Online Fragmentation in Wartime

Abstract

Theorists have long predicted that like-minded individuals will tend to use social media to self-segregate into enclaves and that this tendency toward homophily will increase over time. Many studies have found moment-in-time evidence of network homophily, but very few have been able to directly measure longitudinal changes in the diversity of social media users’ habits. This is due in part to a lack of appropriate tools and methods for such investigations. This study takes a step toward developing those methods. Drawing on the complete historical record of public retweets posted between January 2011 and August 2013, we propose and justify a partial method of measuring increases or decreases in network homophily. We demonstrate that Twitter network communities that focused on Syria are in general highly fragmented and homophilous; however, only one of the nine detected network communities that persisted over time exhibited a clear increase in homophily.