Online communities are an increasingly widespread and influential phenomenon (e.g. Facebook, Digg, YouTube). While economic logic suggests that the reduced communication barriers online should bring people together through exchange of ideas in a global village, group polarization may occur more easily online as like-minded people can find peers regardless of geographical constraints and reinforce each other?s views. Polarization may result in the monopolization of a social network by the majority group. Therefore a key empirical question is how the dynamics of user content generation, consumption, and evaluation leads to the formation of shared or divergent worldviews online.
To address this question, this research team will first empirically measure the evolving heterogeneity of sub-communities in an online social news community of over 30000 users. The team has access to complete data on all interactions since its inception in 2006. They will then identify the factors influencing the dynamics of different sub-communities within this community (growth, contraction, merging, splitting, birth and death of sub-communities), and the factors influencing user participation and join/leave decisions. Interviews and content coding will be combined with statistical methods for the purpose of understanding the underlying mechanisms that govern the dynamics of sub-communities in an online environment.
This study will shed light on the dynamics of polarization and homogenization in online communities, which is an issue of great interest as civil society copes with the changes engendered by new, pervasive, electronic communications media. In addition, in the process of measuring community opinion over time the team will develop and improve upon metrics and analysis methods that should be of value to others who wish to study group beliefs or interactions in online communities.
From the emergence of cultural norms to evolution of public opinion, the reinforcement of prejudices and the construction of tolerance, opinion dynamics are central to many questions researchers and policy makers care about. With the increasing proliferation of digital interactions, new questions and opportunities for research emerge. The current project combines statistical estimation and dynamic modeling to better understand the dynamics of online communities. Besides training graduate students in data science and contributing to workforce development, this project tackles a few research tasks. Specifically: 1) We develop a method for inferring impression (whether users have seen an item online) based on what they have voted for. 2) We design and implement an automated method for extracting user opinions in online communities based on their interaction pattern. 3) We use that method to estimate the online opinion changes. 4) We estimate the underlying decision rules that guide individual participation in online communities including visiting the website, posting stories, and voting for stories. 5) We build an agent-based model that is fully specified based on realistic parameter ranges. The core feedback mechanisms we empirically estimate are simple, but powerful. First, individuals change their opinions as they consume media. In turn, they produce media for the consumption of others (in our case through posting stories they find from the internet) which is closely related to their own opinion. As a result the majority tends to convert more of the community members, further reducing the diversity of opinions expressed in this community. A second mechanism relates to erosion of motivation among the outliers of the community. Observing few attractive stories and getting little support, these individuals are more likely to leave the community, which will further reduce the heterogeneity of the opinions expressed. Analysis of the simulation model shows that the dominant mode of operation in the social news website we analyze is majority domination, where the majority of participants converge to a single region of the opinion space, and the outliers become relatively inactive in the system. We also show that other modes of behavior, including competitively polarized, diversity, and complete consensus are also feasible. An important parameter in determining the dominant mode of behavior is the distance between the posted stories embedded opinion and the opinion held by the individual who posts the story. Current online community structures leverage two alternative designs to prioritize information sharing and avoid overload: Filtering and ranking systems. Ranking systems promote the majority’s point of view (like Balatarin.com or Reddit.com) and leads to majority dominant online environment. Filtering methods create a bubble around similar opinions by personalized filtering (e.g. Netflix). This research can be utilized to design online communities that are better with respect to some social utility measures. For example the filtering of stories presented to individuals in a social news website could be personalized and take into account the impact on the polarization, motivation, and long-term diversity of the community.
