To Persuade or Not to Persuade?

For Democrats in 2014, that is the question.


There is a debate raging in Democratic data circles about how the party can best avoid an electoral wipeout that would see it lose control of the Senate. At its core, this is a debate about voter behavior in midterm elections, and it highlights a stark contrast between practitioners who focus on persuasion (convincing people to switch their vote) and mobilization (getting partisans who don’t normally vote out to the polls).

The divide is cultural, and often visceral, driven by the feudal nature of the political industry. Most campaigns are Teams of Rivals comprised of specialists predisposed towards their specialty. A media specialist will argue for more, better, and catchier TV advertising and message strategy to persuade their way to victory. A turnout specialist will argue that they can knock on enough doors to make the difference in a close race. What should be a measured debate about the relative emphasis of each tool — in truth, both are essential — is being magnified into a low grade holy war. Heed the other side’s advice, and Democrats will fall right off a cliff in November.

The fundamental question is this: When the electorate shifts one way or the other (as it did to Democrats in 2006 and 2008, and to Republicans in 2010) is that because the voters changed their minds, or because different people voted? Is your job as a political professional to rally an unbending universe of partisans, or is it to persuade people to change their minds?

Data can help us answer this question. Over at FiveThirtyEight, Harry Enten suggests that turnout won’t make much of a difference:

[T]urnout isn’t nearly as important as D.C. wags make it out to be. The demographics of who voted in 2012 vs. 2010 were different, but that difference didn’t make much of a difference. The reason Republicans won more votes in 2010 — and likely will in 2014 — is that voters wanted Republicans in office, not that minorities and young people didn’t turn out to vote.

Enten points to turnout changes in various electoral demographics. Yes, the midterm electorate tends to be older and more Republican. But not as much as you might think. If minorities and millennials voted at the same rates in 2010 as they did in 2012, Republicans still would have won the 2010 elections by 3.3 points — three points off their actual margin — but still a dramatic enough shift from 2008 to suggest that a large number of voters actually changed their minds.

The turnout argument is led by Sasha Issenberg, author of The Victory Lab, the bible for the application of the social sciences to campaigns, and Tom Bonier, head of the Democratic targeting firm Clarity Campaign Labs.

This approach relies on “hard” data gathered about individual voters, as opposed to the “soft” polling data that provides insights into groups of voters, which Enten relies on.

Writing in The New Republic, Bonier explains further:

The bottom line is, Enten’s theory doesn’t hold up under the scrutiny of individual vote history. For example, Enten looks at the variation in turnout among younger voters between 2010 and 2012, and then considers the partisan vote share of that demographic in order to assign some sense of partisan impact of these turnout changes. But what he’s missing is an understanding of which younger voters cast a ballot in each year. By using vote history and partisan models, we can gain a better sense of this dynamic. For example, in Ohio in 2012, the average modeled partisanship of registered voters under the age of 30 who cast a ballot was 57.3%. The same statistic for that group for 2010 voters was 50.5%. So while the overall share of the electorate that younger voters comprised in each election could be largely unchanged, that would mask the sub-demographic dynamic that is truly impactful, from a partisan vote perspective.

What seems like a technical debate about how one measures the electorate could have a huge impact on how hundreds of millions of dollars will be spent to shape the midterm electorate this year. If it wasn’t just young voters who vote less in midterms, but specifically Democratic young voters, then trying to convert fickle or undecided voters might be a futile exercise. You would be better off just getting all of the voters who already agree with you to the polls. In Enten’s assessment, there aren’t that many of those voters — enough to give you a 3% boost in the final tally, at best. In Bonier’s view, grounded in more granular individual behavior, the universe of people who skip midterms is actually 15 points more liberal than those who don’t. What seem like dramatic seesaws in momentum (“Obamacare is an albatross around Democrats’ necks!” “Wait, maybe it’s not!”) are actually driven by individual-level turnout decisions. Who votes makes a huge difference.

Mark Mellman is a well-regarded Democratic pollster who bucked his party and accurately called the defeat of the last unsuccessful Democrat to seek the White House, John Kerry. He vigorously defends a persuasion strategy in a series of pieces for The Hill, and on Twitter.

