The ORE of Politics will be available for subscription in late September. Speak to your Oxford representative or contact us to find out more.

Dismiss
Show Summary Details

Page of

PRINTED FROM the OXFORD RESEARCH ENCYCLOPEDIA, POLITICS (politics.oxfordre.com). (c) Oxford University Press USA, 2016. All Rights Reserved. Personal use only; commercial use is strictly prohibited. Please see applicable Privacy Policy and Legal Notice (for details see Privacy Policy).

date: 20 September 2017

Field Experiments on Political Behavior

Summary and Keywords

Field experiments allow researchers on political behavior to test causal relationships between mobilization and a range of outcomes, in particular, voter turnout. These studies have rapidly increased in number since 2000, many assessing the impact of nonpartisan Get-Out-the-Vote (GOTV) campaigns. A more recent wave of experiments assesses ways of persuading voters to change their choice of party or alter their social and political attitudes. Many studies reveal positive impacts for these interventions, especially for GOTV. However, there are far fewer trials carried out outside the United States, which means it is hard to confirm external validity beyond the U.S. context, even though many comparative experiments reproduce U.S. findings. Current studies, both in the United States and elsewhere, are growing in methodological sophistication and are leveraging new ways of measuring political behavior and attitudes.

Keywords: field experiments, Get-Out-the-Vote, voter turnout, campaigns, social pressure, persuasion

Introduction

Field experiments were once a rarity in the study of political behavior, but now they are almost commonplace. There has been a remarkable expansion in the number and range of this kind of study, a development that has taken place over a relatively short space of time. Researchers now regularly test interventions on randomly allocated groups of voters, whose political behavior is compared to those in a control group who do get the intervention. The experimental method resolves some long-running problems of using observational data to establish the causal effects of campaigns and other mobilization strategies, when citizens are observed as being mobilized and participating, but where the relationship may often be confounded. In contrast, experiments can make causal inferences, provided of course their assumptions are respected in the design (Gerber & Green, 2012). In addition, experimental studies also provide precise estimates of the effect sizes for interventions, such as canvassing door-to-door.

The main area of expansion has been nonpartisan Get-Out-the-Vote (GOTV) interventions, mainly targeted at voter turnout but also to a lesser extent to voter registration (for a summaries and citations for the studies cited below, see Green & Gerber, 2015). There are fewer interventions outside GOTV, but recent experiments study vote choice as well as turnout. There is some attention to other acts of political participation, such as party and organizational membership, and also tests for the effectiveness of political persuasion. There has been a growing methodological sophistication over time in the design and analysis of these experiments, partly helped by cross-disciplinary fertilization between economics and political science, with many publications appearing in economics journals. Another development has been studies in jurisdictions outside the United States. This article gives these studies play as they raise classic questions in the study of comparative politics as to whether results found in the United States vary according to institutional and cultural contexts or whether treatment effects have universal application.

Get Out the Vote Studies in the United States

Before the 2000s other methods, in particular random probability surveys, were cheap, convenient to use, and appeared to answer the key questions in the study of political behavior (see Green & Gerber, 2003, p. 96). Randomized controlled trials—that is, randomizing subjects from the field, carrying out an intervention on one group, leaving a control group untouched, and measuring outcomes in both groups afterward—were not commonly used. The few that were carried out had small sample sizes, which limited the generalizations that could be made. In addition, these experiments failed to account for the failure of subjects to take the treatment, which is a feature of door-to-door and other personal contacting interventions, where not all respondents answer the door or pick up the telephone. This limitation means it is not valid to compare the participation of those who took the treatment with those in the control group.

Gerber and Green (2000), in their now famous study of the impact of canvassing, telephone calls, and direct mail, dealt with these problems with a field experiment with a large sample size and a statistical procedure technique that dealt with noncompliance. They selected 30,000 registered voters in New Haven, Connecticut. Nonpartisan Get-Out-the-Vote messages were conveyed through personal canvassing, direct mail, and telephone calls shortly before the November 1998 election. They found that voting was increased substantially by personal canvassing at 8.7%, only slightly by direct mail at .06%, and not at all by telephone calls. They used two-stage regression to estimate the effect of canvassing because not everyone received the treatment, which now is a routine procedure of generating estimates in experimental voter turnout studies. Observed data is also an improvement over self-reports from surveys commonly used by the previous generation of researchers in political behavior.

