Incorrect citations of my research in MSM

Filed to: corrections needed

I recently found some of my research cited incorrectly in two different articles. Neither outlet contacted me at the time to verify if what they were publishing was accurate and both misinterpreted my analysis. One article has been changed (the reference to my research was deleted but no correction was made) and the second article has not been updated at all, despite assurance from an editor that it was corrected.

In October 2017, I worked with Charlie Warzel from BuzzFeed to publish several of my graphs that included the notorious Russian Twitter troll, Ten_GOP.

I had data that showed the Ten_GOP account had tweeted during four divisive events. I did not have evidence to claim how much influence the account wielded and Charlie included several important caveats that put the data I analyzed into perspective:

Source: BuzzFeed

Two news outlets, The Daily Beast and Think Progress, cited the BuzzFeed article incorrectly and misinterpreted my data. Both articles cited my research regarding Ten_GOP’s activity during Unite the Right and wildly exaggerated what my research represented. The goal of each article seemed to be to link Russian Twitter activity with Unite the Right and the tragic murder of Heather Heyer by a white nationalist.

The problem is Ten_GOP was not very active during Unite the Right. The most influential accounts before, during and after Unite the Right were white nationalists. I did an entirely separate analysis, a collaboration with Ed Summers, on Unite the Right which was not cited by either (or any) news outlet.

Statements that my analysis showed a Russian troll was more influential than white nationalists are categorically wrong. Blaming a Russian troll for inciting violence at Unite the Right or exaggerating its influence shifts the blame from white nationalists to Russia — something that I believe is not only careless reporting but also dangerous.

In an article dated October 26, 2017 titled “Russian Twitter Bot Targeted Maxine Waters, a Prominent Trump Impeachment Backer” The Daily Beast cites my analysis in the following paragraph:

Source: The Daily Beast

There are two problems with this paragraph:

  1. BuzzFeed did not do the analysis, I did it.
  2. My analysis did not indicate that Ten_GOP was “one of the most influential Twitter accounts pushing the ‘Unite the Right’ white nationalist rally”

I went back to my Unite the Right dataset to double-check what Ten_GOP looked like in relation to the rest of the UTR network:

User-to-user network of around 23,000 #UniteTheRight tweets from August 12 to August 14 with Ten_GOP highlighted

Ten_GOP tweeted one apolitical tweet and retweeted two tweets — one tweet from polNewsForever and one BakedAlaska tweet — which is why it is near those two nodes in the graph.

This statement in Daily Beast’s article: “Ten_GOP’s account was one of the most influential Twitter accounts pushing the Unite the Right” is false. It was in no way “one of the most influential Twitter accounts” and it did not “push” Unite the Right.

Here is my analysis of #UniteTheRight that showed white nationalists were the most influential and pushed the event in the week leading up to August 12 using automation:

So my work was cited incorrectly and out of context in an article that had nothing to do with the Unite the Right rally.

I contacted an editor at The Daily Beast who responded to me on February 11 and said they fixed the attribution (they did not) and suggested the following text to replace what’s in the article right now:

“According to extensive data analysis by Erin Gallagher posted on BuzzFeed, @Ten_GOP’s account appeared to be one of several influential accounts tweeting during the “Unite the Right” white nationalist rally in Charlottesville, Virginia. While one cannot pinpoint their position on the rally, it did raise the event’s visibility. A white nationalist ran over several protesters at the event, injuring 19 people and killing 32-year-old Heather Heyer.”

I replied on February 11 and corrected the above text because (1.) we can pinpoint exactly what Ten_GOP tweeted during the rally and (2.) my analysis does not support the statement that the account raised the visibility of Unite The Right.

I have not heard back from The Daily Beast since February 11 and as of the time of publishing this post, the article has not been corrected.

The Think Progress article that cited my analysis was dated August 13, 2018 and titled “New database shows how Russian tweets inflamed tensions around Charlottesville.” So at least this article was about Unite the Right and therefore made some sense to include my analysis.

