Victory Lab
The New Science Of Winning Campaigns

Sept. 28 2012 9:38 AM

Man Microtargets for Food, But Sometimes There Must Be a Beverage

Yesterday’s most trafficked piece of political data candy was the above chart of beer brands arrayed according to the political orientation of their drinkers, which ran on National Journal’s website. Former George W. Bush ad strategists Will Feltus and Mike Shannon have become specialists in these displays, which are all based on voluminous studies that Scarborough Research conducts primarily to help advertisers distinguish the consumer habits of viewers of different television shows.

As 2004 approached, Feltus recognized that the Scarborough survey included two questions usually overlooked by commercial marketers, asking respondents about their partisanship and frequency of voting. It quickly became possible to plot TV audiences along those two axes, which Feltus used as a guide to which shows would be the best fit for reaching the Bush campaign’s targets. But the Scarborough surveys offered an unending trove of future insights: one could similarly map any other consumer preference that the firm asks its 200,000 respondents onto political orientation.

These nuggets contributed to one of the persistent mythologies that emerged from the Bush re-election campaign: that Republicans had won because Karl Rove discovered that his base drank Coors and bourbon. It was a notion floated in post-election interviews by some of the consultants who were looking to market this new “microtargeting” to other political clients, and argued extensively in the 2006 book Applebee’s America, by Bush adviser Matthew Dowd, who is a partner of Shannon’s in the firm Vianovo, and Ron Fournier, who now edits National Journal. (The third co-author was former Clinton aide Doug Sosnik.) The book ratified what might be considered the consumer fallacy of 21st-century of politics: that buying and lifestyle habits are the best predictor of political belief and behavior.

In fact, this has rarely been the case. The best predictors of political attitudes tend to be political attachments, and one doesn’t need to mine deeply into marketing research to find characteristics that separate Republicans and Democrats. The Republicans, for instance, tend to register to vote as Republicans, or in the places where that’s not possible—or it’s in greater vogue to call oneself an “independent”—vote in Republican primaries. The best predictor of one likelihood of voting is frequency in having done it before. Furthermore, even those consumer categories that might help to locate voters often don’t cover enough of the population to be terribly useful. After all, what share of voters have been identified as bourbon drinkers by a consumer-research firm?

Indeed, the consultants who had been involved in the Bush microtargeting project acknowledged to me for my book The Victory Lab that they entirely misrepresented the role that consumer variables like liquor preference or car ownership played in segmenting the electorate. “As we talked about this publicly we had to make stuff up,” Fred Wszolek, a Republican operative involved in the Bush microtargeting project, said. “We just wanted to keep them from knowing how accessible it is.”

Buried in the short National Journal essay that accompanied their beer chart, Feltus and Shannon effectively concede the point. They argue the data could offer insight into the consequences Dos Equis faces among consumers following the news that its “Most Interesting Man in the World” spokesman had hosted an Obama fundraiser. Because Dos Equis does not have a sharply defined partisan drinkership, Feltus and Shannon caution, the identification with Obama could prove perilous for the brand. “This is in contrast to its corporate sister Heineken, which as it turns out is the most Democratic beer of all,” they write. “On the other hand, Republicans love their Coors Light and favor Sam Adams, which is brewed just a few miles away from Romney campaign headquarters and whose namesake was an original tea partier.”

Indeed, Feltus and Shannon reach a conclusion that suggests there’s little value in any of these correlations for the political world.

We continue to advise big brands—who spend millions on consumer research—to make the investment to know where their fans stand politically and to put in safeguards to mitigate a political firestorm.

The implication is clear, and long overdue: political targeters shouldn’t waste time trying to translate insights from the beer aisle to the voting booth.

Sept. 27 2012 9:15 AM

The Consulting Specialty Which Must Not Be Named

I wrote earlier this year about Democratic consultant Scott Goodstein, who oversaw Barack Obama’s mobile-phone tactics in 2008 and has since led a crusade for stronger laws restricting campaigns from sending unsolicited text messages to voters.

In the piece, Goodstein battled with Gabriel S. Joseph, the president of the firm ccAdvertising, who claimed to have eluded existing spam regulations by sending “millions” of messages through email servers that delivered them to phones in an SMS format. Joseph refused to talk to me for my piece. Prompted by political text-message spam reaching Virginia voters with attacks on Senate candidate Tim Kaine, the Los Angeles Times on Thursday surveys the legal terrain and discovers Joseph distancing himself from the practice if failing to explain what he exactly he does for paying clients instead.

