Does Social Media cause Polarization? Evidence from access to Twitter Echo Chambers during the 2019 Argentine Presidential Debate
We study how two groups, those inside vs those outside echo chambers, react to a political event when we vary social media status (Twitter). Our treatments mimic two strategies often suggested as a way to limit polarization on social media: they expose people to counter-attitudinal data, and they get people to switch off social media. Our main result is that subjects that started inside echo chambers became more polarized when these two strategies were implemented. The only scenario where they did not become more polarized is when they did not even experience the political event. Interestingly, subjects that were outside echo chambers before our study began experienced no change (or a reduction) in polarization. We also study a group of non-Twitter users in order to have a simple, offline benchmark of the debate’s impact on polarization.
All authors contributed equally to this work. We thank Hunt Alcott, Levi Boxell, James Druckman, Matt Gentzkow, Gary King, Jesse Shapiro, Ro'ee Levy, seminar participants at Facebook Core Data Science and especially Vincent Pons, for very helpful suggestions. We thank Julián Regatky, Lucía Freira, Lucas Soulès and Agustín Garassino for exceptional research assistance. This work was approved by the Ethics Committee of INECO (https://www.ineco.org.ar/), under the protocol titled “Political campaigns, political debates, and social networks” (PI Ernesto Schargrodsky, approved on 09/2019). This work was pre-registered under AEARCTR-0004850. This material is based on work supported by a Facebook Unrestricted Research Gift (“Content Policy Research on Social Media Platforms”). The views expressed herein are those of the authors and do not necessarily reflect the views of the National Bureau of Economic Research.