Taking inspiration from the first Yale experiment, a large number of nonpartisan GOTV interventions have now been conducted in the United States, usually varying the mode of delivery of the treatment. Door-to-door canvassing, telephone calls (either volunteer or commercial phone banks), text messages, direct mail, lawn signs, voting apps, door hangers/flyers/leaflets, radio advertisements, e-mails, and robocalls have all been tested. These studies widened the geographical range to many locations across the United States, and also varied other aspects of their delivery, such as timing of the mobilization, degree of professionalism, and quality of phone calls. The basic finding is that the degree of personalization matters. Methods that make strong personal contacts, such as door knocking, have higher effect sizes than more impersonal methods, which with e-mails reduce to near zero. The degree of professionalism is also important as it can affect the efficacy of these modes, especially for phone banks.

The range of research questions amenable to experimentation has expanded over time, such as the impact of GOTV on subgroups, with many studies testing different target populations, such as young people, African American voters, Latin American voters, Asian American voters, Indian American voters, and public housing residents. In the main, most studies show little variation of the treatment effect according to subgroup, though there are a few studies examining heterogeneous treatment effects.

As the technology of political participation has changed how people access politics, experiments have examined mobilization with social media, such as a large-scale Facebook experiment on encouraging people to vote. Another study placed ads in their Rock the Vote experiment in 2012 of encouragements to vote in the social media feed and in the adverts, an intervention repeated in the following year. Though people “liked” the intervention, there was no impact on turnout.

Most studies concentrated on mode as discussed above, which followed from the early finding that there was no impact of varying the messages in a GOTV campaign (Gerber & Green, 2000). But from the mid-2000s studies showed that message design can affect voter turnout choice, in particular from the deploying social pressure through informing voters of neighbors’ participation, thanking voters for voting, and stimulating feelings of pride or shame by revealing past turnout, though there is no effect for conveying the emotion of hope. Asking voters by telephone if they have a plan for voting works whereas simple encouragement or self-prediction does not.

Experiments have also found that incentives to encourage people to vote are effective. Informing people of their voter registration number increases their likelihood of voting. Telling people of the norm of voting can increase turnout in low turnout groups. Addressing threats to ballot secrecy can increase turnout. There is only weak evidence for asking to mobilize neighbors. Simply informing voters of the date of the next election has no effect and is used as a placebo in several studies. Image motivation—that someone would be motivated to vote because of an anticipation of being asked after the election—was tested positively by DellaVigna, List, Rao, and Malmendier (2016). Other studies have used the experimental design to make further inferences within or across households. It has been shown how the effects generated by GOTV persist over time, a finding that also applies also to social pressure. There have been relatively few studies of voter registration in the United States, which have produced mixed results, either finding voter registration drives can affect voter turnout whereas other attempts at mobilization among students can decrease registration.

Overall, the pace of change of GOTV research in the United States has been very impressive with 187 experiments appearing in Green and Gerber’s (2015) meta-analysis,1 which excludes experiments from outside the United States. The research agenda grew rapidly because of the revival of experimentation more generally across the social sciences, which had gathered pace after tests of social policy interventions in the 1980s and 1990s. Academics in a variety of fields, such as economics, development, social policy, criminology, and education, have carried out experiments, so giving encouragement to political scientists. Another reason is that once the lessons had been learned about how to do a GOTV, the basic techniques of running a field experiment are relatively straightforward and possibly easier to do than other kinds of intervention (see John, 2017, for this argument). There is no need to work under the control of a public agency, which can veto or alter the design of the experiment; canvassing can be organized by a locally organized field force or in telephone banks; and it is relatively easy to match large samples to voter turnout records, at least in the United States, which also have covariates on the voter’s background. The cost is not that high compared to other kinds of experiment, such as social policy interventions. The final reason for the rapid diffusion is that the pioneer experimentalists are great advocates for the method and are generous to scholars in the field, particularly to younger colleagues.

Partisan Field Experiments in the United States

A second main kind of mobilization experiment is normally carried out with political parties, ensuring that party supporters turn out to vote as well as influencing vote choice. Researchers work with a political party to ensure the experiments are genuinely from a partisan source. This kind of project requires close attention to building a partnership as parties determine which experiments are carried out.

