Hi,
I'm interested in developing a marketing mix model, in which I have exogenous variables (marketing inputs) that act directly on a dependent variable (say sales), endogenous variables (or intermediate outcomes) that act directly on sales (customer-initiated actions such as social media posts) and exogenous variables that act on intermediate outcomes. All variables are observed - there are no latents.
Currently, I have GLS models at three levels (Sales~Exogenous), (Sales~Intermediate outcomes) and (Intermediate outcomes~Exogenous variables).
Can I undertake a path analysis with SmartPLS for this model? As far as I can see, an exogenous variable must be linked to a latent variable. I see from the forum discussions that it is possible to model an observed endogenous variable as a latent with one formative indicator. However, I would also like to model an intercept (often interpreted as a base effect in a marketing mix model). One answer to a similar query noted that SmartPLS uses standardised data and so no constant term is computed. Another answer noted that selecting "Original" in Calculate -> PLS Algorithm produces coefficients based on raw data, but I have been unable to figure out how to obtain intercepts.
I was able to obtain unstandardised coefficients, and a constant for each endogenous variable, using AMOS, and was wondering if thereis a way to do it with SmartPLS as well.
Thanks for any help you can provide!