1
$\begingroup$

I want to use factor scores computed from my CFA model to test my hypothesis. I use Maximum Likelihood for CFA and GLS to compute factor scores in EQS. My question is - are the factor scores derived this way supposed to be standardized? The reason I ask this question is I get factor scores that have a mean of 0 but the SD is greater than 1.Hence the confusion.

I just need to confirm if this is normal or I am having an error in my procedure/data.

$\endgroup$

1 Answer 1

1
$\begingroup$

When you define your model, you have to either fix a factor loading to 1, or the variance of the latent variable/factor to be 1. This is to make the model identifiable.

If you have specified the latent variable/factor variance to be 1, the SD should be 1. However, if you have instead fixed one of the factor loadings to be 1 so the model can be identified, then I don't think the SD has to be 1 (though it certainly could).

If you want the factor scores with variance/SD=1, either set the latent variance to 1 and allow all factor loadings to vary freely, or standardize the factor scores after creating them.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.