Evaluation of the degree effect on the work path by a latent variable causal model

We formulate a causal model for the effect of the degree program on the work path of graduate students, which is based on a Markov process to represent latent characteristics of the subjects. A latent Markov model with covariates and mixed-type responses results, which is estimated by an EM algorithm. We illustrate the proposed approach through an application to a dataset deriving from administrative panel data which concern labor market in Lombardy and allows us to evaluate the effectiveness, in terms of education, of certain Universities in Milan