When calculating effect size (f square) for a moderating effect, should the R-square-without-the-moderator exclude the direct effect of the moderator variable too, or just the effect of the IV x Moderator?
I can't seem to answer this based on Henseler & Fassott (2010), Section 30.5.2.