Hello,
I am modelling a complex higher order model, with a higher order mediator variable. I built my model in various steps. First analyzing the effects on the mediator, second measuring the direct effects of the independent varibles (excluding the mediator) on the dependent varible. When combining those two models I get a very high effect size for the mediator, which partly represents other influencing factors as well. So far, I understand the results. But when I analyze the mediators dimensions influence on the dependent variable, I get very high values for R2, but very low effect sizes of the different first order constructs. Is this still due to the mediator function, although the model just refers to the first oder dimensions of the mediator variable and the independent variable?
I would be very glad if anybody perhaps has had the same problem or knows some interpretation of those results.
Thanx a lot,
Verena