Consider first the punishing arithmetic. Let’s start with a very oversimplified case, a model. Imagine a world with 200 voters and 50 percent turnout across the board. Say 40 percent are Democrats, 40 percent are Republicans and 20 percent are independents (or at least voters whose partisanship is uncertain and can’t be mobilized by Democrats without fear of adding to the GOP total). That translates into a 50-50 race, with 40 voters each backing the Democrat and the Republican and 20 votes uncertain.
Now say, through what we will see would be a miraculous application of turnout technology, Democrats generate an increase in turnout of 10 percentage points among their partisans, and Republicans do nothing. You’ll then have 48 people supporting the Democrat, but now you need 55 votes to win. You lose unless you can persuade 35 percent of those independents to vote for you. Alas, you have no money left for persuasion because you’ve followed advice to spend it all on turnout. Whoops.

Reviewing the social science research on the subject, Mellman notes that a 10 percent Democrat-only bump in turnout is unlikely. Research by the Analyst Institute’s Aaron Strauss suggests the bump was at most 2% among the most coveted set of unlikely Obama voters in 2012. Even if Democrats sincerely wanted to transform the electorate, chances are they would see only marginal success. Turnout alone can’t win, and persuasion can’t be overlooked — especially in states like Arkansas or Louisiana that are strongly Republican at the presidential level.

To be sure, Mellman isn’t entirely dismissing the importance of turnout — “I’ve done campaigns where less than 0.8 of a point, or the equivalent of a 1.7- or 2.5 percentage point increase in turnout, determined the outcome, so you’ll never catch me denigrating the importance of turning out voters” — but then adds, a bit acidly, “even as the science indicates its impact is quite limited.”

In reality, though, a turnout program that achieves a 1 to 2 percent lift isn’t a bit player that’s vastly under-delivering on its promises. It’s doing exactly what it was designed to do. This is something that most political professionals well understand. In the trench warfare of competitive campaigns, which are closely contested and with the opportunities for breakouts few and far between, 1 or 2 percent can make all the difference.

Campaigns control precious few things through their own deliberate effort, independent of what their opponent does or doesn’t do. Turnout is one of those things. Persuasion, like the search for the blockbuster drug or summer movie, can yield fantastic results, but is often hit or miss — too often outside the capability of the campaign or the candidate to fully control.

Campaigns control precious few things through their own deliberate effort, independent of what their opponent does or doesn’t do. Turnout is one of those things.

So, what about persuasion? Mellman suggests that he’ll turn to this in his next column, so I don’t want to put words in his mouth. But it’s safe to assume that his assessment will be less skeptical. To be fair, there are some good reasons for this.

Where Persuasion Matters

In this week’s Senate primary in Nebraska, university president Ben Sasse ran away with the Republican nomination, winning 49.5% of the vote, to 22.1% for his closest opponent. None of the late polls predicted this outcome, projecting a close three way race. Sasse convinced a significant chunk of voters to vote for him at the last minute. Examples of this sort of late movement are legion, especially in primaries. And there is little chance that turnout had much to do with it at all.

Primaries are persuasion-rich environments because the ideological differences between the candidates are small, and decisions are made based on temporal considerations (who appears like the stronger candidate most likely to win the general election in the days leading up to the vote). It’s difficult to apply a paint-by-numbers approach in primaries. Rand Paul won the Kentucky Republican primary by 24 points against an opponent from the party’s establishment, but similarly situated Tea Party conservative Matt Bevin looks likely to go down by about the same margin against Mitch McConnell next Tuesday.

Presidential elections operate on the other extreme of the spectrum. In their data-rich assessment of the 2012 election, Lynn Vavreck and John Sides, debunk the commonly-hyped notion of campaign “game-changers” — finding that less than 5 percent of voters changed their minds throughout the course of the campaign, a finding derived from surveying 45,000 voters, interviewed a year before the election, during the campaign, and in its immediate aftermath. Vavreck and Sides argue that the election was decided by economic fundamentals, and there was very little either side could do to persuade voters to deviate from these fixed assessments.

