How to use higher-order scale?

[szidel]

Hello,

I am currently working on the development and validation of a new scale. The process went well so far, and I ended up with a higher order solution (4 subscales/factors on the first level, of which 2 load on a common second-order factor). Since the two first-order factors, which share a common second-order factor, are conceptually quite similar, this made a lot of sense to me. But seeing this result, I don't really know how it may be used? Without the second order factor (and without freeing the covariation between some items of the factors) the model fit is quite bad, so I don't think one could use them just as two separate scale. But one factor, that includes all these items does not fit well, either.

This question can be generalized: How to use scales in research, that have a structure of a higher order? I have absolutely no idea and my Internet research was quite unsuccessful, too...

Thank you!

 

[nbjo]

You use a higher order factor just as any scale

For example
Core self-evaluation is a combination of the scales, Locus of control, Neuroticism, Generalized self-efficacy, and Self-esteem.

Just don't mix them together

 

[szidel]

Hello!

(sorry that I respond so late, it seems my alert is off).

I still don't understand, in your example I see two ways of using it:
A: I use the core self-evaluation scale, i.e. make the mean (or whatever) of all items /subscales. But this would then equal a solution with 1 factor.
B: I use all the subscales on their own, i.e one scale (mean) for locus of control, neuroticism, and so on. But this would equal a solution with 4 first-order factors.

Where do I "see" the second order factor?

Thank you!