The two stage approach (single-item first order constructs)
Posted: Tue Apr 05, 2016 8:46 pm
Dear All,
I have one question related to using the two stage approach to estimate a second order construct. I have a second order exogenous formative construct that have 6 first order constructs. Two first order constructs are formative and one is reflective. The remaining three first order constructs are single item constructs. Again, all of the six first order constructs have a formative relationship with the second order construct.
In the first stage, I need to estimate the scores of the first order constructs. This is done by running a model that includes all first order constructs with each has an arrow pointing to the final endogenous construct. In the second stage, the scores of the first order constructs (obtained in the first stage) are used as indicators for the second order construct.
My question is:
In the first stage, should I estimate the scores of the single item constructs? Or, Would it be enough to include these single item first order constructs in the second stage only?
I am asking this because including them in the first stage will lead me to violate the sample size requirement.
Some might ask why do not I use the repeated indicator approach instead. The problem is that I do not know how to model this second order formative construct using the repeated indicator approach given that one of the first order constructs is reflective and two are formative (note: all first order constructs have a formative relationship with the second order construct).
Any advice is appeciated.
I have one question related to using the two stage approach to estimate a second order construct. I have a second order exogenous formative construct that have 6 first order constructs. Two first order constructs are formative and one is reflective. The remaining three first order constructs are single item constructs. Again, all of the six first order constructs have a formative relationship with the second order construct.
In the first stage, I need to estimate the scores of the first order constructs. This is done by running a model that includes all first order constructs with each has an arrow pointing to the final endogenous construct. In the second stage, the scores of the first order constructs (obtained in the first stage) are used as indicators for the second order construct.
My question is:
In the first stage, should I estimate the scores of the single item constructs? Or, Would it be enough to include these single item first order constructs in the second stage only?
I am asking this because including them in the first stage will lead me to violate the sample size requirement.
Some might ask why do not I use the repeated indicator approach instead. The problem is that I do not know how to model this second order formative construct using the repeated indicator approach given that one of the first order constructs is reflective and two are formative (note: all first order constructs have a formative relationship with the second order construct).
Any advice is appeciated.