Dear Vivek Agrawal

In Partial Least Squares Structural Equation Modeling (PLS-SEM) using SmartPLS or similar software, it's typically inadvisable to use the same indicator for two different constructs, particularly if they're meant to be independent. This approach can blur conceptual boundaries, leading to a lack of clarity in construct representation and creating measurement issues, with discriminant validity (DV ) being a major concern. Statistically, it can result in inflated correlations and multicollinearity, quantifiable through a high Variance Inflation Factor (VIF), which distorts coefficient estimates and complicates model interpretation. Although in certain cases, only in formative constructs, this might be considered, but still requires careful theoretical justification. In your case, since x1 is already part of a reflective construct A, it should not be considered for inclusion in another reflective construct B.