Twitter posts reveal polarization in Congress on COVID-19

Study shows how quickly parties diverged on pandemic

The rapid politicization of the COVID-19 pandemic can be seen in messages members of the U.S. Congress sent about the issue on the social media site Twitter, a new analysis found.

Using artificial intelligence and resources from the Ohio Supercomputer Center, researchers conducted an analysis that covered all 30,887 tweets that members sent about COVID-19 from the first one on Jan. 17 through March 31.

The algorithm they created could correctly classify the political party of the member who sent each tweet 76 percent of the time, based only on the text of the tweet and the date it was sent.

Jon Green

Jon Green

“We found that once the parties started to figure out the political implications of the issue, polarization was evident in the tweets pretty quickly,” said Jon Green, co-author of the study and doctoral student in political science at The Ohio State University.

The study was published today (June 24, 2020) in the journal Science Advances.

“It is remarkable that we could identify partisanship even when members have only 280 characters to send their messages in Twitter,” said study co-author Skyler Cranmer, the Carter Phillips and Sue Henry Professor of Political Science at Ohio State.

Democrats sent out significantly more tweets (19,803) about COVID-19 than did Republicans (11,084), the study showed.

The gap in the number of tweets sent by Democratic versus Republican politicians widened after the first case of community spread was identified in California and grew further following the declaration of a national emergency.

“This suggests Democratic members were sending earlier and stronger signals to their constituents that they should be concerned about the crisis,” Cranmer said.

What Democrats and Republicans tweeted about concerning the pandemic was different, too, results showed.

For example, the word “health” was used in 26 percent of Democratic tweets, but only 15 percent of Republican tweets.

Skyler Cranmer

Skyler Cranmer

Overall, Democrats were more likely to discuss public health and safety, as well as American workers, while Republicans emphasized a general need for national unity, discussed China and business, and framed the pandemic as a war.

As one component of the analysis, the researchers identified members who fell in the range of what they called “partisan overlap.” Congressional members in this overlap area were those whose tweets were more likely to be confused by the algorithm with those of someone from the other party.

Only 31 percent of members fell in this area.

“That means for 69 percent of members, their tweets are more partisan than the most similar member of the other party,” Green said.

Polarization was not constant over time.

In the first full week after the first mention of COVID-19, the algorithm developed by the researchers had relatively low accuracy when trying to determine whether a Democrat or a Republican wrote a particular tweet. That indicates there was little polarization.

However, polarization quickly rose, peaking during the week beginning Feb. 9. It then declined slightly in early-to-mid March before rising again in late March as the parties debated economic relief packages.

The findings suggest that Congress missed an opportunity early in the pandemic to develop a consensus that could have helped the United States respond to the crisis, Cranmer said.

“Something on the scale of COVID-19 requires a large-scale government response. The government can respond much better when it is united in its mission,” he said.

Green and Cranmer, co-authored the study with their Ohio State political science colleagues Jared Edgerton and Daniel Naftel, along with Kelsey Shoub of the University of South Carolina.

 

Contact: Skyler Cranmer, Cranmer.12@osu.edu

Jon Green, Green.1764@osu.edu

Written by Jeff Grabmeier, 614-292-8457; Grabmeier.1@osu.edu

Share this page
Suggested Articles
Libraries, Data, and Privacy

By Alison Armstrong, MALS, and Amanda Rinehart, MLIS and MS, Botany/Plant Biology Librarians have a long history of respecting, protecting, and defending patron[i] confidentiality and privacy; the freedom to read...

TDA awards seed grants to four interdisciplinary teams

Translational Data Analytics has awarded seed grants to four faculty teams tackling challenges ranging from detecting unexploded land mines to predicting dementia. Selected by a group of 18 faculty reviewers...

Data visualization: The new fundamentals

By Lee-Arng Cheng University Libraries Data Visualization Specialist Lee-Arng Cheng With the shear amount of data being collected and analyzed nowadays, the value of understanding and making sense of data...

New research examines interplay between microbes

By Misti Crane Below the surface of systems as large and ancient as an ocean and as small and new as a human baby are communities of viruses and bacteria...

Registration open for TDA Fall Forum Poster Session

Ohio State postdoctoral researchers, graduate and undergraduate students, faculty, and staff will present on their data analytics and decision science research at TDA’s second annual Fall Forum Research Poster Session on...