Instant Runoff Voting and the Renewal of American Democracy
© March 2007 by Alan F. Zundel
We citizens of the United States like to think our nation is a model of democracy for the rest of the world, but in the eyes of the rest of the world this is clearly not true. Democracy has to do with how citizens elect their government officials, and newly democratizing nations have been rejecting our electoral system for alternative systems which they regard as superior—and with good reason. These nations include Greece, Portugal and Spain in the 1970s, many of the nations of Central and Eastern Europe freed by the collapse of the Soviet Union since the late 1980s, and, most recently, even the nations of Afghanistan and Iraq after U.S.-led military action overthrew their previous governments. Not only have newly democratizing nations been rejecting our electoral system, nations that have long had electoral systems similar to ours, such as New Zealand and England, have reformed or have seriously considered reforming their electoral systems as well.
Why the disenchantment with our type of electoral system? The particular system that is used in the U.S. for most of our elections is called single-member plurality voting. In plurality voting systems, whoever gets the most votes—which is not necessarily a majority of the votes—wins. Our plurality voting system generally uses single-member districts where each geographic district elects only one representative, thus the term, “single-member plurality voting.” (Alternatively a plurality system could use multi-member districts, each of which elects several representatives.) Plurality systems were established in England at the dawn of the democratic era and spread from there to England’s now-former colonies, including the United States. Beginning in the mid-nineteenth century British reformers noticed defects in plurality systems and advocated alternative electoral systems such as proportional representation, which by the turn of the century was adopted by most of the nations of continental Europe. While under plurality systems whoever gets the most votes wins, under proportional representation offices are awarded to parties in rough correspondence to the proportion of the total vote each party wins. Plurality electoral systems typically result in two dominant political parties, while proportional representation systems produce a variety of parties able to gain elective office.
In the United States, critics have accused single-member plurality voting as being at the root of many of the problems that plague our political life, such as:
• Restricted choices for voters. Ours is a classic “two-party” system, which means that only candidates from two dominant political parties—the Democrats and the Republicans—have a realistic chance of winning the most votes in an election. If a voter wants to vote for someone from a less-established party or for an independent candidate, not only will her candidate be unlikely to win, the voter might end up helping the dominant party she dislikes more to win by giving up her chance to vote for the other dominant party. Called the “spoiler” effect of plurality systems, this artificially depresses the vote for candidates outside of the two dominant parties. In addition, in many voting districts only one of the two dominant parties has a realistic chance of winning, due to the way district boundaries are deliberately drawn to include or exclude particular geographically-located blocks of voters. This means that voting is a waste of time for almost all of the voters in the district, as that one party’s candidate is all but assured of winning before voting even begins.
• Negative campaigns that obscure candidates’ positions. Because there are only two viable parties and whoever gets the most votes wins, the best strategy for a candidate of a dominant party is to (1) attack the other dominant party’s candidate so as to deter people from voting for that candidate, and (2) be as broad and vague as possible about his or her own positions so as to avoid alienating any voters. This results in campaigns that are negative in tone and empty of informative content. (Because they are so unlikely to win, the campaigns of independent candidates and candidates of other parties are mostly ignored by the dominant party candidates, the media, and the general public, and thus have little effect on the overall tone of election campaign activities.)
• Low voter participation. Very few of those who are eligible to vote in the U.S. actually bother to register and vote: only about half vote in Presidential elections, about a third in Congressional elections, and even fewer in state and local elections. Two reasons for this voter tune-out are the restricted voting choices and uninspiring election campaigns described above. Voters also tune-out because they feel their vote does not make a difference, which is explained further below.
• Diminished government responsiveness to voters. Because of the restricted voting choices, voters have less leverage to get politicians to respond to their concerns. If both the dominant parties ignore some issue, or if both the dominant parties take similar positions on an issue, dissatisfied voters have nowhere to turn. Similarly, if a voter disagrees with one of the dominant parties on one issue and with the other dominant party on another issue, the voter has nowhere else to turn. Elected officials thus feel free to ignore voters on many issues, and instead their actions are influenced by big money campaign contributors or lobbyists and political “insiders” with special access.
