Door knocking for local campaigns

The modern evidence starts with a randomized field experiment in New Haven before the November 1998 election. Gerber and Green assigned about 30,000 registered voters to receive nonpartisan get-out-the-vote contact by canvass, mail, or commercial phone bank, then checked who voted. Turnout was raised substantially by personal canvassing, slightly by direct mail, and not at all by the phone calls.

The finding held in local elections. Before the November 2001 municipal elections, Green, Gerber and Nickerson ran door-to-door experiments in Bridgeport, Columbus, Detroit, Minneapolis, Raleigh, and St. Paul. Canvassers reached about 30% of the voters on their lists, and among voters actually contacted, turnout rose by an average of 7.1 percentage points, roughly one additional vote for every 14 conversations. In Bridgeport, where control-group turnout was 9.9%, the effect of contact was 14.4 points. The lower the baseline turnout, the more a knock is worth.

The effect spreads inside the house. Nickerson found in two experiments that about 60% of the turnout boost passes to the other registered voter in a two-voter household, even though the canvasser only spoke to one of them. Talk to whoever answers; the household gets the message.

Be honest about what doors do not do. Kalla and Broockman's meta-analysis of 49 field experiments found the average persuasive effect of campaign contact in general elections was zero, with exceptions when contact came very early or the candidate held unusual positions. Doors are for finding your supporters and getting them to vote, and for hearing what the district is worried about. Treat any persuasion as a bonus.

Start with the voter file for your district and score every voter on the last four comparable elections; how to read a voter file covers the fields and the scoring. Then decide who you are knocking and why. Before ballots or early voting start, the list is your persuasion and identification universe: likely voters, the 2-of-4 through 4-of-4 tiers, whose support you do not know yet. After identification, the list narrows to supporters and leaners who need a turnout push, the 1-of-4 and 2-of-4 tiers first. How many votes to win a local election explains the tiers.

Collapse voters into households. Two registered voters at one address is one door, and the record should show both names so the canvasser can ask for either. Drop cancelled and deceased records, drop anyone who has already returned a ballot in a vote-by-mail state, and flag inactive registrations so the walker knows the address may be stale.

Order the list by walking order, not alphabetically. Odd numbers up one side of the street and even numbers back down the other is the standard, and a list that follows it saves a volunteer twenty minutes of criss-crossing per shift.

• Filter: active registrations in the district, in the turnout tiers you are targeting, not yet contacted this phase.
• Household: one row per address with every targeted voter listed.
• Exclude: cancelled, deceased, moved, do-not-contact, already returned (in mail states).
• Sort: street by street in walking order.

A turf is one shift's worth of doors in one contiguous area: typically 40 to 80 households for a two- to three-hour shift, depending on density. Cut it so a walker never has to cross a highway, a rail line, or a mile of empty road. Start and end near the parking spot. Keep apartment complexes as their own turfs, because gated and locked buildings need a different plan and often a resident to let you in.

Precincts are a convenient starting unit because past results are reported by precinct, but they are often too big for one shift. Split them along natural lines and name each turf after the neighborhood or the main street, so a volunteer can find it on a map and a manager can talk about it in a meeting. Keep a count of doors and target voters per turf and hand the densest turfs to your fastest walkers.

The whole job at the door is to say who you are, ask one question, and write down the answer. A script that fits in 30 seconds respects the voter's time and gets the canvasser to the next door. Here is a model for the identification phase: Hi, I'm Dana, a volunteer with Maria Lopez's campaign for city council. She's running to fix the drainage on the east side and keep the library open on Sundays. Can she count on your support on November 3? Then stop talking and listen.

Have three follow-ups ready and no more. If the voter is supportive, ask if they would take a yard sign or volunteer. If undecided, ask what issue matters most to them and write it down verbatim. If opposed, thank them and leave; there is no version of an argument on a porch that wins a vote. In the turnout phase, the question changes: Do you know where you vote, and what time works for you on Tuesday? In vote-by-mail states it is: Have you returned your ballot yet? Ballot chase in vote-by-mail states covers that version.

Leave literature at every door, answered or not. The piece needs your state's required disclosure statement; Texas requires the words political advertising or pol. adv. plus the name of who paid, per the Texas Ethics Commission, and California requires Paid for by and the committee name, per the FPPC. Campaign finance basics has more on disclaimers. Never put anything inside a mailbox; it is federal property and postal carriers will remove it.

The data you bring back is the entire return on the shift. At minimum, record a support ID on a simple scale, whether the voter wants a yard sign, whether they will volunteer, and the top issue they raised. Record the outcome for every door, including not home and moved, so the next list is cleaner than this one and so you know your contact rate. A door with no outcome recorded is a door that gets knocked twice.

Use the same codes for everyone. A 1-to-5 support scale where 1 is strong support and 5 is strong opposition is the convention most tools and consultants share, which matters when you compare your numbers to someone else's or hand a list to a phone bank. Record moved and deceased carefully; those are corrections to the voter file and they save your mail budget.

