Hello,
I am estimating a one-factor CFA model and am using the command vce(cluster clustvar) because I have a dataset with a nested structure. I would like to test whether the fit of my CFA model is adequate.
Here is my code:
sem (LATENT -> q4 q5 q6 q7 q8), latent(LATENT) stand vce(cluster mdid) method(mlmv)
I then asked for fit statistics using the command: estat gof, stats(all), and only the Coefficient of determination is report with a message that says: "Note: model was fit with vce(cluster); only stats(residuals) valid." My understanding from this message that the fit statistics are limited due to the fact the model was estimated with vce(cluster).
Does anyone know why it is not appropriate to estimate fit statistics with vce(cluster?) How do I assess whether the fit of the model is adequate?
I found a posting asking a similar question date 05 Apr 2013 by Nicolas Sommet, but I didn't notice any replies to that message.
Thanks in advance for any help!
Caroline
I am estimating a one-factor CFA model and am using the command vce(cluster clustvar) because I have a dataset with a nested structure. I would like to test whether the fit of my CFA model is adequate.
Here is my code:
sem (LATENT -> q4 q5 q6 q7 q8), latent(LATENT) stand vce(cluster mdid) method(mlmv)
I then asked for fit statistics using the command: estat gof, stats(all), and only the Coefficient of determination is report with a message that says: "Note: model was fit with vce(cluster); only stats(residuals) valid." My understanding from this message that the fit statistics are limited due to the fact the model was estimated with vce(cluster).
Does anyone know why it is not appropriate to estimate fit statistics with vce(cluster?) How do I assess whether the fit of the model is adequate?
I found a posting asking a similar question date 05 Apr 2013 by Nicolas Sommet, but I didn't notice any replies to that message.
Thanks in advance for any help!
Caroline
