For many, social media offers a new site of political debate, and a forum in which competing views can and should be exchanged. But, there is a great deal of concern about the existence of “echo chambers”: a dynamic within public political discourse where citizens speak only to, and hear only from others of the same ideological persuasion.

This dynamic is commonly thought to have a negative impact both on democracy as a whole and on the political capacities of individual citizens.

"We have seen the vast majority of political conversations migrate to digital platforms, and the consequence of this migration is that we have discursively balkanized," said DSI Presidential Fellow Colin Kielty, a Ph.D. candidate in Politics. "If you're concerned about the possibility that average citizens can have a part in the political process, the echo chamber thesis is damning."

And for this reason, the vast majority of studies on the topic have been confined to establishing the existence and extent of the echo chamber effect; or, in addition, to determining whether certain ideological groups or issues are more likely to be stuck in an echo chamber than others. This approach, however, may fall prey to circular reasoning by using the same ideological variables to both define the outcome and to causally explain it. Moreover, they do not contribute to our understanding of how citizens can “escape” the echo chamber – by reaching broader and more diverse audiences – save by “toning down” the ideological content of their online speech.

Yet, citizens may take “toning down” to be at cross-purposes with their reasons for engaging in online discourse in the first place, whether those reasons are to in fact to promote an unabashedly ideological cause by reaching the largest possible audience, or to build community around a cause with others of like mind.

The question is not whether echo chambers exist, but if you can break free of them, and how?

For this DSI Presidential Fellows research project, Kielty and fellow Ph.D. candidate Anup Gampa (Psychology) seek to understand online political discourse from the perspective of those engaged in it.

Using a dataset consisting of 3.8 million Twitter users and 150 million tweets, and an additional dataset containing more than 40 million tweets corresponding to 45 keywords centered around the Black Lives Matter movement, the researchers will explore how citizens can expand the audience for their political discourse without modifying the issues or ideology. They provisionally hypothesize that greater exposure to social media users who hold divergent political beliefs will contribute to escape. Psychological research suggests that exposure to different social groups leads to transformed attitudes towards those groups, and research on political deliberation suggests that listening not only leads to greater understanding but also to more effective ways of presenting ideas.

The researchers will assess the effect of exposure for users who are differentially exposed, but are otherwise of the same ideological backgrounds and talking about the same political issues. In this way, they avoid building the implication that citizens should “tone it down” into their analysis of the echo chamber effect.

Further, citizens may not always prioritize escaping from the echo chamber. Instead of reaching a broad and ideologically diverse audience, they instead may be consciously trying to reach others of like mind, establishing ties of support and solidarity – what we label “community.”

"Today people are creating their political identities online," Gampa said. "We want to use the power of data science technologies we're really excited about, like natural language processing, to harness the explosion of data on social media and answer a well-established question: can you change people's implicit attitudes, and how?"

Stay up-to-date on this research

Default placeholder headshot.
Colin Kielty
Ph.D. Candidate
College of Arts and Sciences
Default placeholder headshot.
Anup Gampa
Ph.D. Candidate
College of Arts and Sciences
Brian Nosek
Completed in:
Social Media