US2020: ANOTHER FACEBOOK DISINFORMATION ELECTION?
U.S. flooded with over 158M views of political fake news ahead of the 2020 elections
November 5th, 2019
Download PDF version Back to Disinfo Hub
U.S. flooded with over 158M views of political fake news ahead of the 2020 elections
November 5th, 2019
Download PDF version Back to Disinfo Hub
One year out from the US 2020 elections, a new Avaaz investigation has uncovered a tsunami of political “fake news” flooding US citizens on Facebook. Politically relevant disinformation was found to have reached over 158 million estimated views, enough to reach every reported registered voter1 in the US at least once.
The research team focused on collecting the most viral Facebook posts over a ten-month period, between January 1 and October 31, 2019. Our team then analyzed the top 100 fake news stories about US politics still online on the platform, which, collectively, were posted over 2.3 million times. Avaaz only included examples that had already been fact-checked and debunked by reputable US fact-checking organizations at the time of our study.
The 2016 US elections raised public awareness of disinformation as one of the most serious threats to healthy public discourse, free and fair elections, and democratic stability worldwide. Top Facebook executives, including Mark Zuckerberg, announced immediate action to reduce the spread of harmful disinformation on the platform.2
However, our investigation shows that Facebook’s measures have largely failed to reduce the spread of viral disinformation on the platform. To the contrary, they might have even allowed it to grow: one year before Election Day, the most viral fake news about US politics were able to reach more users than what was reported from three to six months preceding the 2016 elections.3
Without immediate and decisive action, we expect the US 2020 elections to be, once again, severely impacted by disinformation.
One year ahead of the US 2020 elections, the top 100 fact-checked political fake news stories have been posted over 2.3 million times and reached an estimated 158.9 million views and 8.9 million interactions (likes, comments, and shares).
To put this into perspective: Avaaz found enough disinformation to reach each of the 153 million registered voters in the 2018 midterm elections4 once.
The number of views of the 100 fakes across the platform is more than twice the 59.7 million estimated views of the official Facebook pages of the Republican Party (GOP) and the Democratic Party (Democrats)5 in 2019.
The number of views and interactions for the most viral fake news stories spiked in the last 3 months of our research, coinciding with the increased pace of the US 2020 presidential race (August 1 - October 31, 2019):
86 million estimated views of disinformation in the last 3 months, which is more than 3 times as many as during the preceding 3 months (27 million).
When adapting our methodology to only consider the top 20 fake news stories,6 the number of interactions on fake news in the last 3 months - 4.6 million - appears to be 1.5 times more than reported in the period 3 to 6 months before the 2016 elections by the media7 - 3.1 million-.
Negative fake news stories account for 91% of our sample:
62% of the disinformation in our sample were against Democrats/Liberals, reaching over 104 million views out of the 158 million total views.
29% of the disinformation in our sample were against Republicans/ Conservatives, reaching 49 million views.
Only 9% of the disinformation stories in our sample were “positive”. All of them were pro-Republicans/Conservatives, reaching 5 million views.
Most of the fake news sources were individuals (39%) or non-official political pages (35%). Alternative media accounts for 19% of the fakes, politicians for 6%, and mainstream media for 1%.
Some of the most shocking fake news stories analyzed in this report include:
The false claim that Trump’s grandfather was a pimp and tax evader and his father was a member of the KKK (over 29 million estimated views)
The false claim that Nancy Pelosi diverted billions of dollars to cover the costs of the impeachment process (over 24 million estimated views)
The false claim that Democratic Representative Ilhan Oman attended an Al- Qaeda training camp (over 77k estimated views)
The false claim that a Trump supporter was mocking a dead migrant child (over 99k estimated views)
The false claim that Joe Biden called Trump supporters the “Dregs of Society” (over 4 million estimated views)
These findings can only be seen as the tip of the iceberg of disinformation ahead of the US 2020 elections. Avaaz only analyzed the top 100 political fake news on Facebook, that were fact-checked and debunked by reputable US fact-checking organizations. Further analysis into the spread of disinformation on the platform, as well as on YouTube, Twitter, Instagram and WhatsApp should be carried out immediately to reveal more of what lies below the surface of this tidal wave approaching the horizon of the US 2020 elections.
However, the US 2020 elections and other future elections around the world can still be saved from disinformation. To do so, Avaaz is calling on Facebook and all other social media platforms to immediately implement “Correct the Record”, which means working with independent fact- checkers to ensure that EVERY user who has seen or interacted with false information is notified and offered a correction. This solution is proven to work and would tackle disinformation while preserving freedom of expression, as “Correct the Record” provides transparency and facts without deleting any content.
Avaaz’s investigation selected the 100 fake news stories about US politics that had the biggest amount of interactions7 and had already been debunked by major US fact-checking agencies.
