Agent-Based Modeling of Family Formation and Dissolution

Abstract

Patterns of family formation and dissolution are typically assumed to derive from an interplay between the structure of the marriage market and people’s partner preferences. One challenge that family researchers often face is that they have no direct measures of the preferences that have guided the partnering decisions that underlie observed family patterns. This is particularly the case when the interest is in long-term changes that have started before survey data about such preferences became available. This chapter introduces agent-based computational (ABC) modeling as a way to deal with this challenge. In ABC modeling, researchers make explicit assumptions about the constraints and preferences that guide people’s partnering decisions. These assumptions are then translated into formal models that are submitted to computational simulations of familial behavior in potentially large and heterogeneous populations, along with relevant and available empirical information. The results of these simulations make it possible to (1) assess whether a given set of preferences may have plausibly been involved in generating observed familial behavior given the constraints that people face and (2) to assess whether very different sets of preferences may generate similar patterns. We illustrate these capabilities of ABC modeling with three examples from our own research in the areas of assortative mating and divorce.