Does Social Media Predict Elections?

Just as students of public opinion use polling to predict elections, students of online politics use activity counts (viewcounts, likes, comments, retweets, etc.) to predict elections. The fundamental theorem of online politics is that more activity is better, and the candidate with more activity is likely to win the election.

A recent article in The New York Times by technology writer Kevin Roose notes that Trump’s campaign has experienced much more activity on Facebook, Instagram, and YouTube than Biden’s. Trump has had 207 million views of his videos on YouTube in the last 30 days compared to 29 million for Biden’s. In contrast, in the public opinion polls, Biden has maintained a steady 10-point lead. What does this inconsistency signify?

First, a bit of background. A decade ago, candidates posted all their broadcast ads on YouTube and didn’t make much use of other social media. YouTube viewcounts were thus a meaningful indicator of candidate popularity. I found that in the 2015 Canadian election the Liberals had a much higher total viewcount than either the Conservatives or NDP and that in the 2016 US presidential election the total viewcount for Clinton’s ads was slightly ahead of Trump’s until  FBI Director James Comey’s announcement of a second investigation of Clinton’s emails in the last week of the campaign. During the last week, Trump’s ads posted on YouTube received four times as many views as Clinton’s (4 million to 1 million), reflecting the shift in voter preferences that would lead to Trump’s narrow victory.

Now, social media campaigns are much more elaborate. While official campaigns and PACs run ads on broadcast media and repeat them on YouTube, they and their surrogates are constantly sending out ads on social media to targeted voters. So YouTube viewcounts are less likely to be an accurate predictor of voter preferences. An academic studying online campaigning would need a major grant to follow all this activity. That’s no longer me. But I have been following the campaigns on YouTube and have some observations.

An exhaustive study would include the YouTube channels of both the official campaigns and major associated PACs. In Trump’s case, Future45, his major PAC in 2016, still maintains a channel but hasn’t been posting. The most effective PAC working on Biden’s behalf is The Lincoln Project, led by disaffected Republicans. (See Paige William’s profile in The New Yorker).

The Lincoln Project has 687,000 subscribers and has posted 240 videos on its YouTube channel. Almost all are attacks on Trump, his inner circle, or key Republican senators running for re-election. Because The Lincoln Project is doing the attacking, the official Biden campaign can set a more positive tone. In the last month, the Lincoln Project’s videos have had 38 million views, so they are obviously amplifying Biden’s message.

In previous analyses of campaign ads, I’ve referred to three fables: hero, knave, and fool. To this point, attacks on Trump have emphasized his bigotry, racism, sexism, and authoritarianism, all subsumed under the category of knave. In many of its ads, the Lincoln Project has taken a different tack, portraying Trump as a fool. Don the Con argues that he has been ripped off by his own campaign chiefs, who have spent a billion dollars in funds with no evident results. Rats mocks him because he is losing the election and his henchmen and women are leaving the sinking ship and blaming him. The ads activating the “fool” fable are often placed on media Trump is known to watch and use unseen female narration to twist the knife.

The Lincoln Project’s ads are viewed extremely favorably by their audiences, with 99 percent of the votes being likes rather than dislikes.

Joe Biden’s official YouTube channel has 426,000 subscribers and has posted 700 videos. The tone of most is upbeat, focusing on Biden’s program, record, and empathetic personality. Many are narrated by citizens who support Biden. A considerable number of clips from the debates and his speeches have been posted. The recent ad that has received the most attention is ‘Go from There,” an ad that extols Biden as a unifier not a divider, narrated by the actor Sam Elliott. In a week it has received 1.5 million views. Biden’s videos generally receive more likes than dislikes, but nothing like the near unanimity of The Lincoln Project’s.

The Trump campaign’s YouTube channel has 1.5 million subscribers and carries 3400 videos. The number of videos is so large because it is continually posting short clips from Trump’s speeches. The name most often seen on the titles of Trump campaign videos is Joe Biden, as the dominant message is an attack on Biden. Using my trichotomy of knave, fool, and hero, few ads are heroic references to Trump, and the vast majority are split between Biden as fool (confused to the point of senility, dupe of the socialist left wing of the Democratic Party, particularly Kamala Harris) or as knave (enabler of Hunter’s alleged influence-peddling with foreign governments, supporter of higher taxes).

Some of the campaign’s videos are achieving viewcounts of over 5 million almost instantly. The most recent of these include “Do You Trust Joe Biden with your Money?”, “Did Something Happen to Joe Biden?”, and “Joe Biden: All Talk, No Action,” all of which were posted on October 25. The first activates the knave fable regarding taxes, the second the fool fable by zeroing on Biden’s stuttering or verbal gaffes, and the third is a contrast ad aimed at Blacks narrated by the former athlete Herschel Walker. Assuming they are real rather than fabricated, these viewcounts are impressive. These ads have a total of 50,000 to 100,000 likes and dislikes, in a ratio of approximately 2 to 1. So they are not being watched only by Trump supporters.

To return to the question with which I began, why is Trump dominating the social media campaign so strongly when Biden is leading so steadily in the public opinion polls? Here are some hypotheses.

First, thinking of The Lincoln Project as an extension of the Biden campaign reduces the extent of the Trump campaign’s online dominance.

Second, in the 2016 campaign and during his presidency, Trump built up a large, enthusiastic, and committed base. The viewcounts for the 2020 campaign reflect the ongoing support of the base. Even if it is being chipped away, as the public opinion polls suggest, it is still large enough to deliver the viewcounts.

Third, the Biden campaign has enjoyed a huge advantage in funding and has been running many more television ads, particularly in swing states, as a recent analysis in The New York Times shows. Perhaps the Trump campaign, unable to compete on television ads, is disseminating its ads on social media, which is less expensive.

Fourth, I am somewhat suspicious of ads that get millions of views the day they are released and the next day their viewcounts stop increasing. The normal viewcount trajectory isn’t like that: it climbs more gradually before it peaks. Maybe the Trump campaign is fabricating views.

That’s what strikes me today. For the rest of the week leading to the election, I will keep watching the YouTube battle.


Leave a Reply

Your email address will not be published. Required fields are marked *

Subscribe by email

If you are interested in my weekly blog posts about politics and political narrative, as well as updates about my research and teaching, please enter your email address below to receive a free subscription.


Previous Posts