A chain graph multilevel model for the analysis of graduates' employment

The main goal of the present paper is the analysis of the working position of graduates using the potentialities of multilevel models and chain graph models, extended to the case of correlated data. After a brief introduction to multilevel modelling and a description of the conditional independencies implied by the model, the paper describes chain graphs for multilevel models. The proposed model is used to analyse the factors influencing the graduates’ job position, using data on the graduates of the year 2000 of the University of Florence. The work ends with some indications for future research.