Data Integration without conditional independence: a Bayesian Networks approach

Statistical Matching, at a macro level, consists in estimating the joint distribution of variables separately observed in independent samples. As a consequence of the lack of joint information on the variables of interest, uncertainty about the data generating model is the most relevant feature of matching. In the present paper the use of graphical models to deal with the statistical matching uncertainty for multivariate categorical variables is considered, under both a model-based and a model-assisted perspective.