Logo eprints

Collective behaviour in digital societies

Samantray, Abhishek (2020) Collective behaviour in digital societies. Advisor: Riccaboni, Prof. Massimo. Coadvisor: Pin, Prof. Paolo . pp. 205. [IMT PhD Thesis]

[img] Text (Doctoral thesis)
Samantray_phdthesis.pdf - Published Version
Available under License Creative Commons Attribution Non-commercial Share Alike.

Download (4MB)

Abstract

The number of social and economic activities on digital platforms has been increasing since the last two decades, and especially in the last decade. Several such platforms also provide opportunities for interactions among participating users, either as a part of the primary activities on such platforms or as a secondary feature. Such interactions form the basis of the emergence of collective behaviour on individual platforms, wherein users’ interactions affect the aggregate state of the platform and at the same time the aggregate state of platform feeds into how users interact among themselves. Digital platforms turn into digital societies by providing many ways to conduct users’ interactions driven activities, that were traditionally conducted in the physical world, in efficient and scalable ways. Apparently, these two societies – digital and real world – exist simultaneously and have feedback effects in shaping outcomes in both the societies on various factors including behaviour, beliefs, opinions, and others. This thesis contains three essays including (1) peer influence on creation of new projects in the learning environment Scratch, (2) polarization of climate change beliefs due to homophily in interactions on the social media Twitter, and (3) effects on collective attention due to political framing of climate change by the news media Guardian. Particular emphasis is laid on statistical inference of the effects and hypothesizing mechanisms behinds such effects. The production and consumption of projects in the Scratch community, a digital platform developed by MIT Media Lab where users, usually young children, learn to program by creating and sharing projects, is analysed using the data for the first five years after its launch. In particular, investigation is done to discover if users are influenced by the popularity of their peers’ projects and their peers’ preferences for consuming specific baskets of projects. The major challenges in this type of analysis is to provide parsimonious models for complexities of interactions on the platform and to disentangle peer influence from homophily in the vast network of behaviours and friendships. Homophily is a term widely used in social networks studies to describe friendship or tie formations that arise due to similarities in behaviours or common attributes between participating agents in the formation of such ties. The analysis reveals that while Scratchers’ consumption preference is not influenced by their peers, the popularity of their projects is significantly influenced by their peers in short and long terms. A large proportion of the influence from peers is mediated via Scratchers’ creation of new projects, which highlights Scratchers’ subsequent decisions in response to existing popularity of peers’ projects. These insights can potentially help in incorporation of behaviourdriven designs in future educational technologies. Producer-consumer business models is at the heart of several social networking sites. Activities on such sites range from meeting friends, exchanging messages, propagating messages, advertisements, and others. Lack of regulations on information posting and limitations of computer-assisted information checks therefore provide opportunities for people’s beliefs to be polarized due to the spread of fake information in such social networks. Homophily in communication creates groups of people or agents with bounded beliefs about the reality, and hence can polarize a society. Such homophily in digital media has been termed as echo chambers which intuitively promotes the notion that people hear nothing more than what they already believe. Using evidence from 11 years of Twitter conversations on the climate change topic, an empirical analysis is conducted on the effect of homophily in communication patterns on the polarization of beliefs about the reality of climate change. The analysis reveals a counterintuitive result that increasing levels of homophily in communication predicts decreasing levels of polarization in beliefs in the long run. To understand better the mechanism of the effect of homophily on polarization, a model is developed that shows how polarization can emerge due to the joint effects of precision of misinformation propagating in a social network and homophily in communication among agents in the social network with differing beliefs. Credibility of fake news, modelled as precision of misinformation, circulating in the social network can lead to acceptance of the fake news (depending on agents’ susceptibility to it), thereby changing beliefs and creating polarization. The model shows that fake news can not polarize the society unless it has a minimal level of credibility, irrespective of the level of homophily in communication patterns. This throws a light on perhaps the most intuitive but usually the forgotten factor of information – credibility. While the results show that the climate change sceptic exchanges of messages on the social media Twitter do not carry enough credibility to create large scale polarization in society, they also provide useful indications to directly or indirectly quantify the emergence of credibility of information in digital platforms, and also to shift attention of technology from detecting fake stories to detecting fake stories which the society might find them to be credible. Digital news platforms have become outlets for engaging discussions. They provide journalists and publishers with several dimensions to gauge the acceptance of their articles and at the same time provides readers with simple tools to participate in discussions. It has been long acknowledged that news media frame their articles. As an instance of political framing, to understand how the politicization climate change articles influences the collective attention and discussion on such articles, articles published in the Guardian until 2018 are used to examine whether and how politicization influences readers’ collective discussion. The results suggest that (unknown) factors of perception associated to an article when it is categorized in ‘Politics’ section positively impacts the collective attention and engagement received on articles. Estimates also suggest that mentions of political inclinations of the entities within an article impact such collective attention. In particular, a large proportion of such participation is found to be mediated by discussions becoming politicized by past contextual entities related to the article but strictly absent in the article, thereby suggesting a temporal effect of perception. In addition, a large proportion of the impact of political mentions on users’ engagement is mediated by users who join discussion being influenced by past politically oriented contextual entities. Although no evidence is found to support that authors or journalists might be enjoying increasing marginal benefits from collective attention to their articles that result from their choices about mentions of political entities within the main texts of climate change articles, the results highlight that their choices do impact the readers’ perceptions and participations (and potentially the intensity of climate change action in real world). Overall, this series of research has contributed to improving how collective behaviour is shaped in digital societies in relation to peer influence in educational media, polarization of beliefs in social media, and political framing of articles in news media. Hopefully, these results would be of interest to general audience and researchers in fields of social sciences, economics, marketing, media and communications, and applied data science.

Item Type: IMT PhD Thesis
Subjects: H Social Sciences > HB Economic Theory
PhD Course: Economics management and data science
Identification Number: 10.6092/imtlucca/e-theses/319
NBN Number: urn:nbn:it:imtlucca-27033
Date Deposited: 25 Nov 2020 08:50
URI: http://e-theses.imtlucca.it/id/eprint/319

Actions (login required, only for staff repository)

View Item View Item