Abstract for: Connecting Micro Dynamics and Population Distributions in System Dynamics Models

Common system dynamics models capture the mean behavior of groups of indistinguishable population elements (e.g. people, tasks, widgets) aggregated in stock variables. However, many modeling problems require capturing heterogeneity across these elements with respect to some attribute(s) (e.g. weight, errors, price). The representation of heterogeneity could be important for correct characterization of behavior of a system as well as evaluation of policy options. In this paper we develop a method to connect micro-level dynamics (associated with elements in the population) with macro-level population distribution along an attribute of interest. The method enables modelers to efficiently characterize the distribution of attribute of interest without explicitly modeling all the elements in the population. We apply our method for modeling distribution of Body Mass Index and its changes over time in a sample population of 3074 female adults obtained from the National Health and Nutrition Examination Survey (NHANES) data. Comparing our results with the ones obtained from an agent-based model that captures the same phenomena shows that our method offers good precision with computational costs that are significantly less than agent-based models.