Multicategorical factorial design: Dummy or Helmert coding?

Fer_M

New Member
#1
Hi, I have a question about moderated mediation with multicategorical variables, I would like to ask for your advice.
I am not sure whether I should use Dummy coding or Helmert coding.
I haven't find papers with moderated mediation and Helmert coding yet.
My study:
X= 3 levels
W=4 levels
M1,M2, M3, continuous variables
Y= continuous variable
My hypotheses propose that one of the levels of X, level 1, will have the highest impact on Y though the mediators, followed by the next level, which is followed by the next level. The three groups are statistically different in their impact on Y.
Regarding moderator W, I also propose that one of the levels of W, level 1, will have the highest impact on Y. I also propose that the control level of W will have the lowest impact on Y. And I don't have predictions about the two other levels.
The interaction M*W hypothesis proposes that the condition that combines level 1 of both factors will have the highest impact on Y in comparison to all the rest of the conditions.
Would you use Helmert coding for this
?Thanks in advance