Moderator Effect - Standardize, mean-center, or original?

[fairviewpark]

Hi PLS experts,

I have a PLS problem. I searched the forum and did not find a specific answer. Please help here.

I have a moderator which is not included in the "base" model. It moderates the paths from the 2 predictors to the same 1 LV. The indicator items in both the predictors and LV are measured in 7-point Likert scales whereas the indicator items in the moderator are measured in ordinal scales with various points. Scales are 5, 8, 10, or 12-point. All constructs are reflective.

When I build the product indicators for measuring interaction effect, should I standardize, mean-center or use original values for the product indicators?

Thanks,
Fairview