Those stories were shared over 2.3 million times on Facebook, and mainly targeted political figures and parties across the political spectrum. Between January 1 and October 31, 2019, they reached over 158.9 million estimated views.
All of the disinformation pieces analyzed in this report have been fact-checked by independent US fact-checking organizations including Snopes, Politifact, FactCheck.org, and Lead Stories.
To understand the scale of the reach of disinformation detailed in this report, in Image 1 we compare the estimated 158,908,992 views of disinformation we found with
153,066,000 registered voters for the 2018 midterms8, meaning we found enough disinformation to reach every registered voter at least once.
59,717,750 estimated views in 20199 for the official Facebook page of the Republican Party (GOP) (2,162,452 likes) and Democratic Party page (Democratic Party) (1,595,689 likes). The disinformation content we found therefore reached more than twice the total number of views of these two pages combined in 20199.
In Image 2, we analyze how much of this disinformation is targeting Democrats/Liberals or Republicans/Conservatives:
62% of political fake news stories in our sample were against Democrats/Liberals, reaching over 104 million views out of the 158 million total views.
29% of political fake news stories in our sample were against Republicans/Conservatives, reaching 49 million views.
9% of political fake news stories were pro Republicans/ Conservatives, reaching 5 million views.
None of political fake news stories were pro Democrats/ Liberals.
Image 3 shows every fake news item in our database as a circle, labeling in blue those targeting the Democrats, those targeting the Republicans in red, and those that are pro-Republican in orange. The estimated views, the larger the size of the circle.
Image 4 shows the main targets of disinformation. Each circle represents the person or group targeted in the post. Circle sizes correspond to the number of times they were targeted. Blue circles are against Democrats/Liberals, red ones against Republicans/Conservatives, orange ones are Pro-Republicans/Conservatives. We found no fake news positively targeting Democratic targets.
In Image 5, we analyze the sources of political disinformation. We identified 5 main categories: political pages (which includes non official partisan pages and activists pages with a clear political position), politicians (when the source was a person affiliated to a political party or an official party page), alternative media, mainstream media and individuals when the source were personal profiles. Individuals and non-official political pages are the biggest sources for sharing or creating the viral fake news in our database.
In Image 6, we break down the sources of disinformation according to the corresponding political targets. While fake news against Democrats/Liberals appear to be mainly driven by political pages (37%), those against Republicans/Conservatives are mainly posted by individuals (62%)
In Image 7, in order to be able to compare our findings with a similar study by Buzzfeed about the 2016 elections10, we apply the same methodology restricting our analysis to the 20 fake news items with the biggest amount of interactions between February and April, May and July, and August and October. This not only shows a spike of disinformation in the last three months, but also that the number of fake news interactions in this timeframe - 4.6 million - appears to be 1.5 times more than Buzzfeed’s analysis identified in the period 3 to 6 months before the 2016 elections11 - 3.1 million. In 2016, Buzzfeed also reported a further spike in the last 3 months before Election Day, with 8.7 million interactions, which highlights the threat we are facing ahead of 2020 elections.
The table below shows the 20 most-viewed fake news stories for the investigation period. The full list of the 100 most viewed fake news can be found in the Annex.
We have provided the post link of the top sharer of each story so it can be viewed in context. For clarity, the “estimated views” and interactions are based on the cumulative CrowdTangle data for each post across the pages, groups and profiles featuring the post on Facebook. The estimates do not represent the views or interactions solely for the account associated with the link provided for each post.
To see all table data, please, scroll to the right.
This chapter presents a selection of some of both the most shared and the most absurd fake news from this investigation - it also shows the breadth of disinformation analyzed in this report. Each item has a summary of the fake news story, its reach, and a short summary from a fact-checking organization debunking the events depicted in the posts.
Disinformation: This post was shared by “ImpeachTrump” on February 18, 2019. The article13, written by the American Herald Tribune in December 2015, asserts that Trump’s grandfather was a pimp and a tax evader, and that his father was a KKK member.
Reach: 29,202,553 estimated views.
Fact check: rated FAKE NEWS by Lead Stories: “Donald’s Trump’s paternal grandfather Frederick Trump accumulated his initial wealth by owning restaurants and hotels in Seattle and the Yukon. [...] Trump’s father Fred Trump was detained at a KKK protest in the Queens borough of New York City, but he was released without charges, according to a New York Times clipping from June 1927. The report did not say he was a member of the KKK and there is no evidence otherwise that he was in the group. As for tax evasion by the grandfather, the article in question offers no details or sourcing.”
Disinformation: The post, shared by an individual in a public group16, claims that Pelosi stole billions of dollars to cover the costs of Donald’s Trump’s impeachment process. The post shared a link to an article from potatriotsunite.com17.
