The Impact of Personal Conversations on Voter Behavior in France
Electoral campaigns often use door-to-door canvassing as a way to persuade voters. However, little is known about the effect of these discussions on political participation and support. To fill this gap, researchers conducted a randomized evaluation of a countrywide canvassing effort in favor of a French political party before France’s 2012 elections. Results show that canvassing visits did not affect voter turnout, but they increased the vote share of the supported candidate. This impact largely persisted in later elections, suggesting that brief one-on-one discussions can be persuasive in a lasting way
Policy issue
In well-functioning democracies, open discussions around political issues are crucial to encourage participation and ensure citizens’ preferences are part of the political debate. To leverage the power of personal discussions, electoral campaigns increasingly use door-to-door canvassing to influence voters’ beliefs and opinions about candidates and increase voter turnout. However, little is known about how these discussions affect individuals’ vote choice. Recent evidence has highlighted the role of door-to-door canvassing on encouraging voters to “get-out-the-vote”.1 Can door-to-door canvassing influence voters’ opinions on candidates and final voting decisions?
Context of the evaluation
France held elections in 2012 to elect both a new president and a new National Assembly (the lower house of the French Parliament). Presidential elections in France have two rounds, with the two candidates who receive the highest vote shares in the first round advancing to the second. Voter turnout that year was high in both rounds: 79.5 percent and 80.4 percent, respectively. Nicolas Sarkozy of the right-wing Union pour la Majorité Présidentielle (UMP) and François Hollande of the left-wing Parti Socialiste (PS) progressed to the second round and Hollande won the second round with 51.6 percent of the votes. The 2012 parliamentary elections, which also have two rounds, took place about one month after the presidential elections concluded. The PS candidates won in 49 percent of the constituencies. French voters also participated in the 2014 elections to the European Parliament, where the PS suffered a major defeat.
In French electoral campaigns, the government mandates that radio and television channels give equal coverage to each candidate. French parties allocate few resources to the recruitment, training, and coordination of activists, and door-to-door canvassing is historically quite rare. However, the PS decided to use canvassing as a major strategy during their 2012 campaign.
Details of the intervention
The researcher partnered with the PS to conduct a randomized evaluation to test the impact of a national door-to-door canvassing campaign on voter turnout and voter choices. The campaign ran from February to May of 2012, covering both presidential election rounds.
Before canvassing began, the research team divided France into 3,260 territories and each territory into precincts based on postal codes. Within each precinct, they calculated a target number of doors based on the potential to win PS votes. Afterwards, they randomly assigned precincts in each territory with the highest potential to win votes to either receive canvassing visits or serve as the comparison group.
Canvassers were both volunteers and official members of the PS, with varying levels of canvassing experience. The party had little authority over local canvassers, making this a good representation of real-world canvassing campaigns. National trainers traveled across France to deliver an identical training course to field organizers, responsible for coordinating and training the canvassers. The course encouraged canvassers to engage in interactive discussions with voters, provide basic information about the date of the election, the location and opening times of the poll office, and the name of the PS candidate. At the end of the discussion, canvassers gave voters some campaign literature, e.g., a thematic leaflet or a booklet summarizing the candidate’s platform. When no one opened, they left a leaflet or doorhanger on the door
During the campaign, which was the largest door-to-door effort in Europe to date, around 80,000 PS volunteers knocked on a total of five million doors. The research team used data tracked by volunteers and field coordinators to monitor the campaign progress. In addition, they used administrative records from the 2012 presidential and parliamentary elections, as well as the 2014 European election, to measure the impact of the campaign on voter turnout and vote share.
Results and policy lessons
Results show that door-to-door canvassing visits did not affect voter turnout, but they increased the vote share of the PS presidential candidate. This impact largely persisted in later elections, suggesting that brief one-on-one discussions can be persuasive in a lasting way.
Not all canvassers followed the allocation of precincts into different groups: While the canvassing was part of a national campaign covering 3,260 different territories, canvassers only adhered to the randomized list of precincts in one-fourth of territories. The results of the study are, therefore, based on the 791 territories containing 4,674 precincts and 5.02 million voters. Estimates including all 3,260 territories found no significant effects of canvassing on any measured outcome.
Door-to-door canvassing did not impact overall voter turnout, but it increased the share of votes of the PS candidate: Within the territories that did adhere to the randomized list of precincts, the canvassing campaign had no effects on voter turnout in either presidential election round. Canvassing did, however, result in a 3.24 percentage point increase boost to François Hollande’s share of votes in the first round and a 2.75 percentage point boost in the second round, in precincts that were allocated to canvassers. Taking the total number of doors knocked at the national level, the campaign accounted for around one half of Hollande’s lead in the first round and one fourth of his victory margin in the second round.
The increase in PS’s vote share came from lower vote shares for right-wing candidates: Researchers estimate that the door-to-door visits led to a 3.14 percentage point reduction in the vote share of the right-wing candidates Nicolas Sarkozy and Nicolas Dupont-Aignan, in precincts that were allocated to canvassers. The campaign did not affect vote shares for left, center, or far-right wing candidates, suggesting that canvassers persuaded undecided voters to support Hollande by changing their beliefs about the quality of this candidate, rather than changing their policy preferences.
The positive effects of canvassing persisted through time: Similar to the presidential election, door-to-door canvassing had no effect on voter turnout for the 2012 parliamentary elections, which happened one month later. It did, however, lead to increased vote shares for PS candidates in the first and second rounds of the election: 0.94 and 0.73 percentage points, respectively, in precincts visited by canvassers (a 2.9 percent and 1.6 percent increase over the comparison group, respectively). Canvassing also had an impact on the 2014 European elections, boosting the share of PS votes by 0.37 percentage points (a 2.6 percent increase).
Overall, the results suggest that in elections of very high salience, such as presidential elections, voter outreach methods might have little effect on turnout, but that interpersonal discussions can have a large and long-lasting persuasion effect.
Following this evaluation, door-to-door electoral campaigns have widely diffused within France. This included the 2012 parliamentary elections, which took place just one month after the presidential election, with candidates both on the left and on the right knocking on doors to mobilize nonvoters and persuade undecided voters.
Pons, Vincent. 2018. “Will a Five-Minute Discussion Change Your Mind? A Countrywide Experiment on Voter Choice in France.” American Economic Review, 108(6): 1322-1363.