Polarization in Political Rallies: A Markovian Agent-Based Model for Opinion Dynamics

One of the most studied problems among researchers in recent years is how individuals form their opinions. This problem has become more urgent with the advent of social networks, which can easily influence a huge number of followers and have become increasingly pervasive over time. The produced effect is the rise of polarized opinions among different groups of people. Understanding polarization is of great relevance across various application domains, such as economics and politics. Opinion dynamics has often been studied by exploiting the popular Friedkin-–Johnsen model. In this paper, we propose a different modeling approach based on the Markovian agents paradigm for deriving metrics characterizing polarized opinions. The main goal of this work is to demonstrate the potential of the Markovian agent modeling paradigm for the analysis of opinion dynamics. The main advantages of Markovian agents are the ease of setting a large number of behavioral parameters, spatial distribution of agents, scalability, and numerical tractability. We extend our previous work, in which we analyzed a peer assembly and validated it against other commonly used modeling approaches. In our opinion, the Markovian agent approach offers an effective modeling framework due to its scalability and flexibility in handling parameters that describe the behavior of individuals in the opinion formation process. The context we will discuss is inspired by election rallies, where an assembly attends a speech by a political candidate. The crowd consists of individuals with diverse initial political opinions, and the candidate seeks to polarize them toward his/her own political stance.