The following is an edited excerpt from Donald P. Green and Alan S. Gerber's Get Out the Vote: How to Increase Voter Turnout, available for pre-order in its fourth edition Aug. 27.
In 2005 and 2006, Elizabeth Addonizio, Donald Green, and James Glaser set out to explore the feasibility of creating a more festive and community-focused atmosphere at the polls. This line of inquiry turns the usual approach to GOTV on its head. Researchers and policymakers tend to focus on the costs of voting and ways to reduce those costs. But rarely do they address the potential benefits of casting a ballot. This line of research attempted to quantify the impact of imbuing polling places with the attractiveness of a local fair.
These poll parties, while inspired by those of the nineteenth century, departed from their historical models in significant ways. Elections and Election Day activity are highly regulated today, and most state and federal laws prohibit not only vote buying, but also any kind of quid pro quo inducement to vote.
Researchers were therefore informed by local and state officials that the parties had to be advertised and carried out in such a way as not to link the casting of a ballot with the receipt of food and entertainment. Provided that these restrictions were respected, the festivals were well within the bounds of election law. And unlike the social activities surrounding elections in the nineteenth century, which catered to the tastes of male voters, these parties were meant for general audiences, including children. The empirical question the researchers thus addressed was whether these family-friendly, alcohol-free variants of the old-time poll party raise turnout.
Over the course of two years, Addonizio and her co-authors evaluated a series of sixteen experimental festivals. These festivals were held in a wide array of locations, ranging from middle-class, all-white suburbs to poor, largely minority inner cities. All of the festivals followed a common model, which was developed in New Hampshire, where the authors conducted a pilot study before municipal elections held in the spring of 2005. The authors began with two towns—Hooksett and Hanover—that had similar populations and voting rates. A coin was flipped, and the researchers organized a party in Hooksett, leaving Hanover as the control group.
The festival was preceded by a week of publicity and local organizing. Flyers were handed out at town meetings, and posters were displayed in local stores and meeting spots. On the Saturday before Election Day, the regional newspaper included a flyer advertising an “Election Day Poll Party,” giving the location and time. The local paper also advertised the event. On the Sunday before Election Day, a story describing the party appeared in one of the papers. At the same time, three dozen lawn signs advertising the event were planted on busy streets in town. Finally, two prerecorded thirty-second phone calls were directed to 3,000 Hooksett households. Both extended an invitation to the party and gave details about its hours (3:00 p.m. to 7:00 p.m.) and location.
Election Day in Hooksett featured beautiful spring weather, perfect for an outdoor event. The festival took place immediately outside the polling place, on the front lawn of the local middle school. A large tent was set up, surrounded by signs encouraging people to enjoy free snacks, drinks, and raffles. A cotton candy machine, expertly staffed by political science professors, attracted a steady stream of children. People of all ages milled about the party tent, eating, drinking, and listening to music supplied by a local disk jockey. People at the party seemed aware of the event before coming to the polls to vote. They had read the flyer, received the calls, or heard about the various advertised activities from other residents.
Judging from the size of the crowd it attracted and partygoers’ positive evaluation of the event, the party was deemed a success. Hooksett, despite having no contested candidates on the ballot, garnered a higher voter turnout rate than Hanover. After replicating the experiment in a local election in New Haven in 2005, the authors attracted the involvement of Working Assets in 2006, which coordinated local festivals in every region of the country during the primary and general election season. In each site, target precincts were identified, some of which were randomly assigned to a treatment group in which a festival was held. Turnout rates of the treatment and control precincts were compared; because these additional sites were not matched in terms of past voter turnout rates, a multivariate analysis was conducted to control for past turnout differences across precincts.
Across all thirty-eight precincts, the results indicated that festivals—or to be more precise, the festivals and the pre-election publicity surrounding them—increased turnout by approximately 2 percentage points. Because many of these precincts were small, the festivals together generated approximately 960 additional votes at a cost of $33,168 in 2018
dollars. These figures imply a cost per vote of $35, which puts festivals on the same general plane as other face-to-face tactics.
***
Despite the promising results from these early festivals, political campaigns and nonpartisan groups seldom used festivals to bolster turnout between 2005 and 2016. The breakthrough came when a nonpartisan group called Civic Nation took an interest in promoting festivals during the 2016 presidential election. These festivals, coordinated by Edna Ishayik in collaboration with local organizations, equaled or surpassed the earlier festivals in terms of the energy that went into local canvassing to spread the word to local businesses and nearby homes. All of the festivals featured free food, and local sites offered various forms of entertainment: live broadcasts from a local radio station, dance troupes, photo booths, arts and crafts, lawn games, and puppies.
A total of nine festivals took place in an assortment of battleground (North Carolina, Ohio) and non-battleground states (California, Tennessee, Texas). All of the sites enjoyed pleasant weather. The total cost of the festivals, including advertising, food and beverages, staff, venue, and entertainment—as well as the value of in-kind donations from local businesses—was approximately $28,000.