“I don’t know anything about sending text messages," Joseph told the Times’ Kim Geiger. "My company specializes in creating unique ways to be able to do stuff."

Sept. 27 2012 8:07 AM

Big Labor Tries To Make Online Ads Work for Outrage

When it begins showing ads Thursday that come close to calling Virginia Senate candidate George Allen a racist, the AFL-CIO’s new Workers’ Voice super PAC will be using new tools to focus old-fashioned outrage online.

Workers’ Voice is rotating four ads each attesting to a different case study in Allen’s alleged racial intolerance: one about his vote against honoring Martin Luther King Day, one each describing his use of a Confederate flag and a noose as interior-décor elements, and one declaring simply macaca—the nickname he used for an Indian-American tracker during his last Senate campaign. Viewers who click on one of the ads will be taken to a landing page where they will be allowed to vote on which “of these you think is the most deplorable of George Allen’s past actions.”

The design reflects Big Labor’s expanded post-Citizens United reach beyond merely communicating with union members and its desire to brand Workers' Voice as a platform for a new crowdsourced approach to once-centralized tactical decisions like which ads to run. “Voting on them lets us get real-time feedback on which is working the best, which then we can use to change the traffic,” says Workers’ Voice communications director Eddie Vale. “If people think one of them is way worse, we’ll increase the frequency of that one in rotation.”

Above all, the $54,000 online buy exposes the extent to which ads delivered through individually identifiable online cookies may supplant direct mail as political communicators' favorite channel for engaging preselected targets on provocative themes with little fear of mobilizing opponents. The ads will reach core Democratic constituencies—voters in Richmond and the Washington suburbs, along with African-Americans and Latinos statewide—in which Tim Kaine’s allies hope to increase interest and awareness of the Senate contest. It amounts to a universe of infrequent voters predicted to be uncertain to cast a ballot in the Senate race—an apparent tactic for fighting so-called “drop-off” among Democrats who turn out to vote for Barack Obama but may not yet be invested in the Senate race between two former governors.  

These groups cover both union members and nonunion households, although the AFL has different methods of reaching each population. To assemble its own cookie pool of union members, the AFL gathers email lists of members and sends prospecting messages that draw recipients to a Web page where they can be identified and matched through the email address to a real-world name and address. The AFL profiles nonunion households through microtargeting models and has firm DS Political match its list of targets to individual cookies that can be used to serve them ads across a variety of networks. By requiring those who want to vote on the Allen ads to sign in with a name, email address, and ZIP code, Workers’ Voice will expand its own database of Virginia supporters reachable again through other channels. 


Sept. 21 2012 6:43 PM

Why Democrats Think Anti-Romney Ads Work Better in Spanish

Mitt Romney takes part in Univision's show 'Gran Encuentro' with moderators Jorge Ramos and Maria Elena Salinas in Miami on September 19.

Photo by NICHOLAS KAMM/AFP/GettyImages

When Priorities USA and SEIU, two of the leading liberal players in the presidential campaign, decided that they wanted to roll out another round of anti-Romney ads aimed at Hispanic voters, the groups had a big decision to make. Should the ads be in Spanish or English?

At least half of American Hispanics say they consume only English-language media. Yet marketers who specialize in targeting Hispanics regularly pitch political campaigns on the cultural value of airing ads in Spanish-language outlets, from national networks like Univision and Telemundo to individual radio stations. Running ads in Spanish will be taken by voters as a sign of respect, they say.

But one Democratic research outfit has found that, in 2012 at least, the decision of whether to speak to Hispanics in Spanish or English is far from solely an aesthetic choice. Earlier this year, Project New America, a Denver-based consulting firm that develops strategic guidance for Democrats competing in areas that have not been traditional demographic strongholds, built a microtargeting model to profile Hispanic voters nationwide. When firm analysts isolated the characteristics that differentiated voters based on their candidate preference, one stood out: those who consumed news from both Spanish and English sources were predicted to be undecided at a much higher rate than English-language dominant Hispanics.

Project New America shared the finding with its clients, including Priorities USA and SEIU, two of the groups most active in trying to persuade Hispanic voters on Obama’s behalf. When they this week unveiled a new joint ad—featuring testimonials from Latino voters explaining why they would not vote for Romney—the voices were all in Spanish. The ad is now airing on Univision and Telemundo, along with Azteca América and Entravision affiliates.  (According to Project New America’s research, between one-third and half of Hispanics describe themselves as bilingual news consumers, with fewer than ten percent Spanish-only and the remainder English-only, although it varies by state.)

