I am running two CFA analysis were I am investigating a 6-factor correlated model and a 5-factor correlated model in a questionnaire. For this purpose I am using Lavaan (version 0.6-19.2150). I would like to calculate McDonals Omega for my two models, one for the 5-factor model and one for the 6-factor model.

After having run the CFA analysis (code below):

fit_six <- cfa(six_fac, data = data, std.lv = TRUE, estimator = “MLR”)

fit_five <- cfa(five_fac, data = data, std.lv = TRUE, estimator = “MLR”)

I then try to estimate omega with compRelSEM() function in the semTools package and get the following returned for my 6-factor model:

   strategies nonacceptance       impulse         goals     awareness       clarity 
        0.806         0.887         0.911         0.879         0.543         0.793

My two questions:

(1) Given that the latent variables named above are correlated, does the omega value for any one subscale I get from compRelSEM() account for variance explained by the other subscales?

(2) Even thought a higher order factor is not explicitly modelled, is it possible to get the omega for the total score of the scale and how would one go about to achieve this?

Thank you in advance

I have used the psych package function to calculate omega for the total scale (code below)

omega(data[total_index])$omega.tot

However, I do feel insecure of whether this is correct or not.