forum.smartpls.com

Hello Community,

does SmartPLS allows conclusions about the individual proportion of an exogenous latent variable on the R2 of the endogenous latent variable by removing the variable of the equation model and tracking the difference in R2? Would this approach be methodologically correct?

Thanks a lot!!!

We provide the f² value which is very similar to your approach: It takes the individual contribution R² included - R² excluded and then divides this number by 1- R² included. Thus, it is the individual R² contribution relative to the unexplained variance.

You may not want to calculate such a number yourself, because in PLS you are changing the whole model context when deleting a construct. Thus you may get strange results if you really delete the construct and not calculate the difference based on a fixed model.