I have fitted an ADE (threshold) model for two separate groups: an adolescent and an adult group.
The estimates of A, D and E differ between the two groups so I would like to test if this difference is significant.
I thought this might be possible by putting both groups together into one model, and then equating
the A, D & E estimates one by one to see if this significantly deteriorated the fit.
Attached is the script for the ADE model, I have managed to make one model in which I specify the A, D & E separately for adolescents (YNTR) and adults (ANTR), but I am not sure if I have done this correctly.
The model does run, but the estimates of A, D & E are slightly different to the model in which
I included only the adults. More specific, some of the initial estimates are negative while they were
first positive (see attached word file, in yellow). In the end, they are of course squared so it doesn't make a difference,
but it makes me doubt if there is something going wrong in my model?
Thank you in advance!