Hi.
I have a reflective-formative higher-order construct that combines seven reflective first-order constructs. I validated this higher-order construct in a study in which all seven subconstructs showed significant positive path coefficients with this higher-order construct.
In my new study, I included this higher-order construct as an independent variable with other endogenous four reflective constructs as dependent variables. When I run the PLS-SEM algorithm with this model, two of the previously positive path coefficients between the seven subconstructs and the higher-order construct become negative.
Could anyone please explain to me what could be the reason? I am using the same data as the previous study, so how can I get consistent results?
Any help would be greatly appreciated.