If we are aiming to study the impact of an independent variable X on Y and on Z.
At the same time, we want to study the impact of Y on Z.
So the relationships to be studied are :
Effect of X on Y
Effect of X on Z
Effect of Y on Z
Each is in a single research question.
The data of the 3 variables is ordinal (non-parametric), so ordinal regression is the selected test for the 3 research questions according to Cohen's book.
My question:
If I think that Y might be the mediating variable, should I put this as a research question? Like: Does Y mediate the relationship between X and Z?
If so, what is the correct statistical test for this question?
I thought of multiple regression, but here X and Y might be highly correlated (according to literature), and I read that multiple regression is not suitable in this case.