Computer Simulation and Statistical Modeling: Rivals or Complements?

Abstract

The model-based view of science encourages a ‘toolbox’ approach to theory, models, methods, and techniques. Some tools are multipurpose. Some purposes can be served by more than one tool. Some standardization in the use of tools is inevitable, but it is important to avoid stylized analysis, or the rote use of a tool for a given purpose. These ideas help clarify the relationship of two kinds of quantitative analysis, simulation and statistical modeling (what Adrian Raftery, paraphrasing C. P. Snow, has termed ‘the two main cultures of quantitative research’), notably the lack of interaction between them in day-to-day research.