Democratic elections suffer from several shortcomings, including low voter turnout and the effects of inaccurate polling. Sergiu Hart suggests adopting a simple repeat voting procedure that aggregates the results of two equivalent rounds of voting to overcome some of these issues and produce more representative outcomes.

Suppose that it is two weeks after the Brexit vote, and there is a new vote on the same issue—what will the result be? Given the way the original vote went, will people change their minds and vote differently? Will the original results cause people who had not voted to cast their vote in this second round? Will the final result be different? (There are no clear answers to any of these questions, and one can easily provide arguments either way.) Now carry out a similar thought experiment regarding the latest presidential election in the U.S., or whatever your latest favorite, or unsettling, election is.

These are not just academic questions. A case in point is the petition launched before the Brexit vote that called for a second vote in the event of low participation and a narrow winning margin. The petition got only 22 signatures before the vote and more than two million signatures in the two days after the result was announced (ironically, the initiator was a “leave” supporter who believed that “leave” would lose).

That we ask these questions reflects the issues that beset democratic elections. One issue is low voter turnout, which at times is only one-half of the eligible voters or even less. Another issue is excessive reliance on polls: polls affect voters, despite repeatedly turning out to be quite far from accurate. This also relates to the low-turnout issue, as voters may decide not to waste their time voting if their candidate is sure to win or lose. Polls may also lead people not to cast their vote for their preferred candidate, if, for example, they do not want him or her to win by too large a majority, or if they want to voice a certain “protest” through their vote—only to find out that in the end their candidate did not win at all. Yet another issue concerns unexpected events that occur extremely close to election time, too late to be able to be addressed by the candidates. Examples include a terrorist attack, the publication of false information, bad weather, and so on. What is common to many of these situations is that people might want to change their vote or their decision not to participate in the election once they see the actual results and how these came about.

To address these and other issues, I propose the use of the following repeat voting procedure:

  1. Voting is carried out in two rounds.
  2. Every eligible voter is entitled (and encouraged) to vote in each of the two rounds.
  3. All the votes of the two rounds are added up, and the final election result is obtained by applying the current election rules (be they plurality, special majority, electoral college, etc.) to these two-round totals.
  4. The results of the first round are officially counted and published; the second round takes place, say, two weeks after the first round, but no less than one week after the official publication of the first round’s results.

 What are the advantages of repeat voting?

Polls. The first round becomes a de facto giant opinion poll. However, because the votes of the first round count in the final result, it is a much more truthful poll; this is in contrast to the usual pre-election polls, where giving untruthful answers—whether intentionally or not—carries no cost. The combination of the large sample size and incentivized truthfulness makes the results of the first round a significantly more accurate predictor of the electorate’s views. It is thus crucial for the votes of the first round to count no less than the votes of the second round, which explains why we are adding up the votes of the two rounds, rather than having only the second round determine the outcome.

Participation. Voters who do have a preference, but one that is not strong enough to make them vote in the first round, may well be led to vote in the second round because of the results of the first round. Thus, participation in at least one round of the election is expected to increase. It is better that people vote even in one round than not at all. One indirect advantage is that people who vote may feel closer to the elected officials and to the democratic system in general.

In addition, since voters who have strong or extreme positions will most probably vote in both rounds, their relative weight in the final result will decrease when enough additional people are motivated to vote in the second round (which may well happen if such extreme positions get higher shares of the vote in the first round). 

Representative results. The final results may be more representative because the second round makes it possible for the voters as well as for the candidates to “correct” any problems of the first round. These include the effects of wrong predictions by the polls, as well as any special circumstances and events that occurred close to election time (which are less likely to occur in both rounds). All this, again, can only increase the robustness of the results: they become more trustworthy and more accepted.

New reference point. The results of the first round become a new reference point, which may well affect a person’s choice in the second round: imagining a new situation and being in a new situation are not the same thing.

Strategic voting. People seem to be more strategic in their voting than is generally believed, but under current procedures they base their strategic decisions on possibly inaccurate polls. Repeat voting provides a much more solid basis. In close elections it is conceivable that the voting of the second round may be less strategic (and the other way around when there is a large winning margin in the first round). 

In addition, in parliamentary elections where a party must achieve a minimal voter-share threshold to gain representation, many potential entrants try to convince voters that they have support that is higher than the threshold and so voting for them would not be a waste of one’s vote. In many cases, it turns out that these parties do not pass the threshold. Once this is seen in the first round, there will be many fewer such wasted votes in the second round. 

What are the possible disadvantages of repeat voting?

Costs. A second round adds costs (however, in future voting that may be conducted online, these costs would become much smaller). The additional electoral campaign between the two rounds also increases the costs (but one should remember that two rounds are already used in various elections, albeit not two identical rounds as proposed here). One way to reduce costs is to carry out the second round only when the results of the first round are close (for instance, when the winning margin is below a certain threshold that is specified in advance).

Participation. There may be fewer voters in the first round if voters know they will have the chance to vote in the second round.

Bandwagon effect. Voters with strong or extreme positions, who are much more likely to vote in the first round, may have a big effect on the results of the first round, which may then have a bandwagon effect on the whole election. 

One can think of other ways to overcome the problems associated with elections that we pointed out above. For example, one can repeat the vote three times, with the winner having to win at least two rounds (this applies only to two-outcome elections). Another possibility is to make voting mandatory (as some countries do); while this may resolve the participation issue, it does not resolve the significant polls issue discussed above. Yet another possibility is to weight the votes in the two rounds differently (for instance, depending on the total number of votes in each round). At this point, however, it seems best to leave it as simple and straightforward as possible.

In summary, repeat voting is a simple modification of many current election procedures that is capable of increasing voter participation and yielding more accurate and representative results. Everyone deserves a second chance, as the saying goes. Shouldn’t this include voters and candidates?

Acknowledgments. The idea of repeat voting was developed following discussions at the annual conference of the Federmann Center for the Study of Rationality in February 2017. The author thanks Maya Bar-Hillel, Steve Brams, Shachar Kariv, Orit Kedar, Motty Perry, and Richard Thaler, for their comments and suggestions.

Articles represent the opinions of their writers, not necessarily those of the University of Chicago, the Booth School of Business, or its faculty.