Nonetheless, these experiments work in much the same way as a GOTV: Voters are sourced from the electoral register (or in party records); an intervention is carried out such as canvassing and going door-to-door; then researchers match these people to voter turnout or other sources. The difference is that there are no publicly accessible records on party choice, which is the key outcome of partisan mobilization. Researchers and party activists have to carry out surveys of those who have been mobilized and those in the control group, for which they do not get high response rates. They also have to trust self-identified responses to intention to vote questions, which can be unreliable. It is also possible to use precinct data if the study size is large enough to randomize by this unit. Nonetheless, such experiments are being more frequently carried out, partly because political parties have discovered the use of these methods in campaigning and have become more accessible to researchers.

The interesting finding that the key findings by mode from nonpartisan GOTV campaigns do not fully generalize to partisan mobilization. In spite of using many of the same contact methods to increase voter turnout, such as canvassing, telephone calls, and leafleting, studies of partisan campaign interventions have either positive or null or even negative effects. Green, McGrath, and Aronow (2013) conclude that the GOTV effects of nonpartisan mailings and advocacy mailings differ from their nonpartisan equivalents. While nonpartisan direct mail on average has a positive effect on turnout, advocacy mailings has a negative, but statistically insignificant effect. Many studies of turnout target certain groups of voters and these have produced varying results. These apparently contradictory results might partly be explained by the double-edged nature of the appeal both to turnout and to vote for a particular candidate or party. Parties are less interested in voter turnout overall but in altering the composition of the electorate as well as in changing opinions.

Persuasion Experiments in the United States

Outside of political parties, persuasion experiments have become popular. These are designed to change attitudes, which previous research has shown to be hard to shift in a brief intervention. One of the most controversial experiments was published in Science in 2015 using canvassers to persuade citizens’ views on gay marriage; but it was revealed that one of the investigators, Michael Le Cour, had falsified the results (the other investigator, Don Green, was entirely innocent), and the study had to be retracted. But similar experiments, which have been carried out faithfully, have found that persuasion works. One example reduced prejudice toward transgender people (Broockman & Kalla, 2016). In this study, the researchers sent transgender canvassers to deliver either a persuasion message or a placebo on recycling. Other experiments test for opinion change on abortion in an online panel experiment also with a placebo. Facebook has been used on in an online experiment to see if comments made influence vote choice, which it does, and whether Twitter messages can encourage people to sign an online petition through their networks. Overall, persuasion experiments show the potential for interventions using fine-grained behavioral insights. It appears that messages are important and in how they are crafted, whereas the earlier generation of studies indicated that it was mainly the mode of contact, for example, whether door-to-door, that mattered.

Comparative Studies of GOTV

All the studies referred to above have been carried out in the United States, which has the advantage of reducing or holding relatively constant political institutions and cultures, and in allowing research findings to cumulate over time. Comparative work provides additional tests of the claims experimentalists have made in recent years, either to provide greater external validity to the findings of the U.S. studies or to indicate that the effect sizes and results change in different conditions. One limitation, however, is that there are far fewer studies than in the United States, but at the same time there are a large number of countries so the findings are spread across many contexts. This limitation is the classic small-N problem of comparative politics of having too few cases to make inferences from the variations in the institutional variables.

Nevertheless, many insights have been generated even from the small number of studies that have been carried out. Researchers have reproduced Gerber, Green, and others’ work on nonpartisan mobilization. John and Brannan (2008) find strong and comparable results between door-to-door canvassing and telephoning for the United Kingdom. This finding suggests that the hierarchy in the effect by mode was not found outside the United States, perhaps because landlines are less targeted by campaigns in the United Kingdom. Fieldhouse, Cutts, Widdop, and John (2013); and Fieldhouse, Cutts, John, and Widdop (2014) followed up this study with an England-wide two-stage multi-armed trial testing the impact of impersonal modes of mail and telephoning, which found comparable, low effects between modes for the first wave 2009, but where there was a stronger effect in 2010 for telephoning than mail. Like John and Brannan (2008), this experiment showed stronger effects than found in the United States, though in this case it was an in-house run phone bank staffed by students (who were paid) and closely supervised by the academic staff. This study was more complex than the classic U.S. trial allowing for tests of different combinations and repeat treatments. It was a national GOTV that allowed for tests of external validity across jurisdictions in England.

