Posted Aug. 20, 2020
Top photo by Gage Skidmore
As polls show the U.S. Senate race between incumbent Republican Lindsey Graham and Democratic challenger Jaime Harrison may be tighter than many expected, comments on social media suggest Graham is facing a significant negative backlash in his home state while positive buzz about his challenger is growing.
The Social Media Insights Lab at the University of South Carolina analyzed more than 450,000 posts about Graham made in South Carolina since 2014. The lab found that Graham’s popularity has trended significantly downward since 2018. In fact, the top Twitter hashtag related to Graham is #SendLindseyHome. It has been retweeted more than 5,200 times.
“Before Donald Trump took office in 2017, there were not many social media comments about Graham,” said Insights Lab manager Kaitlyn Park. “However, his vocal support of President Trump got people talking about him. While Graham’s positions tended to play well with South Carolina Republicans, they also produced a backlash.”
The Insights Lab analysis shows from 2014 through 2016 there weren’t many social media conversations about Graham and generally they were favorable. When President Trump took office, conversations about Graham started to increase, spiking during three controversial periods of the Trump presidency:
- The nomination of Brett Kavanaugh to the Supreme Court.
- The Congressional testimony of Special Counsel Robert Mueller.
- The complaint by whistleblower Alexander Vindman.
Graham’s full-throated defense of Kavanaugh produced the highest number of both positive and negative comments during the period analyzed. In September 2018, during the confirmation hearings for Kavanaugh, Graham was mentioned in 17,000 South Carolina posts. According to the lab analysis, 66.2 percent were critical of Graham while 28 percent supported his position.
“Donald Trump is controversial, and Sen. Graham, in his defense of the president, has become a lightning rod for Trump’s critics,” Park said.
As Graham has campaigned for re-election, his presence on social media has diminished. While the number of conversations about him has decreased significantly in 2020, the tone of the comments has become increasingly negative, as evident by the top Twitter hashtag #SendLindseyHome.
“Our sentiment analysis shows that about 70 percent of the social media posts this year that mention Lindsey Graham have been negative, which is astonishingly high,” said Park. “This may reflect the impact of the well-funded campaign by his opponent, Jaime Harrison.”
Harrison was virtually unknown to those on social media until he began campaigning in 2019. Since then, the number of posts mentioning him has continued to grow; this month, Harrison actually passed Graham in number of mentions.
“We saw in South Carolina’s Democratic presidential primary the importance of number of mentions or share of voice,” Park said. “The fact that more people this month are talking about Harrison than Graham is significant.”
A sentiment analysis for 2019 and 2020 shows that while Graham has far more total mentions, he has far more negative comments and proportionately fewer positive comments than his challenger.
“South Carolina is a Republican state and it is virtually impossible for a Democrat to win a statewide race,” said Randy Covington, a professor in the School of Journalism and Communications. “But this analysis shows Harrison has a shot.”
About the Social Media Insights Lab
The lab is part of the College of Information and Communications. It is used for teaching, academic research and public reports intended to help people better understand issues of the day.
The Insights Lab software, Crimson Hexagon, uses artificial intelligence to interpret data. View a full list of reports and follow the lab on Twitter at @UofSCInsights.
How is sentiment calculated?
The lab uses software developed by Crimson Hexagon, now known as BrandWatch following
a merger. The software gauges the emotional tone of conversations using auto-sentiment
artificial intelligence technology. This feature is useful for identifying patterns
within large sets of social media data, but it should be noted that auto-sentiment
has its limits. For example, it does not always recognize sarcasm, nor does it account
for posts which may express more than one emotion.