umxSuperModel optimises the fit of its submodels which themselves have objectives. But in was detecting these by whether they have an expectation.
Turns out that while non-group models (like a RAM model) have an expectation, umx twin models themselves optimise a group DON'T have an expectation - only a fitfunction.
Solution: get umxSuperModel to search for fitfunctions, which all optimisable models have, not expectations...
umxSuperModel optimises the fit of its submodels which themselves have objectives. But in was detecting these by whether they have an expectation.
Turns out that while non-group models (like a RAM model) have an expectation, umx twin models themselves optimise a group DON'T have an expectation - only a fitfunction.
Solution: get umxSuperModel to search for fitfunctions, which all optimisable models have, not expectations...