Abstractions, ensembles, and virtualizations simplicity and complexity in agent-based modeling

In this paper we consider the uses political scientists have made of agent-based modeling (ABM) and the challenges associated with designing research at differing levels of complexity. We propose a typology of ABM research designs-investigating abstractions, testing theories comprised of ensembles of simple variables, or implementing virtualizations of complex situations. Our illustrations are drawn from work done on problems pertaining to the evolution of collective identities and norms and to their contribution to collective action. By increasing the visibility of research design questions and clarifying the choices and opportunities associated with them, we seek to expand the scope of responsible methodological uses of ABM techniques and render the increasing variety of that work accessible to wider audiences.