The pivotal mechanism revisited: some evidence on group manipulation