When Project New America polled earlier this summer, at a time when Obama and his allies were aggressively attacking Romney in media, its analysts found that Spanish-speaking Hispanics shifted opinions more dramatically than those who consumed news only in English. They found Romney’s unfavorable rating among Colorado Hispanics increased nine points over six weeks between May and July; among Spanish-language news consumers, it moved by 21 points. There were similar results in Nevada.

The findings are likely to guide other pro-Obama advertisers looking to sway Hispanics to keep their ads popping up between soccer and telenovelas. “This research showed that Spanish-language communication was extremely powerful to fill in the unknowns about Romney,” says Project New America research director David Winkler.

Sept. 21 2012 6:45 AM

Who Can Pull the Plug on the Romney Campaign?

The New York Times ran an odd story yesterday that piggybacked a critique of Mitt Romney’s media strategy onto a smart observation about the accounting device he employed during summer months in which he so fearsomely outpaced Barack Obama in the campaign-finance race. By employing “joint fundraising committees” shared with other Republican players—and being publicly hazy about exactly how the money was distributed—Romney has entered the fall with only limited authority over much of the money he has raised. As we learn more today about the candidates’ and parties’ finances, it looks as though Romney’s lack of advertising parity in August may be the least worrisome consequence of this structure.

Romney’s biggest problem is that more than one-half the money he has collected this spring and summer from his most generous donors is controlled by entities whose priorities may no longer align with his by election day. The campaign has been raising money into a Romney Victory account that allows donors to write one check for up to $75,800 that gets allocated among different committees—all of which are now working to further the nominee’s prospects. Only $5,000 of that goes to Romney’s own fund, while $30,800 goes to the Republican National Committee and $40,000 is divided among various battleground-state Republican parties.

As the Times noted, about $22 million of the RNC’s money can be spent in full coordination with Romney; basically the Boston campaign can directly order up ads from the party headquarters and specify what they should say and where they should appear. The rest of the committee’s money has to be spent on party-building activities not allocated exclusively for Romney’s benefit. “The fact that money is at the RNC significantly hobbles the Romney folk,” a former Federal Election Commission official explains to me. “This Victory money is going to have to be party things that benefit Romney incidentally.”

The money that has been assigned to state parties can’t be treated as Romney dollars, either. It is usually devoted to get-out-the-vote programs that benefit the entire ticket or use generic “vote Republican” ballot messages. Romney has effectively franchised his "ground game"—the operation of identifying and mobilizing individual voters—to such Victory programs. In other words, Romney’s ability to get out his vote is entirely dependent on the continued cooperation of party organizations.

The worst scenario for the Romney campaign would come if the national or state parties begin to see his candidacy as a lost cause, or decide to make congressional majorities—or gubernatorial, state row-office or legislative races—a priority instead. In 1996, Bob Dole’s campaign faced such a reckoning starting after his first debate against Bill Clinton failed to move the race in his favor, and party leaders redirected resources to protect Republican interests elsewhere. Romney is yet far from that point, but a cascade of further bad news and pessimistic polls could leave him quickly isolated.

“If the national party or the state parties make a decision that the presidential race is not working out, they can shift money down-ballot,” says a lawyer who has served as counsel to a presidential campaign. “Which the party has the right to do.”

Sept. 18 2012 4:21 PM

Why Journalists Should Be Forced to Work on Campaigns

Barack Obama talks with supporters during a visit to a campaign field office in Port St. Lucie, Florida, on September 9, 2012.

Photo by SAUL LOEB/AFP/GettyImages

Two weeks ago I wrote an essay in The New York Times arguing that horse-race coverage is bad because journalists don't understand how campaigns work.

Over the last decade, almost entirely out of view, campaigns have modernized their techniques in such a way that nearly every member of the political press now lacks the specialized expertise to interpret what’s going on. Campaign professionals have developed a new conceptual framework for understanding what moves votes. It’s as if restaurant critics remained oblivious to a generation’s worth of new chefs’ tools and techniques and persisted in describing every dish that came out of the kitchen as either “grilled” or “broiled.”

Today I proffered a solution: journalists should work on campaigns.

We need working reporters who have spent time inside a field office and have the comfort with the street-level politics that an engaged activist would develop after a few months of regular volunteer shifts on a modern campaign.

Sept. 18 2012 10:44 AM

Can Crowdsourcing Its Politics Help Big Labor Keep Its Influence?

