Academic Minute
9:13 am
Thu March 7, 2013

Dr. David Schuff, Temple University – Using Social Media to Predict Elections

In today’s Academic Minute, Dr. David Schuff of Temple University explains how your posts to various social media platforms are making traditional political polls obsolete.

David Schuff is an associate professor of management information systems at Temple University where he teaches courses in Java programming, object oriented modeling and development, and networking. His research interests include the management of computer support in large networked organizations and issues surrounding IT valuation. He holds a Ph.D. from Arizona State University.

About Dr. Schuff

Dr. David Schuff – Using Social Media to Predict Elections

Many pundits predicted a tight presidential election. Obama would probably win, they said, but only because of the Electoral College. The popular vote may not go his way.

So Obama’s decisive electoral and popular victory over Mitt Romney surprised many pollsters. But for me and my colleagues, who collect and track social and traditional media activity through our TEMPO campaign media index, Obama’s victory came into focus as early as the October 22 debate.

In the age of social media and big data, TEMPO is the answer to the traditional political poll. When polls emerged decades ago, it was only practical to ask a few hundred people whom they would vote for. But today we can use customized methods to track social media and – through a unique research alliance – the vast media databases of LexisNexis to analyze tens of thousands of data points to track what voters actually do … not just what they say.

Our research aggregates candidates’ blog mentions, Facebook likes, Twitter followers and YouTube views, as well as their media coverage across thousands of websites, broadcast outlets and print publications. The result is a much broader and more detailed sample than polls can offer.

While polls provide a snapshot of who people say they will vote for at a given time, TEMPO can identify trends in their actions. Let’s say thousands of people suddenly start following a candidate’s Facebook page. How long before some of their voting intentions begin to change? TEMPO picks up on these future shifts in voter intention by tracking their behavior from the first Facebook “like.”

Which brings us back to the third presidential debate on October 22. After the two candidates made their closing remarks, Obama quickly attracted 40,000 new Twitter followers, 120,000 new Facebook likes, and 300,000 views on his YouTube channel. Romney’s social media boost was much less impressive.

For those of us behind TEMPO, this signaled that the momentum had turned in favor of Obama, who already displayed a consistent TEMPO lead throughout the campaign. But this turn hadn’t manifested itself in the polls just yet; Romney was still slightly ahead.

So while many pollsters continued predicting a neck-and-neck election, TEMPO indicated that the shift in interest would become a shift in votes. The rest is history.




Production support for the Academic Minute comes from Newman’s Own, giving all profits to charity and pursuing the common good for over 30 years, and from Mount Holyoke College.

Related Program