Is it possible to model only the measurement of a formative construct?
Posted: Tue Mar 10, 2020 8:06 pm
I would like to model Knowledge as a formative composite measure using PLS. I understand that formative measures alone are unidentifiable and unfortunately, I don't have any outcome variables to include in the model. I'm wondering if I could use the repeated indicator approach to model only the measurement of a construct by turning indicators into single-item first-order constructs and the composite a second order construct?
Also, I'm wondering if anyone is familiar with Adanco? Adanco is able to estimate item loadings of a composite construct, although the weights are all equal when using Mode B. Would modeling the variable this way be any more useful than simply calculating item-total correlations?
Also, I'm wondering if anyone is familiar with Adanco? Adanco is able to estimate item loadings of a composite construct, although the weights are all equal when using Mode B. Would modeling the variable this way be any more useful than simply calculating item-total correlations?