Voting "Because Others Will Ask" in the United States

Fieldwork by:
United States of America
13,197 households
2011 - 2012
Target group:
  • Voters
Outcome of interest:
  • Electoral participation
  • Voter Behavior
Intervention type:
  • Community participation
  • Social networks
  • Community monitoring
AEA RCT registration number:
Research papers:

Social image may be a factor in individuals' decision to vote. Researchers conducted a series of randomized evaluations to measure the impacts of a get-out-the vote campaign and the value voters place on telling others they voted. Informing potential voters that they would be asked after the election whether they voted modestly increased voter turnout. Results indicate that individuals greatly dislike admitting they did not vote, and also dislike lying and claiming to be a voter. This provides a motivation to vote simply "because others will ask." 

Policy issue

Interventions to increase voting rates have attracted substantial attention from researchers and policymakers alike. One potentially promising means of increasing voting rates is by leveraging social image—if people care that others think they vote, they may be likelier to vote. For example, previous research demonstrates that letters to individuals comparing their own voting record to that of their neighbors increased voter turnout by eight percentage points.1

Outside of formal interventions, social pressure to vote may occur more naturally through friends and family asking people whether or not they voted. This may be a powerful motivator, and people may vote in part to be able to tell others that they voted. How much do people care about telling others that they have voted, and might this be a means of encouraging them to vote?

Context of the evaluation

In the largely Democratic suburbs of Chicago in which these evaluations were conducted, voter turnout was 60 percent in the 2010 election and 73.1 percent in the 2012 election. The 2010 congressional election in Illinois was disappointing for Democrats, including the loss of the Illinois Senate seat previously held by President Obama. The secondary evaluation involved get out the vote campaigns during both the 2010 congressional election and the 2012 presidential election. The households contacted had a relatively high median income of $92,000, and income is strongly and positively correlated with voter turnout.

Details of the intervention

Researchers conducted a series of randomized evaluations to measure the value voters place on telling others they voted and the impacts of a get-out-the vote campaign leveraging social image to increase voter turnout. Using voting records from the 2010 election, researchers restricted the sample to households where all or no registered voters voted, excluding households with a mixed turnout record in 2010. Researchers then randomly assigned households to receive variations of incentives to complete a survey about their participation in the 2010 congressional elections.

Households received one of four variations of a flyer announcing an upcoming survey, or to serve as a control group:

  • Survey flyer: Households received a flyer stating that researchers would visit to conduct a survey the next day.
  • Election flyer: Households received a flyer indicating the survey would cover households’ voter participation in the 2010 congressional election.
  • Opt-out flyer: Households received the survey flyer with an added do-not-disturb check box allowing them to opt-out of the survey.
  • Election opt-out flyer: Households received an election flyer with an added do-not disturb check box.
  • No flyer

In addition to the variation in fliers, some households were randomly assigned to receive a $10 incentive to complete the survey and researchers varied the announced duration of the survey. Finally, to measure how much voters value telling others they voted, researchers informed some respondents that the duration of the survey would be reduced by eight minutes if they reported not having voted in the 2010 election and offered a $5 incentive for reporting non-voting to others.

For each group, researchers tracked survey completion rates as well as reported voter participation—whether people say they voted or not. Researchers then checked voter responses against actual voter participation records to determine truthfulness.

In a related evaluation, researchers designed a get-out-the-vote program seeking to increase voting rates by leveraging individual’s desire to tell others they voted. During the 2010 and 2012 elections, an additional set of households were randomly assigned to receive flyers that reminded households that there was an election approaching, flyers noting that researchers would contact households shortly after the elections to survey residents about whether they had voted, or to serve as a comparison group. Researchers then evaluated whether informing potential voters that they would be asked about their electoral participation increased voting participation.

Results and policy lessons

The researchers find that individuals much prefer to be seen as a voter relative to a non-voter—by $7 on average each time they are asked. This is because they greatly dislike admitting they did not vote, and also dislike lying in order to falsely claim they voted. On average, for the 2010 Congressional Election, individuals are calculated to have an average incentive to vote of $15 due to being asked about 5 times whether they voted. The related get-out-the-vote intervention—which increases the times asked by at most one—led to modest increases in voter participation in the 2010 Congressional election, and no detectable increase in the 2012 Presidential election. However, both voters and non-voters on average place a negative value on being asked whether they voted. Thus, while such interventions may boost turnout—at least in Congressional elections—and thus have large societal benefits, the targets of the intervention might suffer welfare losses.

In the get-out-the-vote intervention, telling individuals before the election that they would be surveyed about voter participation led to an increase in electoral participation. In 2010, the campaign increased voter turnout by 1.3 percentage points relative to the control group turnout of 60 percentage points. Simply sending a reminder of the upcoming election had no effect on turnout. The campaign had no effect in the 2012 election, which had higher turnout overall.

Survey flyers: Advising nonvoting households that the survey included questions on voter participation led to a six percentage point (20 percent) decline in survey completion. However, providing this information did not change survey completion among voting households. These results suggest that non-voting households feel shame at admitting to not having voted, but that voting households may not feel pride.

Compensation for survey completion: Reducing payment for the survey from ten dollars to nothing significantly decreased survey participation. Informing non-voters that the upcoming survey included questions on voter participation had a similar negative effect. This finding further suggests that non-voters substantially dislike being asked about their voting habits.

Survey length: Non-voters are significantly more willing to alter their answers to the survey question than voters. When incentivized to respond that they did not vote in the 2010 elections, the portion of non-voters who (truthfully) answer that they did not vote in 2010 increases by 12 percentage points. When receiving this same incentive to report non-voting, however, voters do not change their responses significantly. Thus, because voters continue to report that they voted even when incentivized to lie, they reveal a strong preference for saying that they voted.

Gerber, Alan S., Donald P. Green, and Christopher W. Larimer. (2008). “Social Pressure and Voter Turnout: Evidence from a Large-scale Field Experiment,” American Political Science Review, 102(1): 33-48. (