Moderator variable

[plshelpme]

Hi everybody,

I'm testing the UTAUT2 model in a smartwatch context and I have a question about testing age as a moderator. I essentially want to understand what the correct process is for doing so.

I add the moderator and test its effect on each relationship individually. Is it correct to test it as a continuous moderator first, and then split the data into groups if I find a significant result? (spoiler... I have no significant results!)

Also, should I be mean centring my data first?

Would somebody be able to point me towards some literature that guides you through the process of testing a moderator in Smartpls 3? I've read the basic literature (Hair et al 2012), but I can't figure it out.

Thank you!
Sam

[jmbecker]

I add the moderator and test its effect on each relationship individually. Is it correct to test it as a continuous moderator first, and then split the data into groups if I find a significant result? (spoiler... I have no significant results!)

Yes, add it as a continous moderator, but don't split it into groups, even if you have signficant results.

Also, should I be mean centring my data first?

No. You don't need to mean center, because data is standardized in PLS, which includes mean centering. However, you need to be aware that the interpretation of coefficients is slighly different for standardized coefficients.

The book "A Primer on PLS..." by Hair et al. 2017 has a very good chapter on moderation analysis. https://www.smartpls.com/documentation/ ... s-sem-book