Sign up for our newsletter
Students of the 2020 presidential election have a daunting array of information to consider in the final two months before Election Day. There’s traditional public opinion polling, online survey panels, focus groups, perpetual punditry, and more.
Now there are also efforts to determine what the churn of social-media posts might tell us about the course of the political campaigns.
A national study of social-media messages by swing voters indicates that after months of declining fortunes, President Donald Trump has recovered some of the ground he lost during a period of overlapping crises. Even up against issues such as a global pandemic, widespread protests against racially motivated police violence, and the biggest economic challenge since the Great Depression, Trump has continued to make himself the number one campaign issue.
“He sucks the oxygen out of the room,” said former Democratic Senator Heidi Heitkamp of North Dakota. “Trump’s behavior has just driven everything.”
So a study of how swing voters using social media view the president could hold important lessons about the election.
The study was released last week by One Country, a 501(c)4 organization focused on helping Democrats reach rural voters. Heitkamp serves on One Country’s board.
The report, produced by Impact Social, details an increase in anti-Trump sentiment in social-media posts of swing voters this spring and summer. And it also shows that Trump began to recover some of his lost ground in mid-July, just before the Democratic National Convention.
Earlier this year, the Daily Yonder reported on another set of studies conducted by Impact Social that focused on social-media posts in swing states.
This new study looks at a different set of potential voters. Instead of being limited to rural voters in swing states, the new study looks at potential voters (rural or urban) anywhere in the U.S. who meet a set of criteria defining them as “swing.”
The theory is that examining the social-media posts of a diverse and unpredictable group of voters can tell us something about the larger national political trends that will affect the 2020 election.
The Fight for the Middle
Heitkamp said One Country is interested in the national study because it shows that rural and urban voters have a common set of concerns. For example, she said the U.S. Postal Service gets portrayed as a “rural issue,” but concern about the agency is more generalized.
“Where we tend to take the Postal Service issue is to say that it’s going to be worse in rural America because those areas are more dependent on the mail,” she said. “But concern about the Postal Service is much more generic and widespread, both urban and rural.”
Heitkamp said the social-media posts of swing voters show her that a Democrat has a good chance with moderate Republicans who are concerned about the direction the president is taking the Republican Party. She said Vice President Joe Biden stands to pick up some of these voters who previously supported Republicans like Mitt Romney and John McCain.
“I’m convinced we could win some of those people,” Heitkamp said. “With all of the catastrophes that have been the Trump presidency, a lot of those swing voters, those Republicans, could vote for a Democrat” in 2020.
Improving Prospects for Trump
After five weeks of substantial declines, Trump’s fortunes in social media began to improve shortly before the Democratic National Convention. One reason may be that the specifics of a Biden campaign gave Trump supporters more to talk about. And the Republican National Convention also prompted a bump in pro-Trump posts.
Since the convention, Trump has enjoyed a gradually improving “index of net sentiment” (which shows the proportion of anti-Trump posts to pro-Trump posts among swing voters). The index is still negative, meaning swing voters are posting more anti-Trump messages than pro-Trump messages. But Trump’s position has improved by about 15 points since his nadir in late July. In the past two weeks, the improvement has come from an increased number of pro-Trump posts, while the proportion of anti-Trump posts has remained stable.
Impact Social called the Republican convention a “partial success” for Trump. The convention led to an increase in posts about Trump’s pre-Covid-19 economic record. And there were increased numbers of posts against Biden.
But Trump’s theme of law and order was only partially successful, the study said. “Despite the passion and anger expressed in posts [by pro-Trump swing voters], the overall volume of this conversation was not significantly greater than in previous weeks,” the study said. “This implies the issue may be of less importance than has been suggested in determining the outcome of the election.”
In the weeks after the political conventions, swing voters continued to talk about civil unrest. Some swing voters said they are “appalled at Trump’s reactions to the riots, charging him with whipping up hatred and sowing division for his own ends.” But others agreed with his law-and-order message and his claims that Democratic leaders in major cities are failing to restore order.
To conduct the study, Impact Social created a roster of approximately 20,000 social-media users based on the contents of their posts. Social-media users identified as swing voters include people who say they may switch party allegiance from 2016 to 2020, who consider themselves to be independents or third-party supporters, who say they are Republicans but will not support Trump, who indicate they are Democrats disillusioned with the choice of Joe Biden as the party nominee, and other kinds of voters who represent volatile parts of the electorate.
Since the social-media users are posting publicly and use their name, Twitter handle, or other unique identifier, researchers can return to sample this same group of social-media posters over time to see how the conversation is changing. The posts are filtered to include only original posts from local citizens, the researchers say. They remove reposts, shares, retweets, or other repetitions.
Each week, Impact Social conducts a random sample of the posts originating from the group of 20,000 voters. Readers categorize the statements as pro-Trump, anti-Trump, or neutral. They also group the posts around themes to see which issues and messages are making their way into the conversation.
The Impact Social study is different from a public opinion poll, which seeks to predict broad voter preference based on a scientific sampling of voters. The study of social-media posts shows the range of opinion and the relative number of posts, but it does not directly reflect voter preferences.