Be a Voter: Identity, Self-Image, and the Words That Invoke It

This is the third post in our “What Actually Persuades” series. We’ve covered social pressure and gratitude and plan-making. This week is identity: the idea that you can raise turnout by getting a person to think of themselves as a voter. It’s a great story. Part of it is true.

The finding everyone loved

In 2011, Christopher Bryan, Gregory Walton, Todd Rogers, and Carol Dweck published a paper in PNAS with a finding that seemed too good to be true. They gave registered voters a short survey the day before an election. Half the surveys used verb phrasing: “How important is it to you to vote?” Half used noun phrasing: “How important is it to you to be a voter?” Then they checked the official turnout records.

In the 2008 election in California, the verb group turned out at 81.8 percent and the noun group at 95.5 percent. In the 2009 New Jersey governor’s race, 79 percent versus 89.9 percent. Eleven to fourteen points from swapping a verb for a noun. The authors called it one of the largest experimental effects ever observed on turnout, and the idea spread fast. It turned up in books, in trainings, on yard signs. The hero image on one of our own posts says “Be a Voter.” We’re not above it.

The replication

The two experiments had 88 and 214 participants. That’s small, and the turnout in the verb groups was already around 80 percent, which is not what most of our lists look like.

In 2016, Alan Gerber, Gregory Huber, Daniel Biggers, and David Hendry ran the same noun-versus-verb survey on a much larger sample ahead of the 2014 primaries in Michigan, Missouri, and Tennessee. They found nothing. Bryan, Walton, and Dweck replied that low-salience primaries were the wrong setting, because a person’s identity as a voter only matters when the election feels like it matters. So Gerber and colleagues ran it again in 2015, in contested governor’s races in Kentucky, Louisiana, and Mississippi, and in a Houston mayoral race. Nothing. Then they ran a preregistered version during the 2016 presidential election, including competitive states, where salience could not have been higher. Still nothing.

That’s the straight version of the story. One small study found a huge effect. Three large studies, designed to answer each objection in turn, found none. You should treat the “be a voter” wording as harmless and unproven, not as a tool. The general lesson is worth keeping too: when a tiny study reports an effect bigger than door-to-door canvassing, wait for the big study.

What does hold up

Identity still matters. The part that fails is the grammar trick. The part that holds is telling people, truthfully, that they belong to a group of people who vote.

Alan Gerber and Todd Rogers tested this directly in a 2009 paper. Campaigns love to say that turnout is low, because it sounds urgent: “Most people won’t vote in this election. Don’t let them decide for you.” Gerber and Rogers found that messages emphasizing high expected turnout motivated people more than messages emphasizing low turnout, and that the difference showed up mainly among people who vote infrequently or occasionally. Which is to say, among exactly the voters our programs are written for.

The logic is simple once you see it. Telling someone that most people like them don’t vote tells them that not voting is normal. They were looking for permission, and you just gave it to them. We touched on this in the low-propensity post: civic-duty scolding confirms that politics is for other people. Descriptive norms do the opposite, but only if the norm you describe is participation.

The second thing that holds up is the neighborhood. The strongest effects in last week’s social-pressure research came from the block, not the nation. People care what the people near them do. A handwritten card from someone who signs off as a neighbor in the same county is already an identity appeal: you and I live in the same place, and people here vote.

Safe language and cute language

Here’s a way to sort it. Safe identity language describes a real group the voter belongs to and makes a true claim about it. Cute identity language tries to make the voter feel something with a slogan. Safe language has evidence behind it. Cute language takes up room.

Safe:

People in this neighborhood vote, even in the midterms. I’m one of them, and I hope you’ll join us on November 3.

You voted in 2024, so you already know where your polling place is. This year matters just as much.

I’m writing from the other side of the county. Folks out here are voting this fall, and I wanted to make sure you knew the date.

Cute:

Be a voter!

Your voice matters.

Voters like you make the difference.

None of the cute lines will hurt you. But a postcard has about sixty words on it, and the reading science says the voter will read the first sentence and skim the rest. Spend those sixty words on things with evidence behind them: a thank-you, a plan, a true statement that the people around them vote.

Two rules for the true statement. Never write that turnout is low, even when it is, and never write that turnout is high when it isn’t. “Most people on your street voted last time” is a claim about a specific street, and you probably can’t back it up. “People in this neighborhood vote” is a claim you can stand behind anywhere, because some of them do, and you’re one of them.

Next week: persuading someone who doesn’t already agree, and why the calendar decides whether that’s possible.