���߲ݴ�ý

When U.S. President Barack Obama declared his support for same-sex marriage on May 9 of this year, the Twitterverse went into overdrive. Social media analysts reported that Twitter saw 1.6 million #gaymarriage tweets immediately after Obama’s announcement. The pace peaked at 7,347 tweets per minute that afternoon.

That’s a lot of talk about an issue that could well have a tangible impact on the 2012 U.S. presidential campaign.

But how do political scientists like U of T Scarborough professor Chris Cochrane make sense of the actual messages being articulated in this wave of opinion?

Welcome to political analysis in the age of big data.

“Computing has changed the nature of research in any discipline and it’s making a huge impact on how we keep track of what’s going on in politics,” says Cochrane, who specializes in the left-right divide and how that plays out with issues such as abortion, capital punishment and same-sex marriage.

One of Cochrane’s key interests is in how political parties get their manifestos in the minds of the broad population and then into the conversations that influence how people vote. “Twitter has become an excellent source of data about those political conversations. It’s just staggering how much data is available. Analyzing it and then storing it requires computational power that wouldn’t have been possible only a few years ago.”

The power of high-performance computing has also enabled researchers to “expand the horizons of what we are able to investigate.” Cochrane points to a project that has been underway for a number of years that has gathered and analyzed the election platforms of every political party from every democratic country since World War II. “The sheer amount of data this kind of a project generates and the extent of coordination it demands wouldn’t be possible without contemporary computing. And this analysis is providing us with superb detail on politics worldwide.”

Cochrane notes that modern content analysis software enables political analysts to look at surveys or masses of tweets and determine the zeitgeist on a certain subject by examining such factors as the frequency at which certain words appear or the way in which words are used in combination with others.

Still, as powerful as computers can be, at a certain point in political analysis the human element is needed to look at the subtleties computers can’t detect. “There’s a term, ‘dog whistle politics,’ which is about words that indicate meaning to political party supporters but don’t actually say anything. These are difficult to pick up using computing. But it’s tedious and expensive to hire people to do this work.”

Cochrane’s current work is focused on making it easier to nail those subtleties. “Professional survey software enables us to ask questions which structure their answers in very particular directions and that I can code into the categories I want to look at with comparable answers from any number of people. This is called ‘crowdsourcing.’ And it wouldn’t be possible without modern computing technology.”

As an example, Cochrane points to a key question in his research — do political parties actually disagree with each other on issues such as capital punishment? “Is it the case that the right and left compete because the right supports capital punishment and the left opposes it? Or do the right and left compete because they really talk about entirely different issues, such as economic growth on the right and environmental protection on the left? Computing enables me to first gather the content and store it and then have people analyze that content and understand exactly how political messaging is working.”