Exposure to Twitter lowered Republican vote share in last two presidential elections, study claims

Exposure to Twitter lowered the Republican vote share slightly in the presidential elections of 2016 and 2020, a new study from Princeton University claims.

The 13 April study states that Twitter had “limited effects” on vote shares in House and Senate races as well as in previous presidential elections.

The authors write that the effects are “driven by independent and moderate voters”. They add that their “results are consistent with Twitter’s relatively liberal content persuading voters to alter their views”.

The study estimates that an increase of 10 per cent in a county’s number of Twitter users resulted in a lowering of former President Donald Trump’s vote share by 0.2 per cent in 2016 and 2020.

But the research also found that the convincing power of Twitter is not as strong as the Republican-boosting effects of watching Fox News or the rise in Democratic sympathies as a result of readingThe Washington Post. 

The effects on the elections of 2008 and 2012 are “statistically indistinguishable from zero,” the study says. In the 2016 and 2020 elections, voters joining Twitter “lowered Trump’s vote share but did not do so for Republican candidates in congressional races in the same election”.

Twitter’s effect on the 2016 election was “driven by independents and moderates switching their votes towards the Democratic candidate,” former New York senator and secretary of state Hillary Clinton.

“Moderates presumably have less extreme priors and are thus more likely to be persuaded,” the authors posited.

Eric Levitz of New York Magazine tweeted that the findings are also “consistent with the hypothesis that that Trump would have been more popular if Twitter had banned him in mid-2016”.

But he also noted that it was “possible that this is 100 per cent correlation being mistaken for causation”.

Mr Trump, an infamously boisterous presence online, was banned from Twitter and several other social media platforms after the Capitol riot on 6 January.

In relation to Gallup approval ratings, the researchers found that “Twitter decreased Trump’s approval ratings and increased Clinton’s with only small effects on relatively more moderate Republican candidates”.

The study also said that political content on the social media platform “has a clear pro-Democrat slant,” adding that they found “a striking difference in the number of tweets mentioning Trump relative to Clinton and Joe Biden”.

They estimate that there were “70 per cent more tweets about Trump with a Democratic slant than tweets with a Republican slant” in 2016 and 2020.

The results of the study match Twitter’s “pro-Democratic content” supposedly “persuading voters with moderate views that Clinton was preferable to Trump without inducing a more general negative effect on Republicans”.

Hillary Clinton won the 2016 popular vote, getting 48.2 per cent to Mr Trump’s 46.1 per cent, but she lost the Electoral College 227 to 304.

President Joe Biden won both the popular vote and the electoral college in 2020, beating Mr Trump with 51.3 per cent of the popular vote to the Republican’s 46.9 per cent, and winning the electoral college 306 to 232.

The authors stressed that their findings don’t imply anything about other social media platforms, such as Facebook.

They also wrote that their research “cannot separate the effect of particular types of social media content on Twitter, such as the potential role of foreign governments or misinformation”.

Since their research is based on county-level data, they also pointed out that their study “cannot address whether Twitter had a national-level effect on the election,” such as Mr Trump’s tweets driving the national news cycle.