Some Notes on the Use of Cluster-Weighted Modeling in Effectiveness Studies