Elsevier

Electoral Studies

Volume 47, June 2017, Pages 1-12
Electoral Studies

Internet voting and turnout: Evidence from Switzerland

https://doi.org/10.1016/j.electstud.2017.03.001Get rights and content

Highlights

  • We investigate the effect of Internet voting on turnout in two Swiss cantons.

  • In both cantons, Internet voting was introduced in the presence postal voting.

  • Drawing on difference-in-differences estimation, we find no turnout effect.

Abstract

Internet voting (i-voting) is often discussed as a potential remedy against declining turnout rates. This paper presents new evidence on the causal effect of i-voting on turnout, drawing on trials conducted in two Swiss cantons: Geneva and Zurich. Both Geneva and Zurich constitute hard cases for i-voting, given that i-voting was introduced in the presence of postal voting. However, this setting allows us to test some of the more optimistic claims regarding i-voting's ability to increase turnout. Empirically, we exploit the advantageous circumstance that federal legislation created a situation coming close to a natural experiment, with some of Geneva's and Zurich's municipalities participating in i-voting trials and others not. Using difference-in-differences estimation, we find that i-voting did not increase turnout in the cantons of Geneva and Zurich.

Introduction

In many advanced democracies, turnout rates have decreased over the past few decades. To counter this trend, many countries are forging ahead with convenience voting reforms, such as Electoral Day registration and postal voting, that aim to simplify the voting process and thus increase political participation (Gronke et al., 2008). Internet voting (i-voting), a voting method that allows voters to cast their vote remotely over the Internet, is increasingly discussed in this context.1

I-voting is less than two decades old, but it already has a turbulent history behind it. Cyber enthusiasm was widespread around the turn of the millennium, and i-voting widely heralded as the ‘magic ballot’ that would entice many more people to vote (Alvarez et al., 2009, Gibson, 2005, Norris, 2005). With great hopes, i-voting was for the first time trialled in the context of a binding election in 2000 at the occasion of the Arizona Democratic Primaries. Soon experimentation started in other countries as well. In 2002, the first i-voting experiments were conducted in the UK, and in 2003 i-voting experimentation began in Canada and in Switzerland. Estonia followed in 2005.

Against the initial enthusiasm, most of today's polities continue to vote offline. Several countries have abandoned experimentation with i-voting after the first few trials. Examples include the US, the UK, the Netherlands, Austria, and Norway. The reasons are diverse and include concerns related to the security of voting online, but also more prosaic worries that i-voting might help political opponents (Hall, 2015, Mendez and Trechsel, 2005, Mendez, 2010). Furthermore, initial, often wildly exaggerated hopes of immense turnout increases have clearly not materialized (Bochsler, 2010, Goodman, 2014).

Nevertheless, i-voting continues to be practiced in several corners of this world. Estonia has fully generalized i-voting and uses it in all national and local elections. In several other countries, experimentation with i-voting continues at a more limited scale, typically in local or regional settings. Examples include Canada, Australia, France, Brazil, and Switzerland. Further, many citizens continue to view online elections favorably, and arguments in favor of i-voting continue to be frequently made by both journalists and politicians (see e.g. Castella, 2015, Goodman, 2014, Hall, 2015, Milic et al., 2016). There are several reasons. In a world where people shop, date, and bank online, paper-and-pencil elections appear somewhat archaic. Online elections promise increased speed, efficiency, and accuracy in terms of the vote counting, but also increased accessibility for the disabled and elderly. And, perhaps most importantly, many continue to argue that i-voting increases turnout, if maybe not to the phenomenal rates suggested by some in the early days of i-voting (Pammett and Goodman, 2013).

Given that discussions about i-voting are likely to stay with us in the foreseeable future, academics should provide policymakers and the general public with evidence regarding the pros and cons of i-voting. Clearly, i-voting raises security issues, making it important that computer scientists investigate the risks associated with voting online and how they can be tackled (see e.g. Simons and Jones, 2012). The main contribution political scientists can deliver is evidence regarding the turnout effect of i-voting. How large a turnout increase can we expect if i-voting is introduced? Or can we even expect any?

Existing evidence points in different directions. Some have found large turnout increases, others more moderate ones. Still others found no turnout effect at all, thus questioning one of the key justifications for i-voting. However, the lessons that can be drawn from existing studies are limited due to methodological problems. In this study, we present new evidence on the effect of i-voting on turnout, drawing on i-voting experiments conducted in two Swiss cantons: Geneva (2003–ongoing) and Zurich (2005–2011). These two cases offer an advantageous setting to learn about the causal effect of i-voting on turnout. That said, they notably constitute difficult terrain for i-voting. In both Geneva and Zurich, i-voting was introduced in the presence of postal voting, which decreases chances for an i-voting turnout effect. Still, this setting allows us to test some of the more optimistic claims regarding i-voting's ability to increase turnout.

Empirically, we exploit federal legislation that limited the number of citizens allowed to take part in i-voting trials and thus created a situation resembling a natural experiment, with some of Geneva's and Zurich's municipalities taking part in trials and others not. The resulting within-canton variation in the availability of i-voting allows us to circumvent many of the challenges to causal inference prevalent in previous studies. Drawing on difference-in-differences (DID) estimation, we find that the introduction of i-voting did not affect turnout in the Swiss cantons of Geneva and Zurich. Implications for policymaking as well as future research are discussed in the conclusion.

