Sasha Issenberg studies how campaigns use behavioral and analytical data they have about you to influence your vote. Whether it's a call to arms from the President himself at the top of your inbox or a creepy, unmarked letter signed by a fictional...
Sasha Issenberg seems an eccentric among American political reporters. He is not especially interested in campaign rallies, convention speeches, televised debates, or celebrity surrogates. He doesn't pay mind to the topics that dominate the 24-hour cable news networks, and yet he might better attuned to the way campaigns work than anyone else in the Washington press corps.
Issenberg studies how campaigns use data they have about you to influence your vote. His new book, The Victory Lab: The Secret Science of Winning Campaigns, charts the rise of behavioral psychology and quantitative research in Democratic and Republican get-out-the-vote operations. Whether it's a call to arms from the President himself at the top of your inbox or a creepy, unmarked letter signed by a fictional sender who writes that she looks forward to talking about who you're voting for, political campaigns are rapidly developing new ways to motivate, and in many cases manipulate, voters.
The Victory Lab isn't yet required reading, but in five years it could be seen as the book that reshaped the way we view campaign politics. Issenberg presents a series of propulsive narratives, balancing the stories of basement-dwelling nerds on their way to the top of the political foodchain with concise analysis of the sophisticated social and behavioral science experiments those nerds have unleashed on the voting public. It's really good.
Sasha Issenberg is a columnist for Slate and the Washington correspondent for Monocle. He was a political reporter in the Washington Bureau of the Boston Globe during the 2008 election. I talked to him about his work and what he thinks about the growing importance of the behavioral sciences in our democracy.
VICE: When did you realize that big presidential campaign rallies were really just a veneer for the campaign stat nerds who really run the show?
Sasha Issenberg: I had been aware that there were academic political scientists running randomized field experiments to test the effectiveness of campaign techniques. But it wasn’t until after 2008 that I appreciated the extent to which people inside campaigns were learning from these studies and running their own experiments from within campaigns, using voters as guinea pigs. I learned about the Analyst Institute, a left-leaning consulting firm in Washington that acted like a cross between a think tank and a secret society that increasingly found its inspiration for new political experiments from behavioral psychology.
You compare the way campaigns test tactics on voters to drug trials. One really interesting approach you also highlight comes from behavioral scientists and activists who mailed people their neighbor's voting record and threatened to do the same after Election Day. The goal was to shame them into voting for fear of embarrassment in the community. This stuff works?
Almost all the interesting stuff works to play with the social dynamics around voting. We have a secret ballot, and so people have traditionally thought of voting as a private act. Threatening to shame non-voters—which increased turnout by 20 percent in the Michigan experiment you mention—is probably the bluntest form this approach has taken. The guy who sent the letters got death threats! But it turns out that even abandoning the usual language about voting as a civic duty can have an impact. Experiments have found that talking about how high turnout is expected to be—in other words, making it seem like a popular activity—makes people more likely to want to fall in line with their neighbors than does the usual handwringing about how few citizens participate.
Besides threatening people in Michigan, what do campaigns do with the data they have on us? How has that changed the way they operate compared to 30 years ago?
Campaigns often have access to thousands of data points on an individual voter and use it as grist for microtargeting algorithms that can automate the process of sorting through the electorate. It allows them to be a lot smarter about deciding which voters to engage, when, and how.
In the 1980s, campaigns wrote off whole sections of states because they didn’t know enough about the people who lived (and voted) there. Campaigns allocated resources based on a simple geographic triage: Democrats, for example, would stay out of areas that voted overwhelmingly Republican and indiscriminately mobilize people who lived in their strongholds. Then they would deploy volunteer canvassers and paid call centers to individually interview as many voters as they could reach. But they could never get to everyone, and so campaigns would often find themselves writing off whole sections of states because they didn’t have enough knowledge to intelligently engage the people who lived there.
Since 2000, campaigns have been able to bring in all sorts of new intelligence on voters—much of it purchased from databases established by commercial marketers—and use statistical-modeling techniques to make refined, unique predictions about each individual. So if there are two people on a block worth mobilizing and three who are susceptible to persuasion, a Democratic campaign can pick them out even if it’s in a Republican neighborhood. This process has restored the individual as the fundamental unit of American politics.
Photograph by the Romney Campaign
Are there moral hazards to treating voters as data?
I don’t think they treat voters as data as much as use data to understand who voters are, what they are about, and how politically active they are. Any social-scientific research that you’d characterize as helping to manipulate voters is focused more on modifying behavior than changing opinions—whether a person votes rather than how a person votes and always in the direction of trying to nudge non-voters to the polls. We still know very little about how voters make up their minds, and I don’t think social science has offered any shortcuts to persuasion.
Should voters be more skeptical of what they're hearing directly from campaigns, now that messages are designed specifically for them?
I actually think it’s almost entirely the opposite. Campaigns have more information about you, so they’re more likely to take on delicate issues in direct contact like mail, phone calls, or targeted web ads. Not long ago it was too risky to send mail about abortion or gay marriage because the risk of mistargeting and provoking a backlash was too great. Now campaigns have a lot more confidence that they know what you believe. They can delicately calibrate their degree of certainty about whom to contact. It should mean fewer of the big themes we see in TV ads and far more precise, specific, provocative content in direct contact. I generally think that’s a good thing.
