Getting Out the Vote in Local Elections in the United States

Researchers:
David W. Nickerson
Fieldwork by:
Location:
Bridgeport, Columbus, Detroit, Minneapolis, Raleigh, and St. Paul, United States
Sample:
18,933 voters
Chronologie:
2001 - 2001
Target group:
  • Urban population
  • Voters
Outcome of interest:
  • Electoral participation
Intervention type:
  • Nudges and reminders
Partners:

Low rates of voter participation are common in US elections, especially in local elections despite the fact they are often directly relevant to voters. Researchers examined the impact of door-to-door canvassing on voter turnout in the 2001 local elections in six US cities. The results indicate that voters who were contacted by canvassers were significantly more likely to vote in a wide variety of settings.

Policy issue

Americans are given more opportunities to cast ballots than citizens of any other country, including in state, municipal and primary elections. However, voter participation rates in the US are very low, particularly in off-year elections when local and state-level candidates are on the ballot. The typical US municipal election draws only one-fifth to one-half of all the registered voters. A previous evaluation conducted in New Haven, Connecticut, found that door-to-door canvassing significantly increased voter turnout. In this study, researchers evaluated the impact of similar door-to-door canvassing campaigns in six US cities: Bridgeport, Columbus, Detroit, Minneapolis, Raleigh, and St. Paul.

Context of the evaluation

The door-to-door canvassing campaigns evaluated in this study took place in six different political and demographic settings. Some elections were tightly contested; others were devoid of meaningful competition. Some sites have large populations of voters of color; others are predominantly white.

  • Bridgeport, Connecticut is a racially diverse, low-income city that votes overwhelmingly for Democratic candidates. The November 2011 election featured only local electoral races with little competition, and turnout was expected to be low.
  • In Columbus, Ohio, researchers canvassed areas near the local university where many residents were students and most were under 26 years old. Again, the November 2011 elections were local, and turnout was expected to be low.
  • Detroit, Michigan is a large city with many low-income residents. There was a closely contested mayoral election in November 2011, and greater voter interest than in other sites.
  • In Minneapolis, Minnesota, canvassers visited neighborhoods with diverse, primarily working-class residents. Several citywide electoral races were on the ballot in the November 2011 election, and turnout was expected to be moderate.
  • In St. Paul, Minnesota, the 2011 mayoral race was very closely contested. Two neighborhoods were canvassed—a racially diverse, low-income neighborhood and a primarily white, middle-income neighborhood.
  • In Raleigh, North Carolina, which also had closely contested elections, canvassers visited two neighborhoods, one racially-mixed and the other predominately white.

The aim in drawing from such a diverse collection of sites was two fold: to better gauge the average treatment effect of canvassing, and to examine whether the treatment effects varied systematically with electoral competitiveness or other characteristics of the sites.

Details of the intervention

Researchers used a randomized evaluation to measure the impact of door-to-door canvassing on voter turnout across six different US cities. From a database of registered voters, households were randomly assigned to either the treatment or comparison group. Canvassers were recruited from local chapters of the national group ACORN in Bridgeport and Detroit, and from PIRG in the other four sites. Experience among canvassers varied greatly across sites. In one case (Raleigh), canvassers were paid for their efforts; in other cases, canvassers were volunteers. Canvassers visited households in the treatment group in the days leading up to the election. They were given scripts containing a reminder about the election and non-partisan encouragements to vote. In some cases, a non-partisan flyer was also delivered.

Results and policy lessons

In every site, the treatment group had higher turnout rates than the comparison group. For example, in Detroit, where 43 percent of the control group cast ballots, turnout increased in the assigned treatment group by 2.4 percentage points. In Bridgeport, where turnout in the comparison group was very low at 9.9 percent, turnout in the treatment group was 4.0 percentage points higher. The only outlier was Raleigh where the canvassing efforts had no significant impact on voter turnout. On average, across the six sites, the researchers found that face-to-face canvassing increased turnout in the treatment group by a 2.1 percentage points. Canvassers frequently found no one at home, could not locate the address, discovered they had the wrong address, or were told to go away before they could deliver their script. In the end, they were only able to contact between 14 and 45 percent of households in the treatment group, depending on the site. Among those who were reachable at the door, canvassing led to a 7.1 percentage point increase in voter participation. This is consistent with the previous study conducted in New Haven. The researchers found that face-to-face canvassing was an effective means of mobilizing voters in a number of settings, ranging from competitive to non-competitive races and across starkly different economic and demographic voter compositions.

Green, Donald P., Alan S. Gerber and David W. Nickerson. 2003. "Getting out the Vote in Local Elections: Results from Six Door-to-Door Canvassing Experiments." The Journal of Politics 65 (4): 1083-1096.