Drawing on the concept of explanation as it is currently used in naturalist political science, I discuss how the idea of a model as a goal-driven representation or "mental image" of a system fits into this paradigm and how it relates to simulation understood as the imitation of one process through another. Using the field of agent-based modeling as an example, I locate the terms modeling and simulation within the regular bathtub-structure of social explanation used in methodological individualism. I then ask what it takes for an agent-based simulation to be regarded as a successful explanation and how the representation of individuals through software agents might enrich our understanding of social phenomena, in particular electoral behavior.
