Computer simulation is widely regarded as a useful activity during various phases of research. However, depending on its context, the meaning, definition, and focus of the term can vary: In traffic planning, for example, simulation is used to determine useful configurations of a road network, thus focusing on the environment. An entirely different perspective is used within multi-agent systems. In such settings, the environment of the agents remains static, while the interesting research questions concern the behavior of the agents themselves. The research focuses on the microscopic level and the resulting emergent behavior. This article puts such diverse meanings in the context of a research process that treats descriptive and prescriptive research as two sides of the same coin. We develop a framework to classify different types of simulation, based on the actual research activity they are intended to be used for. Two case studies supplement the framework.