Why Do Twitter Birds of a Feather Flock Together? - An ABM Approach

• In Event: Social Media and Polarization

Abstract

It may be argued that one thorny problem vexing America today is a polarization of public opinion on controversial issues such as climate change, free speech, environmental policy, among others. A polarized public opinion will create great obstacles to Habermas’ deliberative democracy. For deliberative democracy to function, citizens need a vibrant public sphere where it permits an unfettered flow of information, ideas, and debates.
In an age laden with advanced ICT’s, social media such as Twitter has rapidly become an important channel via which information is exchanged, public debates conducted, and even activism and political campaigns carried out. Indeed, “Twitter has become one of the most important communication arenas in daily politics” (Barbera, 2015, p.76).
So, have those digital platforms empowered by ICT’s promoted a “truth-tracking” deliberation (in Habermas’ words) and constructive civic debates? A large body of research seems to suggest the opposite - instead of nurturing a reflective citizenry, internet reinforces citizens’ prior political views. Increasingly, we hear scholars argue that there is an echo chamber effect on social media that leads to worrisome phenomena such as ideological segregation or polarization of public opinion. Echo chamber refers to an environment in which a person encounters only beliefs or opinions that coincide with their own so that their existing views are reinforced, and alternative ideas are not considered.
The echo chamber thesis is not without challenge. Scholars also found mixed results regarding extremism in an experimental setting or inconsistent evidence for reinforcement of preexisting attitudes. Some demonstrated that the echo chamber effect could be mediated by “strong social endorsements” that could increase exposure to heterogeneous perspectives.
Suffice to say, first, social sciences have thus far produced a prodigious body of findings on the aforementioned subjects. Second, literatures are mixed, as will be further demonstrated in the present paper. I suspect that methodology plays an important role in this mixed picture. Briefly, I would like to point out two limitations the traditional research methodology suffers. First, social scientists often treat the objects of their research in aggregates rather than individuals. Therefore, when studying system-level problems, they use approaches such as differential equations to map how the whole system change. But when dealing with problems of emergence, such a method tends to muffle underneath mechanism. Second, variables are treated mainly in stationary and in separation. By that, I mean political scientists, for instance, study institutional factors while holding individual behaviors constant, and vice versa. But problems of emergence often require us to both look at what “happens to the system because of what its individuals do and what happens to the individuals because of what the system does.” (Railsback & Grimm, 2019, p.31).
The formation of polarized public opinion is a problem of emergence. This polarization is a global pattern that emerges from various local interactions. This complex picture is entangled with both human and structural factors such as confirmation bias, cognitive dissonance, selective disclosure, and filter bubbles, just to name a few. To adequately approach a complex phenomenon as such, we need a holistic perspective that takes into account various entities of a system, their linkages, and the feedback loop between individual agents. This alternative approach requires a different conceptual framework and analytical tool that go beyond the classic linear models.
This project, therefore, constitutes a modest effort to explore a new approach to analyzing the formation of polarized public opinion on Twitter. An agent-based model will be presented to explore what environmental factors and individual choices, as well as their interaction, would generate global patterns such as the segregation of public opinion. This model aims to observe what global-level patterns emerge from what local actions through virtual experiments and to conduct a series of quantitative analyses based on data produced by computer simulations. For instance, we can analyze the relations between the number of retweets and the size of an echo-chamber. How long (i.e., how many retweets) would it take for an individual to join an echo chamber? The relations between the distance an agent migrated along the ideological line and its preconception score.

Author

• Qian Zhang, Purdue University