Rejoinder to Discussions on: Data-driven confounder selection via Markov and Bayesian networks.