ApolloChoiceModelling forum

Hi all. Hope you are doing well.

In a typical SEM, you could use an LV as a regressor or independent variable of another LV. I am wondering if in an ICLV this is possible. Can you use an LV as an independent variable of another LV that influences the choice? If so, how would you write it?

I understand that both LVs should need their gamma objective values, but I am curious about how to create the relationship between latent variables.

I imagine something like this in "conceptual" terms, but I do not know if it makes sense:

LV_endogenous = gamma1*X1 + gamma2*X2 + gamma3*X3 + Blv*LV_exogenous + eta_end

LV_exogenous = gamma4*X4 + gamma5*X5 + eta_exo

Hi

yes, this is fine, but a few things to take into account:

1 - for the model to be identified, you will need separate measurement model components for both latent variables, i.e. they need to be used with different indicators
2 - you'll need to change the order below, so that exogenous is defined first

Stephane