What is the difference between reflective and formative constructs in PLS SEM? and how these constructs can make difference in analysis part?
In PLS, which is a method of composites, the distinction between formative and reflective is the distinction between regression weights, which take account of collinearity among predictors in the outer weights, and correlation weights, which ignore collinearity.
Article Rethinking Partial Least Squares Path Modeling: In Praise of...
The type of relationship that, you theoretically assume, exists between a construct and its indicators decides whether the construct should modelled as a reflective or a formative one. In reflective cases, the construct causes its indicators whereas in formative cases, the construct is caused by its indicators. It follows from this that, indicators of a reflective construct should be highly correlated while indicators of a formative construct do not have to be correlated. If you model an inherently formative construct as reflective, this will lead to a psychometrically weak measurement model. Further, if you leave out even one indicator from a formative construct, this may lead to biased estimates in the structural part of your model.
I recommend
Article Estimation issues with PLS and CBSEM: Where the bias lies!
In PLS, which is a method of composites, the distinction between formative and reflective is the distinction between regression weights, which take account of collinearity among predictors in the outer weights, and correlation weights, which ignore collinearity.
Article Rethinking Partial Least Squares Path Modeling: In Praise of...
For clear understanding -
Formative- To find the individual weight of every variables and item.
Reflective- Is the weight and correlation among variables and items.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Structural Equation Modeling