In Election Outcome, Data Looms Large

Reuters Content Solutions

The merging of new-school data tactics with old-school campaigning will decide the 2016 presidential election, says “Dean of Big Data” expert and author Bill Schmarzo.

Eight years after candidate Barack Obama led the way in the use of Big Data, analytics have moved from the vanguard to the mainstream. Since 2012, candidates and their staffs have become a lot smarter in their use, and the tools available to data scientists are also far more advanced, notes Schmarzo, CTO for EMC’s Global Services Big Data practice. ”The campaign that will be most effective is the one that combines Big Data with face-to-face interactions.”

Campaigns today rely upon constantly updated analytics, compiled from social media, polls, door-to-door canvassing, mobile apps, and public databases to determine when and where to buy TV ads or invest in more field staff, and how to hone messages to different audiences. And that’s just the start.

“You have to know what it is that you are trying to measure,” says Schmarzo. If he was running a campaign, he says he’d capture as much data on voters as possible, especially social media data, to determine their positions on issues, who’s in their network, and how influential they can be with other voters.The goal: build an analytical profile for individual voters, mapping all of their preferences, opinions, affiliations, and alliances.

The task is gargantuan. It requires a “data lake” of technologies including storage, servers, and data science platforms like Hadoop, Python, and DataRPM along with proper security and governance to protect the data. The data lake provides a place where people can dump data and where analysts can test ideas and share the results with others.

Before building the technology foundation, campaigns, just like companies, need to establish the strategy. “Not all voters are of equal value, and not all issues are equal,” Schmarzo says. “Big Data forces organizations to do what they don’t really love to do, which is to focus and prioritize.”

In the 2016 election season, the following data strategies would have been most effective in helping Clinton and Trump get ahead, according to Schmarzo.

1. Understand swing-state voters, plus their friends and family

At this stage in the game, campaigns are spending their time and money trying to reach voters in swing states including Ohio, Pennsylvania, Iowa, Colorado, Florida, North Carolina, and Wisconsin. It makes little sense to crunch the data on Californians, since that state has voted blue since 1988. “I might look at data from Johnson County, Iowa, where there are lots of college kids,” says Schmarzo. “By doing so, I can get a good sense of the sentiment for my campaign there, and then I would target those at risk for not voting for me or those who are undecided.”

Using cluster-analysis techniques, campaigns can compile segments of “lookalike” voters with shared sentiments and common demographics and then determine how to best reach them and with what messages.

A voter’s ability to influence others is also important in swing-state politics. Targeting a voter in Arizona who has family and friends in Iowa would be one smart strategy. Measuring someone’s ability to influence is difficult yet doable with today’s technologies. “It takes a solid understanding of analytics and a bounty of data but, especially, [it takes] a well thought-out process of how to test and learn,” Schmarzo adds.

2. Embrace one-to-one marketing

Personalized marketing is all the rage in consumer products and in politics too, thanks to the power of data. “The key differentiator is not the big data, but the small [data], down to the level of the individual,” says Schmarzo. In addition to collecting data online, campaigns still need to spend ample time and money meeting people at their homes and in coffee shops or shopping malls. Observations from those interactions, such as the cars people drive, the books they read, facial expressions, and even the foods they eat, deliver deeper insight to those voter profiles, he adds. Such granular analytics can help a campaign identify power voters—those with a large influence on key voter segments in a swing state.

3. Measure on issues not candidates

In politics, unlike other sectors, past results are poor predictors. That’s because elections turn on the idiosyncrasies of the candidates and the issues of the moment. With the 2016 field consisting of the first female presidential nominee and a Republican nominee who has defied almost every campaign norm, the climate is volatile, to say the least. “Every cycle is so different, but what doesn’t change as much are people’s sentiments and passions,” says Schmarzo. “So you focus less on the candidates and more on the core issues.” He recommends that candidates build a scoring system to measure the importance of key issues for individual voters or voter segments.

4. Use social media, but carefully

Social media has been vital to elections since 2008, and today there are many tools available to collect and analyze it. Social sentiment and behavior helps data scientists build richer profiles on those highly prized voters who haven’t made up their minds already. However, older voters are less likely than younger voters to use social media on a regular basis, which limits its influence. There is also mixed evidence about the real influence of social-media conversations on election results. As a gauge of voter attitudes to help guide campaign messaging and tactics, though, it’s critical, says Schmarzo.

Things have changed quite a bit since 2004, when information technology was an adjunct to campaigns. Now, technology is central to nearly everything that candidates do on the campaign trail.

Even so, candidates will always need to press the flesh. “You’ve got to know the scores to build, which tells you what data to test, and you need an environment where you can test ideas quickly,” says Schmarzo. “And while Big Data from social media gives you the insights to focus, it cannot replace the human touch of those canvassers who go door to door.”

For more on technology and the upcoming election, see Election Night Coverage Gets Out Of Comfort Zone.


About Reuters Content Solutions

Reuters Solutions develops tailored, multimedia content with a journalistic approach. Reuters Solutions operates independently from Reuters editorial.