Explaining Inference On A Population Of Independent Agents Using Bayesian Networks