A case for simulated data and simulation-based models in organizational network research

Abstract

Social networks shape innovation dynamics both within- and across organizations. Unfortunately, obtaining relevant and high-quality data on social networks is often a challenge. We argue that simulated networks and simulation-based models can be a valuable complement, and even a viable substitute, to real-world network data in innovation research and beyond. We draw on a review of network simulation models and methods to illustrate how researchers can utilize simulations in ways that are grounded in empirical best practice. Furthermore, we explain how simulation models can be used to build new and richer networks, either from scratch or by using existing real networks as the point of departure. As an illustration, we compare four widely used empirical organizational networks with their simulated counterparts to show that simulations can indeed be used to mimic certain core properties of real-world networks. At the same time, we also emphasize that domain expertise from researchers is critical for model selection, specification, and tuning. Finally, we offer a prescriptive framework on the generation, modeling, estimation, and validation of simulation procedures, to help researchers make greater use of simulated data and simulation-based models in empirical innovation research.