THE
REPUBLIC
BELOW
← all the receipts

the receipts · 3 of 7

two doors down

A short conversation with a neighbor goes better than people expect, and listening first is what moves anyone.

Controlled15 minuteslights the neighbor's windows

Why it's in the game

People consistently predict that talking to someone they barely know will be awkward or unwelcome. When researchers actually made them do it, it went better than they guessed, for both people.

The field experiments on persuasion found the same shape. Arguments alone did nothing. A short exchange where each person listened and shared something real shifted views for months.

The effects are small, and they come from trained canvassers talking to strangers. This action borrows the method for a neighbor and one agenda item, which nobody has tested.

the study that started it

The quiet train car

Nicholas Epley and Juliana Schroeder asked commuters at a train station outside Chicago to do one of three things on the ride in. Some were told to talk to a stranger, some to sit alone, and some to ride as usual. The talkers reported the most pleasant commute, though people had predicted the opposite.

The receipts

Every source was checked against the original on September 25, 2026. Each one links to it.

  1. d = 0.63talking vs. sitting alone

    Epley & Schroeder (2014) Journal of Experimental Psychology: General 143, 1980–1999

    Randomized field study · 97 commuters

    Talking with a stranger made for a more pleasant commute than sitting alone.

    read it at faculty.haas.berkeley.edu ↗
  2. Sandstrom & Dunn (2014) Personality and Social Psychology Bulletin 40(7), 910–922

    Diary studies

    Students felt happier and more like they belonged on days they talked with more classmates than usual.

    Observational, and the abstract gives no sample sizes.

    read it at journals.sagepub.com ↗
  3. 60%of the turnout effect spread to the other voter

    Nickerson (2008) American Political Science Review 102(1), 49–57

    Placebo-controlled canvassing · two-voter households

    When one person in a household was canvassed, most of the effect passed to the person who didn't answer the door.

    read it at cambridge.org ↗
  4. 3 monthshow long the change held

    Broockman & Kalla (2016) Science 352(6282), 220–224

    Randomized canvassing trial · 501 voters

    Roughly ten-minute conversations reduced transphobia, and the change lasted.

    read it at escholarship.org ↗
  5. d = 0.04–0.08small, but durable

    Kalla & Broockman (2020) American Political Science Review 114(2), 410–425

    Three field experiments · 6,869 voters

    Arguments alone did nothing. Adding a nonjudgmental exchange of personal stories reduced exclusionary attitudes for at least four months.

    read it at isps.yale.edu ↗

The honest part

The well-being effects are about strangers and classmates. The persuasion effects come from trained canvassers at strangers' doors, and the durable effects are small.

The 60% spillover is about turnout inside a household, not neighbors next door. The 2016 study came after an earlier, similar paper by other authors was retracted, so this line of work has been checked hard. Nobody has tested one unscripted talk between neighbors about a local agenda item.

What you do in the game

  1. The neighbor's door on the street. Knock twice and it opens on the script: open with the item, ask one question, listen first.
  2. Don't correct anything, and don't pitch. If it goes badly, hold the door: "Thanks, I'll let you go" is a whole sentence.
  3. Not today? Pick a time, put it on your calendar and tap I'll knock. Already talked? Write one line about what they said. It stays on your phone.
  4. Their windows light up.

How strong is the evidence?

Controlled, since the conversation itself was randomized in several studies, though in different settings.

Practice
Case histories and experience. Nobody has measured it with a control group.
Field
Real-world data: a natural experiment, a panel, or careful observation. Strong, but not randomized.
Controlled
People were randomly assigned to do it or not, and the difference was measured.

Back to the game