Abstract We study a combinatorial problem formulated in terms of the following group-formation scenario. Given some agents, where each agent has preferences over the set of potential group leaders, the task is to partition the agents into groups and assign a group leader to each of them, so that the group leaders have as high support as possible from the groups they are assigned to lead. We model this scenario as a voting problem, where the goal is to partition a set of voters into a prescribed number of groups so that each group elects its leader, i.e., their leader is a unique winner in the corresponding election. We study the computational complexity of this problem (and several of its variants) for Approval elections.