Juliette Quenioux
Pronouns: She/Her/Hers
Research Mentor(s): Justine Davis
Co-Presenter:
Research Mentor School/College/Department: DAAS / LSA
Presentation Date: April 20
Presentation Type: Poster
Session: Session 2 – 11am – 11:50am
Room: League Ballroom
Authors: Justine Davis, Juliette Quenioux
Presenter: 64
Abstract
Despite the prevalence of election violence across the globe, we have yet to systematically investigate the use of social media in countries that have experienced or are expected to experience election-related violence. This paper aims to fill the gap in our understanding of popular narratives around elections in contexts of violence. What predicts discussions of election violence in social media? To answer this question, we conduct computational text analysis on over 1.2 million posts from 171 public groups supporting major political candidates on Facebook in Côte d’Ivoire from 2019-2020. In order to analyze photos and videos, we are using hand-coding, pre-trained convolutional neural network models, and machine learning techniques. We have found that discussions of violence-related topics are more prevalent among opposition-supporting groups. Opposition supporters are not only more likely to discuss violence, but they are also more likely to use hostile, xenophobic, and incendiary language, and are more likely to describe specific episodes of violence than incumbent supporters. Alarmingly, we find that violence-related topics are associated with more shares on Facebook. This research has much larger implications than Côte d’Ivoire; we hope to demonstrate how leveraging social media can provide insight into the processes underpinning the use of and support for violence and polarization in elections globally.
Arts and Humanities, Interdisciplinary, Social Sciences
