Inside the Victory Lab with Sasha Issenberg

Using Data, Experiments and Analytics to Make Campaigns Smarter

Sasha Issenberg

Thursday, 4:00-5:30pm


There’s far more to winning office today than meets the eye. Thanks to an analytical revolution whose impact on politics has been likened to the effect Moneyball has had on sports, 21st-century campaigns are not just merely a parade of speeches, debates, rallies and TV ads. Instead, campaigns have been transformed by a pair of methodological breakthroughs that have wormed their way into the campaign brain from other fields: field experiments, essentially drug trials for politics that treat voters as unwitting guinea pigs, and microtargeting statistical models that lead campaigns to believe they know how you’ll vote even before you do. The most influential figures in today’s campaigns often never appear on television, and scheme assiduously to keep their machinations out of the newspapers. But we’ll bring them in to do show-and-tell: not just to describe their work in the abstract but to demonstrate what exactly their work entails, how they visualize problems and the techniques they use to solve them. Students will emerge not only with a rudimentary understanding of the mechanics of the modern war room, but a familiarity with the way today’s tacticians talk and think — and how it is often academic research tools that can be applied to solve practical problems for campaigners. Some of the best jobs in politics, after all, now belong to those with expertise in applied mathematics, the data sciences, and behavioral psychology.

Week One (9/26): What campaigns do The basic challenge every campaign faces is deciding which voters it needs to engage and which it should ignore — and then among them where, when, and how to allocate resources and direct messages. We’ll start by stepping back, to understand the decision-making behind the most most important strategic commitment a campaign makes — even if the media never really cover or understand it — and how it informs every they do through election day.

Guest: Dave Carney, former White House political director and chief strategist to Texas Governor Rick Perry


Week Two (10/3): What campaigns know about voters People running for office have far more information about the people whose support they’re trying to win than they did a decade ago, sometimes thousands of data points about an individual voter. It comes from sources as varied as voter-registration forms, warranty forms, subscriber lists, and conversations with canvassers knocking at the door or phone-bank callers ringing during dinner. We’ll get a tour of the data available about an individual citizen, understand where it comes from and how it ends up in campaign databases — and what use it is for a candidate to know if you’re a gun owner or regular cruise-goer or cast a ballot by mail in a 2010 Republican primary.

Guest: Bryan Whitaker, chief operating officer of NGP VAN and former Democratic National Committee technology director


Week 3 (10/10): How campaigns make assumptions about voters How do campaigns sort through all the disparate data they’ve gathered about individuals? We’ll get an introduction to the predictive-modeling techniques often called microtargeting, and an understanding of how algorithms can churn through thousands of data points to help campaigns draw conclusions about people — both their past attitudes and future behaviors — without ever interacting with them directly.

Guest: Tom Bonier, co-founder, Clarity Campaign Labs


Week 4 (10/17): How campaigns know what works Using randomized-control experiments, academics and political operatives can now empirically measure the effectiveness of campaign programs and test new techniques in the field — and have begun to translate insights from behavioral psychology to challenge long-held assumptions about what actually motivates people to vote. What have we learned from election experiments thus far, what is left to learn, and how has our understanding of the voter’s brain changed in the process?

Guest: Todd Rogers, professor Harvard Kennedy School; founding executive director, Analyst Institute


Week 5 (10/24): When experiments and microtargeting merge The cutting edge of campaign analytics comes from integrating experimental designs with rich individual-level voter data to provide freshly granular insights on the impact of political communication. We will get a primer on the quest to isolate what statisticians call “heterogeneous treatment effects,” and understand how the Obama campaign used them to find new answers to basic questions: what types of voters respond best to different persuasion messages or get-out-the-vote contacts?

Guest: Dan Wagner, former chief analytics officer, Obama for America; founder, Civis Analytics


Week 6 (11/7): What analytics can explain about TV Most campaign spending and strategic attention remains devoted to paid advertising and free press coverage, areas that lend them themselves to only crude targeting and limited opportunities for rigorous experimentation. The 2012 Romney campaign devoted more of its analytical energies to understanding how it can use the media to drive public opinion, and we’ll get the head of the Romney’s team to introduce their methods and findings.

Guest: Alex Lundry, senior vice president, TargetPoint Consulting; co-founder, Deep Root Analytics; former director of data sciences, Romney for President


Week 7 (11/14): What analytics can show us online In many respects, the friendliest place to approach politics in scientific terms is online, where every click is trackable, every user followable. But in certain respects the old New Yorker cartoon caption On the Internet no one knows you’re a dog applies: linking voters’ online activities to identities offline, where people are registered and cast ballots, remains a major challenge. What can campaigns test and measure online and how can they draw conclusions from it about politics IRL?

Guest: Amelia Showalter, former director of online analytics, Obama for America;


Week 8 (11/21): Mystery science politics We’ll take a perplexing piece of real-world campaign research — the results of an experiment conducted within the context of presidential election in a swing state — and, drawing on all of the tools we’ve reviewed throughout the semester, work through what it might mean. We’ll be joined by a veteran of one of the campaigns involved to see how our conclusions match up with the view from inside the war room.

Guest: Michael Simon, president, HaystaqDNA; former national targeting lead, Obama for America