Electronic Voting Systems

by

Greg Vonnahme

• LAST REVIEWED: 15 January 2019
• LAST MODIFIED: 15 January 2019
• DOI: 10.1093/obo/9780199756223-0259

Introduction

In 2001, Wand and colleagues published a paper titled “The Butterfly Did It” (see Wand, et al. 2001, cited under Voting System Neutrality) in which they argue that Palm Beach County’s butterfly ballot caused enough errors to decide the 2000 election for George W. Bush. The butterfly ballot also helped launch significant new research initiatives into voting systems and prompted new federal legislation through the Help America Vote Act of 2002, which served to modernize American voting systems. Along with Internet voting, these developments account for most contemporary research on electronic voting systems. Research on electronic voting systems is now at a crossroads. Much of the research following the 2000 election evaluated technology including lever and punch-card machines that are now largely obsolete (Stewart 2011, cited under History and Development of Voting Systems). Current and future research is moving in the direction of issues of security, Internet voting, ballot design, usability, efficiency, and cost of electronic voting systems. All voting systems in the United States today are electronic to a degree. Ansolabehere and Persily 2010 (cited under Empirical and Legal Evaluation of Voting Systems) identifies three discrete parts to voting systems: voter authentication, vote preparation, and vote management. Electronic voting technology can facilitate any of these steps. The term “electronic voting” is polysemous. Electronic voting (or e-voting) variously describes direct-recording electronic voting, electronic vote tabulation, or Internet voting among others. This document defines electronic voting as any voting system that uses electronic technology at any step in the voting process. Fully electronic voting systems use DREs (direct-recording electronic machines), in which ballots are electronically generated, prepared, and counted. Hybrid types of electronic voting are optically scanned ballots (precinct or centrally counted) or ballot mark devices (BMDs), which the voter completes manually and submits but is electronically counted. Electronic voting systems can also include Internet voting in which voters receive, prepare, and submit ballots online. The 2000 presidential election precipitated the most sweeping changes to voting systems, and we continue to see officials adopt new voting systems and Internet voting pilot programs, such as those in Estonia, Canada, Brazil, and Switzerland. Voting systems, particularly Internet voting, are a source of controversy in the United States and abroad. Debates over security and ease of use involve complex technologies and core democratic principles about the rights and responsibilities of citizens. Elections are also, at least in a narrow sense and especially in the United States, zero-sum. Only one person can hold an office, and any change in voting systems that helps one candidate or party necessarily harms the electoral prospects of others. At best, this leads officials to closely scrutinize new voting systems. At worst, it can lead to irreconcilable and unprincipled polarization over questions of voting technology. E-voting involves issues of technology, democratic participation, and electoral politics. This creates a rich environment for research on voting systems.

Residual Votes and the Performance of Voting Systems

A residual vote occurs when a voter casts a ballot, but their preference is not registered for a particular race. Most studies of voting system performance analyze residual votes in a presidential contest. It is “the most rough and ready measure of performance” (Alvarez, et al. 2008, cited under Confidence in Voting Systems) and “the de facto standard for post-hoc evaluations of voting system effectiveness (Campbell and Byrne 2009, cited under Usability Studies and Experiments). Researchers can also assess the performance of individual voting machines using residual votes (Stewart 2014). The most consistent finding is that residual votes are higher with punch-card voting systems (Kimball and Kropf 2008; Knack and Kropf 2003; Ansolabehere and Stewart 2005; Hanmer, et al. 2008; Bullock and Hood 2002; Alvarez, et al. 2011). One study in this section reports an experiment involving a mock election in which they evaluate residual votes across three types of voting systems (Shocket, et al. 1992). Studies also compare lever voting machines, paper ballots, direct-recording electronic voting systems (DREs) optical scan ballots (Thomas 1968; Ansolabehere and Stewart 2005; Knack and Kropf 2003; Nichols and Strizek 1995; Hanmer, et al. 2008) and vote-by-mail (Alvarez, et al. 2011).

• Alvarez, R. Michael, Dustin Beckett, and Charles Stewart III. “Voting Technology, Vote-by-Mail, and Residual Votes in California, 1990–2010.” Political Research Quarterly 66.3 (2011): 658–670.

  DOI: 10.1177/1065912912467085

  This study compares residual votes in California and finds that optical scan ballots reduced the residual vote rate compared to punchcard systems. DREs reduced residual votes for downballot races, but not for federal elections. Voting by mail increased residual vote rates.

• Ansolabehere, Stephen, and Charles Stewart III. “Residual Votes Attributable to Technology.” Journal of Politics 67.2 (2005): 365–389.

  DOI: 10.1111/j.1468-2508.2005.00321.x

  This study compares residual votes from 1988 to 2000 presidential, gubernatorial, and senate elections. They find that paper ballots produced the lowest residual vote rate in presidential elections, followed by optical scan, lever machines, DREs, and then punch cards. While they find differences in the residual vote rate, 60 percent of the variation is attributable to county characteristics, indicating that local characteristics have a much greater effect on uncounted votes than technology.

