MU professors partner with RJI in Twitter analysis
Researchers have focused on national trends and regional data during the debates.
Oct. 19, 2012
Among the 65.6 million Americans who tuned into the second presidential debate Tuesday night were a team of MU researchers who were as glued to their laptops as they were to the TV.
The researchers, who were working in the Reynolds Journalism Institute's Microsoft Application Development Lab, are studying the use of social media during presidential debates.
As candidates Mitt Romney and Barack Obama gave opening remarks, the team, including professors and graduate students, listened to the candidates but were eagerly waiting to see how viewers would react on Twitter.
Mitchell McKinney and Brian Houston, fellows at the institute and associate professors of communication, are trying to make sense of the chorus of tweets. They are heading a team of graduate student researchers in order to study how debate viewers use Twitter to respond to candidates’ performances. The professors and researchers will study all three presidential debates as well as the vice presidential debate.
The professors hope to eventually use the tweets to study how social media usage affects political engagement.
“There seems to be something going on where this capacity to use social media while watching these events is engaging, getting people to pay attention and getting people to participate in a way that we haven’t had before,” Houston said. “That seems to be quite promising.”
The team kept an eye on the computer monitor that displayed Greenwich Mean Time as they listened for 'zingers' that might excite Twitter users. When President Obama brought up Big Bird, a reference to Romney’s plan to cut PBS funding, a team member groaned as he waited for another spike in tweets. Romney’s Oct. 3 mention of Big Bird brought in 21,124 tweets, making it the most tweeted-about moment of the first debate.
In addition to examining national trends, the research team has partnered with the Dallas Morning News, the Seattle Times and the Florida Times Union in Jacksonville, Fla., to specifically study tweets sent from these areas. Each newspaper has its own hashtag, which it promotes to its readers. The researchers analyze tweets with the newspaper hashtags, compare the regional tweets with national trends and then send reports on regional tweet activity back to each newspaper.
As the debate began, researchers were excited as tweets containing the Dallas hashtag, #DMNdebate, poured in almost immediately. They continued to flow steadily throughout the 90-minute debate. By the end of the night, Dallas had sent 979 tweets.
The partnership with newspapers in Texas, Washington and Florida was deliberately set up so the team could examine how tweets in a Republican-dominated state, a Democrat-dominated state and a swing state might differ. So far, the professors have found that certain topics excite tweeters in certain regions more than others.
During Tuesday's debate, tweeters in Dallas sent the most tweets about gas prices, while nationwide gas prices were the second most tweeted about topic. In Jacksonville, a comment Romney made about China generated the biggest response of the night. Nationwide, Obama’s 47 percent remark at the end of the debate excited Twitter users the most, according to the researchers’ latest summary of findings.
McKinney said tweets nationwide are sometimes less serious than tweets coming from specific regions.
“Some of the national spikes tend to be more around off-the-cuff moments that may be regarded as humorous," he said. "In the local communities, they tend to stick more to issue discussion."
The researchers are particularly interested in tweets and conversations that focus on campaign issues, but those tweets can be hard to find amid reactionary tweets that dominate the national feed.
“(Reaction tweets) are the easiest thing to see,” Houston said. “The challenge is to try to dig deeper and find instances and examples of when people are really talking about policies and proposals related to the debate.”
To sift through the millions of tweets sent during each debate, the team uses Topsy and DataSift, two data mining programs that allow users to track and analyze specific terms and hashtags.
At the debate’s conclusion, the team gave tweeters three minutes to tweet final thoughts, and then stopped collecting data. Though the programs stopped running, the researchers were just beginning their jobs for the night. They immediately began to search for the terms that made their feeds explode that night: pensions, women and Libya, among many others. Each word that produced a spike was recorded, along with the time that the spike occurred and the number of tweets it produced. The information would be used in the next day’s reports.
McKinney and Houston said they were both surprised by the volume of tweets sent at each debate. Nationally, Twitter reported that more than 10 million tweets were sent during the first presidential debate, and more than 7 million were sent in the second.
"The overall levels of activity have been huge," Houston said. "It probably isn't surprising … but it is impressive nonetheless."