Mode A for formative indicators

[janschreier]

In Rigdon, E. E. (2012): „Rethinking Partial Least Squares Path Modeling: In Praise of Simple Methods“. In: Long Range Planning. 45 (5-6), S. 341–358.
Rigdon's proposal is to calculate formative constructs with Mode A (correlations instead of regressions).

My questions are now:
1) what do other experts think on this approach?
2) If I use it, do I have to run my checks for this construct as if these constructs were formative (e. g. check cronbach's alpha, composite reliability etc.) or should I check it like I check formative constructs (i. e. convergent validity, outer VIF, etc).

For 2) my view would be that the checks of constructs only depend on the constructs definition, thus if I define a construct formative I should check it formative, regardless if I use Mode A or Mode B.

Thanks for your help!
Jan

[jmbecker]

I agree with Ed Rigdon that considering Mode A for formative constructs is worthwhile. In Becker, Rai & Rigdon (2013) we also show why and when it might be good choice to use Mode A for formative constructs. It is not always the better choice, but in quite many situations.

As for 2), you are correct: You should still treat your constructs as formative and assess construct quality based on formative assessment methods.


Becker, Jan-Michael, Arun Rai and Edward E. Rigdon (2013): Predictive Validity and Formative Measurement in Structural Equation Modeling: Embracing Practical Relevance, in: Proceedings of the International Conference on Information Systems (Milan, Italy).

[janschreier]

Thank you, Jan-Michael!