By Lloyd Marino
Forget the debates. They’re a lot of talk, but change few minds. Speeches and rallies gather the faithful, but bring in few new supporters. Newspapers and TV news are too easily distracted by the latest scandal and much of direct mail is thrown out unread. Instead, it is big data that will elect our next president, at least according to Wired Magazine.
Consequently, both major presidential candidates are striving to tap big data most successfully and learn the most from campaigns’ experience with big data in the election of 2012.
Big Data in 2012 and Today
Big data was a major factor in Obama’s 2012 victory. Obama had a 100 person analytic team to wade through the data and classify every potential voter in swing states by their likelihood of showing up at the polls and voting for Obama. The campaign used that number for data modeling through a SQL Massively Parallel Processing database called Narwhal that could target likely voters with customized messages for their “get out the vote” efforts. They matched potential voters to similar volunteers who would be most persuasive on a person-to-person level.
The Republicans have greatly scaled up their big data operations. In 2012, Orca, their data modeling system crashed on Election Day. This year, Republicans are determined not to make the same mistakes. They have teamed with Deep Roots Analytics to target their television advertising to specific groups. Meanwhile TargetSmart helps Democrats optimize phone calls, fundraising, voter registration, direct mail, social media, and door-to-door canvassing with a national voter profiling database.
How Big Data Helps Campaigns
Both parties monitor social media through big data tools to find people’s concerns and adjust candidates’ stances accordingly. Big data also helps politicians target the best places for voter registration drives, best people to contact to remind them to vote, and best neighborhoods to organize rides to the polls. Using big data, political campaigns can screen out those most likely to vote for opponents so they can focus attention on those who can be swayed. Digital behavioral tracking identifies the voters in swing states who could be convinced to support a candidate and what would be the best way of reaching them. And ad targeting over the Internet ensures that each audience sees the most convincing ad.
Each state compiles a publicly available official voter file with name, gender, birthdate, address, and phone as well as political registration and how frequently the person votes. While the file does not say who the voter selected, even that can be estimated using polling data. Knowing the address, political data experts can supplement this record with census data on average levels of education, income, and race/ethnicity for that zip code, as well as property tax information. Private polling also brings in valuable information. Additional information on interests and concerns can be found on social media. Other data is volunteered by people joining email lists or signing petitions.
Big data also monitors the progress of campaigns. Nate Silver at FiveThirtyEight amalgamates polls, the economy, and historical trends to weigh the odds of victory on Election Day. While polling sometimes seems as much an art as a science, as witness the constant complaints about skewed polls, Nate Silver uses big data techniques to calculate a more accurate model than any one poll can provide.
Big Data and the Future of American Politics
I am confident that in the future Big Data will encompass an even larger role. Right now, candidates use polling to set and adjust their positions. In the future, they can mine social media for issue areas, allowing them to create more nuanced targeted messages and how to adjust their messages for different audiences. It will allow greater monitoring of what potential voters want to hear from a candidate and could supplement poll information on how candidates perform. For instance, currently campaigns and news media have audience response meters to gauge a sample audience’s reactions during a debate and conduct a quick poll afterwards. But big data analytics on twitter feeds produce a much larger sample of opinion. In the future, I expect less polling and more data analytics.
As more data is collected on each of us, political campaigns will be able to do more. If candidates had access to Amazon’s information on what you buy, they could send endorsements from your favorite authors and target content more exactly. Google records on your most searched topics could inform what position papers they send you. And tracking how much you spend could help set suggested amounts when fundraising.
Big data tools improve the efficiency of a campaign at convincing more people to vote for the candidate. Big data now has the power to model individuals’ behavior to target those most likely to respond and the best ways of reaching them. If big data can help businesses convince consumers to buy Pepsi instead of Coke they certainly can sway potential voters to go to the polls and choose a specific candidate.
I don’t generally use my blog to make an appeal. But elections determine the future of our nation. I implore all my readers to go out and vote on November 8th. And don’t forget the races for House, Senate, and state and local positions that may not get as much attention, but are still vital for our well-being.
Image By: Melissa Fox