Being Both Too Liberal and Too Conservative: The Perils of Treating Grouped Data as though They Were Independent | Synapse