The Partisanship, Propaganda, & Disinformation: Online Media & the 2016 U.S. Presidential Election — Click here to download PDF of the study.
This Street Prophets Coffee Hour is brought to you by The Berkman Klein Center for Internet & Society. Today’s article, The 2016 U.S. Presidential Election, is part 11 of 15 in a series about figuring out just what is going on in American politics. It will be about how we got to where we are now. And hopefully present a story of where we should be going. Along the way we will take a look at Russia, the U.S. 2016 Presidential election, Memes and Fiction, Network Propaganda, soft warfare, and cyberwarfare.
This is an Open Thread and all topics of conversation are welcome. What is for dinner? How are you doing? What is on your mind. If you are new to Street Prophets please introduce yourself below in a comment.
I was shocked by the success of Trump’s win of the 2016 presidential election. I thought Clinton would win. What Happened? Why was the Democratic Party machine defeated? What should we do about it? Can we do anything at all? All of these questions prompted me to try and figure out what was going on. Ultimately it lead to me researching and writing this series.
My starting place was this study from Harvard: Partisanship, Propaganda, & Disinformation: Online Media & the 2016 U.S. Presidential Election. This study was followed up by another study from Harvard released in 2018 called Network Propaganda that builds on the data collected in the first study and goes in to a far more detail analysis just what transpired.
Below is an explanation of how the data was collected for the first study.
“The starting point was a collection of approximately two million stories identified on the open web that were relevant to the election. The story collection process began by running a seed query on the Media Cloud archive for stories published between May, 1 2015 and November, 7 2017…”
“We collected and analyzed data from Twitter using several approaches. Using Crimson Hexagon, we drew a random sample of 4.5 million tweets that matched the same search terms described above for the large election set: tweets that included the name of any of the major candidates.”
“To evaluate the sharing of stories on Facebook, we queried the Facebook API to acquire data on the number of times each of the stories in our data set had been shared on Facebook.”
From Harvard Study. Appendix 1: Methods, Selected snips from pages 128 — 129.
Figure 11. The link-based network map of election media sources. Network map based on open web media from May 1, 2015, to November 7, 2016. Click to enlarge or explore this map in higher resolution PDF.
The Link-Based network map of election media sources was created by collecting and categorizing millions of sentences shared on media outlets, websites, and social media during the 2016 Presidential election.
Looking at Figure 11 is rather disappointing other than to get a sense of size. It is much better to click this link, Figure 11 PDF, and zoom in on the graphic using a PDF viewer. One can play around with the scroll bars and the zoom function while it is rendering. In the viewer I was using It zoomed in to 1000%. In the PDF viewer the rendering process starts with a point and then draws the in link connections.
The Media Landscape With Location of Daily Kos in lower left corner. Click to enlarge.
These lines represent inlinks to the media source. Someone or something external to the media source made reference to the media source in a web hyperlink. The size of the circle represents the number of inlinks. Popularity or number of inlinks is represented by the size of the circles. The larger the circle the more inlinks .
There is a color code, Partisanship Attention Color Code, for the network maps that equates dark blue with the political left and dark Red with the political Right. Pink is center Right and light blue is center Left. Green is for center. The gray circled media sources were not ranked Left or Right in the study.
From Executive Summary
In this study, we analyze both mainstream and social media coverage of the 2016 United States presidential election. We document that the majority of mainstream media coverage was negative for both candidates, but largely followed Donald Trump’s agenda: when reporting on Hillary Clinton, coverage primarily focused on the various scandals related to the Clinton Foundation and emails. When focused on Trump, major substantive issues, primarily immigration, were prominent.
We find that the structure and composition of media on the right and left are quite different. The leading media on the right and left are rooted in different traditions and journalistic practices.
On the conservative side, more attention was paid to pro-Trump, highly partisan media outlets. On the liberal side, by contrast, the center of gravity was made up largely of long-standing media organizations steeped in the traditions and practices of objective journalism.
Our data supports lines of research on polarization in American politics that focus on the asymmetric patterns between the left and the right, rather than studies that see polarization as a general historical phenomenon, driven by technology or other mechanisms that apply across the partisan divide.
