More about the autor

Morning Feature – The Victory Lab, Part III: Who Voted, and Why? (Non-Cynical Saturday)

November 17, 2012

Morning Feature

Morning Feature – The Victory Lab, Part III: Who Voted, and Why? (Non-Cynical Saturday)

President Obama and Democrats clearly had the stronger field operation in 2012, with more dynamic and effective micro-targeting and far superior GOTV campaigns. Were these the keys to this year’s victories? (More)

The Victory Lab, Part III: Who Voted, and Why? (Non-Cynical Saturday)

This week Morning Feature discusses Sasha Issenberg’s The Victory Lab: The Secret Science of Winning Campaigns. Thursday we looked at the familiar, precinct-based analysis and planning. Yesterday we saw the rise of data-mining and campaign plans that target individual voters. Today we ask whether micro-targeted voter contacts provided the Obama campaign’s decisive edge.

Sasha Issenberg writes the Victory Lab column for Slate and is the Washington correspondent for Monocle, where he covers politics, business, diplomacy, and culture. He covered the 2008 presidential election for the Boston Globe, and has also written for New York, The New York Times Magazine, The Washington Monthly, The Atlantic, and other magazines.

A Clear Advantage

President Obama and the Democratic Party had a clear advantage this year: “A Vast, Left-Wing Conspiracy” that used data-mining, sophisticated mathematical analysis, and randomized empirical trials to identify and register likely supporters. They then ensured supporters had the requisite ID and pushed the psychological buttons that turned mere support into ballots cast.

By contrast, the Romney campaign outsourced their data analysis, betting that competition among private vendors and consultants would defeat President Obama’s and Democrats’ more centralized organization. That strategy reflected Romney’s and the Republican Party’s ideology.

Among the results were a mailer sent and at least a dozen calls made to Casa Crissie. As I’m the Vice Chair of my county’s Democratic Party and my partner is a precinct committeewoman, these were a waste of Romney campaign resources. It wasn’t just me. Every county Democratic Party member I spoke with had received at least one mailer and at least one phone call. Issenberg’s Slate Victory Lab articles, linked above, report similar stories from OFA and Democratic activists across the country.

In fairness, Casa Crissie also received dozens of calls from OFA, and we repeatedly reminded them that we were both Democratic Party activists and that their time would be better spent calling other people. OFA were very good, and we local Democrats tried to model their approach in our voter outreach efforts … but neither we nor they were as breathtakingly efficient as Issenberg’s writing suggests.

A Different Theory of Victory

A glance at the popular vote track at Nate Silver’s New York Times website suggests a different theory for 2012. Silver offered his first popular vote forecast on May 31st, projecting President Obama to win 50.6% and Romney 48.3% of the vote. While the track opened somewhat – to 51.6% vs. 47.3% after the Democratic National Convention and again after Romney’s infamous “47% comments” – the final projection almost exactly matched where Silver started. The only difference was the margin of error, which shrank from ±6% on May 31st to ±2% on election eve.

In an article published November 8th, Silver argued that the Republican Party faces “a structural disadvantage” due to demographics and the distribution of GOP supporters among the states. Simply, Republican voters are clustered in the south in smaller farm belt and mountain states that total too few Electoral College votes to win an election.

As it happened, Silver’s model got one call very wrong this year. He projected Ohio to be the tipping point state, and with a high degree of confidence. In fact the tipping point state was Colorado, which President Obama won by 4.7% … on a nationwide popular vote margin of 2.5%. In order to win this year, Silver writes, Mitt Romney would have had to win the popular vote by at least three points.

No Republican has done that since 1988, when George H.W. Bush defeated Michael Dukakis, and 85% of voters were non-Hispanic whites. In 2012, only 72% of voters were non-Hispanic whites, and President Obama and Democrats won overwhelming margins people of color.

A Foregone Conclusion?

Data like those point toward the 2012 election being almost a foregone conclusion. In that theory – popular among some political science academics – the election hinges on a few key economic numbers and the incumbent party’s approval ratings. The ads, debates, stump speeches, registration drives, canvassing, phone-banking, and rides to polls are all but irrelevant. Yet Silver doesn’t go that far, noting:

Finally, some of the Democrats’ apparent advantage in the swing states may reflect Mr. Obama’s voter targeting and turnout operations – which were superior, by most accounts, to John McCain’s in 2008 and Mr. Romney’s in 2012.

A comparison of the 2008, 2010, and 2012 electorates drives home that point. While Republican voters are always more likely to turn out in midterms, for a variety of reasons, the key elements of the Democratic coalition did not drop off from 2008 to 2012. Indeed – except for African Americans, whose turnout was the same as four years ago – the Democratic coalition grew across every demographic: Hispanics, Asians, women, and young voters.

That wasn’t mere population growth; the percentage of women or young voters has not increased since 2008. It also wasn’t a statistical fluke. It was partly Romney and the Republican Party insulting everyone except rich white men. But mostly it was, in fact, the sophisticated planning and tens of millions of man-hours invested by OFA and local Democratic Party activists across the country.

The 2012 election was a victory of state-of-the-art data-mining and mathematical analysis mated with old-fashioned, one-to-one voter contacts. As one staffer put it, “We ran a national election like a school board race.”

Yes, they did. Yes, we did.


Happy Saturday!

  • winterbanyan

    As one staffer put it, “We ran a national election like a school board race.”

    To me that sums it up best. We had boots on the ground, a huge volunteer base, and we reached out as much as possible one-on-one, as a local candidate would do. And while that doesn’t guarantee success, it certainly impresses voters that you considered them, individually, important enough to spend time with.

