Tom Booth posted on Friday, October 28, 2016 - 12:43 pm
I am attempting to fit a 2 group model which contains both ESEM and CFA factors. I have established metric invariance for the factors across groups, and I subsequently want to test a mediation hypothesis among the latent variables, and test whether these structural parameters are equivalent across my two groups.
However, when I try and constrain structural parameters related to the ESEM factors, the model throws up errors.
I have tried doing this with factor labels, and with simply dropping the lines from the model for my second group - neither work.
Right, this is not generally available for ESEM. After estimating the ESEM model you can translate the ESEM model into an EFA in a CFA framework, imposing the right number of EFA restrictions (m^2 for m factors). See our short course handout for Topic 1. Then you can use Model Constraint to test what you want.
Tom Booth posted on Sunday, October 30, 2016 - 12:22 am
Thank you Bengt.
Sorry if that question has been asked many times.
Would you recommend this approach over say taking factor scores from the ESEM metric invariant measurement model?