Does Door Knocking Work? What the Research Says
Yes, door knocking works well for one job that researchers have measured many times: getting people to vote. In randomized field experiments, where researchers choose at random which households receive a visit and then check official voting records, a face-to-face conversation with a canvasser has raised turnout in a range of elections. The evidence on changing people's minds is weaker. A large review of general election experiments found that the average persuasive effect of campaign contact, door-to-door canvassing included, was zero.
That gives a canvassing program a practical starting point. Plan door knocking around mobilization (reminding and encouraging people to vote), expect modest persuasion in general elections, and measure what your own program does. Below are four published studies and what they mean for planning. For the full planning process, see our guide to planning and running a canvassing campaign.
How researchers test door knocking
In a field experiment, researchers divide a list of registered voters at random into two groups. Canvassers visit the first group (the treatment group), and nobody visits the second (the control group). After the election, the researchers check official records, which show whether each person voted but never how. Because chance alone decided who got a visit, a difference in turnout between the groups can be credited to the canvassing.
Two terms come up in every study. The contact rate is the share of the treatment group that canvassers actually reached. The effect is usually reported in percentage points, the plain difference between two percentages. For example, if turnout rises from 45 percent to 50 percent, that is a 5-point increase.
New Haven, 1998: canvassing, mail and phone calls
Alan Gerber and Donald Green ran the study that started much of this research, published in the American Political Science Review in 2000. They worked with about 30,000 registered voters in New Haven, Connecticut, and sent nonpartisan get-out-the-vote messages by personal canvassing, direct mail or telephone before the November 1998 election.
Turnout rose substantially with personal canvassing, slightly with direct mail, and not at all with the telephone calls in the study. The authors estimate that among the people canvassers actually reached, a face-to-face conversation raised the probability of voting by 8.7 percentage points. Canvassers reached about 28 percent of the people they set out to visit, so finding people at home was the main limit on the program.
The canvassers used three nonpartisan appeals: civic duty, neighborhood solidarity and the chance of a close election. The authors could not show that any one of these worked better than the others.
Six cities, 2001: the same test in local elections
Donald Green, Alan Gerber and David Nickerson repeated the design in six cities before the November 6, 2001 elections and published the results in the Journal of Politics in 2003. The cities were Bridgeport, Columbus, Detroit, Minneapolis, Raleigh and St. Paul, and the canvassers came from nonpartisan student and community groups.
Combining all six sites, the authors estimate that a conversation with a canvasser raised the chance of voting by about 7.1 percentage points among the people reached. The results varied by city. In Raleigh the effect was close to zero, while four sites showed larger estimates than New Haven, although some of those estimates were imprecise. Canvassers reached about 30 percent of the people on their lists, so the increase across everyone assigned to a visit was smaller: about 2.1 percentage points.
The authors also offer a planning figure. In these campaigns, about 12 successful face-to-face conversations produced one additional vote. Your own number will differ. The figure is still a useful reminder that turnout depends on real conversations, so reaching people at home matters as much as covering ground.
Persuasion in general elections: 49 experiments
Persuasion has proved much harder to show in these experiments. Joshua Kalla and David Broockman reviewed the evidence in "The Minimal Persuasive Effects of Campaign Contact in General Elections," published in the American Political Science Review in 2018.
First, they combined 40 earlier field experiments in a meta-analysis (a study that pools the results of many studies). The average effect of campaign contact on candidate choice in general elections was zero. Second, they ran nine original experiments focused on door-to-door persuasion in general elections. Their average effect was also zero.
The authors found two rare exceptions. Persuasion appeared when candidates took unusually unpopular positions and campaigns invested heavily in finding persuadable voters. It also appeared when campaigns contacted voters long before Election Day and measured the effect right away, but that early persuasion faded before the election. The authors also report that campaign contact could have persuasive effects in primaries and ballot measure elections, where voters do not have a party label to go on. Their evidence does not say how a campaign should split its effort among registration, turnout and persuasion.
A conversation method: the 2016 study in Science
A fourth study tested a way of talking at the door rather than a campaign for a candidate. David Broockman and Joshua Kalla published "Durably reducing transphobia: A field experiment on door-to-door canvassing" in Science in 2016. After Miami-Dade County passed an ordinance protecting transgender people from discrimination, 56 canvassers talked with 501 voters in South Florida about their attitudes toward transgender people. The conversations lasted about 10 minutes and asked voters to take another person's perspective, in part by recalling a time when they had been judged negatively themselves.