Such oft-decried problems as those described above are not due to massive numbers of unethical politicians and apathetic citizens, but are a by-product of our single-member plurality system. Although there have long been activists and intellectuals who have pointed this out and called for replacing our system with proportional representation, these efforts have never gotten much traction. The problem is not only that the two dominant political parties wish to retain their dominance—although that is certainly an obstacle—but that proportional representation would be such a radical change in the U.S., for example, abolishing single-member districts and the kind of geographic representation that U.S. citizens have grown used to. But there is another type of electoral system that has a better chance of being instituted here, and in fact has been rapidly gaining attention in the last few years.
Instant Runoff Voting
In 1856 the British reformer Thomas Hare proposed that voters be given the opportunity to rank candidates in the order of each voter’s preferences, and that this ranking be factored into the distribution of seats in England’s multi-member districts. This procedure was rejected as too complicated, but the American professor W.R. Ware saw that Hare’s ranked preference voting could be adapted to single-member districts as well. This idea received little attention in the United States or England at the time, but at the beginning of the twentieth century an Australian professor advocated it for elections to their House of Representatives and it was eventually adopted and has continued to be used for elections to that body since 1918. This voting system, usually referred to as the alternative vote, has also been used in Australian state house elections, the Irish presidential election, and a few other places.
The distinctive characteristics of the alternative vote are that it requires a majority of votes rather than a plurality to elect someone, and it allows voters to rank candidates in the order of each voter’s preferences. If a majority of votes is not obtained by any candidate when all the votes for first-ranked candidates are counted, the candidate with the fewest of these votes is eliminated and the votes are recounted using the highest-ranked still-eligible candidate on each voter’s ballot. This process of elimination and recounting is repeated until a majority is obtained. In the U.S., the term “instant runoff voting” caught on for this type of voting system, because it eliminates the need for a traditional runoff election. Traditional runoff elections are found in places that have a majority requirement for electing a candidate, as some U.S. electoral districts do, but in which voters cannot rank candidates. In these situations, if no candidate receives a majority of the votes then a second, runoff election must be held between the two top voting-getting candidates from the initial election. The ability to rank preferences for candidates on the instant runoff ballot makes the second election superfluous, and thus does away with the added effort and expense involved in holding it.
Instant runoff voting (IRV) also frees voters to vote for less-established party or independent candidates without causing their less preferred dominant party candidate to win, thus eliminating the spoiler effect. Because only a candidate who obtains a majority of votes can win, a candidate cannot be elected simply because the voters who oppose him or her had split their votes between other candidates. Rectifying the spoiler effect has been one of the principal motivating factors in efforts to institute IRV in the United States. An early example is Ann Arbor, Michigan during the 1970s, where conservatives were winning the mayoral election because the larger liberal vote was split between the Democrats and the Human Rights Party. The liberals got IRV adopted and began winning, but the conservatives got it repealed in 1976.
With the number of non-majority winners in U.S. elections increasing, a wider movement to institute IRV began gaining momentum in the late 1990s, with legislative efforts begun in Texas in 1997, New Mexico, California and Vermont in 1998, and North Carolina and Alaska in 1999. The Presidential campaigns of Ralph Nader as a Green Party and an independent candidate during the very close Presidential elections of 2000 and 2004 brought renewed attention to the problem of the spoiler effect in the U.S. voting system, spurring the IRV movement to an increasing number of successes at the local government level. IRV measures were passed in Santa Clara County CA in 1998, Vancouver WA in 1999, San Leandro CA in 2000, Burlington VT and San Francisco CA in 2002, Ferndale MI and Berkeley CA in 2004, Tacoma Park MD in 2005, and Davis CA, Oakland CA, Minneapolis MN, and Pierce County WA in 2006. IRV has not yet been implemented in most of these localities due to factors such as not having IRV-compatible voting machines or apparent conflict with state election laws, but implementation took place in San Francisco beginning in 2004 and in Burlington beginning in 2006. In addition, the states of Arkansas and Louisiana now provide IRV for local and statewide elections on the absentee ballots for overseas military personnel from their states, bills have been introduced in a number of state legislatures to establish IRV for all state-wide and federal elections in those states, and IRV-related bills have been introduced in the U.S. House of Representatives.