Of all the unintended consequences of Citizens United, this may be the most remarkable: the AFL-CIO crowdsourcing its political program.  I wrote from Wisconsin before the gubernatorial recall there about the Supreme Court decision had changed labor politics, by knocking down sixty-year old restraints on the ability of unions to communicate with voters. The AFL now runs programs through multiple channels, but can know work on mobilizing and persuading working-class voters regardless of whether they live in union households.

Today the AFL’s super-PAC Worker’s Voice will unveil a new platform it calls RePurpose. (Perhaps the finest stylist working in broadcast campaign ads, Jimmy Siegel, did the promotional video above.  He was also responsible for the gorgeous spots in Eliot Spitzer’s 2006 campaign—evidence of the aesthetic risks one can take when running for reelection without serious opposition.)  On the front end, RePurpose resembles a loyalty program, allowing Worker’s Voice supporters to accumulate points for basic activities, from one for for signing a petition to seven for knocking on a voter’s door as part of a canvass.

Where RePurpose really deviates from usual practice, though, is that it allows supporters to effectively cash in their points to sponsor particular Worker’s Voice activities—choosing which states and candidates they want to back and the method or mode of contact.  “Different items cost different amounts, depending on a few different factors, such as the amount of staff time needed to create or change something (such as a phonebank script or a canvass turf) or the actual cost of the item (such as an online ad or direct mail piece),” according to an explainer.

As I wrote about other Worker’s Voice efforts in June, the platform keeps labor active as a broker of others’ activities: supporters can spend their points only on candidates endorsed by the AFL or its affiliates.

Sept. 17 2012 7:53 PM

What Would Mitt Romney the Management Consultant Think About the Way Mitt Romney Hires Political Consultants?

Mitt Romney and consultant Stuart Stevens during a sound check at the Tampa Bay Times Forum during the Republican National Convention.

Photo by STAN HONDA/AFP/GettyImages

Ben Wallace-Wells has already written magisterially about the role Mitt Romney played in inventing the practice of management consulting, and his contribution to the relatively modern ambition of rending the corporation “perfect, to strip of its inefficiencies until it might function as a perfectly frictionless economic unit,” as Wallace-Wells memorably put it in New York.  Since Romney joined Bain a generation ago, the consultant has come to represent at once two different (and ultimately incompatible) styles of fixer.  There is the scientist of business, capable of dispassionate measurement and objective decision-making. Then there is the magical stranger who can be invited into any corporate mess and sort it out.

The assumption has long been that Romney’s campaigns are, as the overused and under-demonstrated term of art would put it, “data-driven.”  His 2002 campaign for governor was indeed (as I relate in my book The Victory Lab) a pioneer in using new statistical-modeling techniques to model voters.  But when it comes to the culture of decision-making within his headquarters about campaign strategy, a wisdom entirely detached from empiricism is being celebrated—as two delicious quotes that have emerged in the last 24 hours demonstrate:

“I have a very good team of extraordinarily experienced, highly successful consultants, a couple of people in particular who have done races around the world. I didn't realize it. These guys in the US—the Karl Rove equivalents—they do races all over the world: in Armenia, in Africa, in Israel. I mean, they work for Bibi Netanyahu in his race. So they do these races and they see which ads work, and which processes work best, and we have ideas about what we do over the course of the campaign. I'd tell them to you, but I'd have to shoot you.”
—Mitt Romney

That’s from Mother Jones’s recently unearthed video of Romney’s remarks at a springtime fundraiser. And then here’s one of those consultants who has worked all over the world—the chief strategist and top wordsmith of the Romney-Ryan effort—in his self-justification to Politico last night:

“Politics is like sports…A lot of people have ideas, and there’s no right or wrong. You just have to chart a course, and stay on that course.”
—Stuart Stevens

Put aside the glaring failures of analogy in each of these.  (Sports is a relativist contest of ideas in which all competitors are equal?  And why would knowing what ads worked in a past Israeli parliamentary election be of any great assistance in trying to unseat an incumbent American president?)  What’s clear from them is that, as his roles changed, Romney started to see the consultant in radically different terms.

Mitt Romney the consultant may have been a scientist of business.  Mitt Romney the candidate wanted to hire a magical stranger.  And he got his man.

Sept. 12 2012 6:08 PM

The RNC's Phony "Ground Game" Video

Despite its efforts at a cinema-vérité style, this video released today by the Republican National Committee “highlighting our ground game after contacting our 20 millionth voter last weekend,” as a press release put it, falls a bit short when it comes to verisimilitude.  The interaction shown with a voter is one likely that would have never taken place without a camera present.