Executive Summary From Harvard Study
The Key Takeaway ideas from the Harvard Study are presented after the fold. If you want to jump ahead and read the study click here to Download the PDF.
Lastly, I would like to thank the Berkman Klein Center for releasing this study under a Creative Commons License.
This article is continued after the fold. The comments for this article are are after community links and its sub thread.
Please limit community links to one per user because we want to encourage greater diversity of submissions of community links from different users. In the case of a needed fundraiser, either the member in need will write a diary or someone will do it on their behalf. That diary is then linked to in the comments. Please only post community links in community links comment provided.
The 2016 U.S. Presidential Election Key Takeaways
- Donald Trump succeeded in shaping the election agenda. Coverage of Trump overwhelmingly outperformed coverage of Clinton. Clinton’s coverage was focused on scandals, while Trump’s coverage focused on his core issues.
Figure 1B: Number of Sentences by Topic and Candidate. Click to enlarge.
Attempts by the Clinton campaign to define her campaign on competence, experience, and policy positions were drowned out by coverage of alleged improprieties associated with the Clinton Foundation and emails. Coverage of Trump associated with immigration, jobs, and trade was greater than that on his personal scandals.
Some of the key words from the topics tracked for the Clinton campaign were Foundation, Email, Benghazi, Trade, Immigration, and Jobs. For Trump it was Jobs, Immigration, Trade, Taxes, Women, University, Foundation, and Russia. In addition the data was classified if it was a scandal or issue.
- Immigration and Muslims/Islam were the two most widely covered substantive issues of the campaign.
Figure 2: Number of sentences by substantive topic and candidate from media on the open web. Click to Enlarge
Immigration emerged as the leading substantive issue of the campaign. Initially, the Trump campaign used a hard-line anti-immigration stance to distinguish Trump from the field of GOP contenders. Later, immigration was a wedge issue between the left and the right. Pro-Trump media sources supported this with sensationalist, race-centric coverage of immigration focused on crime, terrorism, fear of Muslims, and disease.
- While coverage of his candidacy was largely critical, Trump dominated media coverage.
Figure 3: Valence and focus of the 100 most linked-to stories. Stories were hand-coded for topic and tone. Click to Enlarge
- The media landscape is distinctly asymmetric.
The structure of the overall media landscape shows media systems on the left and right operate differently. The asymmetric polarization of media is evident in both open web linking and social media sharing measures. Prominent media on the left are well distributed across the center, center-left, and left. On the right, prominent media are highly partisan.
Figure 4: Partisan distribution of top 250 most-linked-to media sources by total inlinks* *“Inlinks” refers to the incoming cross-media hyperlinks to stories and media sources. Click to Enlarge
Twitter is a more partisan environment than the open web media landscape.
Figure 5: Partisan distribution of top 250 media sites by Twitter shares Click to Enlarge
Facebook is more partisan than Twitter.
Figure 6: Partisan distribution of top 250 media sites by Facebook shares. Click to Enlarge
From all of these perspectives, conservative media is more partisan and more insular than the left.
- The center-left and the far right are the principal poles of the media landscape.
The center of gravity of the overall landscape is the center-left. Partisan media sources on the left are integrated into this landscape and are of lesser importance than the major media outlets of the center-left. The center of attention and influence for conservative media is on the far right. The center-right is of minor importance and is the least represented portion of the media spectrum.
Link-based network map of election media sources. Network map based on open web media from May 1, 2015, to November 7, 2016. Click to enlarge or explore this map in higher resolution PDF.
- Conservative media disrupted.
Breitbart emerges as the nexus of conservative media. The Wall Street Journal is treated by social media users as centrist and less influential. The rising prominence of Breitbart along with relatively new outlets such as the Daily Caller marks a significant reshaping of the conservative media landscape over the past several years.
Figure 8: Network map based on Twitter media sharing from May 1, 2015, to November 7, 2016 with nodes sized by number of Twitter shares. Click to Enlarge or explore this map as a PDF in higher resolution.
- On the partisan left and right, the popularity of media sources varies significantly across the different platforms.
