Blindfolding - when and what?

[i_laerdal]

Hi,

I have searched the forum in order to get a better understanding of the Blindfolding approach, and I have also read the Tenenhaus (2005)-article that is often referred to on this matter.

However, I still have some questions and I hope someone can help me out:

1) When should the blindfolding procedure be used, i.e. what are the main outcomes of the procedure and what can you use it for?

2) I understand that Blindfolding provides estimates of crossvalidated redundancy- and communality which should be used to evaluate the quality of the model. Are there any agreed-upon criteria for how high the cv-measures should be? I cannot find this in the literature.

3) I have read that G=7 is recommended. Should the blindfolding procedure be run for all latent variables simultanously, or should the blindfolding analysis be split up and only some latent variables be estimated at the time?

I would be grateful for any feedback on this!

Regards,

Ingrid Lærdal

[bergsorensen]

I used it for my entire model (based on the EPSI framework).

values above 2 are considered significant (t>2 wheer df=inifinit)

If you have more questions -> bergsorensen@gmail.com