MU researcher looks into political polarization, lack of moderate voices on social media
Michael Kearney developed his own software called rtweet to collect mass amounts of user data from Twitter.
Apr. 16, 2019
MU research has found that although partisan Twitter users became more polarized leading up to the 2016 election, polarization as a whole may be exaggerated since many moderates remained uninvolved in political conversations.
In January, Michael Kearney, assistant professor in the MU School of Journalism with a joint employment in the Informatics Institute, published his research into political polarization and social media in the peer-reviewed journal “New Media & Society.”
Kearney’s research won the 2017 Lynda Lee Kaid Outstanding Dissertation Award, an award from the National Communication Association.
“It’s a very prestigious award, very competitive,” Ben Warner, associate professor in political communication and a member of Kearney’s dissertation committee, said. “I think it’s no surprise that he won the award. His dissertation was unconventional.”
The research centered around how partisan and nonpartisan social media users communicated and connected leading up to the election.
“The main goals were to see if political polarization and user networks increased in relation to proximity to the elections,” Kearney said. “As [the election] got closer, did we see more polarization, and also, was there a difference in users who followed political accounts versus users who tended to follow more entertainment-oriented accounts?”
For this research, Kearney collected data from 3,000 Twitter users who either followed Democrat partisans, Republican partisans, or non-political, entertainment-based accounts, seeing how the amount of partisan users they followed changed over a seven-month period.
Twitter makes its data openly available to the public, but because of Twitter’s rules about how a person can get data and how much data they can access, the process of collecting mass amounts of data would take a long time. As such, Kearney developed software called “rtweet” to make this mass data collection feasible.
“Because of the rate limits that Twitter puts on the amount of certain data you can get, each time I wanted to get a snapshot of what the user networks looked like took 50 hours,” Kearney said. “I did that 20 or more times. I basically had to write the computer code to do that so I could let it run for 50 hours.”
The software has been used by other academics to study Twitter data, and according to the rtweet website, it has over 74,000 downloads.
The fact that Kearney developed his own software is unique for communications-based research, Warner said.
“It’s not an overstatement to say very few people in the department of communication in any university write dissertations where one of the chapters is software programming,” Warner said. “He didn’t really come from a software background, he came at this wanting to study Twitter and teaching himself how to code so that he could study Twitter.”
There were many factors to Kearney’s decision to use Twitter for this research, including Twitter’s openness with its data and the platform’s focus on information, Kearney said.
“On Facebook, [user connections are] symmetrical, meaning two people actually know each other, agree to be friends, and then we see each other’s content,” Kearney said. “On Twitter, it’s asymmetrical. I can choose to follow you, and you don’t have to follow me back, or you can follow your favorite celebrities, see what they’re up to, follow organizations or mascots or sports. That type of asymmetry actually makes the platform more attractive for information seeking.”
One of Kearney’s findings was that partisans followed more partisan accounts leading up to the election.
“People weren’t magically balancing out their media diet,” Kearney said. “As the election got closer, they just continued to find more people on their side that they agree with and follow them.”
Additionally, Kearney found that the moderate, entertainment-focused accounts continued to not engage with partisan accounts or become polarized.
“I think a lot of attention gets put on that there is growing division between Republicans and Democrats, and in some respects that might be true,” Kearney said. “I think that division is always there and maybe will always be there when there’s two competing groups. But the division that might be increasing at a real rate is the division between the haves and have nots, or the politically interested and not politically interested.”
This group of not politically engaged users is related to the amount of other options for people who are uninterested in politics, Kearney said.
“Fifty years ago, you had three TV channels, and if there’s a public address, you watched it, you watched the president speak,” Kearney said. “Today, we have hundreds of cable channels, the internet, Netflix, Hulu and video on demand. We can choose anything we want, and if we don’t like politics, we have an infinite number of options at our disposal to avoid it.”
Because of many users’ choice to not engage in political conversations on social media, the political polarization that seems apparent when one looks at social media may be exaggerated and not indicative of the public’s true opinions.
“If you were to imagine every user on Twitter entering a political conversation, and then you were to imagine the actual political conversation that occurs on Twitter, the latter seems a lot more polarized because the only people in that conversation are the partisans,” Kearney said. “If everyone were in the conversation, you would have a lot more people expressing uncertainty, not really sure, or maybe ambivalent.”
Since this means that social media political discourse is not completely representative of the public, Kearney’s findings may be comforting to those alarmed by political polarization.
“I think it’s somewhat optimistic findings if your knee jerk reaction to Twitter is to say, ‘Oh no, where’s this country going? It’s polarized, and people are just yelling and talking past each other,’” Kearney said. “You can rest a little easier knowing that’s what happens when you get the most extreme of both sides.”
Also, Kearney’s findings may be cautionary for those who tend to only see and interact with content they agree with.
“Another takeaway is that we still tend to select our networks and our content based on existing preferences, and our decisions tend to reinforce what we currently think to be true,” Kearney said. “So, it might be a bit of caution for people who try to supply themselves the healthy diversity of information to think about the choices they make in that regard.”
Edited by Emily Wolf | firstname.lastname@example.org