How to generate the factor correlation matrix after CFA?

[Friederike Eggert]

Dear fellow Stata users,

I was wondering whether it is possible and what the respective commands in Stata would be to produce the factor correlation matrix after a CFA for which I made use of the SEM builder? I learnt that in EFA this is possible because one can predict the factors after conducting factor analysis and rotation, and subsequently can also conduct regressions with having the factors predicted and stored as variables.

I am just curious to look at the correlations between the latent constructs (factors) to see whether there is a problem with multicollinearity or else.
Overall, I am applying SEM, and in the subsequent path analysis the issue is that the indirect paths of one of my independent latent constructs (exogenous) via 3 other latent constructs (which are both endogeneous and exogenous at the same time) to the dependent latent construct are significant but the direct path to the dependent latent construct is insignificant (very high p-value) and with a reversed sign in the path coefficient.

So, I was wondering if this just shows that there is absolutely no relationship (which would not make sense from a theoretical perspective), or if there might be an issue with surpression effects, multicollinearity, etc.?

I am looking forward to your responses.

Thank you in advance for your support and best regards,
Friederike

[Weiwen Ng]

I’m not familiar with the builder. However, a SEM will report the covariances between the latent variables at the end of the output table. If you add the , standardize option, SEM will report these as correlations.