2nd order subconstructs for further analysis?

[wgaenswein]

Dear all,

I am a PLS-beginner and currently setting up my first SEM. Apart from some calculation issues to which I've found very helpful advise in this forum, I have a so far unresolved problem concerning the set-up of my model:

I use 4 similar reflective variables which measure "scope of information" from four different kinds of information sources (e.g. from management information system, personal sources etc.).

In a first step, I like to combine these four variables to get a measure for the "total information scope" and then integrate it into an SEM as exogenous variable.

In a second step, I like to integrate each of this four variables (i.e. sources) as exogenous variables to find out which sources have which effect on the endogenous variable.

Now I wonder about the following questions for which I hope to get some adive:
* Can I use 2nd order modelling for aggregating the four variables?
* I thought of just summing the variables up - which on the other hand is a special kind of 2nd order model?
* If I combine them in a 2nd order model, could I still use the individual 1st order variables for the path analysis in the second step?
* Can anyone name any references in the areas of strategic management, management accounting or organizational behavior?

I am looking forward to receiving any thoughts on that!

Many thanks,
Wolfgang