On the left, the Huffington Post, MSNBC, and Vox are prominent on all platforms. On the right, Breitbart, Fox News, the Daily Caller, and the New York Post are popular across platforms.
Table 1: Most popular media on the right from May 1, 2015, to November 7, 2016. Click to Enlarge
Table 2: Most popular media on the left from May 1, 2015, to November 7, 2016. Click to Enlarge
- On the most widely covered topic of the election, immigration, Breitbart was the most prominent site. On Twitter, Breitbart is far above the rest.
Figure 10: Network map based on Twitter sharing for the topic of immigration. Click to Enlarge
Breitbart’s key role in the media landscape during the election was particularly pronounced in coverage of immigration. On Twitter, Breitbart stories on immigration were shared more than twice as often as stories from the Guardian, which ranked second.
- Disinformation and propaganda are rooted in partisanship and are more prevalent on social media.
Table 3: Seven media sources that receive substantially more attention
on social media than they receive inlinks from open web media
The most obvious forms of disinformation are most prevalent on social media and in the most partisan fringes of the media landscape. Greater popularity on social media than attention from media peers is a strong
indicator of reporting that is partisan and, in some cases, dubious.
Among the set of top 100 media sources by inlinks or social media shares,1 seven sources, all from the partisan right or partisan left, receive substantially more attention on social media than links from other media outlets.
These sites do not necessarily all engage in misleading or false reporting, but they are clearly highly partisan. In this group, Gateway Pundit is in a class of its own, known for “publishing falsehoods and spreading hoaxes.
- Disproportionate popularity on Facebook is a strong indicator of highly partisan and unreliable media.
Table 4: Thirteen media sources that receive more attention from Facebook than on Twitter or through open web media inlinks
A distinct set of websites receive a disproportionate amount of attention from Facebook compared with Twitter and media inlinks. From the list of the most prominent media, 13 sites fall into this category.
Many of these sites are cited by independent sources and media reporting as progenitors of inaccurate if not blatantly false reporting. Both in form and substance, the majority of these sites are aptly described as political clickbait.
Again, this does not imply equivalence across these sites. Ending the Fed is often cited as the prototypical example of a media source that published false stories.
The Onion is an outlier in this group, in that it is explicitly satirical and ironic, rather than, as is the case with the others, engaging in highly partisan and dubious reporting without explicit irony.
- Asymmetric vulnerabilities: The right and left were subject to media manipulation in different ways.
The more insulated right-wing media ecosystem was susceptible to sustained network propaganda and disinformation, particularly misleading negative claims about Hillary Clinton. Traditional media accountability mechanisms—for example, fact-checking sites, media watchdog groups, and cross-media criticism—appear to have wielded little influence on the insular conservative media sphere. Claims aimed for “internal” consumption within the right-wing media ecosystem were more extreme, less internally coherent, and appealed more to the “paranoid style” of American politics than claims intended to affect mainstream media reporting.
The institutional commitment to impartiality of media sources at the core of attention on the left meant that hyperpartisan, unreliable sources on the left did not receive the same amplification that equivalent sites on the right did.
These same standard journalistic practices were successfully manipulated by media and activists on the right to inject anti-Clinton narratives into the mainstream media narrative. A key example is the use of the leaked Democratic National Committee’s emails and her campaign chairman John Podesta’s emails, released through WikiLeaks, and the sustained series of stories written around email-based accusations of influence peddling. Another example is the book and movie release of Clinton Cash together with the sustained campaign that followed, making the Clinton Foundation the major post-convention story. By developing plausible narratives and documentation susceptible to negative coverage, parallel to the more paranoid narrative lines intended for internal consumption within the right-wing media ecosystem, and by “working the refs,” demanding mainstream coverage of anti-Clinton stories, right-wing media played a key role in setting the agenda of mainstream, center-left media. We document these dynamics in the Clinton Foundation case study section of this report.
Our clearest and most significant observation is that the American political system has seen not a symmetrical polarization of the two sides of the political map, but rather the emergence of a discrete and relatively insular right-wing media ecosystem whose shape and communications practices differ sharply from the rest of the media ecosystem, ranging from the center-right to the left.