Another example of GOTV studies outside the United States and the United Kingdom is Guan and Green (2006) who carried out a field experiment in a Peking University precinct during the 2003 election, showing how GOTV can apply outside traditional democratic countries. In a Spanish local election, Ramiro, Morales, and Jiménez-Buedo (2012) tested four kinds of leaflet, but which did not raise turnout. Bhatti, Dahlgaard, Hansen, and Hansen (2014) test for the impact of text messages in Denmark in two experiments, which they show increased turnout.

In another study, John, Evans, and Sanders (2015) examined whether incentives in the form of a lottery could encourage people to register to vote, which they did. In France, Pons and Liegey (2013) randomly assigned citizens to receive visits from political activists during the lead-up of the 2010 French regional elections. The treatment increased the turnout of immigrants without affecting nonimmigrants, while turnout was roughly equal in the control group. In the 2012 French presidential election, Braconnier, Dormagen, and Pons’s (2016) experiment show that home registration visits had a higher impact than the information-only ones, indicating that both information costs and administrative barriers impede registration. Home registration did not reduce turnout among those who would have registered anyway.

Like in the United States, there are studies of the longitudinal effects of GOTV. With John and Brannan’s (2008) intervention, about half of those mobilized in 2005 voted in 2006. There are similar extensions to GOTV as in U.S. studies, such as examining spillover effects in the household. As in the U.S. studies, accounts of the subgroups are important, such as Bhatti et al.’s (2014) intervention with newly enfranchised voters, which showed that more vivid communications encourage voting.

Comparative Partisan Experiments

The main area of expansion for non-U.S. experimental studies on political behavior is partisan interventions. Prominent is Pons’s (2016) research, which shows positive results of door-to-door canvassing campaigns affiliated with the French Socialist Party in the 2012 presidential elections. In a countrywide precinct-level randomized experiment, 22,500 precincts and 17.1 million citizens were randomly allocated to either a control group or a treatment group. Canvassers targeted treatment precincts supporting François Hollande, the Socialist candidate. The visits did not a raise voter turnout, but reduced the vote share of right-wing candidates while increasing Hollande’s. Foos and John (2015) randomly assigned voters in a Conservative-held constituency, working with the local Conservative Party to receive a home visit and leaflet and just a leaflet, finding that the leaflet decreased voter turnout of the opposing Labour Party’s voters. Cantoni and Pons (2016) randomly assigned volunteers and party workers to canvass in the Italian municipal elections, finding that only volunteers could raise turnout, but party workers could shift vote choice in favor of their party.

Some experiments have implemented more sophisticated message-based interventions. Foos, Cunnigham, and John (2015) carried out a randomized field experiment, embedded in a Labour Constituency Party’s leafleting campaign during the 2014 local and European elections. They randomly assigned whether subjects received a leaflet emphasizing an issue owned by the Labour Party (health care), one owned by the governing Conservative Party (crime and policing), or no leaflet. A posttreatment telephone survey shows that the health-care leaflet significantly increased the salience of the issue to voters, as well as increasing turnout. The crime leaflet in contrast, did not have any effects on turnout. Dewan, Humphreys, and Rubenson (2014) sought to persuade voters of the benefits of the new electoral system in the campaign in British Columbia for Single Transferable Vote finding evidence for campaigns and endorsement, effects that do not vary by canvasser. Kendall, Nannicini, and Trebbi’s (2015) study of persuading voters, in a mayoral campaign in Italy about the candidate’s valence or ideology, showed a positive impact of valence for the incumbent in terms of vote share though not of turnout when the message is conveyed by phone. Brown, Perrella, and Kay (2010) randomized a leafleting campaign of a Green Party candidate in the 2007 Ontario elections in Canada with modest effects. Lam and Peyton (2013) target voters in Australia with a message that successfully reduced votes for the Coalition government. In general, these studies show mixed findings, which confirms the picture revealed in the United States, both for voter turnout and vote choice.

Making a partisan appeal poses challenges for parties on both sides of the Atlantic as it is hard to predict the results when voters are receiving messages they may not welcome. Nonpartisan GOTV is easier to understand, where results vary by mode, even if the hierarchy of effect sizes varies somewhat from the non-U.S. context. Voters around the world respond to calls to turnout at elections, whether made door-to-door, or by the many other contacts made. In fact, much of this research shows how traditional forms of mobilization, such as posters, still have a lot of mileage even in the Internet age.