How Would U.S. Politics Change Under IRV?
The existence of the traditional runoff election in some places in the U.S. should remind us that despite the current predominance of single-member plurality voting, there has been variation in our electoral system as well as significant change across time. The United States has been in a fairly continuous state of electoral evolution since its beginnings: for example, extending the vote to the unpropertied in the early nineteenth century and to women in the early twentieth; introducing the direct election of Senators, the initiative, the referendum and the recall during the Progressive era; insuring the voting rights of black Americans in the civil rights era; and opening the vote to 18-to-21 year olds at the time of the Vietnam War. As for variation, multi-member districts were once common in state legislative elections, multi-member Congressional districts could be found up until the 1960s, proportional representation and ranked preference voting have been used in dozens of localities at one time or another, and cumulative voting (a form of ranked preference voting) has recently been making inroads as a solution for cases where issues of race are implicated in the drawing of district boundaries. Clearly, change in the U.S. voting system is not unprecedented. If instant runoff voting was to be implemented as our next major electoral reform in the U.S., what would the effect be? Let us examine this question from the standpoint of the four major problems of our current electoral system described above.
First of all, IRV frees voters to vote for less-established party or independent candidates if they would like to, due to the elimination of the spoiler effect; they can still show a preference for one of the two dominant parties by choosing it as a second choice in case there is no majority winner in the first round. The two-party system would be altered by giving voters this freedom, although it is not likely to be ended by it. Single-member districts, which we would still have, tend to produce two dominant parties, but single-member districts with ranked preference ballots generally have a larger number of active parties than single-member plurality systems. For example, the Australian House of Representatives tends toward a centrist, bipolar party system, composed of the Labor Party versus a coalition of the Liberal Party and the National Party (the latter two are urban conservative and rural conservative parties, respectively); other party candidates are rarely elected. But minor party candidates have, at times, wielded considerable influence on a major party’s policy positions by bargaining with their voters’ second-choice preference ranking, and in both Australia and Ireland minor parties often link up with one of the dominant parties by asking their voters to cast their second-choice vote for their partner party. Thus under IRV in the U.S. we would expect Democrats and Republicans to continue to dominate, but other parties to play a more substantive role in our electoral system. The Democratic and Republican parties would continue to play the role of “large-tent” parties aiming at broad voter appeal and winning most of the elective offices, while smaller parties would seek out ideological or issue niches with smaller sets of voters that they would use as bargaining chips in negotiations with a dominant party partner. This arrangement would give voters more electoral choices in which to express their political preferences (and their expression of preferences would be more clearly reflected in the initial election results), addressing the first of the four problems discussed above.
