Moderating effect - original or standardized indicators?
Posted: Sun Jun 14, 2020 7:09 am
Hi to everyone,
I am using SmartPLS 2 and I want to performe moderating effect.
When I create moderating effect there are 3 option for interaction effect term generation:
- use orginal indicator values for moltiplication
- standadize indicator values before moltiplication
- mean-center indicator values before moltiplication
What is the real difference between them?
In my model my moderator variable and my predictor variable are on different scale, and when I run pls algorithm I use data metric with mean 0 and variance 1.
So, which of the 3 options for interaction should I choose?
I am using SmartPLS 2 and I want to performe moderating effect.
When I create moderating effect there are 3 option for interaction effect term generation:
- use orginal indicator values for moltiplication
- standadize indicator values before moltiplication
- mean-center indicator values before moltiplication
What is the real difference between them?
In my model my moderator variable and my predictor variable are on different scale, and when I run pls algorithm I use data metric with mean 0 and variance 1.
So, which of the 3 options for interaction should I choose?