Modelling and social science : problems and promises

Formalised models are simplified representations of empirical phenomena that help to abstract away essential mechanisms from details and contexts. Although (mathematical or computational) modelling has always had a contested status in the social sciences, the use of formalised models is key to integrate abstract theorisation and inductive empiricism. This is especially true for agent-based modelling (ABM), which is a computational method which allows social scientists to study aggregate patterns as consequences of complex agent interaction. Unlike standard mathematical and statistical models, ABM permits us to consider heterogeneity, autonomy and local interaction, as well as the effect of institutional, structural or spatial environmental constraints. Simulations are then performed to observe and visualise aggregate properties and understand complex time-space dynamics at micro and macro scales. Considering the (ethical and economic) constraints on experiments in the social sciences, modelling and simulation are instrumental to test the logical coherence of theories, scale up microscopic observations and perform counterfactual analysis when scenario manipulations are difficult or impossible to perform in reality.