Reach: 24,606,644 estimated views.
Fact check: rated PANTS ON FIRE by Politifact: “No, Nancy Pelosi didn’t divert Social Security money for the impeachment inquiry. House Speaker Nancy Pelosi recently drew the ire of some Facebook users as they reacted to a headline claiming she was tapping Social Security funds for the impeachment inquiry. We searched for real news coverage of Pelosi diverting Social Security money for the impeachment inquiry, but none exists.”
Disinformation: The post, which is a screenshot from a tweet, claims that the woman in the picture checking her gun is the Somalia-born Congresswoman Ilhan Omar. It reads: “Here is your precious Muslim congresswoman at a Al’ Qaida training camp in Somalia” in an attempt to to connect Omar to the terrorist group.
Reach: 77,313 estimated views.
Fact check: rated PANTS ON FIRE by Politifact: “The photo, while authentic, was taken in 1978, about three years before Omar was born in 1981. The Associated Press image that appears in the Facebook post was taken at a military training campus in the city on Feb. 25, 1978.”
The Avaaz team is aware this post has a small label incorporating “Satire”, but it is in a font size so small that it barely registers. This and the two other similar examples have been deliberately included on this report because the clear danger they present to national democractic debates. The fact that all three examples went viral and major fact checkers were compelled to debunk them as “false”, clearly indicates that once viral on social media, so-called “satire” can easily be interpreted as real news. The use of “satire” 13 can yet be another tactic of disinformation, as it has been widely reported by BBC, Poynter, and Punditfact. http://archive.fo/O5eOv
Disinformation: On July 5 2019, a Facebook user posted a photo claiming a Trump supporter was mocking the death by drowning of two migrants.
Reach: 99,204 estimated views.
Fact check: rated FALSE by Politifact: “This Facebook post claims that the woman in the picture is a Trump supporter “mocking a dead child.” We don’t know the woman’s political affiliation. But it’s clear from reporting and her own words that she is not mocking a dead migrant child. Rather, she talks about how she’s saddened by what’s happening at the border, and how we should celebrate the United States as a country of immigrants.”
Disinformation: The post by Life News article21 claims that Biden called Trump supporters the “dregs of society” during a speech held in September 2018.
Reach: 4,495,064 estimated views.
Fact check: rated FAKE NEWS by Lead Stories: “Did Joe Biden call Trump supporters the “dregs of society”? No, that’s not true: The former vice president referred to “forces of intolerance” who “remain determined to undermine and roll back the progress” as the “dregs of society” in a speech to the annual Human Rights Campaign dinner in Washington, D.C., in September 2018. But Biden said they were “a small percentage of the American people,” and he said nothing to indicate he was talking about a large block of Donald Trump supporters.”
Avaaz’s team of data scientists and researchers used a combination of Facebook’s public data, analytics tools like CrowdTangle, independent US fact-checking organizations, and statistical modelling to arrive at what we believe provides a credible, though conservative, picture of the reach and impact of the most shared and factchecked political fake news stories in the US one year ahead of the US 2020 elections
Time frame: We focused our Facebook research on content that was still being shared by, or available on public pages, groups, or profiles, from January 1 to October 31, 2019.
Scope: Research, collection, and analysis of the top 100 political fake news items based on the number of interactions, fact-checked by reputable US fact-checking organizations, stories focused on political candidates and parties across the political spectrum in the United States. The team examined Facebook posts shared by pages, groups, and profiles, containing text, images, videos, or links to external web pages.
Fake news sources and selection criteria definition: To identify the fake news for our analysis, we collected posts related to US politics that contained “verifiably false or misleading information”, as assessed by reputable factchecking organizations, “with the potential to cause public harm, for example by undermining democracy or public health or encouraging discrimination or hate speech.” The US factchecking organizations included: Snopes, Politifact, FactCheck.org, Lead Stories and Truth or Fiction.
Assessment of disinformation and fake news included in this study: Independent US fact-checking organizations provide differing levels of assessment and differing labels22. In our study, we included the 100 top fake news stories which were assessed as false, mostly false, or had similar ratings to this effect. We only included “mostly false” stories if they met our definition above, for example the exaggerated claims that connected Hillary Clinton to Russian funding and uranium deals, and the claims that the whole Trump family has been ‘disallowed’ from operating charities in New York. These types of posts are based on shreds of truth, with so much exaggeration and false elaboration thrown in that both the headlines and the posts’ content are deeply misleading. We also included content assessed as “miscaptioned” or “misquoted” where the miscaptioning or misquoting would mislead - such as the infamous viral picture23 of a young man drunkenly peeing his pants, falsely identified in the caption as Beto O’Rourke. If content has received an “unproven”24 label by one fact-checking organization, we have examined the full context of the post, and only included these where they met the selection criteria defined above25.