Section snippets

Existing evidence

The effect of i-voting on electoral turnout remains an open question. Several existing studies have examined the nexus between i-voting and turnout, but the conclusions vary dramatically from study to study. Some have come to highly optimistic conclusions. For example, Spada et al. (2015) analyze an i-voting experiment in the context of a budget referendum in Brazil and find that i-voting increased turnout by more than 8 percentage points. Solop (2001) comes to a similarly optimistic conclusion

Theory

As argued above, in both cases under study i-voting was introduced in the presence of postal voting. Can i-voting increase turnout even in the presence of postal voting? Optimistic proponents of i-voting would suggest so. Indeed, increasing turnout constituted an important motive for i-voting experimentation in the two cantons we analyze (e.g. Beroggi, 2014). The primary argument of the i-voting optimists is that i-voting constitutes “the ultimate in convenience voting” (Alvarez et al., 2009,

Internet voting in Switzerland

Switzerland's approach to i-voting is best described as cautious and strongly shaped by its decentralized structures (Gerlach and Gasser, 2009, Mendez, 2010). In the initial phase, i-voting experimentation was limited to three cantons: Geneva, Neuchâtel, and Zurich. In the spirit of a decentralized approach, each of the three pilot cantons developed its own i-voting system. However, in effect Geneva's and Zurich's solutions are highly similar: Voters access the i-voting web page, enter the

Research design

This study aims to establish the causal effect of the introduction of Internet voting in the Swiss cantons of Geneva and Zurich on voter turnout in federal referendums. Ideally, we would measure the difference in turnout rates under i-voting and no i-voting for the same units and the same referendums. Since this is not possible, we need to impute a credible counterfactual. As noted above, i-voting was constrained to selected municipalities in both Geneva and Zurich. This allows us to exploit

Main results

Table 1 presents our main results. The coefficients are based on linear regressions of turnout in federal referendums on the i-voting identifier with municipality and referendum day fixed effects and quadratic municipality time trends. Standard errors are clustered at the municipality level (Bertrand et al., 2004). Column 1 shows the results for the canton of Geneva. We find that the introduction of Internet voting did not increase voter turnout in the presence of postal voting. The

Conclusion

This article presented evidence that the introduction of i-voting did not raise turnout in extended trials in two Swiss cantons, Geneva and Zurich, where i-voting was made available in addition to voting at the polling station and postal voting. To estimate the causal effect of i-voting on turnout we used a difference-in-differences approach that exploits federal legislation which in both Geneva and Zurich resulted in some municipalities taking part in the trials and others not. Since this

Acknowledgments

We thank Daniel Bochsler, Fernando Mendez, Kristjan Vassil, Jonathan Wheatley, and the three anonymous reviewers for helpful discussions and comments at various stages of this project. We also thank Maja Harris and Christoph Wellig for excellent research assistance. The authors gratefully acknowledge financial support through the e-Democracy project (2012-000413) funded by the Swiss cantons of Argovia (main contributor), Basel-City, Geneva, Grisons, and Schaffhausen, as well as by the Swiss

References (45)

  • András Nemeslaki et al.

    Could on-line voting boost desire to vote? Technology acceptance perceptions of young Hungarian citizens

    Gov. Inf. Q.

    (2016)
  • Anne Powell et al.

    E-Voting intent: a comparison of young and elderly voters

    Gov. Inf. Q.

    (2012)
  • R. Alvarez et al.

    Internet voting in comparative perspective: the case of Estonia

    PS Political Sci. Polit.

    (2009)
  • Joshua D. Angrist et al.

    Mostly Harmless Econometrics: an Empiricist's Companion

    (2009)
  • Orley Ashenfelter

    Estimating the effect of training programs on earnings

    Rev. Econ. Stat.

    (1978)
  • Michael M. Bechtel et al.

    Does compulsory voting increase support for leftist policy?

    Am. J. Political Sci.

    (2016)
  • Giampiero E.G. Beroggi

    Internet voting: an empirical evaluation

    Computer

    (2014)
  • Marianne Bertrand et al.

    How much should we trust differences-in-differences estimates?

    Q. J. Econ.

    (2004)
  • Sarah Birch

    Perceptions of electoral fairness and voter turnout

    Comp. Polit. Stud.

    (2010)
  • Richard Blundell et al.

    Evaluation methods for non-experimental data

    Fisc. Stud.

    (2000)
  • Daniel Bochsler

    Can Internet voting increase political Participation? Remote electronic voting and turnout in the Estonian 2007 parliamentary elections

  • Thomas Brambor et al.

    Understanding interaction models: improving empirical analyses

    Polit. Anal.

    (2006)
  • A. Cameron et al.

    Robust inference with multiway clustering

    J. Bus. Econ. Stat.

    (2011)
  • David Card et al.

    Minimum wages and employment: a case study of the fast-food industry in New Jersey and Pennsylvania

    Am. Econ. Rev.

    (1994)
  • Tom de Castella

    Election 2015: How Feasible Would it Be to Introduce Online Voting?

    (2015)
  • Alan S. Gerber et al.

    Do perceptions of ballot secrecy influence Turnout? Results from a field experiment

    Am. J. Political Sci.

    (2013)
  • Jan Gerlach et al.

    Three Case Studies from Switzerland: E-voting

    (2009)
  • Micha Germann et al.

    Internet voting for expatriates: the Swiss case

    eJournal eDemocracy Open Gov.

    (2014)
  • Rachel K. Gibson

    Internet voting and the European parliament elections: problems and prospects

  • Brad T. Gomez et al.

    The republicans should pray for rain: weather, turnout, and voting in U.S. Presidential elections

    J. Polit.

    (2007)
  • Nicole J. Goodman

    Internet voting in a local election in Canada

  • Paul Gronke et al.

    Convenience voting

    Annu. Rev. Political Sci.

    (2008)
  • Cited by (0)

    View full text