Developing World Experiments

The developing world context has been the site for considerable number of experiments, and experimentalists have carried out a wide range of interventions. For example, Vicente (2014) distributed voter education messages to voters in Sao Tome and Principe (STP), a two-island country in West Africa, which increased confidence in the political process and reduced turnout for the challenger showing evidence or reduced vote buying. Ocantos and Nickerson (2014) carried out survey experiments in five Latin American countries to show the conditionality of responses to vote buying. Wantchekon (2003) randomly distributed clientelist and policy platforms to randomly selected villages in Benin, which shows how policy messages can affect support for the opposition candidate. Collier and Vicente (2014) carried out a nationwide field experiment in Nigeria based on randomized antiviolence grassroots campaigning. They find that voter intimidation is effective in reducing voter turnout.

These experimenters face more varied and challenging conditions than their colleagues working in the developed world, implementing randomized evaluations across large, often inaccessible rural areas, some of which are torn by civil war and ethnic conflict. At the same time these experiments, in the range and intensity of the interventions, are often more ambitious than those done in the United States and in Western Europe, testing concepts in political economy, such as vote buying and measures to limit corruption. Perhaps these experimenters do not fear public opinion or political backlash as much their counterparts in developed countries.

Comparative Experiments on Political Participation and Recruitment

It is easy to forget there is a range of outcomes in political behavior that can be subject to experimental manipulations, such as party membership and standing for office. But these kinds of experimental interventions have been quite rare, though with more done in comparative contexts than in the United States. Such studies tend to be small scale because they do not have access to large population lists (like voter lists in the GOTV studies). For example, Vissers, Hooghe, Stole, and Mahéo (2012) compare online and offline mobilization methods in Belgium finding that online mobilization encourages mobilization online and offline. A final vein of work examines attempts to get citizens to stand for office, either from within political parties, or from the general public. Such studies are on the fringes of the topic for this entry, closer to elite experiments, which involve manipulation of political institutions through contacts with politicians and bureaucrats. Ryan et al. (2015) target local councils to encourage them to recruit local people for office (which was not effective). Other interventions are within political parties themselves, such as Preece and Stoddard’s (2015), which randomized messages to Republican Party’s caucus-goers to encourage them to attend a candidate-information seminar, which attracted more women to attend though there was a very low response rate to this message. There is a growing interest in the experience of participating online, which is amenable to experimentation, such randomly assigning citizens to Facebook pages, as done in one experiment in Romania, which showed positive effects on civic activism (Foos, Kostadinov, Marinov, & Schimmelfennig, 2015).

Conclusions

The research agenda using field experiments on political behavior continues to expand. Early work was mainly on nonpartisan GOTV experiments, usually done in the United States, leading to a profusion of studies, largely about the mode of contact on voter turnout. But it did not take long for new topics to be examined, such as the behavioral cues behind a successful mobilization, and how the application of social pressure and the provision of norms affect voter turnout. A rich vein of work has been mined, seeking to understand the subtle cues that can persuade voters or citizens to act or change beliefs.

The early focus on voter turnout was understandable, partly because of the relative ease of running localized GOTV experiments and of identifying the outcome variable from voter records. But researchers have been more successful in finding outcome variables that can overcome the problems of survey attrition or using party records more effectively. And one of the characteristics of the current generation of experimental researchers in political behavior is greater sophistication in the methodologies used in designing and analyzing interventions.

If research in the United States has taken a natural arc from early GOTVs to more sophisticated persuasion experiments, the comparative experience has been more patchy. The comparative field provides a greater range of contexts for greater external validity, but so far the number and range of studies make this opportunity hard to grasp, with barely more than a handful for each country. Nevertheless, extant studies show how many findings generalize across national boundaries, but there are not enough studies yet to ascertain whether effect sizes vary from studies carried out in the United States.

It is conventional in any review to ask for more research, but experiments are an area where more studies do massively add to knowledge, by replicating findings and allowing for the carrying out of meta-analyses. These replications and summations are the main building blocks for advances in knowledge in the medical and health sciences. This opportunity is where comparative research with its range of subjects and contexts could offer so much. In political science, however, there is not much funding and incentives for carrying out replications and extensions of existing studies. Even though replications have recently been seen as more desirable and can be subject of advanced methodological treatments, an original study is still more likely to be published in a high-citation journal and offer career advancement to young researchers. Just like the United States, the comparative context is more likely to offer innovations and new kinds of treatments rather than replications, much for the same reason. The future, both in the United States and elsewhere, largely remains with the current generation of researchers who are delivering fine-tuned and precise interventions designed to leverage changes in attitudes and behaviors in new ways.

