Cat Variables Dummy-Coding in SPSS for onward use in SmartPLS

[dcwilliams]

Hi forum members. Please I need your kind advice with respect the above topic. My model includes mediating and moderating relationships. The problem is that three of my independent variables are non-ordered categories, specifically each is nominal with 5 indicators. Being guided by the work of Falk and Miller (1992), I came to a point where I needed to dummy-code the categorical variables, but observed that there are two sub-menus in SPSS under "Transform" menu, namely, "Recode into different variables" and "Create dummy variables".

So (1) which of these two sub-menus do I use for dummy-coding? I observed alternative use of these two by some authors, but not sure if either of them would suffice.

(2) In transforming each of these manifest variables to binary, via dummy coding, how do I treat the referent group? Do I exclude (drop) it from entering SmartPLS model? Falk and Miller (1992) talk about dropping it, but they also included it in their final model, though with zero value. Precisely, what's the literal meaning of "drop" here?

(3) I am also considering weighting the manifest variables, and redirecting the arrows of each of the blocks inward. Pls am I on the right track or drifting off-course? Are these current practices, and accommodated in SmartPLS . Thanks for your response.