Why might this be? Opinions of the incumbent president is well formed, and media coverage of a Presidential election is at saturation levels for more than a year before the vote. Campaign efforts from both sides on television (and in the field) tend to cancel each other out. Even “game changers” like Mitt Romney’s “47 percent” gaffe are found to made hardly any difference at all in the final outcome. To be sure, Vavreck and Sides found that more television advertising could move votes (by around one percent per 700 gross ratings points advantage), and that a massive television advertising advantage could have propelled Mitt Romney to victory. But in practice, such spending advantages are rare, and unlikely to materialize in a way that would make more than a percentage point’s difference. Using a cost-per-vote calculation, turnout may be a more efficient way to maximize your odds.

But midterms are different from Presidential campaigns. They are also different from primaries and lower-turnout elections where small nudges in one direction or the other can make a big difference. How might this play out in 2014?

Midterm Persuasion in the Age of the SuperPAC

Persuasion — defined as the act of a voter carefully considering the different alternatives and making a choice not dictated by partisan considerations — happens all the time, for reasons often unrelated to campaign activity. One example, as we have seen is primaries (which are by definition more volatile because of their intramural nature). Another example is in lopsided or uncontested elections. Softer partisans may cut a popular incumbent a break when their vote won’t make a difference, while rallying ‘round the flag in the races that really matter. Oftentimes, elections in deep red or blue states can be closely contested because the incumbent is seen as more moderate. Wanting to be seen as fair-minded, voters will break with their party to support the incumbent, but will be cross-pressured by party demands if the race is pivotal to national control of Congress.

To see how this might play out, let’s take the archetypical midterm Senate race, Democrat Mark Pryor’s bid for re-election in deep red Arkansas. Clearly, Pryor has persuaded some Mitt Romney-voting Republicans to support him, based on the fact that he isn’t losing by 20 points. But any further persuasion Pryor might do from now until November is limited by the fact that heightened interest in the race based on heavy media coverage and early TV advertising may have already hardened voters’ perceptions of him and his opponent, Republican Rep. Tom Cotton.

Both campaigns will be very well financed to roughly the same degree, and outside SuperPACs will be more so. As long as the race remains close, there is little chance that one side or the other will gain a lasting advantage due to TV advertisements. Politics is an efficient marketplace when it comes to allocating resources to the most competitive races, and if a campaign falls behind, a SuperPAC will invariably pick up the slack. And the more dollars that are spent on TV, the less likely it is that the incremental next dollar will make any difference at all.

This means that any hope for campaigns to gain some sort of structural advantage will need to be more tactical and under-the-radar. This is particularly the case if the fundamental dynamics of the race don’t shift between now and November.

Assume Cotton, a talented candidate, runs a relatively error-free race, and Obamacare and the president’s low approval numbers continue to dog Democrats. There may be few outside “game-changers” to throw off the dynamics of the Pryor-Cotton race, thus persuading large numbers of voters to switch sides. While these moments can and do happen, they happen unevenly and unpredictably (through a clever ad that really caught on, or a major gaffe or scandal — like George Allen’s “macaca” moment in 2006). Campaigns would be foolish to count on events outside their control to save them. In this frenzied environment, campaigns can only execute patiently against their own plan, and focus on the things they can control. The attraction of investing in turnout is that it’s a dependable way to gain small but potentially decisive advantages without attracting much attention or response from the other side. TV ads are very public; turnout and data operations are under the radar, and often hard to replicate and understand, even in this age of obsessive Big Data coverage.

Campaigns can only execute patiently against their own plan, and focus on the things they can control.

The good news is that the question of persuasion and turnout is not a zero-sum game, though it is often made out to be. Doing persuasion is not a question of making a difficult tradeoff vis-a-vis mobilization; it’s a matter of criminal incompetence if you don’t. All candidates have an obligation to make themselves appealing enough to deliver a winning coalition on Election Day. Campaigns will use various tactics— earned media being the most cost-effective — to enhance their natural appeal to persuadable voters.

This is not an argument for turnout over persuasion in campaigns, but an argument for balance and for sophisticated execution on all fronts. Right now, the budgetary balance in elections is tilted in one direction — towards paid persuasion. An approach that diversifies risk by investing more evenly in all both persuasion and turnout must be tested against more one-dimensional approaches.

When different groups of consultants argue from dubious all-or-nothing scenarios that will never happen (or attribute these views to the other side), we lose hope of getting to a rational and scientific understanding what actually works, defined as the lowest cost-per-vote at scale. In a world of probability, there are no absolutes or silver bullets.

Flickr photo from Michael Lovitt.