Predictor for First Order Variable

[miked]

Hi,

Thanks in advance for any help. There seems to be alot of active and wise contributors to this forum, and I hope you can help me.

I have a Hierarchical Component model that has the following features:
- One Second Order Construct (Formative, in addition to two reflective items) (EDIT: the HOC)
- Four First Order Constructs (All Reflective, with 4-5 items each) (EDIT: the LOCs)
- Several predictor variables (either factor scores or reflective latent variables with 4-5 items) that predict some of the first-order variables. (NOT third order constructs as they don't try to capture the complete domain, just predictors.

My questions:

1 - I can't find anywhere else that has predictor variables that try to predict each of the first order variables individually. Is this possible, or did I model something that can't be tested?
2 - If this is possible, what would be the recommended way to process this? Two-stage or hierarchical components?
3 - Can you think of any samples that come close to this?

Thanks again in advance.

M

EDIT:

Some clarifications: This should be considered a "reflective-formative" model. The predictors I refer to are attempting to predict the LOC variables.

[Hengkov]

I think you need to describe relationships between variables, both 1st order and 2nd order more clearly.
From the above explanation, I catch that 2nd variable order is not influenced by other variables, only two-dimensional shape of formative. If so, then your model can not be tested. Tell me if this was wrong.

[miked]

Hi Hengkov,

Thank you for your response!

I have latent variables that are antecedent to the Lower Order Constructs, as opposed to the Higher Order Construct, in the same manner as described in this paper (The Higher Order Construct is a cause to other factors):

Cadogan, J. W., & Lee, N. (2013). Improper use of formative endogenous variables. Journal of Business Research, 66(2), 233–241. Retrieved from https://research.aston.ac.uk/portal/en/ ... f77d).html

I have found some researchers doing this in a few articles:

Pavlou, P. A., & Fygenson, M. (2006). Understanding and Predicting Electronic Commerce Adoption: An Extension of the Theory of Planned Behavior. Management Information Systems Quarterly, 30(1), 115–143.

Grgecic, D., Holten, R., & Rosenkranz, C. (2015). The Impact of Functional Affordances and Symbolic Expressions on the Formation of Beliefs. Journal of the Association for Information Systems, 16(7).

I do realize that there are some arguments against this (I'm in the process of going through all of this) but is this not possible at all?

Thank you!

M