Campaign Data Collection and Analysis is First Project in iSchool’s New BITS Lab
As Election Day approaches, candidate advertisements and campaign messages consume the broadcast airwaves. But it’s another kind of political chatter—social in nature, occurring in bits and bytes, by and between candidates and among voters online—that several faculty members at the School of Information Studies (iSchool) are studying in a newly established research lab.
In a multi-disciplinary team, faculty and students in the iSchool’s BITS (Behavior – Information – Technology – Society) Lab are collecting, saving and assessing the tweets, hashtags, mentions and chat streams on Twitter, plus the posts, likes, shares and follower rates on Facebook in 36 gubernatorial races this year across the United States. They’re looking at what that social messaging reveals based on what people are saying about the candidates, what candidates and campaigns are saying about themselves and each other, and how messages are shifting during the election cycle.
What’s the point?
Right now, the data-laden project “is just an experiment that’s big, fun, loose and messy, and that’s how a lot of research is,” explains associate professor Jenny Stromer-Galley, an information science expert in how people interact and communicate on the Internet and in political communication. However, assessments from the data is something she hopes “can answer some enduring social science questions about the differences in strategies that campaigns produce.” It’s also possible, the professor said, “that the tools that will come out of it will help put us in shape to do a pretty comprehensive analysis of the presidential election in 2016”—producing insights and information about messaging and communication tactics that can be very useful to, and highly valued by, political campaigns.
Stromer-Galley’s collaboration with assistant professors Jeff Hemsley, whose expertise is in big data collection and analysis and social media networks; and Bryan Seamaan, who is both a social ethnographer and a computer scientist, provides the trio a breadth of research opportunity within their own disciplines. It also means fertile ground for students who want both research opportunities and hands-on application experience in a wide range of information science disciplines.
Semaan sees the lab as a social media research center “where the intersection of behavior, information, technology and society all are studied to see how people actually are using technology and information, and what the impacts of technology and information are on society.” For Hemsley, BITS “is a place where I can apply some computation social science big data techniques to questions having to do with how people behave online and in political and social movement settings.”
The data collection now underway uses a data-gathering tool Hemsley initially built, then expanded upon with National Science Foundation funding, while at the University of Washington’s SoMe Lab.
Called the “Syracuse Social Media Collection Toolkit,” the tool is available as open source code on the Git Hub repository. It provides a much faster, cleaner and more efficient way of collecting data than typical “scraping tools” can, since it interfaces directly with Web 2.0-type social network application programming in publicly available ways to access data, Hemsley explains.
Stomer-Galley, Hemsley and Semaan are working with other iSchool faculty (research associate Professor Nancy McCracken and assistant professor Bei Yu), plus associates at other academic institutions on the project. They include Warren Allen, a Florida State University faculty member who is collecting images and videos from Instragram and YouTube; associates at the University of Arizona and the University at Albany; a Ph.D. student from Brazil; and a team of about a dozen iSchool graduate and doctoral students.
The effort has collected about 700,000 tweets and many thousands of Facebook comments in the last few weeks. Every week, the team compiles a basic qualitative analysis to report on trends and campaign developments, and deciphers what new messaging is emerging. After the election, a concerted and multi-pronged data analysis will be conducted, focused around data base management, cleaning large-scale data sets and some sentiment analysis, Stromer-Galley says.
Results-Tweeting Election Night
In addition to current data collection and after-election analyses, the BITS Lab team will be live-tweeting results in the gubernatorial races it is studying on election night. The public can follow those results Tuesday starting at 8 p.m. on Twitter via the hashtag, “#govstudy.”