I am exploring the bidirectional association between two variables over three timepoints (T1, T2, T3), and am considering using R's lavaan package to run a cross-lagged panel model. One of the dependent variables (DVs) is continuous and the other is dichotomous. And ideally I'd like to adjust the model for some covariates (some dichotomous, others categorical with more than two levels). All variables are observed (no latent variables). The below diagram is roughly what I'm interested in fitting.
My question is - can lavaan theoretically support this combination of continuous, binary and exogenous nominal variables in a panel model? My understanding is that, provided the WLSMV estimator is used, it should be able to handle it.
