Control variables in interaction models // Purification?

[stephan.kramer]

Dear all,

I have two important questions I am struggling with: I have a basic interaction setting with two reflective IV's (A, B), an interaction term (AxB) and a DV (C). In addition, I have a control variable (D) that is hypothesized to have an effect on A,B and C.

Now my big question is whether I have to control for the effect of D as well in the interaction term AxB??? Does this make sense at all? I realized that it is impossible in SmartPLS to link any other variables to the interaction term which already might indicate that this should be avoided right?

Another question I have is whether continous control varibales have to satisfy the quality criteria for the measurement model as well, and thus item purification is required for control variables.

I would really appreciate any answers or opinions concerning this topic.

Best regards

Stephan