bifactor factor scores

[Sebastian Marin]

Hi, 

I ran a bifactor model with 3 specific traits and one broad trait. When I look at the factor correlations, they do not correlate. However, when I extract the factor scores per trait and correlate those, there seem to be correlations. I want to use these factor scores as criteria in a regression equation, but I wanted to have them completed uncorrelated.  I was wondering why there is a discrepancy, or if I should even be worried? 

Best, Sebastian 

[Phil Chalmers]

Hi Sebastian,

I've mentioned this a few times on this forum in the past. Correlations between point-estimated factor scores are possible even if the underlying structure assumes the latent traits are uncorrelated, because this is not a constraint imposed at the time of estimating the individual latent trait scores (hence, shrinkage plays an effect). The only reasonable way to retain the latent covariance structure is to use plausible value imputation methods, but these don't give central tendency estimates of the individual trait estimates unless they are averaged over (in which case they will give the EAP estimates in expectation). HTH.

Phil

--
You received this message because you are subscribed to the Google Groups "mirt-package" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mirt-package...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/mirt-package/7a26c760-9188-4a23-a836-bf85d68e33b3n%40googlegroups.com.