Twitter Censorship Paper Review
Americans overwhelmingly believe, and have reported for years, that social media sites are politically biased.(1) As is the case with other perceptions of media bias, these concerns often reflect direct personal observations and experiences, and are not lightly dismissed.
Nonetheless, a non-peer reviewed “junk science” paper by Paul M. Barrett and J. Grant Sims entitled “False Accusation: The Unfounded Claim that Social Media Companies Censor Conservatives” attempted to do just that, claiming that there is no censorship of conservatives on social media.(2)
Undisclosed Conflicts of Interest
False Accusation was funded by a Biden mega-donor, Craig Newmark.(3) While the Barrett and Sims cited his support, they failed to acknowledge the donor's political connections, concealing the conflict of interest from the vast majority of readers. Just as pharmaceutical research funded by a drug company would be regarded as less credible, the public has a right to have real or potential conflicts of interest disclosed in clear language.
Failure to Answer Research Question
The paper offers no solid information regarding the number of individuals with various political viewpoints banned or censored on social media. Instead, it cites "engagement data" which is largely irrelevant to the topic of censorship, and provides long-running opinion and commentary. Various anecdotes are discussed, but no systematic data on censorship by political affiliation are provided which would answer the central question: are certain ideological viewpoints censored on social media more frequently than others?
The piece acknowledges: "Conservatives do get suspended or banned...Pinning down precise proportions is impossible because Twitter doesn’t release sufficient data." Any fair-minded scholar at that point would note that these limitations do not permit the question to be definitively answered, and call for the release of relevant data. We simply don't know what the number and proportion of conservatives and liberals censored by social media companies is, as the companies themselves refuse to provide this information.
Arguments from Ignorance
Undeterred, Barrett and Sims proceed to employ vast inductive leaps, claiming that there is "no evidence" of censorship bias against conservatives, and then sweepingly concluding that claims of bias represent a "false accusation."
In the absence of specific data from Twitter regarding the political viewpoints and alleged offenses of who it bans, their conclusion boils down to a logical fallacy called an argument from ignorance: Twitter refuses to release data on who it bans, therefore, there is no bias.
The argument from ignorance exists in both a positive ("There is no evidence that p is true. Therefore p is false,") and a negative form ("There is no evidence that p is false. Therefore, p is true.")(4) Since "Twitter doesn't release sufficient data," the rules of logic do not permit claims that:
"there is no evidence that conservatives are disproportionately censored, therefore, conservatives are not disproportionately censored,"
"there is no evidence that conservatives are not disproportionately censored, therefore, conservatives are disproportionately censored."
The first is a positive and the second is a negative "argument from ignorance." Both are invalid. In making false claims not supported by evidence, Barrett and Sims have done precisely what they accuse their political opponents of doing. Extrapolating personal biases into the void of information gaps is not scholarship.
Astonishingly, Barrett and Sims do not call upon Twitter to release the relevant data, nor for reforms mandating transparency and fairness. They are perfectly content to present arguments from ignorance in an informational void, and suggest no need or interest regarding the missing data.
The implications of "arguments from ignorance" for obfuscating arbitrary conduct and supporting authoritarianism are disturbing. Barrett and Sims' same logic would argue that disappearances of missing persons, extrajudicial killings in Venezuela, or death camps in Stalin's USSR, represent "disinformation" and "false accusations" of the political opposition because the regime did not release official figures. Credible voices would raise awareness, demand transparency, and call for additional information.
Propaganda, not Scholarship
Barrett and Sim's False Accusation is a partisan polemic which does not meet basic scholarly standards. It is not peer-reviewed or published in an academic journal. It lacks formal methodology. Besides a single reference to a Pew Research Center essay, no scholarly articles are cited in the references, many of which are drawn from left-leaning news and opinion sites like Vox and BuzzFeed. The authors repeatedly attempt to pass off partisan narratives and far-left conspiracy theories as fact. A lengthy section alleges a "coordinated Republican disinformation drive" since 2019. This and other claims are not supported by the evidence presented. Conservative sites are vociferously criticized but no articles discussing censorship from these sites are referenced. They are dismissed with polemical charges and their data and arguments are ignored.
The paper's lack of methodology, extreme partisan bias, and pervasive logical fallacies would be rejected by any credible peer-review process. Limited feedback from like-minded fellow-travelers cited in the piece's acknowledgments is in no way to be conflated with the actual impartial, blinded peer review process conducted by credible scholarly publications.
"False Accusation" is neither a true "research study" nor even-handed scholarship. It. is propaganda funded by a political donor which strings together limited data and flawed analysis to push a partisan narrative. Scholarly review would reveal the papers' authors to be the ones making false, unsubstantiated accusations. The authors, apparently aware of this, merely posted their "research" on their institutional website with no attempt to pursue publication through the peer review process.
