Victory Lab

Obama’s Michigan Machinations

Cagey tactics that will make life difficult for Republicans on Tuesday.

Obama in Michigan
The Obama campaign can confound Republicans by discouraging voter turnout in the Michigan primary

Bill Pugliano/Getty Images.

On Tuesday night, Barack Obama’s name will be on a Michigan ballot, but his campaign is doing everything it can to keep his supporters away from the polls. On previous Tuesdays when Republicans were voting, the re-election campaign made high-profile efforts to encourage Democrats to turn out and participate in party nominating contests in which Obama faced no serious competition. Obama organizers lured voters to Iowa caucus locations with a promise that the president would address attendees by teleconference, and Joe Biden did the same for New Hampshire house parties at 8 p.m. on primary night, an hour after the polls closed in most of the states. In both cases, campaign leadership pointed to participation rates in the contests as evidence of a successful dry run for Obama’s operations in November. “The results showed an incredible amount of enthusiasm for the president’s re-election,” campaign manager Jim Messina told the press after Iowa. “The Republicans should not expect they can build something like this overnight.”

In Michigan, however, the campaign is planning no such show of force. The state legislature scheduled simultaneous primaries for both parties, against the wishes of Democratic Party officials, who have chosen to award delegates through a caucus in May. There were no efforts to dispatch canvassers over the weekend or to leave literature on the doorknobs of voters whom Obama’s analysts have identified as turnout targets. There will be primary-night house parties tomorrow, but the campaign’s scheduling tool allows supporters to begin them as early as 6:30 pm and recommends “starting your event around 7:30”—while polls are still open, and often a prime voting time for the working-class and student constituencies Obama considers his base.

In other words, Obama is helping to depress participation in Tuesday’s nonbinding primary. Some loyal Democrats will surely turn out, but marginal voters—whom hundreds of randomized field experiments have shown can be mobilized through targeted efforts like mailers, phone calls, and canvassing—will get no extra push from Obama. “We’re having a ‘day of action,’ ” says campaign spokesman Frank Benenati, “not to turn out the vote but to build up our organization.”

Some Obama supporters will choose to cast a ballot in the other party’s primary, perhaps with the goal of humiliating erstwhile favorite son Mitt Romney—much as Rush Limbaugh did in 2008 when he encouraged his listeners in Ohio to take advantage of open primaries by supporting Hillary Clinton to derail Obama’s march to the nomination, a project Limbaugh called Operation Chaos. Such efforts last week received the blessing of Michigan Democratic Party Chairman Mark Brewer, who sent an email to his party list letting Democrats know they were eligible to participate in the opposition’s nominating contest. “If Democratic crossover votes affect the results of the GOP presidential primary next Tuesday, the Republicans will only have themselves to blame,” Brewer wrote.

But the Democrats’ efforts to discourage voting in their own primary may represent a more enduring mischief. Their passivity will not trigger the immediate glee Democrats might feel if Rick Santorum were to claim victory in Romney’s home state on Tuesday, but could confound whoever wins the GOP nomination by depriving them of valuable data that could help decode the battleground state’s particularly opaque electorate this fall.

Michigan offers some of the most challenging terrain on which to perform the basic triage of separating likely supporters from likely opponents among those voters who are still up for grabs. It is among a handful of states that does not permit voters to register by party, typically the first attribute campaigns use to categorize individuals. Without any formal indicator of partisan affiliation, targeters look to past participation in primaries for cues—considering those who chose to vote in one party’s nominating contest as either “behavioral Democrats” or “behavioral Republicans.” But not enough people participate in primaries for this to be as comprehensive an indicator as party registration is in other states.

Michigan Democrats have a leg up in finding their targets: They enjoy overwhelming numerical advantages in urban and college-town precincts they can mobilize en masse. Republican votes are more dispersed, often coming from working-class whites—like the famous Reagan Democrats whom pollster Stan Greenberg identified in suburban Macomb County—whose social conservatism cannot be easily profiled through the few demographic variables listed in individual voter-registration records or aggregated by census tract. Historically, when they’ve wanted to identify which of Michigan’s 5 million potential voters lean right, Republican organizers have had to contact them one by one, a task that—in a party without union allies or a culture of volunteer organizing—typically demanded costly phone banks or paid canvassers.

In 2001, Michigan Republican Party Executive Director Michael Meyers asked pollster Alex Gage if there was a more efficient way to sort through Michigan’s electorate. By merging available political, demographic, and commercial information with large-scale surveys, Gage was able to develop a statistical model to identify the formative attributes of Republican voters—and then mine the voter rolls to identify others who would make promising targets for the party’s outreach even if they lived in Democratic areas and didn’t participate in Republican primaries. Gage considered his objective to develop the statistical tools to support a “search-and-rescue” operation identifying Republican sympathizers stranded in enemy turf.

Gage was simultaneously advising Mitt Romney’s gubernatorial campaign, and afterward he and Meyers developed a firm, TargetPoint Consulting, whose most prominent client is the former Massachusetts governor. Meyers and Gage now have access to thousands of variables they can include in microtargeting algorithms. But party affiliation and past turnout patterns are still the best single predictors of an individual’s general-election behavior: whether he or she will vote, and for whom.

Michigan’s voter file has major lacunae when it comes to those variables for its 7.2 million registrants. The 870,000 people who chose to vote in the 2008 Republican primary, in which Romney defeated John McCain, offer a good base for profiling the characteristics of Republican general-election targets. But there’s little similar intel about who belongs to the other side: Michigan has not had a real Democratic presidential primary since 1992. In 2004, the party held a caucus, and in 2008 the state was boycotted by candidates because of disputes over scheduling. It’s been a decade since Democrats had a high-intensity gubernatorial primary, and 18 years since there’s been any contest for a Senate nomination. As a result, there’s little publicly available information for strategists of either party looking to differentiate reliable Democrats from swing voters.

Tomorrow’s turnout may help to fill in some of those blanks, but Democratic machinations—encouraging party loyalists to either stay home or disingenuously meddle in the Republican primary—mean that any new data about individual partisan loyalties that enters the voter file this month could roil more than it enlightens. “The data is perfectly meaningful statistically but it’s asymmetric,” says Mark Grebner, an East Lansing consultant who built the state’s first voter file and sells it to candidates from both parties. “If a person votes in the Democratic primary under these circumstances it’s strong evidence that they’re really a Democrat. But for anyone who votes in the Republican primary, the evidence about their party affiliation is fairly weak.”

Obama’s campaign is confident that it will know who its voters are, through the voluminous individual contact that began in 2008 and will continue in the primary-day phone banks, where the campaign has planned to keep its top volunteers—not at the polls—on primary day. Indeed, even as they blow off Tuesday’s vote, Michigan Democrats say they are likely to treat the binding May caucus as an actual turnout exercise—and the records of who participates in those will not be a matter of public record. If Romney or Santorum wants to know who the state’s active Democrats are, he will have to figure it out for himself.