The mailer that changed political science.

In the summer of 2006, researchers from the University of Michigan and Yale mailed 344,000 registered voters in Michigan a simple letter.

Some received a civic duty message: voting is important, your community depends on you. Some received a message that they were being studied. Some received their own voting history — a factual record of which elections they had and hadn't voted in. And some received something more unusual: their own record, alongside the voting records of their neighbors, with a note that said the records would be updated after the August primary.

The result, published two years later in the American Political Science Review, was the largest effect on voter turnout ever documented in a randomized experiment. The social pressure mailer — the one showing neighbors' records — increased turnout by 8.1 percentage points compared to the control group. The civic duty message: 1.8 points. The Hawthorne effect (being studied): 2.5 points. The self-disclosure: 4.9 points.

8.1 points from a piece of mail.

What the number means

To appreciate why political scientists were stunned, it helps to know what they had been finding before.

Throughout the 1990s, researchers had been running randomized experiments on voter turnout — testing phone banks, direct mail, and door-to-door canvassing. The canonical finding from that work, by Gerber and Green, was that in-person canvassing increased turnout by about 6 to 9 points. Phone calls produced roughly 3 points. Direct mail: almost nothing.

In-person canvassing was considered the gold standard — expensive, labor-intensive, and effective because it was human. The Michigan social pressure mailer matched or exceeded that benchmark using a piece of paper that cost cents to produce.

The mechanism wasn't inspiration or persuasion. It was accountability. Showing voters that their neighbors could see their participation history activated something more basic than civic motivation: the desire not to be seen as the person who doesn't vote.

One hundred replications

The experiment was replicated almost immediately. Then again. Then again.

By 2015, researchers had run more than 100 independent replications of social pressure GOTV interventions across different elections, states, and populations. The effect sizes were consistent: social pressure messages reliably produce turnout increases in the 4–8 point range. The finding held across primary elections, general elections, low-salience elections, and competitive elections. It held for new voters and experienced voters. It held in states where voting was easy and states where it was harder.

Within two election cycles, social pressure mailers had become standard tools in competitive campaign operations. Barack Obama's 2008 campaign used them. Nonpartisan voter mobilization groups deployed them at scale. State Secretaries of State experimented with official versions.

Then the ethics questions arrived

The backlash was swift and, in retrospect, inevitable.

The social pressure mailer worked precisely because it made voters feel watched. It created a kind of ambient accountability — a sense that civic behavior was being observed and recorded, and that inaction would be visible to neighbors. This is not a neutral nudge. It uses social surveillance as a motivation mechanism.

Critics raised several objections, each legitimate:

**Embarrassment as a tool.** The mailing's implicit message — your neighbors can see that you didn't vote — is not persuasion. It is a low-grade form of social coercion. It works by activating shame. Whether democratic participation should be produced through shame is a genuine question.

**Privacy and accuracy.** Voting history is public record, but most voters don't know that. Confronting people with the fact that their neighbors can access this information felt, to many recipients, like a violation. Complaint rates were significantly higher for the social pressure arm than for any other condition.

**The agency question.** A civic institution using social pressure to produce a particular behavior is in a different moral category than providing accurate information and letting people decide. The mailer is optimizing for turnout, not for informed participation.

Researchers themselves were divided. Gerber, Green, and Larimer acknowledged the ethical tensions in their original paper. Subsequent researchers ran experiments specifically on whether voters objected to the technique after the fact — they did, at substantially higher rates than for other GOTV approaches.

What this means for civic practice

The Michigan experiment does not resolve cleanly into a simple lesson.

If the goal is maximum turnout at minimum cost, social pressure mailers are among the most effective tools ever tested. They are also among the most coercive. These two facts are not contradictory — they are the same fact, described differently.

For practitioners designing civic behavior change campaigns, the experiment points toward a set of questions that are worth asking before reaching for any powerful tool:

**Is the behavior we're encouraging one the community has chosen?** Voting is a right, not an obligation in the United States. A campaign that uses social pressure to induce voting is making an implicit normative judgment that non-participation is a failure. Some communities will agree. Others won't.

**Would the mechanism work if it were transparent?** Most social pressure mailers do not explain, at the top, that they are designed to activate social shame. If the design had to be disclosed to be used, would it still feel appropriate?

**Are there alternatives with comparable effectiveness?** Door-to-door canvassing produces similar turnout effects through a fundamentally different mechanism — human connection and genuine conversation, not accountability and surveillance. It is more expensive and less scalable. Whether the difference in mechanism matters is a values question, not an empirical one.

The broader lesson

The Michigan social pressure experiment is the most replicated finding in civic behavioral science. That alone makes it worth understanding.

But its deeper value is what it illustrates about the relationship between effectiveness and appropriateness. The most powerful interventions often work through mechanisms that deserve scrutiny: social pressure, loss aversion, default settings, anchoring effects. They produce outcomes we may want. They do so in ways we may not have chosen deliberately.

Civic institutions that want to use behavioral science well need to do both things at once: understand what works, and ask whether how it works is consistent with the kind of civic culture they are trying to build.

An institution that increases voter turnout through shame may produce higher participation numbers while eroding something else — the sense that civic participation is freely chosen rather than socially coerced. Whether that trade-off is worth making is not a question behavioral science can answer.

Next issue: Why most job training programs show no effect — and what the ones that work have in common.