Pitched to the Press, not Scholars
Pitching ostensible "research" to the press rather than the scholarly community has been widely recognized as a hallmark of junk science. Notwithstanding undisclosed conflicts of interest, no formal methodology, basic logical fallacies, and no formal peer review, False Accusation was uncritically promoted by left-leaning media.(5) The study was subsequently cited by Twitter as scholarly "proof" of its lack of bias, even as the company refused to release relevant data.(6)
Forbes offers a more honest appraisal, noting that "there's no solid evidence" for censorship, while also acknowledging that the question remained open. Forbes reported that some claims of bias and "shadow banning" have been overblown, while also citing outside data and perspectives.(7) However, the paper's deeper flaws went unreported. Coverage of this non-peer-reviewed piece as if it represented serious scholarship requires explanation from any reputable news organization. Outlets which cite the paper's findings uncritically discredit themselves.
Not content with breaches of scholarly standards and basic logic, Barrett evoked an additional conflict of interest by reporting on his own research for the New York Daily News.(8) Unchallenged by the critical scrutiny of ethical journalism, Barrett's piece is replete with inflammatory rhetoric and unsubstantiated claims attacking allegations of censorship by conservatives as "fake news" and "disinformation." Such conduct lays bare the authors' status as partisan activists and not dispassionate scholars following the facts.
Dignifying Barrett and Sims' polemic as a "study" overstates its rigor and merit. Yet Barrett is effuse in self-praise of his own work, describing it in the piece's title as "a careful review of evidence." Even fair-minded scholars experience difficulty in accurately appraising their own work. These concerns are amplified in the setting of partisan funding, conflicts of interest and ideological agenda.
Barrett and Sims' claims are unsubstantiated by the "research" they point to, and thus false by fact-checking criteria. Such misdirection illuminates the gaslighting employed by contemporary authoritarians, who obstruct or withhold access to potentially inconvenient data. Ideological allies invoke "arguments from ignorance" to claim that the problem is non-existent or even a fraudulent contrivance of political enemies, who are then shouted down as spreading "disinformation" to be censored or suppressed. Partisan "true believers" attempt to elevate dubious political narratives to the status of "fact" through "fact-checking fraud" and junk "research" conducted by biased partisans. In this manner, they seek to control the narrative and suppress dissent without having to disclose or honestly engage the data. Such conduct is anti-democratic and anti-intellectual.
The extent to which Twitter censorship may or may not reflect ideological bias cannot be empirically answered with available data. The question remains open. Concerns raised by individuals on the basis of personal experience and observation remain important and relevant. Aggregate data are needed to provide a broader picture. Further investigation is warranted.
While we do not have detailed empirical data to assess whether the authors' assertion that allegations that "big tech...censors conservatives...is fake news," we can conclude from the lack of methodological rigor, logical fallacies, and pervasive partisan bias that “False Accusation: The Unfounded Claim that Social Media Companies Censor Conservatives” is fake research. Its authors spread disinformation under the pretense of countering it.
1. Jukam, Kelsey. “Most Americans Say Social Media Sites Are Biased.” Courthouse News Service, June 28, 2018. https://www.courthousenews.com/most-americans-say-social-media-sites-are-biased/
2. Barrett, Paul M. and J. Grant Sims. “False Accusation:The Unfounded Claim that Social Media Companies Censor Conservatives.” New York University, February 2021. https://static1.squarespace.com/static/5b6df958f8370af3217d4178/t/60187b5f45762e708708c8e9/1612217185240/NYU+False+Accusation_2.pdf
3. Flood, Brian. “Study dismissing conservative concerns about Big Tech bias was funded by far-left Biden mega-donor.” Fox News, February 2, 2021. https://www.foxnews.com/media/study-dismissing-conservative-concerns-big-tech-bias-funded-joe-biden-mega-donor
4. Cummings, Louise. (2015) Argument from Ignorance. In: Reasoning and Public Health: New Ways of Coping with Uncertainty. Springer, Cham. https://doi.org/10.1007/978-3-319-15013-0_3
5. Guynn, Jessica. "Do Facebook, Twitter and YouTube censor conservatives? Claims 'not supported by the facts,' new research says." USA Today, February 1, 2021. https://www.usatoday.com/story/tech/2021/02/01/censorship-conservatives-trump-facebook-twitter-youtube/4316155001/
6. Carlson, Tucker. “Tucker Carlson: NYU's farce of a study on Big Tech censorship of conservatives.” Fox News, February 2, 2021. https://www.foxnews.com/opinion/tucker-carlson-big-tech-censorship-nyu-study-craig-newmark
7. Durkee, Alison. "Are Social Media Companies Biased Against Conservatives? There’s No Solid Evidence, Report Concludes." Forbes, February 1, 2021. https://www.forbes.com/sites/alisondurkee/2021/02/01/are-social-media-companies-biased-against-conservatives-theres-no-solid-evidence-report-concludes/
8. Barrett, Paul M. "Big tech doesn’t censor conservatives: A careful review of evidence shows that’s fake news." New York Daily News, February 2, 2021. https://www.nydailynews.com/opinion/ny-oped-no-big-tech-doesnt-censor-conservatives-20210202-mwro7qspxrapjatqcevf6lrroa-story.html