A recent study by researchers from Aalto University, the University of Helsinki and Qatar Computing Research Institute confirms the existence of echo chambers on Twitter.

“An echo chamber exists if the leaning of the content received by Twitter users is in part with the leaning of the content they share,” co-author Aristides Gionis, a professor in the Department of Computer Science at Aalto University, said in a statement. “An opinion echoes back to the user when it is being shared by others in the chamber, the social network around the user.”

Through an analysis of more than 2.7 billion tweets from 2009 to 2016, the researchers found that social media users primarily share, and are exposed to, political opinions that correspond to their own biases.

The study is the most comprehensive analysis of the structure of social media echo chambers. It demonstrates a strong correlation between the political leanings in content that social media users produce and consume.

The findings could provide one explanation for increasing political polarization in recent years.

The Study

The researchers based their analysis mainly on long-term political topics that they identified as particularly polarizing in the U.S., according to co-author Kiran Garimella, a post-doctoral researcher currently working at EPFL in Lausanne, Switzerland. These included high-profile issues such as “Obamacare,” abortion and gun control.

In addition, they also studied “a very large dataset (containing billions of tweets) from politically active users discussing politics on Twitter,” he said.

The researchers categorized social media users based on their role in echo chambers.

They identified three types of users: partisan users, gatekeepers, and bipartisan users. Each type plays a distinct part in the formation of echo chambers.

“Partisan users” consume and produce exclusively one-sided content.

“Gatekeepers” consume content from both sides of the aisle, but produce only one-sided content.

“Bipartisan users” consume and produce content from both the left and right.

To identify which category a given user falls into, the researchers created machine-learning algorithms that were able to predict partisan users with 80 percent accuracy and gatekeepers with 70 percent accuracy.

The relative success of each type of user in their social networks, in terms of their standings in network position, community connections and content endorsement, reveals a great deal about how and why echo chambers form.

The researchers measured network centrality in the underlying social network of users and endorsement based on reposts and positive ratings by other users.

They found that the most successful users are gatekeepers, while bipartisan users receive less attention in social networks than either gatekeepers or partisan users.

In other words, social media rewards partisanship and discourages more neutral politics.

While the study did not focus on a causal relation, the researchers believe it possible to hypothesize, given the current political climate in the U.S., that “being extremely partisan helps as witnessed by the popularity of partisan websites and tv shows,” he said.

“Gatekeepers are a small group of users who have higher than average network centrality,” Garimella said in a statement. “However, the users that gatekeepers follow are not always connected with each other; instead they belong to opposing sides. Finding and connecting the gatekeepers would help spread information to both sides, even though they are more challenging to identify than partisan users.”

The researchers were careful to eliminate any potential bot accounts from the study to ensure that only real users were included. This included accounts that were active for less than a year or unusually high numbers of connections.

Implications

There is no concrete solution to the problem of social media echo chambers, but the study hints at different strategies that might be helpful, said Garimella.

“There have been a lot of attempts by researchers (including us) to tackle this problem,” he said. “For instance, educating users about their presence in these echo chambers, connecting social media users to something (people or content) outside of their bubble, etc. Different studies report different extremes of success. For instance, we find that connecting people with opposing viewpoints is not necessarily a good thing, because it might backfire and tend to push people to even more politically extreme positions.”

Garimella believes that the research provides a deeper understanding of political echo chambers existing on Twitter and that such insights could be used to develop tools to help social media users break out of their bubbles.

“For instance, we find that there are ‘gatekeeper’ users who get information from both sides of the political spectrum but only share information on one side,” he said. “A next step would be to try to convince these users to share information on both sides.”

Another step in the right direction is an “understanding the phenomenon of the price of bipartisanship,” he added.