Fifteen years ago, I drove a crew of college students from Fresno north through California’s Central Valley to Dos Palos, a sleepy farming town surrounded by cattle pastures and almond groves. Our goal was to increase Hispanic voter participation, which we planned to do by canvassing door-to-door.
We didn’t know how well our plan would work—this was a time when voter mobilization campaigns were dominated by robocalls, television commercials and direct mail—but we knew that it was important to try.
It got off to an ominous start. As we were wrapping up our morning launch meeting at a local gas station, the police rolled up. An officer ran our plates and found out we were from out of town. He wanted to know what we were doing in Dos Palos, which translates as two trees. Our explanation satisfied him; he drove off and we started walking streets with names like Oak, Almond and Fir. Most of the time, we were well received.
At one house, the target voter was out farming in the fields and the woman who answered the door insisted on taking Bailey the Beagle (my canvassing partner) and I out in her golf cart to track him down and let me deliver my get-out-the-vote script.
We had a successful two weekends in the field, and contacted more than 75 percent of our targeted voters. More importantly, it worked. Turnout increased dramatically among contacted registered voters, most notably among Hispanic Democrats (those most similar to my student canvassers), increasing their likelihood of participating in the November 2001 school board election from 34 to 50 percent.
Since then, I have conducted more than 300 GOTV field experiments around the country, mostly targeting low-frequency Hispanic, black and Asian-American voters and often in cooperation with local community organizations. Low-frequency voters in communities of color can effectively (and cost-effectively) be encouraged to vote when best practices gleaned from that work are followed.
Those best practices are surprisingly simple and straightforward. First, make the pitch personal. Once that personal appeal is made, follow it up at least a second time. The message matters during these contacts, as does the messenger. These prescriptions are based on field experiments that use randomized assignment to treatment to isolate the effect of being invited to participate in an election.
Prior to contacting voters, lists of registered voters were randomly divided into two lists: one to be targeted for contact and encouraged to vote, and one to be left alone or contacted with a non-electoral message like “please recycle.” After each election, I used validated data from local county registrars to see who on the lists had voted and to calculate the effect of each GOTV effort.
Make it Personal
In our book “Mobilizing Inclusion: Transforming the Electorate Through Get-Out-the-Vote Campaigns,” Berkeley professor Lisa García Bedolla and I described 268 GOTV experiments conducted across six electoral cycles between 2006-2008. When we launched the project, which was part of the James Irvine Foundation’s California Votes Initiative, we believed (as did most other academics and practitioners) that indirect contact was a cost-effective method of reaching large numbers of voters. Many of the community organizations we partnered with that year used low-cost methods such as mailers and postcards. But those efforts almost all failed to move voters to the polls.
Instead, we found dramatic effects from personal contact efforts. Well-conducted phone bank and door-to-door efforts, particularly those that incorporated some of the additional components noted below, were proven to be powerful ways to increase turnout, often by double digits. In one door-to-door effort in June 2006, turnout increased from 11.1 percent among those in the control group to 54.2 percent among contacted voters. Phone banking by another group, in November 2006, increased turnout from 34.3 to 43.6 percent. A February 2008 phone bank that asked friends to call friends increased turnout from 34.9 to 60.1 percent. Low-cost email, text message and other indirect methods can increase turnout slightly, but big effects require recruiting enough canvassers to talk to folks one-on-one.
More recent efforts have increased the power of indirect methods through the use of social pressure (also sometimes called social monitoring) or ethnic cues. This includes successful efforts using Spanish-language (or bilingual) postcards and advertisements on Spanish-language radio. But the increases in turnout generated by these tactics are much smaller. Making a big difference means bringing out the big guns: Door-to-door canvassers and live phone bankers.
Two Contacts are Better Than One
In addition to the clear evidence from those early California Votes Initiative experiments that indirect methods were not effective, we had a big surprise in our findings from the direct method efforts. In several communities where our organizational partners used two contacts instead of one—following up with a second telephone call with voters who had previously been contacted and committed to vote—they were generating large increases in turnout.
The effect was first apparent in analysis of the November 2006 effort by the Southwest Voter Registration Education Project (SVREP). While other groups using phone banks were reporting increases in turnout of 2-4 percent, SVREP increased turnout among Hispanics in Los Angeles in that election by 9.3 percent. Additional efforts in subsequent elections confirmed the effect was real: Two contacts are better than one.
The Message Matters
In some cases, varying the message delivered to targeted voters can make a huge difference. Door-to-door canvassing efforts by the now-defunct Association of Community Organizations for Reform Now (ACORN) in Bridgeport, Connecticut and Phoenix, Arizona in 2001 and 2003 showed that Hispanics were strongly motivated to vote when appeals to participate were linked to a specific community concern such as keeping a local hospital open or instituting a higher local minimum wage. Their door-to-door campaigns increased turnout among contacted voters by double digits.