    People hang up on robocalls all the time. It’s easy. You don’t even have to be polite. As for mailers…glossy postcards tend to go the way of most junk mail, especially when they multiply.

    Running every campaign as a local campaign seems like a good idea to me. I don’t have any stats, but it sure feels right. 😉

    • NCrissieB

      Studies agree that one-to-one, face-to-face contacts are by far the most effective in mobilizing voters. Those contacts are also the most time-intensive, unless you’re canvassing a very dense neighborhood or you can increase contact density by canvassing where lots of supporters are likely to gather.

      That density advantage helps Democrats, at least for GOTV efforts, because densely-populated areas are more likely to lean blue. Republicans and right-leaning voters are more likely to live in suburbs, exurbs, and rural areas that make door-to-door canvassing very time-inefficient.

      Another big advantage that I didn’t have space to explore in the article is that OFA and the Democratic Party had a more dynamic view of data. Both Democratic and Republican data-crunchers ultimately aim for two numbers for each voter: a support score (how likely this person will favor us) and a turnout score (how likely this person will vote).

      If OFA gave Velma Voter a support score of 80% and a turnout score of 70%, their data showed that President Obama would get 560 votes and Romney 140 votes from a random pool of 1000 people like Velma. (The other 300 Velmas wouldn’t vote.) If OFA learned that Velma had moved, changed jobs, married, joined a civic group, etc., her support and turnout scores would be updated based on statistical inferences about her new data.

      The Romney campaign seems to have calculated those support and turnout scores once and stuck with them throughout the campaign. Using static data may have blinded them to changes in the electorate as the campaign developed.

      Good morning! ::hugggggs::

  • addisnana

    This has been an informative series. It makes me proud to be a Democrat. I am stuck on this:

    Republican voters are always more likely to turn out in midterms, for a variety of reasons,

    I want to make sure that Senator Al Franken, Governor Mark Dayton and the new state house and senate majorities get reelected and hold in 2014. After the holidays, I’d be willing to start phoning for volunteers.

    I hope the glow of winning doesn’t stop Democrats from analyzing how to improve mid term election results.

    • NCrissieB

      I agree that we need to turn more Democratic “federal voters” (who vote only in presidential elections and often only the high-profile races at the top of the ballot) into “super voters” (who vote in every election and down-ballot races).

      Some of that may be helping more Democratic voters better understand how their state and local governments affect their everyday lives. But data suggest information-based appeals don’t influence turnout much. Motivating voters to vote is a function of psychology and social identity, and we need to work on it from that perspective. We also need to work on motivating more local party members and other activists to work as hard in midterm years as we did in 2012 … and that too is a function of psychology and social identity more than mere information.

      Good morning! ::hugggggs::

  • Jim W

    ” [T]he election hinges on a few key economic numbers and the incumbent party’s approval ratings.” Another way of saying “The party that is governing best.”

    Are there two ways to lose an election? Governing or Campaigning.

    Republicans lost 2012 on both counts.

    • NCrissieB

      I’m suspicious of those ‘fundamentals-based’ models, Jim. With over 40,000 economic data points to choose from – just among variables reported by the federal government! – the probability that statisticians can find correlations between a handful of economic data points and electoral outcomes … just by random chance … is almost certain.

      That makes it very difficult to parse out whether the economic data points have a real correlation to electoral outcomes, or whether a statistician merely tumbled upon a handful of variables that, coincidentally, happened to match up with recent elections.

      Moreover, even a sound ‘fundamentals-based’ model has to assume “all other things being equal,” and that’s a risky assumption in Realworldia. All other things are rarely equal. Events during an election year do matter – see the changes in Nate Silver’s forecast tracks – and so do the planning and work of grassroots activists. It’s easy to look back and say “The final numbers were the same as his May 31 forecast,” but we need to beware of hindsight bias.

      Good morning! ::hugggggs::

      • One of the worst “predictions” came from a political science model which attempted to use the economics data, etc. for this election. It “worked” when applied to past elections, where the outcome was already known. In this election, it bombed terribly, being off by 6% (they called it for Romney) 😆

  • There’s a great article over at The Atlantic about the tech team that OFA assembled for this election, not only what they did, but how they worked. There’s also this letter over at TPM about the ground game in Ohio.

    All of which are things that helped get Democrats to victory in this cycle, and I think the “takeaway lesson” from this should be “technology is nice, but boots on the ground works.” Now, the real challenge is to be able to turn out like that (or even half-way) in the mid-terms.

    While there’s a lot of talk about the “structural advantage” Democrats have, it’s only apparent in national election years, not in the mid-terms. In the mid-terms, it’s the Republicans who have the “structural advantage,” in that their base is far more likely to appear at the polls. 😥

    • NCrissieB

      I almost agree with your takeway, Norbrook. I’d change it to “technology is essential, and so are boots on the ground.” Neither, on its own, is enough against a well-organized opponent. High-tech information without activists to carry out the grassroots campaign will not reach enough voters, and activists without high-tech targeting and message planning will waste a lot of time and effort on voters they can’t reach with the messages they’re using.

      And this …

      While there’s a lot of talk about the “structural advantage” Democrats have, it’s only apparent in national election years, not in the mid-terms. In the mid-terms, it’s the Republicans who have the “structural advantage,” in that their base is far more likely to appear at the polls.

      … is all too true. We need to ensure the lessons, skills, and organizational energy of 2012 lie dormant until the next presidential election cycle.

      Good afternoon! ::hugggggs::