• Bullock, Charles III, and M. V. Hood III. “One Person—No Vote; One Vote; Two Votes: Voting Methods, Ballot Types, and Undervote Frequency in the 2000 Presidential Election.” Social Science Quarterly 83.4 (2002): 981–993.

  DOI: 10.1111/1540-6237.00127

  Examines undervotes in Georgia during the 2000 presidential election. It compares lever machines, punch cards, optical scan, and hand-counted paper ballots. They find that lever machines and optical scan ballots, particularly “connect the arrow” designs, reduce undervotes.

• Hanmer, Michael, Won-Ho Park, Michael Traugott, Richard Niemi, Paul Herrnson, Benjamin Bederson, and Frederick Conrad. “Losing Fewer Votes: The Impact of Changing Voting Systems on Residual Votes.” Political Research Quarterly 63.1 (2008): 129–142.

  DOI: 10.1177/1065912908324201

  Examines voting systems and residual votes. It mostly compares residual votes with three types of electronic voting systems: punch cards, optical scan, and DREs. The study examines voting systems in Florida and Michigan from 2000 to 2004, during which time many officials modernized voting systems. They find that residual votes decreased generally from 2000 to 2004, and the reduction was greatest for shifts to optical scan ballots.

• Kimball, David, and Martha Kropf. “Voting Technology, Ballot Measures, and Residual Votes.” American Politics Research 36.4 (2008): 479–509.

  DOI: 10.1177/1532673X08320405

  Comparing residual votes on the top and bottom of the ballot, Kimball and Kropf finds that punch cards increased residual votes by about 0.5 percent, levers decreased it by 0.5 percent, optical scan (precinct) decreased it by 0.7 percent, and DREs did not matter compared to centrally counted ballots in the presidential vote. For ballot measures, residual votes increased by 1.5 percent with punch cards, 19.5 percent with levers, and 4 percent for full-face DREs.

• Knack, Stephen, and Martha Kropf. “Voided Ballots in the 1996 Presidential Election: A County-Level Analysis.” Journal of Politics 65.3 (2003): 881–897.

  DOI: 10.1111/1468-2508.00217

  This study analyzes how voting equipment affected voided votes in the 1996 presidential election. Finds that voided ballots are significantly greater in counties with large Latino and black populations but not in counties that use DREs or lever machines. Across all counties, lever machines, optical scan, and hand-counted paper ballots were associated with lower voided ballot rates than counties using punch cards.

• Nichols, Stephen, and Gregory Strizek. “Electronic Voting Machines and Ballot Roll-Off.” American Politics Research 23.3 (1995): 300–318.

  DOI: 10.1177/1532673X9502300303

  Rolloff is closely related to residual votes. This study operationalizes rolloff as the difference between presidential votes and lower ballot contests. They find significantly lower rolloff with the electronic voting machines (7–23 percent). Rolloff for non-partisan judicial races was especially lower.

• Shocket, Peter, Neil Heighberger, and Clyde Brown. “The Effect of Voting Technology on Voting Behavior in a Simulated Multi-Candidate City Council Election: A Political Experiment of Ballot Transparency.” Political Research Quarterly 45.2 (1992): 521–537.

  DOI: 10.1177/106591299204500213

  This study reports on an experiment with a community sample. The context was a simulated multi-candidate local race, in which voters could select up to nine (of eighteen) candidates. Subjects were randomly assigned to use a DRE, punch card, or paper ballot. Undervoting was far more common with punch card ballots (60 percent) than paper ballots or DREs (34 percent). Overvoting was also significantly higher with punch cards.

• Stewart, Charles III. “The Performance of Election Machines and the Decline of Residual Votes in the US.” In Measuring American Elections. Edited by Barry Burden and Charles Stewart III, 223–247. Cambridge, UK: Cambridge University Press, 2014.

  This chapter reviews the residual vote measure and discusses alternative measures, such as usability studies and wait times. For example, DREs increased time-to-vote by 3.5 minutes compared to optical scan ballots. It estimates that in presidential elections, there is a 0.5 percent lower bound on the residual vote rate and provides an example of how the residual vote rate can be used to assess the performance of individual voting machines.

• Thomas, Norman. “Voting Machines and Voter Participation in Four Michigan Constitutional Revision Referenda.” Western Political Quarterly 21.3 (1968): 409–419.

  This is the earliest study in this bibliography. It examines turnout in the 1958–1963 Michigan state constitutional referenda, by comparing turnout in two hundred matched paper and machine ballot precincts. The study finds that voter participation was 10–26 percent higher in paper-ballot precincts across the four referenda. The difference was larger in 1958 and 1960, which were concurrent with congressional and presidential elections, suggesting that rolloff accounts for some of the difference.