Dear all,
the topic of validity assessment for formative constructs is discussed quite controversial in literature.
Some authors say, that items showing insignificant regressors are not allowed for elimination as each item is believed to represent a facet adding to the whole of the formative construct (e.g., Huber, F., et al. (2007): "Kausalmodellierung mit Partial Least Square", Wiesbaden: Gabler, p:97).
In contrast, other authors quite clearly propose to eliminate insignificant items, as "they cannot represent valid indicators of the construct" (Diamantopoulos, A., et al. (2008): "Advancing formative measurement models", in: Journal of Business Research, forthcoming: p.13, see also Bollen, K. A. 1989): "Structural equations with latent variables", New York: Wiley)
Can anybody provide a valid explanation what suggestion to follow?
Thank you very much in advance, looking forward to your suggestions.
Florian
Eliminantion of insignificant items for formative constructs
Eliminantion of insignificant items for formative constructs
Hallo, I already have one pending registration with my McKinsey-email-address. I just wanted to hereby indicate that I am actually a PhD at Hamburg University of Technology - however still use my McK account for reasons of convenience.