Formative vs reflective

[MaKauTZ]

Hi,
I'm confused.
Let's say I have 4 highly correlating likert scale questions for the topic "good communication" (communication is considered a success factor).

Are my indicators now formative or reflective?
Ive read literature that says that highly correlated items mean that I should model it with reflective indicators, whereas it also says that success factors usually consist of formative indicators..

My questions are like:
-communication is important
-communication is not important
-communication is a big success factor
-communication benefits the xyz process

Any hints?
Thank you

[Diogenes]

Hi,

The example that you gave is reflective:
- High correlations.
- Items 1 to 3 are the same (unusuful redundancy = we’ll have high correlations, high alphas’s Cronbach, high composite reliability, but low content validity)

Suggestions:
1) Create new items (and pre-test for content validity), keep just one of the first three items.

2) I think that “communication is considered a success factor” of other latent variable (create a structural model).

Best regards,

Bido

[MaKauTZ]

Hi,

The example that you gave is reflective:
- High correlations.
- Items 1 to 3 are the same (unusuful redundancy = we’ll have high correlations, high alphas’s Cronbach, high composite reliability, but low content validity)

Suggestions:
1) Create new items (and pre-test for content validity), keep just one of the first three items.

2) I think that “communication is considered a success factor” of other latent variable (create a structural model).

Best regards,

Bido

Thank you very much :)