At last week’s “Data-Crunched Democracy” conference, Obama 2012 Chief Scientist Rayid Ghani pointed out a real difference between the data political campaigns work with and what commercial marketers have at their fingertips: volume.
Whether you’re trying to build models of different segments of the electorate or of the consumer market, you’re relying on the past. For just one example, what’s the best predictor of someone’s willingness to donate to a political campaign? Whether they’ve donated to a campaign in the past. Likewise, what’s a good predictor of someone’s proclivity to vote in an upcoming election? Whether they’ve turned out in the past.
Commercial marketers usually have a wealth of data to work with, from demographics to credit history to homeownership, but when he started working in politics, Ghani was struck by the fact that political campaigns are trying to build voter models based on a handful of data points. The biggest factor: the fact that we simply don’t vote very often. For federal elections, data-driven campaigns get two data points on voting behavior every two years: whether someone voted in a primary (and which one) and whether they voted in the general election. If past is prologue, campaigns have very little concrete history to work with when it comes to individual voters — which is one reason that canvassing and phone-banking are so important, since each one has the potential to yield significantly more information about a voter’s beliefs and intentions than a campaign will otherwise have access to.
As a result, he said that the Obama campaign’s voter file, vast as it was, was the smallest data set he’d ever worked with professionally. Add this to the practical “duct tape and bailing wire” problems Ethan Roeder has mentioned, and you begin to see why the members of Obama’s data team frequently portray their operation as less of an unstoppable juggernaut than it appeared to many outside observers. If the Obama data people have a post-election mantra, at least judging from the folks I’ve talked with or heard speak, it might be “don’t believe the hype.”