References

Bhatti, Y., Dahlgaard, J. O., Hansen, J. H., & Hansen, K. M. (2014). Moving the campaign from the front door to the front pocket: Field experimental evidence on the effect of phrasing and timing of text messages on voter turnout. Retrieved from http://cvap.polsci.ku.dk/publikationer/arbejdspapirer/2016/Text_Message_Field_Experiments.pdf

Braconnier, C., Dormagen, J.-E., & Pons, V. (2016, March). Voter registration costs and disenfranchisement: Experimental evidence from France. (Working paper, Harvard Business School, No. 16–098). Retrieved from http://www.hbs.edu/faculty/Publication%20Files/16-098_13f55f16-697d-4f42-9448-cf5e3dd2f52b.pdf

Broockman, D., & Kalla, J. (2016). Durably reducing transphobia: A field experiment on door-to-door canvassing. Science, 352, 220–224.Find this resource:

Brown, S. D., Perrella, A. M. L., & Kay, B. J. (2010). Revisiting local campaign effects: An experiment involving literature mail drops in the 2007 Ontario election. Canadian Journal of Political Science, 43, 49–67.Find this resource:

Cantoni, E., & Pons, V. (2016). Do interactions with candidates increase voter support and participation? Experimental evidence from Italy. Unpublished paper. Retrieved from http://www.hbs.edu/faculty/Publication%20Files/16-080_559d110b-4b31-467c-9d14-5cd08083a527.pdf

Collier, P., & Vicente, P. C. (2014). Is vote buying effective? Evidence from a field experiment in West Africa. The Economic Journal, 124, 356–387.Find this resource:

DellaVigna, S, List, J. A., Rao, G., & Malmendier, U. (2016). Voting to tell others. Review of Economic Studies. Retrieved from http://scholar.harvard.edu/rao/publications/voting-tell-othersFind this resource:

Dewan, T., Humphreys, M., & Rubenson, D. (2014). The elements of political persuasion: Content, charisma and cue. The Economic Journal, 124, 257–292.Find this resource:

Fieldhouse, E., Cutts, D., & John, P. (2009). Is voting habit forming? The longitudinal impact of a GOTV campaign in the UK. Journal of Elections, Public Opinion and Parties, 19, 251–263.Find this resource:

Fieldhouse, E., Cutts, D., John, P., & Widdop, P. (2014). When context matters: Assessing geographical heterogeneity of get-out-the-vote treatment effects using a population based field experiment. Political Behavior, 36, 77–97.Find this resource:

Fieldhouse, E., Cutts, D., Widdop, P., & John, P. (2013). Do impersonal mobilisation methods work? Evidence from a nationwide get-out-the-vote experiment in England. Electoral Studies, 32, 113–123.Find this resource:

Foos, F., Cunningham, K., & John, P. (2015). How issue ownership shapes campaign effects: A field experiment in the 2014 UK elections. Unpublished paper. Summary available at http://www.policylab.ch/resources/Issue_Ownership_APSA_presentation.pdf, Accessed January 2, 2017.

Foos, F., & John, P. (2015). Parties are no civic charities: Campaigns, demobilisation, and the changing composition of the electorate. Political Science Research Methods, 1–16. Retrieved from https://www.cambridge.org/core/journals/political-science-research-and-methods/article/parties-are-no-civic-charities-voter-contact-and-the-changing-partisan-composition-of-the-electorate/569E7C8D80B9F5B3D37B90AF205730BF, Accessed January 2, 2017.Find this resource:

Foos, F. and Kostadinov, L., Marinov, N., & Schimmelfennig, F. (2015, December 16). Does social media promote civic activism? A field experiment with a civic campaign. Available at SSRN http://ssrn.com/abstract=2704356

Gerber, A. S., & Green, D. P. (2000). The effects of canvassing, telephone calls, and direct mail on voter turnout. American Political Science Review, 94, 653–663.Find this resource:

