VALIDATION AND AGENT-BASED MODELING: A PRACTICE OF CONTRASTING SIMULATION RESULTS WITH EMPIRICAL DATA
As an emerging approach to explore the dynamics of voter preference, agent-based modeling (ABM) highlights new opportunities for intellectual exchange across disciplines, such as mathematics, political science, communication studies, and computer science. By aiming to contribute to cross-disciplinary communication for a better application of this approach, this paper summarizes what scholars have done about internal and external validation and presents a comparison between statistical analysis based on datasets generated in a laboratory and analysis based on corresponding empirical datasets. The results of the comparison suggest that, although there is no perfect matching, the comparison reveals some similarities in terms of increase or decrease in the proportion of different types of agents. This result further implies that an internally valid ABM model may lead to a certain level of external validity.