Second, the style of election campaigns should change for the better at least to some degree, although the evidence is sketchy and mixed at this point. IRV proponents contend that candidates in IRV elections will run more positive, issue-oriented campaigns, because they will not want to alienate potential second-choice votes from rival candidates’ supporters. On the one hand, news reports indicate that some rival candidates in IRV elections in San Francisco have run unusually positive and cooperative campaigns, endorsing each other as second choices and even co-hosting campaign fund-raisers together. On the other hand, polls of voters in the first San Francisco IRV election found that while 15% thought the campaign was less negative than usual, 14% thought it was more negative, with the majority seeing no difference. In addition, election campaigns for seats in the Australian House have been characterized as displaying a “confrontational” style of politics, with the dominant liberal and conservative parties facing off against each other. Putting these clues together, it is likely that there will be more positive and cooperative relations between the campaigns of candidates who are closer on the ideological spectrum (such as between Libertarians and Republicans, or between Greens and Democrats, or between rivals in the primary of a particular political party), due to the seeking of second-choice votes, but that campaigns will continue to be negative between the dominant party candidates (Democrats versus Republicans), because these candidates would be the most likely to reach the final round of vote-counting and at that point next-choice votes will be irrelevant. But it is also possible that the latter campaigns will become more positive as well, if the dominant parties find that this attracts second-choice votes from voters in the middle of the ideological spectrum (for example, supporters of “centrist” independent candidates). In sum, there should be at least a little, and possibly a good amount, of improvement in the quality of election campaigns.
Third, voter participation will likely rise. This certainly will be true when turnout under IRV is compared to turnout for second elections in traditional runoffs, as the latter tend to have very low turnout rates. For example, a study of the 2005 San Francisco Assessor-Recorder election found a dramatic increase in turnout under IRV when compared to the traditional runoff (in some areas tripling it), especially in the poorest and most racially diverse neighborhoods. Whether voter turnout under IRV will be higher than turnout in a non-IRV general election is less certain, but still likely, based on the fact that voters will have more real choices, independent and “minor” party candidates will be able to campaign more effectively once freed of the spoiler effect, and dominant party candidates will have to work harder for votes because of the increased competition. There is some evidence to support this expectation. In the Ann Arbor IRV elections of 1973 and 1975, voter turnout was 28% higher than in the elections of 1971 and 1977 without IRV. Other forms of ranked preference voting in the U.S. have been associated with higher voter participation as well. In Cambridge MA, which has a multi-member ranked preference electoral system, voting turnout rates have been in the 49-60% range, much higher than most municipalities. And a study of a switch to cumulative voting (another form of multi-member ranked preference voting) in several U.S. communities found turnout increased by about five percentage points and sustained this higher turnout over time.
Finally, and perhaps most importantly, the government should become more accountable to voters. According to Farrell and McAllister, “studies based on surveys of politicians have demonstrated how electoral systems that are characterized by a candidate-orientation in politics [voting for a particular candidate rather than for a particular party] and high degrees of preferential voting tend to produce greater attention to personal vote chasing by politicians and the maintenance of close links with their electorates.” In other words, where elections are candidate-oriented, as they are in the U.S., and voters can indicate relative preferences for candidates, as under IRV, candidates and elected officials pay more attention to the voters and what the voters’ concerns are. I regard this as the most important of the likely effects of instituting IRV in U.S. elections: tipping the balance of political influence away from big money campaign contributors and political insiders with special access, and back toward the common citizen. This should have collateral effects as the common citizen’s increased importance encourages more efforts at civic education and grassroots organizing, improving the functioning of our democracy overall. As noted by Farrell and McAllister in their analysis of data from a number of nations, ranked preference voting is associated with higher levels of voter satisfaction with democracy.
Many other arguments both pro and con have been raised regarding IRV, and I will address a few of the more prominent of them in this section.
Are office-holders who win with a majority of the votes, as under IRV, regarded as more legitimate than those who win without a majority, as can happen under a plurality voting system?
An argument frequently used in support of IRV is that by requiring a majority to elect a candidate, an elected official will have greater democratic legitimacy than if elected by less than a majority, an occurrence which has become more frequent in our plurality elections in recent decades. As far as I can tell, the alleged illegitimacy of non-majority winners is not much of a problem, as U.S. citizens commonly accept the legitimacy of officials elected with less than a majority under current rules. For example, even after the extremely close Presidential election of 2000, with all of its contention over ballot-counting methods and court decisions, 83% of U.S. citizens said they would accept Bush as a legitimate President, including 70% of Gore supporters, even though more than 70% of Gore supporters said the Bush v. Gore Supreme Court decision was unfair. But there is evidence that voters regard IRV elections as more fair than majorities obtained under traditional runoff elections. In San Francisco, 37% of the voters in the 2005 election found IRV to be more fair than the previous traditional runoff system, more than two times the 15% who said the previous system was more fair. If this is true, they would probably see IRV as more fair than plurality elections as well.
