Hi,
I used PLS in my working paper and a reviewer raised an issue within it.
"it becomes clear that the measures used contain measurement error, which by definition is not useful variance to be explained"
My model is complex, and I have less than 150 sample size, which is ok when I do the post-hoc power analysis (above .80). However, I recently realized that smatpls has provided a new method (HTMT ) to check out the discriminate validity. with this method I observe discriminate validity problem between some constructs. I am going to make the model simpler (removing some constructs/indicators), do you think that this may be the solution for the reviewer’s comment, or I still have "the measures used contain measurement error"?
Thanks