Bi-partisan News Coverage of COVID-19
COVID-19 has obviously been a dominant part of the American news cycle since the country’s first positive case was confirmed in late January of 2020. As it is with any major event, people of various different backgrounds are looking to outlets of comfort for information on this ever-changing phenomenon. And of course, the media’s agenda setting effects are being magnified, especially in the United States where the news media has been highly partisan.
Considering this, our team was interested in understanding how a media outlet’s political partisanship will impact the words, phrases and rhetoric they use when reporting on COVID-19?
To do so, we decided to analyse news content produced by 9 news outlets, between January 1, 2020 - March 31, 2020. When identifying outlets, we based identification on Faris et al’s (2017) analysis of outlets that Clinton and Trump supporters referred to during the 2016 Presidential election. We decided on the Huffington Post and Slate (Left); the New York Times and Washington Post (Centre-Left); the Hill and the Wall Street Journal (Centre) and Fox News, BrietBart, and Daily Caller (Right). Using structural topic modeling then narrowing the results, we identified fifteen recurring topics, and conducted analyses based on both topic and partisanship. Here is a quick overview of our 31,552 articles, and a topic-wise breakdown:
The top five most recurring topics were:
The economic impact of COVID-19
Forced closures of businesses
Local government response to COVID-19 (green)
Human interest stories
Presidential communications (purple)
It was interesting to see which groups of outlets focused on which topics.
There were some obvious findings. For example, most of the content about the economic impact was generated by center-leaning outlets. Spikes in coverage coincided with the two stock market drops: one at the end of February, and one in the middle of March.
There were also some interesting topics, that were covered very briefly over our window of consideration but deserve a call out, such as the impact of COVID-19 on sports.
Most of the coverage was generated by center-left-leaning and right-leaning outlets. Of note, the content peak seems to predate the announcement regarding the cancellation of March Madness.
To analyse the difference in the kinds of words being used by outlets belonging to different partisan groups, we decided to focus left vs center-left-leaning outlets and right vs left-leaning outlets.
An interesting note is that there seems to be very little overlap between the keywords being utilised by the left, left-centre and right-leaning outlets. Also, as depicted by the data visualisation, there seems to be a significant difference in the subject matter priorities of the different outlet groups.