Sure, but doesn't that mean that every individual voter receives a partial sketch of candidates in their mailbox? If I'm against same sex marriage, I'm never going to hear a contrary argument from the President, even though he is advocating for the opposite policy, because his campaign will deliberately shy away from discussing the issue with me. And doesn't that mean we're obligated to be more aggressive than ever before in seeking out candidate information?
Yeah, but what's the alternative? It's not like before microtargeting we had these wonderful campaigns thad broadcast rich conversation about complicated issues. They usually just avoided those topics entirely, because they didn't feel they knew enough about their audience to shape an effective, sharp message. So we got gauzy TV ads that said little and candidates who would discuss grand themes in speeches and debates but rarely with much substantive detail on delicate policy topics.
When campaigns were more specific and direct, the information and messaging were laundered through outside allies—labor-union canvassers on the left, church voter guides on the right—who knew their memberships but gave the candidates plausible deniability to claim they weren't the ones talking. It's not the Greek agora, but at least campaigns have the confidence to deliver those messages themselves to people they think are likely to be interested.
When Barack Obama and Mitt Romney make gaffes, political reporters tell us that the campaign narrative has been completely reshaped. Quantitative research, however, shows that electorates aren't that volatile. Have political reporters fallen behind the data-obsessed campaigns they are covering?
We’ve learned so little about how this stuff works, but generally I tend to believe that there’s probably a lot less room for big movement in voters’ opinions, especially in presidential general elections. The media may overestimate the fluidity of the electorate. An individual voter’s partisanship is the best predictor of how he or she will vote.
Reporters and the public rely on media and academic polls when we muse about whether undecideds are moving because of the latest gaffe or ad. But there are a lot of reasons people describe themselves as undecideds, sometimes because they’re not paying attention to the campaign or trying to show off how open-minded they are. But campaigns, with lots of data about the past political behavior and other attributes of each voter, might conclude, “She says she’s undecided, but she’s a registered Democrat, votes in Democratic primaries and in every way looks like someone who always votes Democratic.” In the book I quote Ken Strasma, Obama’s top microtargeting consultant in 2008, who says, “We knew who these people were going to vote for before they decided.” Journalists simply don’t have the statistical tools or access to the data necessary to make those determinations.
Photograph by Scout Tufankjian for Obama for America
You profile Dave Carney who was on George H.W. Bush's losing presidential campaign in 1992 and later helped Rick Perry win the governorship of Texas by encouraging Perry's campaign to recruit behavioral scientists. Why did Carney, for example, embrace these new techniques?
Carney had been a general consultant, who helps the campaign hire all the specialist consultants and vendors. He’d bring on the mail guy, the phone guy, the guy who makes the TV ads, the guy who prints the lawn signs, and others. He knew they were all trying to sell their services and would make claims about why the campaign should spent more money on phone calls and less money on mailers. But he despaired that he didn’t have any method to assess what actually moved votes, and the consultants got paid the same whether the campaign won or lost — so it wasn’t even in their economic interest to do what was best for the candidate who paid them. When he heard about these political scientists running randomized experiments that could disentangle cause and effect, he was thrilled. Now he could audit all the consultants bidding for his business.
I vote in Maine, where we have a viable independent candidate for Senate, Angus King. Will third party and independent candidates ever be able to consistently compete with the sophisticated operations that the Democrats and Republicans have built?
This will probably be good for well funded, well organized, disciplined independent candidates, because we have much better data and tools to isolate potentially sympathetic voters and build a coalition that crosses traditional partisan, demographic, and geographic lines. But they can’t inherit that infrastructure from a party the way that Democrats and Republicans do; the national parties are increasingly the repositories of this data and are starting to build and deliver the microtargeting models for their candidates to use. So whatever a candidate like King has he will have to build from scratch, which requires money, expertise, and good planning.
The Democrats spent most of the 2000s in opposition. Losing requires soul searching and often makes a party more flexible to change. Did losing so much push the Democrats to get their act together?
There’s a great quote in the book from Mark Steitz, a Democratic consultant who helped build a lot of the left’s new geek infrastructure in the post-2004 period when they were desperate to catch up with what they saw as the Republicans’ advantage that led to Bush’s reelection. A lot of this required a culture shift among Democratic consultants: Pollsters and data analysts who competed against one another selling services to campaigns came together to collaborate on experiments and share their findings. “Losing is a great tonic for internecine bullshit,” Steitz told me. “We had been beaten badly enough and repetitively enough that lots of people were willing to sit down with one another who previously would not have.”
The one thing I found in re-reporting this period was the extent to which the mythology around Rove and the Republican advantage was exaggerated or straight-out inaccurate. But at the time both sides were invested in the myth: Rove to be credited as a tactical genius, Republican consultants to promote their microtargeting services for sale to new clients, and the Democratic geek world to scare rich backers into believing they were so far behind that the only way to catch up was by writing big checks to create an intellectual infrastructure politics had never seen before.
How do the Obama and Romney geek operations this year size up?
Obama has a huge lead in technique. One can simply be more ambitious as a president seeking reelection—this more than anything else may explain why Bush made such strides in 2004—because you have four years to lay out a research agenda and invest in innovation the way a corporation might approach it. But there also is a huge gap in analytical sophistication between the two sides. The left is asking more interesting questions, often informed by the social sciences, and Obama benefits from that intellectual culture.