression (multiple regression with an interaction effect) using Process by Andy F. Hayes, even if my moderator variable has 3 categories?
(Please no complicated mathematical explanations, I'm only an undegrad)