The Real Story Behind Obama's Election Victory
Business + Economy

The Real Story Behind Obama's Election Victory

REUTERS/John Gress

Elections hang by a thinner thread than you think.

By now you probably know that Barack Obama’s 2012 campaign for a second term “moneyballed” the election, employing a team of over 50 analytics experts.

You may also know that the huge volume of contentious and costly presidential campaign tactics – executed in the eleventh hour in pursuit of the world’s most powerful job – ultimately served only to sway a thin slice of the electorate: swing voters within swing states.

RELATED:  How You're Shaping the Future Through Big Data

But what most people don’t realize is that presidential campaigns must focus even more narrowly than that, taking micro-targeting to a whole new level. The Obama campaign got this one right, breaking ground for election cycles to come by applying an advanced form of predictive analytics that pinpoints rare gems: truly persuadable voters.

This is the new microcosmic battleground.

We’ve heard a great deal about Nate Silver lately. Silver has soared past the ranks of sexy scientist to become the face of prediction itself. If mathematical “tomorrow vision” has a name, it’s Nate. Even before his forecasts were vindicated by the election results, it was hard to find a talk show host who hadn’t enjoyed a visit from Silver.

But an election poll does not constitute prognostic technology – it is plainly the act of voters explicitly telling you what they’re going to do. It’s a mini-election dry run. There’s a craft to aggregating polls, as Silver has mastered so adeptly, but even he admits it’s no miracle of clairvoyance. “It’s not really that complicated,” he told late-night talk show host Stephen Colbert the day before the election. “There are many things that are much more complicated than looking at the polls and taking an average . . . and counting to 270, right?”

Moving beyond poll analysis, true power comes in influencing the future rather than speculating on it. Nate Silver publicly competed to win election forecasting – while Obama’s analytics team quietly competed to win the election itself.

This reflects the difference between forecasting and predictive analytics. Forecasting calculates an aggregate view for each U.S. state – but predictive analytics delivers action-oriented insight: predictions for each individual voter.

The concept of swing voters is ill-defined and subjective. The Democratic National Committee (DNC), in one approach, labels as “not very partisan” those voters who self-reported as independent, or for whom their party is (for any reason) unknown.

Despite this information about them, many such voters have indeed made up their minds and are unswingable.

Instead of mythical swing voters – or unicorns, for that matter – what matters to a campaign is concrete and yet quite narrow: Who will be influenced to vote for our candidate by a call, door knock, flyer, or TV ad? That is, who is persuadable?

With this in mind, Obama’s team predicted a completely new thing. Beyond predicting which way a constituent was destined to vote, they predicted whether each individual voter would be persuaded by campaign contact. Why? Because the best way to do persuasion is to predict it.

RELATED:  Why Romney's $10K Bet Matters Even More to Election 2016

Scientifically identifying persuadable voters empowered the Obama campaign to tweak more optimally and compete more ruthlessly. The notion of expending resources, such as a paid mailing or a campaign volunteer’s precious time, to contact a constituent who was already going to vote your way is abhorrent. Even worse, it’s known that in some cases contact will inadvertently change the voter’s mind in the wrong direction: It can backfire and cause them to vote for the other candidate.

It was only a matter of time before political campaigns began predicting their potential to influence each voter in order to optimize that influence – as achieved with a particular form of predictive analytics known as persuasion modeling (or uplift modeling or net lift modeling).

Enter Rayid Ghani, chief data scientist for Obama for America 2012. He was the man for the job. With a master’s degree in machine learning from Carnegie Mellon – the first university to have a Machine Learning Department – plus 10 years of research work at the labs of consulting behemoth Accenture, Rayid had rare and sought-after experience with persuasion modeling. 

His background included research determining which medical treatment will provide the best outcome for each patient – and which price will provide the best profit from each retail customer. At Obama for America, he helped determine whether campaign contact would provide the right vote from each constituent.

Persuasion models tackle a particularly intricate form of prediction. Beyond identifying voters who will come out for Obama if contacted, these models had to distinguish those voters who would come out for Obama in any case (sure things) – as well as those who in fact were at risk of being turned off by campaign contact and switching over to vote for Mitt Romney.

This last component is key. It averts cases where a knock on the door or a flyer tucked into a mailbox may backfire. 

For this initiative, the campaign needed to collect not donations, but data. No matter how smart, the brains on Obama’s staff could only tackle the persuasion problem with the right data sets. To this end, they tested across thousands of swing state voters the very actions they would later decide on for millions.

Batches of voters received campaign contact – door knocks, flyers, and phone calls – and, critically, other batches received no contact at all (the control groups). All the batches were then later polled to see whether they would support Obama in the voting booth.

Rayid’s team faced the ultimate campaign imperative: Learn to discriminate, voter by voter, whether contact would persuade.

This is where persuasion modeling came in and took storm. “Our modeling team built persuasion models for each swing state,” Rayid said, “to predict the potential to persuade for each of millions of individuals in swing states. It told us which were most likely to be won over to Obama’s side – and which we should avoid contacting entirely.”

To tweak these models, the team experimented extensively across avant-garde techniques for persuasion modeling. Although they’ve not disclosed which types of voter data made the difference for detecting “persuadability,” their related effort predicting a constituent’s propensity to vote for Obama (regardless of campaign contact) employed more than 80 fields – including  demographics, voting history, and magazine subscriptions.

The campaign’s most cherished source was the DNC’s database, which includes notes regarding each voter’s observed response to door knocks – welcoming or door-slamming – during prior presidential election cycles.

The potential “persuadability” of each voter predicted by these models guided the massive army of campaign volunteers as they pounded the pavement and dialed phone numbers. When knocking on a door, the volunteer wasn’t simply canvassing the local neighborhood: This very voter had been predictively targeted as persuadable.

As Daniel Wagner, the campaign’s chief analytics officer, told the Los Angeles Times, “White suburban women? They’re not all the same. The Latino community is very diverse with very different interests.” This form of micro-targeting even brought volunteers to specific houses within the thick of GOP neighborhoods.

Flyers also targeted the persuadables. As with door knocks, a voter received the flyer only if he or she were predicted to be influenced. Traditional marketing sends direct mail to those expected to buy in light of contact, rather than because of it. It’s a subtle but critical difference.

Put another way: Rather than determining whether contact is a good idea, persuasion modeling determines whether contact is a better idea than not contacting.

The method convinced more voters to choose Obama, in comparison to traditional campaign targeting. All of this is advanced, it’s analytical – but it’s not arcane. Having changed the game, persuasion modeling promises to begin a whole new chapter for political elections.

Excerpted with permission of the publisher, Wiley, from Predictive Analytics : The Power to Predict Who Will Click, Buy, Lie, or Die (February 2013) by Eric Siegel, PhD. Copyright (c) by Eric Siegel.