However the article cited my analysis as “prior evidence” that Ten_GOP “hyped the violence” and my analysis did not indicate anything of the sort.

Here’s the paragraph where my research was originally mentioned:

Source: Think Progress via Wayback Machine

Once again my analysis was attributed to “BuzzFeed” instead of to me. Besides the misattribution, the article used my analysis to add weight a claim made by Rep. Tom Garrett (R-VA) who said he learned from the FBI director that “Russian meddling efforts played a role in ‘fomenting the flames of what happened in Charlottesville.’”

Rep Garrett is citing classified information which can’t be confirmed or falsified, so my research is brought in without context to back up his claim in this paragraph. My analysis that was published in BuzzFeed does not support such a claim.

I contacted the author of the Think Progress article, Casey Michel and to his credit, he responded within 24 hours and made the following update at the end of the article:

Source: Think Progress

The Think Progress article now links to a separate analysis from BuzzFeed by Peter Aldhous to add weight to Rep. Tom Garrett’s claim about Russian social media activity hyping the violence in Charlottesville.

Peter Aldhous’s article titled “Russian Trolls Swarmed The Charlottesville March — Then Twitter Cracked Down” relies on a dataset from researchers at Clemson University that was published in FiveThirtyEight in July 2018. It claims “August 2017 was the peak time for pro-Trump Russian Twitter trolls linked to the shadowy Internet Research Agency in St. Petersburg.”

But in early 2019, Twitter revised its list of IRA accounts and said that accounts they had previously attributed to the Russian IRA were actually from a different trolling group in Venezuela.

Bloomburg published a correction to their own reporting about the Clemson dataset and updated an older article with the changes after Twitter revised its IRA list.

“Twitter Inc. this month made a significant revision to its public database of more than 3,000 accounts that it has linked to Russia’s Internet Research Agency, fundamentally altering the record about the group’s trolling efforts in the U.S. in the year following the contentious 2016 presidential election.

Twitter’s changes invalidate central portions of a Bloomberg News article published last August that had analyzed the Russian troll farm’s activity in 2017, according to researchers at Clemson University who have compiled and published a database of the IRA’s Twitter activity.” — Bloomburg

I asked Peter Aldhous if his article was also affected by Twitter’s revision but he said he was on vacation and never wrote me back. Since I’m still trying to get citations to my own research corrected, I haven’t made further efforts to pursue possible corrections to someone else’s research. 🤷🏻

I learned that I had been cited incorrectly in MSM by accident. I stumbled across these two articles while doing a review of articles published about Russian bots and trolls to see which researchers were cited most often. If I hadn’t been curious about that, I probably would have never known that my research had been misrepresented.

In total, I manually reviewed 144 news articles from March 2017 to February 2019 from mainstream media outlets ranging from the New York Times, Washington Post, the BBC, ABC, NBC, CNN and many others. I probably missed a few stories but in my sample of 144 articles, some trends emerged.

2 of the 3 times I was cited were the errors I mentioned in this blog

The three most often cited researchers in Russian bot articles in the sample I checked were Hamilton 68, New Knowledge and DFRLab. Each of these sources is problematic. Here are some articles to learn more about them.

Hamilton 68 has never revealed which accounts it was tracking or how it determined they were “Russian influence” therefore their claims can’t be confirmed or falsified. New Knowledge is a for-profit company that attracts business any time they are cited in the media and the company is now associated with a disinformation operation targeting American voters, a clear breach in ethics for any research team. The funding behind DFRLab, which is part of DC think tank The Atlantic Council, should be more than enough evidence to consider that possible bias exists in their analyses.

I’m concerned that reporting on Russian social media influence has been exaggerated. I’m even more concerned that social media influence and violence in meatspace that should be attributed to white nationalists has been misattributed to Russian trolls. It’s very dangerous to downplay the influence of white nationalists online. Judging by my personal experience trying to get reporting that wrongly cited my own analysis fixed, I don’t have much confidence that any previous reporting on this topic will be reviewed or corrected.

Social media researcher, multimedia artist