The randomized trial found that the conversations changed attitudes and that the change lasted at least three months. The conversations also increased support for a nondiscrimination law, even after voters were shown arguments against it.
For a field program, the finding of interest concerns method. A single conversation built around listening and perspective-taking produced a lasting change in attitudes in this setting. It was one issue in one place, so treat it as evidence about the method rather than a promise for every topic. Practitioners often call this approach deep canvassing, and our post on deep canvassing explains how those conversations are structured.
What the research means for your canvassing plan
- Use door knocking where turnout is the goal. The most consistent finding in these studies is that a visit raises the chance that a person votes.
- Plan for contact as well as coverage. In the two turnout studies, canvassers reached roughly 28 to 30 percent of the people on their lists. Plan return visits at different times of day to reach more of them. Our field plan guide shows how to turn this into a door count and a schedule.
- Be realistic about persuasion. Expect small effects close to a general election. Primaries, ballot measures and longer listening conversations are where the research has found persuasion more often.
- Measure your own program. Every study above reports averages from particular places and years.
What about door-to-door sales and nonprofits?
The published experiments in this post are about voting and political attitudes. They do not measure sales, appointments, sign-ups or donations, so they cannot tell a business or a nonprofit how well door knocking will work for it.
What a business or nonprofit can borrow is the method. To test a new script or time of day fairly, use it on some streets and not on similar streets nearby, and then compare the results.
Measure your own results in Beacon
You can only learn your own contact rate if every door is recorded. In Beacon, a door-to-door canvassing app that canvassers open on their phones, the steps are:
- Draw your turf. Turf is the set of streets one canvasser covers. Draw it on the map and assign it.
- Send out walk lists. A walk list is the list of doors to visit. When a canvasser downloads it, the stops are in walking order by distance.
- Record a result at every door. Canvassers choose from eight result codes (contacted, not home, refused, moved, deceased, inaccessible, come back later and left literature) and can add a 1 to 5 rating and a note. Your contact rate is the share of doors marked "contacted".
- Put one question in the script, such as "Planning to vote? Yes / Not sure / No", and record the answer in the notes the same way at every door.
- Watch progress. Turf completion percentages, a leaderboard and heat views of the results show where conversations are happening and where a return visit is needed.
- Export your results. Download door results as CSV or PDF, and compare them with voter history after the election, when it becomes available.
Our page on canvassing software for political campaigns shows how these steps fit into a full campaign.
Common questions
Is door knocking better than phone calls?
In the New Haven experiment, personal canvassing raised turnout and the telephone calls tested did not. That is one study of one calling program, so it cannot tell you how every phone bank performs.
How many doors does it take to gain one vote?
It depends on how many people you reach. In the six-city study, the authors estimated that about 12 successful face-to-face conversations produced one additional vote. Because canvassers reached only about 30 percent of the people on their lists, that took many more than 12 knocks.
Does door knocking change minds?
Rarely in general elections, where the average persuasive effect across the 49 experiments was zero. Persuasion has been found more often in primaries, ballot measures and longer listening conversations.
Measure what works on your own streets
Every study above reports averages from particular elections. To learn what works where you canvass, start a 14-day Beacon trial and record the results of your first shift. The trial includes the Starter plan's features, and you do not need a card to begin.
Sources
- Alan S. Gerber and Donald P. Green, "The Effects of Canvassing, Telephone Calls, and Direct Mail on Voter Turnout: A Field Experiment," American Political Science Review 94, no. 3 (2000), pages 653 to 663. Journal page. Free copy (Yale ISPS).
- Donald P. Green, Alan S. Gerber and David W. Nickerson, "Getting Out the Vote in Local Elections: Results from Six Door-to-Door Canvassing Experiments," Journal of Politics 65, no. 4 (2003), pages 1083 to 1096. Journal page. Free copy (Yale ISPS).
- Joshua L. Kalla and David E. Broockman, "The Minimal Persuasive Effects of Campaign Contact in General Elections: Evidence from 49 Field Experiments," American Political Science Review 112, no. 1 (2018), pages 148 to 166. Journal page.
- David Broockman and Joshua Kalla, "Durably reducing transphobia: A field experiment on door-to-door canvassing," Science 352, no. 6282 (2016), pages 220 to 224. Journal page.