Canvassers are guests on private property, and every interaction reflects on the candidate. Walk in pairs in unfamiliar areas, especially for the first shift and after dark. Carry a phone, a charged battery, water, and the campaign's number. Stay on the walkway, knock or ring once, step back from the door, and give people space. Never enter a home, even when invited, on a campaign shift. Skip any door with a no soliciting sign if your local rules or your judgment say to, and always skip a door where someone is clearly not in a state to talk.

Respect refusals instantly. Mark the voter as do not contact if they ask not to be contacted again, and make sure that flag actually suppresses them from the next list and from mail and phones; a voter who asks to be left alone and gets a mailer the next week tells everyone they know. Wear a name tag or campaign shirt, do not argue, do not comment on the house or the neighborhood, and thank people who are opposed as warmly as people who are supportive. Dogs, sprinklers, and porch cameras are all part of the job; record the dog for the next walker.

• Pairs in new turf; check in with a captain at the start and end of every shift.
• Knock or ring once, step back, wait, leave literature, move on.
• Never go inside. Never argue. Never enter a mailbox.
• Honor do-not-contact everywhere, not just on the walk list.
• Report any threatening encounter to the campaign the same day.

People are home on weekend afternoons and weekday evenings. The productive windows most campaigns use are Saturday and Sunday from about 10 a.m. to 6 p.m., and weekdays from about 5 p.m. until dark. Avoid dinner hour if you can, avoid anything before 10 a.m. on a weekend, and stop when it is dark enough that a stranger on the porch is unwelcome. In hot climates shift the weekend window earlier; in the last two weeks before an election, every daylight hour counts.

Contact rates and doors per hour are the two numbers that turn a walk list into a schedule. Green, Gerber and Nickerson's canvassers reached about 30% of the voters assigned to them across six cities, ranging from 14% in Columbus to 45% in Raleigh, so plan on one conversation for every three doors and celebrate anything better. Doors per hour depends on density; the figures below are practitioner rules of thumb, not experimental findings, and your own first two shifts will replace them with real numbers.

Paper works. A printed walk list with a column for each code, a clipboard, and a pen has run winning campaigns for a century, and it never runs out of battery. Its costs come afterward: somebody has to type every sheet into the voter file, errors creep in, and a lost clipboard is a lost shift. Paper is the right choice for a small race with a few volunteers and one organized person doing data entry the same night.

A mobile app puts the list, the map, and the script on the volunteer's phone, records each answer as it happens, and writes it back to the voter record so the next list already reflects today's doors. Some apps verify location at each stop, which tells the manager which doors were actually knocked and is the only real defense against a volunteer who fills in a sheet from a coffee shop. The costs are onboarding and the assumption that every volunteer has a charged smartphone and signal. Whatever you choose, the test is the same: is every answer in the voter file within 24 hours?

• Paper: no setup, no battery, no signal; slow and error-prone data entry.
• App: instant sync, maps, verified visits; needs onboarding and a phone per walker.
• Either: same codes, same script, same rule that data comes back the same day.

The knock is the beginning of a relationship, not the end of one. Within a day or two, send every supporter a thank-you and every yard-sign request to the sign crew. Call every volunteer prospect within 48 hours while they still remember the conversation. Put every issue note in front of the candidate weekly; the best lines in a campaign's mail come from what people said on their porches.

Then plan the second pass. Voters who were not home get another knock on a different day and time. Undecideds get a piece of mail on the issue they named and, if the budget allows, a second visit from the candidate personally. Supporters move onto the turnout list and, in the final week, get a reminder by text, phone, or a third knock. Every pass is smaller and more targeted than the last, which is the whole point of recording the outcomes. How local elections work covers where those final-week pushes fit in the calendar.

1. 1.Gerber and Green (2000), The Effects of Canvassing, Telephone Calls, and Direct Mail on Voter Turnout, American Political Science Review
2. 2.Green, Gerber and Nickerson (2003), Getting Out the Vote in Local Elections, Journal of Politics (contact rates, 7.1-point effect)
3. 3.Yale ISPS, data and publication record for the six-city canvassing experiments
4. 4.Nickerson (2008), Is Voting Contagious? Evidence from Two Field Experiments, American Political Science Review
5. 5.Kalla and Broockman (2018), The Minimal Persuasive Effects of Campaign Contact in General Elections, American Political Science Review
6. 6.Green and Gerber, Get Out the Vote: How to Increase Voter Turnout (Brookings, 2019 edition)
7. 7.Texas Ethics Commission, Political Advertising: What You Need to Know (disclosure on literature)
8. 8.California FPPC, campaign advertising requirements and restrictions
9. 9.NCSL, Access to and Use of Voter Registration Lists (building the walk list from the file)