In a series of experiments in California in 2010 and Texas in 2012, Princeton professor Ali A. Valenzuela and I found that ethnic identity messages—messages that reach out to Hispanic voters as Hispanics rather than as Americans—are significantly more effective among voters in poorer communities and among voters who are less acculturated into U.S. society (those who are immigrants and who prefer to use Spanish instead of English). Messages that reach out to Hispanic voters as Americans are more effective among Hispanic voters in wealthier neighborhoods and among individuals who are more acculturated.
These experiments confirm that mobilizing Hispanics to vote is more effective when messages are tailored to the local context and the individual voter. The message does matter, and its influence depends on context.
Blame the Messenger
Multiple experiments from the California Votes Initiative and in my subsequent work find that individuals are more likely to heed an encouragement to vote if the message is delivered by a trusted source—either a neighbor or a member of a well-known local organization. This is true for door-to-door canvassing, phone banking and even for indirect methods such as email.
In one door-to-door effort in Los Angeles in November 2006, a single door-to-door visit from a member of the local organization increased turnout by more than five points, but the effect was increased to 8.5 percent among individuals contacted by a neighbor, defined by someone from the same ZIP code.
In the small, northern Sacramento Valley city of Winters, California, prior to the February 2008 election, a local community organization sent out as their door-to-door canvassers four Hispanics who were in the process of becoming U.S. citizens. Personally unable to vote, the canvassers made very compelling advocates for participation, generating a 12.9 percentage-point increase in turnout. Two phone bank campaigns conducted in cooperation with a Catholic church in Long Beach, California, proved the power of using personal friends to encourage voting in the February and June 2008 elections. In the February 2008 election, turnout was increased from 34.9 percent in the control group to a likelihood of voting of 47.7 percent among contacted voters, and 60.1 percent among those contacted by their personal friends. Similar results were found in a follow-up effort for the June 2008 election.
Making the Magic Happen
One early California Votes Initiative experiment, back in June 2006, put together all of these ingredients, and cooked up a remarkably powerful GOTV effect. The effort was conducted by the Center for Community Action and Environmental Justice (CCAEJ), in the northwest corner of Riverside County, California—a rural area that came with challenges of gated communities, guard dogs, livestock and a general lack of sidewalks in the target areas. Moreover, because many houses in the neighborhood did not include targeted voters, there was often a considerable distance to walk (or drive) between homes. But CCAEJ, drawing on their decades of experience in community organizing in the neighborhood, met these challenges and did everything right.
For one, they started early (but not too early), hitting the sweet spot by launching their door-to-door campaign about four weeks before the election. Then, on Election Day, starting at 5 a.m., they began visiting the homes of individuals they had previously contacted, intending to leave reminder doorhangers on their front doors. Instead, they inadvertently made second face-to-face contacts with voters as those individuals were taking out their garbage cans, walking their dogs, or just going out to work. Voters trusted the messengers because they came from CCAEJ, which has had 25 years of local experience on environmental justice issues and community organizing. Canvassers were experienced members of the group who knew the community, and contacted voters knew the group and its history of local activism.
The CCAEJ script included the line, “We are fighting to improve the quality of life for local residents.” Contacted individuals were asked to sign a pledge card promising participation in the upcoming election. The card reminded individuals of the importance of their vote to the community, the country and the future, and asked individuals if they wanted to volunteer in the GOTV effort. They were also asked to complete a short survey, including questions about their level of satisfaction with elected officials and their opinion of the most important public issue in their community.
The effect of this effort was truly magical. Turnout among those randomly assigned to the control group was 11.1 percent, compared to 19.6 percent among those randomly assigned to be contacted. When CCAEJ canvassers actually contacted a person on the target list, they increased his or her probability of voting from 11.1 percent to 54.2 percent—the largest estimated treatment effect ever in a voter mobilization field experiment. Other door-to-door experiments have not replicated that enormous effect, but the lessons learned from the CCAEJ effort did persist in future efforts.
As the 2014 midterm elections approach, Hispanic voters are frequently dismissed as unlikely to participate. “Frankly,” one observer wrote recently in The Hill newspaper, “Hispanic voters seem to take a collective vacation during midterm elections, consistently failing to show up.” But that doesn’t have to be the case. Following the four-point strategy noted above can get Hispanics to the polls this November, and possibly change the outcome of some close races.
Melissa R. Michelson is a professor of political science at Menlo College with a focus on Latino politics and get-out-the-vote efforts.