The study of public social-media posts in rural parts of six swing states categorizes posts into three major groups: political, citizen discussion, and impact. This circle shows the political discussion. A majority of political posts analyzed in the study criticized Donald Trump for his handling of the pandemic. (One Country graphic)

Sign up for our newsletter

People posting public social media about the coronavirus in rural parts of swing states continued to criticize the actions of President Donald Trump in the last week, while voicing new concerns about personal finances and other topics.

More than half of the political statements in the study of rural social media criticized President Donald Trump for his handling of the pandemic. But a journalism scholar cautioned that studies of public social-media posts should not be mistaken for scientific public-opinion polling.

The analysis is commissioned by One Country, a 501(c)4 organization that says it is trying to get Democrats to do a better job reaching rural voters. The organization is headed by former Democratic Senators Heidi Heitkamp of North Dakota and Joe Donnelly of Indiana. 

The study is analyzing publicly posted social media about Covid-19 in rural counties in Iowa, Michigan, Minnesota, New Hampshire, Pennsylvania, and Wisconsin – all states that are expected to be competitive in the 2020 presidential election. In the counties that are part of the study, President Trump outperformed Democrat Hillary Clinton by 25 points in 2016.

This is the fourth week of the study. The research is conducted by Impact Social, a firm that specializes in social-media analysis.


DOWNLOAD THE STUDIES


The proportion of comments that criticize President Trump has risen during the crisis, from about a third of all political comments a month ago to 57% in the most recent study period, April 13-20. 

A “net Trump sentiment” graph included in this week’s report started at minus 18 in mid-March and has dropped each week. This week, “net sentiment” is minus 52. The number is the difference between pro- and anti-Trump social-media statements.

“President Trump continues to face the wrath of voters who blame his administration for failing to adequately prepare for Covid-19,” the report says. 

Limitations of the Study

But Nikki Usher, Ph.D., an associate professor of journalism at University of Illinois Urbana-Champaign, said quantifying public social-media posts is not a scientifically valid way to judge public opinion or voter preferences.

“The population of people who post publicly is a specific cut,” she said. “It’s likely not representative of the rest of the population. And I think that would make an even bigger difference in rural areas, which are lower in broadband penetration.”

She said there can be value in studying what social-media users are saying to a public audience, but such studies shouldn’t be equated with scientifically randomized public opinion polls. 

New Kinds of Messages

The most recent report from One Country picked up on some new themes, as social-media users began to focus more intently on the financial repercussions of the pandemic and the impact of business closures and physical-distancing guidelines.

“The financial impact of the shutdown is worsening with some feeling let down by the government as they fail to qualify for the stimulus,” the report says. “More tragic posts are being made as a result of job losses and/or broken businesses.” 

A discussion about the protests over lockdowns also entered the conversation for the first time. “Anger is expressed at people protesting the lockdown, and protesters voice their right to work and call for society to reopen.”

More social-media posts are also expressing confusion about the accuracy of fatality rates. “Some accuse authorities of double-counting and thereby exaggerating the severity of the disease.” Other expressions of confusion focus on what constitutes social distancing, inconsistencies in messaging, and whether they should wear a mask.

The number of posts espousing conspiracy theories has increased over the last month. Sixteen percent of the posts discussing the impact of the pandemic fall in this category, including comments that China purposefully loosed the virus on the world. Other rumors say incorrectly that there’s an approved vaccine about to be released.

Other Comments

Other analysis in this week’s report includes the following:

  • A new type of post emerged in the last week focusing on the medical preparedness and capabilities of states and regions. The social-media users “ask about the level of medical expertise, the number of ICU units/respirators, testing capability, or the availability of face masks.”
  • “While less aggressive and vocal [than opponents of lockdowns], some rural people are happy to show their approval of the lockdown and applaud their leaders for showing courage in the face of aggression and abuse.” 
  • “A small but determined group express the need for a mail ballot at the next election and warn that the GOP’s reluctance to implement this reflects their desire to suppress turnout in November.”
  • “Mental health problems are becoming a real issue. Citizens openly state they are thinking of ending their lives as the crisis overwhelms them. Some reach out for help but feel their cries go unheard due to the lockdown.” (EDITOR’S NOTE: Help is available through the National Suicide Prevention Lifeline at 800-273-8255.)
  • “Plenty of people go online and cheer one another up with positive thoughts about the future. Brighter discussions also exist in the form of scientific breakthroughs where people readily share information of vaccine testing.”

How the Study Was Conducted

The study was conducted for One Country by Impact Social. The firm’s methodology combines big data and algorithms with randomized sampling and human analysis.

The study uses publicly available social-media posts from blogs and big platforms like Facebook and Twitter that are geocoded to the rural counties in the six swing states. Geocoding, which documents the location from which a post was made, is taken from information gathered by the social-media platforms for each post. Posts that are not geocoded as coming from one of the rural counties in Iowa, Michigan, Minnesota, New Hampshire, Pennsylvania, or Wisconsin are not included in the analysis.

The researchers filter the posts to “focus solely on the views of local citizens,” according to the report.

From this universe of posts, the researchers pull a randomized sample. In the final stage, humans read the sample of social-media posts and categorize them based on their content.