Public Confidence in Elections
2.1 THE RELATIONSHIP BETWEEN DEMOCRACY AND ELECTIONS
A fundamental characteristic of democracy—perhaps its defining characteristic—is that government derives its legitimacy from elections. For example, the American Heritage Dictionary of the English Language (fourth edition) defines democracy as “government by the people, exercised either directly or through elected representatives.”1
Given the central importance of elections to democracy, it is axiomatic that elections are high-stakes affairs. The stakes are further increased by the majority-rule nature of most elections in the United States—in principle, even one vote out of tens of millions cast can determine the outcome of an election, because victory depends only on a candidate winning a majority (or a plurality) of the votes cast.
2.2 LEGITIMACY IN A DEMOCRACY
Democracies derive their legitimacy from elections that the people collectively can trust. In turn, legitimacy is important for the long-term functioning of a democratic society, because it is what underpins the willingness of the losers in an election to abide by policies set by the winners (with whom the election losers are likely to disagree). In other
words, although elections do determine in the short run who will be the next political leaders of a nation (or state or county or city), they play an even greater role in the long run in establishing the foundation for the long-term governance of a society. Absent legitimacy, democratic government, which is derived from the will of the people, has no mandate to govern.
While many factors contribute to the legitimacy of a government,2 one sine qua non is undoubtedly that elections are perceived by both winners and losers as free and fair. Indeed, it is often said that the main purpose of election fairness is to convince the loser that he or she lost the election fair and square—winners rarely complain about the fairness of an election. Perhaps more important, these comments apply even more strongly to the electorate supporting the losing candidate.
Of course, the process is greatly complicated by the fact that the electoral process will undoubtedly yield some sore losers—individuals who disguise their unhappiness over the outcome of an election with complaints about unfair process, even if the election was conducted under the fairest of circumstances and rules and procedures. Similarly, winners and especially their supporters are likely to invoke the spectre of sore losers, even if complaints about election fairness have some reasonable factual basis. Finally, an important psychological issue is that as a general rule, individuals—that is, voters—tend to associate with like-minded individuals and to read newspapers and other information sources that reinforce their own predispositions. This tendency reinforces their perceptions of being in the majority. Thus, they are likely to see an election loss more as the result of election chicanery than as a fair loss.
The political environment of today compounds the issues described above. Perhaps most significantly, political campaigns and debates today are rancorous and bitter, a throwback to the political climate that existed in the United States over 100 years ago.
This rancor sets the tone for much of the following:
Most governors and state officials are elected from the ranks of one party or another. They are thus partisan officials by definition, and these officials are ultimately responsible for state operations, including the conduct of elections. When such officials make decisions that benefit—or can be seen to benefit—candidates from their party, suspicion on the part of
the opposition is natural. In today’s highly charged political environment, these tendencies are sometimes accentuated, and there is often little shared trust that partisan officials can make nonpartisan decisions.
Close elections—much more likely when the electorate is about evenly divided—are breeding grounds for postelection suspicion, on the theory that even a small amount of deliberate fraud or accident or mishap or improperly followed procedure might have tipped the election the other way. While the presidential election of 2000 is perhaps the most salient example, outcomes in other close races have been very closely scrutinized by supporters of the losing side for irregularities.3
The cost of political campaigns has risen. In the primary elections of some jurisdictions, it exceeds $100 per vote received and has led some analysts to wonder if it raises the incentive to cheat.
Vendors of electronic voting systems have not always been seen as politically neutral. In an environment in which questions are raised about whether such systems are actually trustworthy, partisanship manifested in the vendors of these systems is likely to raise suspicion.
In such an environment, where the perceptions of fairness can depend on whether your side won or lost, a more reasonable objective is the notion of a “trusted” election, where “trust” entails a factual basis for that trust. That is, a trusted election process is one that works, can be shown to have worked after the election has been held, can be shown to have not been manipulated and to have not led to a large number of mistaken or lost votes, and can be shown to reflect the intent of the voters. To the extent that there is a provable and factual basis for calling an election trusted, there is at least a chance that more people will consider the election fair, even if their side lost.