Gerber, A. S., & Green, D. P. (2012). Field experiments: Design, analysis, and interpretation. New Haven, CT: Yale University Press.Find this resource:

Green, D. P., & Gerber, A. S. (2003). The underprovision of experiments in political science. The ANNALS of the American Academy of Political and Social Science, 589, 94–112.Find this resource:

Green, D. P., & Gerber, A. S. (2015). Get out the vote: How to increase voter turnout (3d ed.). Washington, DC: Brookings Institution Press.Find this resource:

Green, D. P., McGrath, M. C., & Aronow, P. (2013). Field experiments and the study of voter turnout. Journal of Elections, Public Opinion and Parties, 23, 27–48.Find this resource:

Guan, M., & Green, D. P. (2006). Noncoercive mobilization in state-controlled elections. An experimental study in Beijing. Comparative Political Studies, 39, 1175–1193.Find this resource:

John, P. (2017). Field trials in political science and public policy. Practical lessons for the delivery of randomized experiments. New York: Routledge.Find this resource:

John, P., & Brannan, T. (2008). How different are telephoning and canvassing? Results from a get out the vote field experiment in the British 2005 general election. British Journal of Political Science, 38, 565–574.Find this resource:

John, P., Evans, E., & Sanders, M. (2015). Targeting voter registration with incentives: A randomized controlled trial of a lottery in a London borough. Electoral Studies, 40, 170–175.Find this resource:

Kendall, C., Nannicini, T., & Trebbi, F. (2015). How do voters respond to information? Evidence from a randomized campaign. American Economic Review, 105, 322–353.Find this resource:

Lam, P., & Peyton, L. (2013). Voter persuasion in compulsory electorates: Evidence from a field experiment in Australia. Unpublished paper. Available at http://polmeth.wustl.edu/files/polmeth/ausexp.pdf

Ocantos, E. G., & Nickerson, D. W. (2014). The conditionality of vote-buying norms: Experimental evidence from Latin America. American Journal of Political Science, 58, 197–211.Find this resource:

Pons, V. (2016, January). Does door to door canvassing affect vote shares: Evidence from a countrywide field experiment in France. Working paper, No. 16–079, Harvard Business School. Retrieved from http://www.hbs.edu/faculty/Pages/profile.aspx?facId=736825&facInfo=pub

Pons, V., & Liegey, G. (2013). Increasing the electoral participation of immigrants. Experimental evidence from France. Working paper, MIT. Available at http://economics.mit.edu/files/8885

Preece, J. R., & Stoddard, O. B. (2015). Does the message matter? A field experiment on political party recruitment. Journal of Experimental Political Science, 2, 26–35.Find this resource:

Ryan, M., Stoker, G., John, P., Moseley, A, James, O., Richardson, L., & Vannoni, M. (2015, August 29). Remaking heaven in small-scale democracy: A randomised experiment to encourage contested elections and greater representativeness in English local councils. Paper presented at the ECPR General Conference 2015, Montreal. Section: Experimental Political Science, Panel: Electoral Systems and Elections. Accessed January 2, 2017, from https://ecpr.eu/Filestore/PaperProposal/1bff16ca-a4d0-4b6f-909d-1d3102b1d888.pdfFind this resource:

Ramiro, Luis, Laura Morales, and María Jiménez-Buedo. (2012). “The Effects of Party Mobi- lization on Electoral Results. An Experimental Study of the 2011 Spanish Local Elections.” Paper prepared for the IPSA conference, July 2012. Accessed February 26 2017, from http://people.bu.edu/tboas/ramiro.pdfFind this resource:

Vicente, P. C. (2014). Is vote buying effective? Evidence from a field experiment in West Africa. Economic Journal, 124(574), 356–387.Find this resource:

Vissers, S., Hooghe, M.Stolle, D., & Mahéo, V-A. (2012). The impact of mobilization media on off-line and online participation: Are the impact of mobilization medium specific. Social Science Computer Review, 30, 152–169.Find this resource:

Wantchekon, L. (2003). Clientelism and voting behavior: Evidence from a field experiment in Benin. World Politics, 55(3), 399–422.Find this resource:

Notes:

(1.) This figure adds together Green and Gerber’s (2015) report of 51 door-to-door canvassing experiments, 85 direct mails, 28 volunteer phone banks, 19 commercial phone banks, and 4 robocalls.