Quick question, I am really stuck...
So, my measurement model in AMOS has great goodness-of fit indicators, and the coefficients are highly significant. However, the error variances are immense. As a result, the convergent and discriminant validity are affected (some AVE values are lower than 0.4).
I know that this usually due to multicollinearity. But the correlations between my constructs are not very high (the biggest is about 0.75).
Is there a way to reduce these errors? Why did they appear? It is the first time in my life when I see such big error variances.
Thank you!
Bogdan
All the usual GOF indicators: CFI, TLI, GFI, AGFI, RMSEA, SRMR....
Smaller factor loadings result in larger error variances. Error variances are the portions of variance in each measurement that do not covary with the latent factor. These are interesting in as much as they can indicate "good" and "bad" measures of a latent factor. I would check the measures for the latent variables. Ensure that the measures are reliable and are measuring what are supposed to measure (think of the conceptual model).
I'm referring to your indicator variables. As mentioned above i would suggest to go your conceptual model.
It can also be due to the size of the data or the large number of variables, I think that if its factor loads are greater than .40 it counts to evidence the model, for example in the case of the AVE it takes into consideration the factor loads that if they are made with WLSVM analysis and ordinal data it can improve and show an acceptable AVE, and the AMOS uses ML which is not an estimator that adapts mostly to the data considering that it should have a normal distribution but to eliminate atypical data.
- Badges
- Science topic
- Similar topics
- Psychology