Influence among preferences and its transformation to behaviors in groups: An agent-based modeling and simulation of fertility intention and behavior

Abstract

We consider settings of group decision and negotiation where agents’ preferences (such as intentions, beliefs and opinions) are influenced by each other and thus their behaviors are possibly changed. We build a multi-agent system (MASIITIB) in the context of social networks to model the mutual influence among agents’ intentions and the transformation from agents’ intentions to their behaviors in groups. On the micro level of individual agents, we construct the self-evolution rule of agents’ intentions and the generation, constraint and termination rule of agents’ behaviors; on the macro level of networked structure, we describe the mutual influence on intentions among agents, which can be diversified in both strength and polarity. We detail this multi-agent system and run experiments and simulations using the interaction of fertility intentions and the generation of fertility behaviors in families as example. Two experimental programs are designed: one is to adjust the initial fertility intentions of prospective parents, and the other is to adjust the range of weight of influence among family members, to investigate the effects on the childbearing behavior and the number of newborn children in the long-term. This study intends to provide modeling bases for the dynamics of preferences and behaviors due to mutual influence among agents in groups, particularly for the study of fertility intention and behavior in families, and more broadly, the forecast of population growth and effects of fertility policy. It is an innovative try of distributed artificial intelligence (multi-agent system) in the field of demography and public policy, and provides with a new bottom-to-up perspective and unconventional agent-based method.