“That script at the door was completely staged,” says a Republican operative familiar with the encounter.  “They just ad-libbed the ‘why are you voting for Romney?’ because they already knew he was voting for Romney.”

On the day canvassers were filmed interviewing voters in Northern Virginia, their data-collection scripts included four possible questions.  One each asked which candidate the voter intended to back in races for president, Senate, and the House of Representatives.  When a voter announces support for the Republican ticket, he is asked whether he wants to request an absentee ballot.  The data from those interactions allows the campaign to sort voters depending on whether they require additional persuasive attention or an extra push to turn out, and can be fed back into the microtargeting models that help to make predictions about the predispositions of voters that canvassers are unable to reach.  

The question that canvassers are shown posing to the guy in a red shirt—Can you tell us exactly why you’re supporting Mitt Romney?—is one that they are not encouraged to ask. Campaigns don’t typically ask open-ended questions in these settings, since the type of material that comes out of voters mouths (“It’s about communities out there, it’s not about people in Washington D.C.,” Mr. Red Shirt says) can’t be slotted in easily to the databases campaigns use to manage voter information.  (I wrote at the beginning of the year about a fitful effort by the Obama campaign, code-named Dreamcatcher, to use text analytics to interpret this type of soft data shared by voters.)

On their rounds, the Virginia canvassers spotted a Romney sign on a house and asked the why you’re supporting Mitt question anticipating the response of an enthusiastic backer.  “When our video crew was following volunteers they were asking that question to people who ID’d as Mitt supporters for video content,” says RNC spokesman Tim Miller.

This is the second time in a week that the RNC, in an effort to trumpet its “ground game,” has misrepresented the nature of its field activities.  This weekend, when the party was running a “Super Saturday” voter-outreach program, political director Rick Wiley wrote on Twitter: “ As I type this tweet we have over 2,000 volunteers on the phone with undecided voters at this second.”  In fact the voters weren’t necessarily undecided: the whole point of the calls was to reach voters and ask whether they had made up their minds about which candidate to support.  “Undecideds meaning unidentified,” Wiley later acknowledged in a later email.

Sept. 10 2012 11:00 AM

If Obama Is Bouncing, Which Voters Are Moving?

Obama is picked up by Scott Van Duzer during a visit toBig Apple Pizza and Pasta Italian Restaurant in Fort Pierce, Florida.

Photo by SAUL LOEB/AFP/GettyImages

So voters finally seem to be moving, part of what’s being called an Obama convention bounce. But who exactly is doing the moving?

I recently wrote about one of the “PocketTrial” lab experiments run by Adam Schaeffer of the Republican opinion-research firm Evolving Strategies.  Schaeffer randomly assigned an online sample of voters to watch either a Romney or an Obama campaign video, and then attributed change in each candidate’s support to the video’s influence.

The most interesting finding from the experiment was that male viewers were more easily susceptible to persuasion than female ones, shifting their opinion in response to both ads while women remained relatively stable. “A larger portion of men are decided, but the proportion that are conflicted are more variable,” Schaeffer says.

Schaeffer then looked at another dataset to see if it showed the same gender split.  He looked at the last eight samples from the Cooperative Campaign Analysis Project, for which the pollster YouGov surveys 1,000 respondents in weekly wave, stretching back from early July to just before the conventions. Each time, between 200 and 400 voters in the sample did not identify strongly with a party.

Schaeffer split that sub-sample by gender, and calculated the average share of undecided voters—in YouGov’s polls they’re categorized “not sure”—across the eight-week period.  He then looked at how much the number of undecided fluctuated week to week, by comparing the average to the wave in which it was highest and the wave it was lowest.  Among women, the mean “not sure” was 26 percent, an average between a minimum of 22 percent one week and 32 percent another.  Fewer of the men were undecided, but they swung more, from a minimum of 8 percent to a maximum of 22 percent around a mean of 15 percent. (In statistical terms, that means the number of standard deviations from the mean is 46 percent higher among men than women.)

The findings that the independent male vote is more volatile raises few possibilities.  Men could be moving more in this election, as Schaeffer’s lab experiment suggested: they’re more susceptible to persuasive messages for and against candidates.  But there could be a behavioral explanation, as well: what if men are more ready to commit to attach themselves to a new opinion after forming it—like, say, if inspired by by a welll-executed convention—and women are more tentative about making such a commitment?