Values Before Facts: Persuading Someone Who Doesn't Already Agree

This is the last post in our “What Actually Persuades” series. The first three were about turnout: social pressure and gratitude, plan-making, and identity and norms. This one is about persuasion, changing what someone thinks rather than whether they show up. Most volunteers want to do it, and the evidence is least kind to it.

The number nobody wants to hear

In 2018, Joshua Kalla and David Broockman published a meta-analysis in the American Political Science Review of 40 field experiments on campaign contact, plus nine new ones they ran themselves, 49 in all. The question was whether door knocks, phone calls, mail, and ads change who people vote for in a general election. Their best estimate of the average effect was zero.

They found two exceptions. Persuasion showed up when campaigns contacted voters months before the election and measured the effect right away, and even then it faded by Election Day. It also showed up when a candidate held an unusually unpopular position and the campaign spent heavily to find the few voters who could be moved. Separately, Kalla and Broockman note that primaries and ballot measures are different. Without a party label to lean on, voters are more open, and contact can move them.

The reason is partisanship. In a general election, most people know which side they’re on before anyone writes to them. A card that tells them to vote the other way is asking them to change a piece of who they are. They don’t.

What works when it works

Two lines of research show persuasion happening.

The first is deep canvassing. In a 2016 study in Science, Broockman and Kalla sent canvassers to doors in Miami to talk about transgender rights. They asked voters to remember a time they were judged for being different, and then connected that memory to what a transgender person goes through. Ten minutes. The reduction in prejudice was large, it held for at least three months, and it was bigger than the shift in attitudes toward gay and lesbian people that took the country more than a decade. In 2020, People’s Action ran deep-canvass phone calls in six swing states and had Broockman and Kalla evaluate them; the calls moved the presidential margin by about three points among people reached, using a before-and-after survey design that is less airtight than a randomized trial but still one of the few positive persuasion results on a candidate race.

The second is moral reframing. Matthew Feinberg and Robb Willer showed across six studies that an argument lands when it’s built on the values of the person hearing it, not the person making it. Conservatives were more persuaded to support same-sex marriage by an argument about patriotism and loyalty than by one about fairness. Liberals were more persuaded on military spending by an argument about fighting inequality. People almost never do this naturally. We argue from our own values because they feel like the only real ones.

The shared shape: start where the other person is, not where you are. Listen or at least acknowledge before you assert. Use a story instead of a statistic, a point we made in Why Stories Beat Statistics.

A letter in July is not a letter in October

Put Kalla and Broockman’s two exceptions next to each other and you get a calendar.

A letter written in July and mailed in August is early contact. The evidence says it can move someone, and the evidence also says the effect decays. So a July persuasion letter is a first touch, not a closer. Its job is to plant a value and a story with a voter who isn’t yet hearing from anyone else, and it needs a second card in the fall to keep the seed alive. If you can only afford one contact per voter, don’t spend it on persuasion in July.

A letter written in October is arriving in the loudest month of the cycle, to a voter whose mind is mostly made up. Don’t persuade. Write the turnout messages from the first three posts in this series, to voters who already agree with you.

The exception is a ballot measure. Voters don’t have a party cue on Question 3, and persuasion on measures holds up later in the calendar. If your state has one, that’s the place for an October persuasion card.

What a postcard can carry

A postcard cannot carry a ten-minute conversation. It can carry three things: one value, named in the voter’s language; one story that shows the value; and one ask. For a ballot measure on school funding, written to a voter whose file says they’re not on our side:

Hi James, I grew up here and I stayed here because people in this town look after their own. My daughter’s third-grade class has 31 kids and one teacher who buys the pencils. Measure 7 keeps that money local, under the school board we elect. I’m voting yes, and I hope you’ll take a look before you decide.

The value is loyalty to place and local control, not equity or fairness. The story is one classroom. The facts are two numbers, and they’re in the story, not in a list. The ask is to take a look, not to agree. That’s the whole budget.

What a postcard cannot carry: a rebuttal, a list of policy points, anything that begins “Studies show.” If you find yourself writing the case, you’re writing for the board meeting, not the voter.

One last note. We’re not aware of a field experiment showing that a handwritten letter changes candidate choice in a general election, and the literature above says the effect, if it exists, is small. The measurable return on a letter program is turnout. Persuasion is a bet, placed early, on issues, with a story. Place it that way or don’t place it.

That’s the series. If you’re building your fall program now, start at the beginning and work forward.