One of the reasons Tesla tycoon Elon Musk says he wants to take over Twitter is that the social media platform censors conservatives.
“I’m worried about de facto bias in ‘the Twitter algorithm’ having a major effect on public discourse,” he tweeted not long before launching his hostile $43 billion bid for the company.
Leading conservatives, who for years have accused Twitter and other social media platforms like Facebook and YouTube of censoring right-leaning views, have applauded Musk’s bid. “Let’s hope this goes through and we have free speech on Twitter again,” Rep. Jim Jordan (R-Ohio) said in a Fox Business interview that he posted on Twitter.
Accepted as gospel on the right, the notion that conservatives are marginalized or “canceled” by social media companies has always had an odd ring, given that conservative pundits and politicians (like Jordan) are notably prolific and enjoy high levels of engagement on the major platforms. Last year, the NYU Stern Center for Business and Human Rights published a paper I wrote that observed the absence of any trustworthy studies finding that platforms were removing conservative content for ideological reasons.
Even anecdotal evidence of supposed bias tends to evaporate under close examination. Yes, Facebook, Twitter and YouTube barred former President Donald Trump after the Jan. 6, 2021, U.S. Capitol insurrection. But while unprecedented, this response was justified by Trump’s repeated violation of platform rules against baselessly undermining election results and inciting violence.
Now, researchers from the Massachusetts Institute of Technology, Yale and the University of Exeter have released the first study that I’m aware of that takes an independent, empirical look at the bias claim. Twitter, they found, does tend to suspend Republican users far more frequently than Democrats. But that’s not necessarily because of partisan favoritism. Instead, it’s just as likely that Republicans were suspended because they tend to spread far more misinformation.
The researchers identified 9,000 politically engaged Twitter users in October 2020, half Democratic and half Republican. They tracked the sample for seven months after the November 2020 election. During that period, Twitter suspended 7.7 percent of the Democrats. In contrast, the platform suspended 35.6 percent of the Republicans — a more-than-fourfold difference.
Evidence that Twitter tilts against the right? Not so fast, the researchers said.
The study team scrutinized their subjects’ appetite for sharing misinformation, meaning provably false content such as phony COVID-19 cures or QAnon conspiracy theories. The researchers found that the Republican Twitter users shared “substantially more news from misinformation sites.” (Misinformation sites had been labeled as untrustworthy by either professional fact-checking organizations or surveys of politically balanced and demographically representative laypeople.)
“Critically,” the researchers said, “we found that users’ misinformation sharing was as predictive of suspension as was their political orientation. Thus, the observation that Republicans were more likely to be suspended than Democrats provides no support for the claim that Twitter showed political bias in its suspension practices. Instead, the observed asymmetry could be explained entirely by the tendency of Republicans to share more misinformation.”
In addition to tracking Twitter suspensions, the researchers surveyed 4,900 Americans on their attitude toward misinformation, finding “strong bipartisan support” for social media platforms taking action against provably false content. Nevertheless, on Twitter, Republicans are more prone to amplify misinformation. “As a result,” the researchers said, “our study shows that it is inappropriate to make inferences about political bias from asymmetries in suspension rates.”
Asymmetry is a key word when thinking about misleading and polarizing content online. Previous studies have determined that conservatives tend to share dramatically more content from low-quality “fake news” sites than liberals and were more prone to visit such sites. Research has also shown that Republican Twitter users are exposed to more misinformation from their political leaders than Democrats.
However Musk’s run at Twitter gets resolved — the company’s board of directors is deploying a “poison pill” defense to try to thwart him — the question of social media bias will remain front and center as the November midterms approach and beyond them, the 2024 presidential election. Republicans are repeating the unfounded claim in an attempt to taint not just the platforms, but Democrats as well.
“Big Tech has censored posts of mine about biological sex, COVID, Hunter Biden, and even about kitchen table issues like energy prices,” Rep. Jim Banks (R-Ind.), chair of the Republican Study Committee, told Axios. “Democrats’ electoral strategy now relies on censorship.”
The newly published study ought to blunt this line of attack. But in an online environment where proclivities for misinformation are asymmetrical, the bias claim is unlikely to fade anytime soon.
Paul M. Barrett, the senior research scholar and deputy director of the NYU Stern Center for Business and Human Rights, writes about social media’s effects on democracy. Find him on Twitter @authorpmbarrett.