I am analysing my Data using PLS, and I have constructed my model as show in the image below. This is somehow a complex model and I really need your help and advice in this matter.
The IS-impact model (formative using reflective indicators (37 items) and formative dimensions (4 dimensions)), the IS-Support (formative using reflective indicators (12 items) and formative dimensions (4 dimensions)), and Satisfaction Reflective (4 items).
I understand that reflective construct validity differ from formative construct validity, where
Reflective constructs require (Jorg Henseler et al 2009):
• Composite reliability
• Indicator reliability
• AVE
• Fornell-Larcker criterion
• Cross Loading.
Formative constructs require (Jorg Henseler et al 2009):
• Nomological validity
• External validity
• Significance of weights
• Multicollinearity (VIF)
And for the structural Model (Jorg Henseler et al 2009):
• R2 of endogenous latent variables
• Estimates for path coefficients
• Effect size f2
• Prediction relevance (Q2 and q2)
As my design consist of a mix of formative and reflective constructs and the multi-dimensional constructs, IS-Impact and IS-support, are formative using reflective indicators and formative dimensions, I am lost !!!
I would much appreciate if you can help me or point me to some literature that address this issue.
Much appreciated
Regards
Rabaa’i