Viewership calculation: Facebook discloses the number of views for videos, but for posts containing only text and image content the platform displays only the number of shares, likes and comments. Therefore, in order to estimate viewership for text and image content, we designed a metric based on the publicly available statistics of the Facebook pages creating or sharing the disinformation pieces in our report. For each page, we took into account the total number of owned and shared video views between January 1 and October 31, 2019, and then divided it by the total number of owned and shared video interactions.
Facebook reports a “video view” after only three seconds, while an image or text can be considered as “viewed” and having an actual impact in under three seconds. Therefore, the estimation of total views in this study is probably lower than the content’s actual total viewership.
Step 1: Views/interaction ratio calculation: We computed a view/interaction ratio for all the pages that shared the selected disinformation stories and had published at least 10 videos (both owned and shared) between January 1 and October 31, 2019. The ratio was computed by dividing the total video views by total video interactions. After computing all the individual ratios, we then computed a global views/interaction ratio of 17.83 through a weighted average of all the pages’s ratio, where the weight was proportional to the interactions gathered by the fake news published by that page.
Step 2: Total views calculation: For each piece of disinformation, we multiplied our views/ interaction ratio of 17.83 per the number of interactions for that posts or web links provided by CrowdTangle. Adding all those estimated views together we obtain the final estimate of 158,908,992 views.
The disinformation uncovered in this report, as well as the tidal wave of fake news undoubtedly still to come, will wreak havoc on the electoral process much like in 2016. The vast majority of the millions of US citizens exposed to it will never know that they have seen fake news, or see factual corrections.
But the solution to this problem is simple: Platforms must themselves work with factcheckers to “Correct the Record” by distributing independent third party corrections to EVERY SINGLE PERSON who saw the false information in the first place. Newspapers publish corrections on their own pages, television stations on their own airwaves; platforms should do the same on their own channels. No one else can do it. This solution would tackle disinformation while preserving freedom of expression, as Correct the Record only adds factually corrected information, and does not require the platforms to delete any content.
Facebook also claims to slow down the dissemination of fake news once fact-checked26, but this report clearly shows that this is not sufficient to prevent lies from going viral. Correct the Record can fill this gap by going back to the users who have seen the false information and offering them factually correct information on the issue concerned.
Avaaz believes that Correct the Record is likely the most powerful action online platforms can take ahead of the upcoming US elections to restore the public’s trust and keep the tidal wave of disinformation at bay. TIME Magazine has called it a “radical new proposal [that] could curb fake news on social media.”
Multiple peer-reviewed studies prove that effective corrections can reduce and even eliminate the effects of disinformation. Studies attempting to replicate the often discussed “backfire effect” - where corrections further entrenched false beliefs - have instead found the opposite to be true. Meanwhile, researchers are converging best practices for effective corrections.
1- Define: The obligation to correct the record would be triggered where:
Independent fact checkers verify that content is false or misleading;
A significant number of people - e.g., 10,000 - viewed the content.
2 - Detect: Platforms must:
Proactively use technology such as AI to detect potential disinformation with significant reach that could be flagged for fact-checkers;
Deploy an accessible and prominent mechanism for users to report disinformation;
Provide independent fact checkers with access to content that has reached a wide audience - e.g., 10,000 or more people.
3 - Verify: Platforms must work with independent, third-party verified fact-checkers to determine whether reported content is disinformation.
4 - Alert: Each user exposed to verified disinformation should be notified using the platform’s most visible and effective notification standard.
5 - Correct: Each user exposed to disinformation should receive a correction from the fact-checkers that is of at least equal prominence to the original content, and that follows best practices.
Online platforms should also, within a reasonable time after receiving a report from an independent fact-checker regarding disinformation with significant reach, take steps to curtail its further spread.
Avaaz has deployed anti-disinformation teams in Europe, Brazil, India (the Assam region), and Canada with the goal of detecting and making public disinformation on social networks and the internet. Our work on disinformation is rooted in the firm belief that fake news proliferating on social media is a threat to democracies, the health and well-being of communities, and the security of vulnerable people. In the context of the upcoming 2020 elections, the Avaaz anti-disinformation team is now also monitoring disinformation in the US, and these are the first findings from those investigative efforts. Avaaz reports openly on what it finds, so it can alert and educate social media platforms, regulators, and the public, and so it can more effectively advocate for smart solutions to defend the integrity of elections and our democracies from disinformation.
Note: If any mistakes were found after reading this report, please do let us know by getting in touch with our media team:
Andrew Legon: +44 742 986 5292 / andrew.legon@avaaz.org and
Oscar Soria: +1 415 217 9392 / oscar@avaaz.org
Endnotes
Tell Your Friends