Dear smartPLS community,
I have the most basic question that I cannot figure out from reading work on the estimation of PLS-SEM models.
My concerns is whether “shared variance” of constructs (IVs) are controlled for when for example estimating a simple additive model with direct effects of 4 IVs on a DV? Or do you need to control for their potential common variance (as OLS does) by linking them with paths?
Differently put, are the estimations of the 4 IVs on the DV a series of bivariate correlations or is it “multivariate” regression? Is it the direct and unique effect (controlled for the other 3 IVs) that is estimated?
BR
Tobias