Hello everyone,
I'm performing my firsts CFA in Amos and I come here with a doubt. I've tested 2 models: an unidimensional model and second-order factor model and the fit indices are exactly the same ones for both of them. How is this possible? I've thought the small sample (n = 200) could be part of the explanation, but I'm not sure at all... Any suggestion?
The two models that you specified may be mathematically equivalent. Do they have the same degrees of freedom?
Model 2 (the second-order factor model) seems misspecified/odd. It looks like you did not include residuals for the first-order factors, which is typically done, since the second-order factor usually cannot explain 100% of the variance in the first-order factors. There is usually some residual variance for each factor that is estimated as a free parameter.
In addition, the covariances between the error terms that pertain to the same first-order factor (e.g., E1 and E2) do not make sense. The first-order factor should account for the covariance among its indicators (e.g., D1.1 and D1.2), so it is not clear why there should be an additional error covariance.
Hi Christian,
First of all, thank you so much for your help, I really appreciate it.
Both models have the same degrees of freedom (31).
I did not include residuals for the first-order factors because I'm testing models that were tested in previous studies and they did not include them.
Wȩziak-Białowolska, D., McNeely, E. y VanderWeele, T. (2017). Flourish Index and Secure Flourish Index Development and Validation. SSRN Electronic Journal. 6(1). https://doi.org/10.2139/ssrn.3145336
Moreover, amos does not tell me the model is misspecified despite I do not include them.
Regarding the covariances, I included them because a friend of mine showed me it is viable when the modification index is higher than 10, but only with the error that pertain to the same first order factors. The goal was to get better fit indices.
Oort, F. J. (1998). Simulation study of item bias detection with
restricted factor analysis. Structural Equation Modeling: A
Multidisciplinary Journal, 5, 107–124. https://doi.org/10.1080/
10705519809540095
Perry, J. L., Nicholls, A. R., Clough, P. J., & Crust, L. (2015). Assessing model fit: Caveats and recommendations for confirmatory
factor analysis and exploratory structural equation modeling. Measurement in Physical Education and Exercise Science, 19(1),
12–21. https://doi.org/10.1080/1091367X.2014.952370