I want to interpret the coefficients of a SEM model with binary endogenous variables (estimator = WLSMV).

I have found online that, since my variables are binary, the link function is probit so coefficients indicate amount of change in z-scores. This is hard to interpret, so I want to turn this z-scores into probability. I found the “margins” package and margins function can do this, but they do not support lavaan objects. Does anyone know another way to transform my coefficients?

Thanks in advance!