Put differently, the fact that in the U.S. system of government, partisan office-holders are ultimately responsible for the conduct of elections (or can exert strong influence over elections) makes very important indeed the existence of procedures and practices that demonstrably minimize the possibility that these officials will be able to improperly affect election outcomes. To the extent that public trust in the integrity of elections is diminishing, the importance of such procedures is magnified.
2.3 DESIDERATA FOR ELECTIONS
With the foregoing in mind, consider the goals that elections must serve. The committee believes that there would be little disagreement about the following as election principles or goals:
Voters are entitled to secrecy in the ballots they cast, both as they cast them and in any subsequent counting of votes.4 (With voting secrecy, voter coercion becomes effectively impossible.)
A voter may cast only the number of votes in any given race for any given office or any given ballot proposition to which he or she is legally entitled. In general, this is one vote per race, although there are exceptions to this rule—for example, where voters can cast more than one vote for more than one candidate on a list of more than two or in instant runoff elections.
A voter may cast a vote only for offices or propositions for which he or she is legally entitled to vote.
A voter may not sell or trade his or her vote.
All voters legally entitled to vote, but only those voters and no one else, should be allowed to vote.
All cast ballots should be counted accurately.
An eligible voter will not face undue obstacles in casting his or her ballot, regardless of her or her personal circumstances (e.g., level of literacy, physical or cognitive disabilities, education, place of residence).
The system on which voters cast ballots will be operable for the entire time that the polling place is open.
Audit trails and other records will be kept to monitor the extent to which these principles are honored (but not at the cost of violating voter secrecy).
The election system will produce an unambiguous and definite winner even in close races (Box 2.1).
Though these desiderata are widely accepted, they are not, in practice, of equal importance. While very few election officials and administrators would admit to breaches of voter secrecy on even a small scale, most would acknowledge that proper procedures may not have been
The requirement that election systems produce a definite winner has historic roots. Elections determined by simple majorities (or pluralities) made sense when the number of voters participating in elections was small—with small elections, errors could be minimized enough that recounts could be expected to result in more accurate vote counts even in very close elections. But as the number of voters in an election increases, it is inevitable that the potential for miscounts of some sort will also grow. Good election technologies and procedures can reduce the magnitude of the likely error in the vote count, but it is virtually impossible to believe that the error can be reduced to zero consistently in all elections.
Today, some states mandate that margins of less than a certain percentage (e.g., 1 percent) trigger an automatic recount. Recounts triggered under such conditions recognize that margins of victory under a certain percentage are inherently clouded, and that measures need to be taken under such circumstances to validate the legitimacy of the election.
If one denotes the magnitude of the irreducible error as x percent of the total vote, an election that produces vote totals that are within x percent of each other is for all practical purposes a tie, and no amount of recounting or auditing will discern the intent of the voters more accurately. Thus, although a mandate to decide such elections by lottery or tossing a coin would be highly controversial (and the committee is silent on the ultimate overall desirability of such a mandate), it would be more faithful to the underlying reality that some degree of irreducible error inevitably exists.
How should an appropriate value for x be determined? To be sure, statistical analysis plays an important role here, as does historical and operational experience. But ultimately policy makers will have to determine the appropriate value. Perhaps of more importance would be an agreement by all candidates—in advance of the election and as a condition for being allowed to run in the election—to abide by a requirement to settle the election by lottery should this “statistical” tie occur.
followed to the letter on a given Election Day, that some properly registered individuals may have been turned away at the polls, that some votes cast may not have been recorded, that ballots cast do not reconcile with votes tallied, and so on. They would further argue that with limited resources, they do the best they can—and that with more generous resource allocations they would be able to do better.
The desiderata described above provide a framework for understanding electronic voting systems and how they fit into the larger societal, organizational, and institutional context of election administration. For example, they drive many of the technical requirements for electronic voting systems, as well as the opposition to electronic voting systems from many quarters.