Donald Green and Oliver McClellan designed an evaluation patterned after the earlier studies. Each local organization that planned to throw a party was asked to nominate more than one feasible site. The researchers then randomly assigned some of the sites to a treatment group and others to a control group. Despite the fact that the study involved just eighteen
voting sites (nine treatment and nine control), the statistical results were surprisingly decisive. Turnout was 3.8 percentage points higher in treated locations, and the chances of obtaining an estimate this large simply by chance was about one in forty-five. This boost in turnout implies a cost-per-vote of $34, which is quite good given the challenges of increasing turnout in a presidential election.
Coincidentally, an advocacy organization also took an interest in festivals during the waning days of the 2016 presidential election. This 501(c)(4) organization sought to increase turnout among college students in the battleground state of Pennsylvania. Most of its festivals were not evaluated experimentally, but five were. Following the same experimental protocol as the Civic Nation study, this group picked two possible voting sites, and the researchers randomly selected one to receive treatment. In contrast to the Civic Nation festivals, which focused on getting the word out in advance to the local community, these festivals popped up suddenly in off-campus precincts where students vote and primarily consisted of free food and shuttles from campus. Turnout was higher in treatment sites in three of the five pairs, but the average effect was almost exactly zero.
Civic Nation coordinated a larger festivals program in 2017, which featured an assortment of state and local elections. Civic Nation developed training materials to introduce local groups to the legal requirements of hosting a nonpartisan festival and to provide pointers for planning the festival and getting the word out. This decentralized model included small grants to local organizations to defray the costs of food and advertising.
The character of the festivals varied. Some were relatively low-key affairs consisting of pizza, soft drinks, and a boom box; others reflected far greater community involvement, such as local entertainer performances and catering from well-known local eateries. Perhaps more important, advance work leading up to the festivals also varied. Some local groups publicized the event through door-to-door canvassing and posters in local
shops, but most relied on flyers, robo-calls, or direct mail. This variation in pre–Election Day advertising was reflected in highly variable festival attendance. While the smallest parties received as few as fifteen or twenty attendees, larger parties received hundreds of party-goers. The weather was no help—rain afflicted 85 percent of the festival sites, which were outdoors. When the results were tallied, precinct-level voting rates were
on average higher in treatment sites than in their control group counterparts, but only by 0.9 of a percentage point, implying a cost-per-vote that was four times higher than in 2016. However, the researchers took note of the fact that while the treatment has only a weakly positive effect on turnout in the sites that were hit by rain, the sites where no rain fell experienced a substantial increase in turnout on par with the results from 2016.
By 2018, Election Day festivals had begun to attract attention among civic leaders and donors. The organization #VoteTogether, a spin-off of Civic Nation and under the leadership of Angie Jean-Marie, set out to orchestrate hundreds of parties in advance of the midterm elections. #VoteTogether coordinated grants to train local organizations, publicize the idea of hosting voting events and defray the costs of throwing festivals.
A total of 1,946 events received #VoteTogether’s encouragement or assistance. The vast majority of these parties took place outside any experimental evaluation, but some groups were willing to do the extra legwork to pick more than one feasible site and let researchers randomly select which site would receive a festival. In all, a total of 56 sites were assigned to treatment and were each paired with a control site. Drawing
on the lessons from 2017, #VoteTogether provided clear guidelines to local groups about the importance of community outreach in advance of a festival, but the weather was again uncooperative, with rainfall in 77 percent of the sites. Attendance was uneven, perhaps due to the lack of personalized outreach efforts in many sites. At the time of this writing, preliminary results suggest that the turnout boost was negligible, at more than $200 per vote.
The results from 2017 and 2018 serve as a reminder that the main challenge is drawing a crowd. In effect, the turnout problem shifts from getting people to vote to getting people to attend (and then vote).
Building festival attendance is as yet more art than science, but the trick seems to be to maximize outreach, community involvement, and economies of scale. Future experiments will offer guidance on many untried tactics for building attendance.
For example, a lineup of all-school musical performances serves the dual purpose of providing performers and an appreciative audience of family members. Jurisdictions that permit voting by mail allow festival planners to hold their parties after church or as part of a prelude to a weekend concert or sporting event. In places that have consolidated voting locations to accommodate early voting, one could hold a festival on a weekend and advertise to the entire city.
House or office parties can bring friends together to congregate for drinks and snacks before walking en masse to a nearby polling place. Variations on the festival theme are almost limitless. The challenge of putting together a successful festival stems from the fact that contemporary Americans do not associate festivals with elections, and some election officials worry that even a nonpartisan festival might appear to serve some partisan purpose. It might take a series of festivals before locals see social gatherings as an expected and attractive aspect of coming to vote.