Is IRV a violation of the “one person, one vote” rule?
No—the “one person, one vote” rule means that each voter should have an equal influence on the outcome of an election, and in the final round of counting under IRV each voter cast only one of the votes. The confusion seems to come from misconceiving IRV as though it is like a plurality election in which someone “should have” won on the first count, but some voters get to cast new votes after the fact to change the results. IRV is more accurately conceptualized, as its name implies, like a traditional runoff election in which the runoff vote (if a runoff is necessary) is cast immediately rather than at a later date.
Is IRV too confusing for voters?
U.S. voters should have no problem understanding IRV. Polls of voters using the new system in San Francisco in 2004 found that 86% said they understood the new voting system “perfectly well” or “fairly well,” and 61% preferred the new system versus 13% who preferred the old. In 2005, 87% said they understood the new system perfectly or fairly well, 46% said the system was “easy” or “fairly easy” to use compared to 16% who found it “difficult” or “fairly difficult,” and 55% preferred the new system versus 17% who preferred the old. The cumulative vote and the single-transferable vote are both preferential choice voting systems and at least as complex as IRV, yet studies of their use in the U.S. show that these systems have not seemed hard to understand for most voters, have not depressed turnout, and have not resulted in a higher than usual number of incorrectly completed ballots. Studies of significant electoral changes in other nations have, in general, shown an “impressive” ability of voters to develop an understanding of new voting systems.
Is IRV constitutional?
I am not aware of any conflicts between IRV and the U.S. Constitution, but Langan makes the argument that IRV is in conflict with some states’ statutory or constitutional law. In some of these states, the law mandates election by a plurality of votes. In others, the law requires a majority winner for certain elections but “majority” has been (or is likely to be) defined by the courts to preclude a “preferential majority” using second, third, and so forth, choices. (That is, a majority must occur in the first choice of voters in the initial election, or in a traditional runoff election.) This is certainly a problem for implementation of IRV in these states, but the solution, although it may be politically difficult, is obvious: a change in those states’ election statutes or constitutions. Langan’s case therefore is not really an argument against IRV, unless reasons can be given why the electoral law or relevant constitutional provisions in those states should not be changed. Without such reasons, this argument simply begs the question of whether IRV is a good idea.
Will IRV cost more, or less, than our current voting system?
The answer to this question depends upon a number of contingencies. First, if a district has a history of traditional runoffs, IRV will almost certainly save a great deal of money by eliminating the second election. Second, it depends on the way votes are tabulated. If a district hand-counts ballots and there is not a majority in the first round, IRV vote-counting will take longer and thus cost more; if a district has vote-counting machines that are compatible with IRV, the switch will cost very little if anything; if a district has voting-counting machines that are not compatible with IRV, it will cost money to reprogram or replace the machines. Finally, there are the start-up costs in switching to IRV of educating voters and retraining administrators and election workers for the new system. The cost, or savings, of switching to IRV thus will vary from one place to another.
Does IRV favor some political parties over others?
In general, no; in specific circumstances, it might. There is nothing in IRV as a voting system that gives an advantage to Democrats over Republicans, or Republicans over Democrats, but in a specific location, switching to IRV might advantage either the Democrats or the Republicans. For example, where Democrats have been winning with less than a majority because the conservative vote is split between Republicans and Libertarians (as in Alaska), the Republicans would benefit from switching to IRV. But where Republicans have been winning with less than a majority because the liberal vote is split between the Democrats and the Green Party (as in New Mexico), Democrats would benefit by the switch to IRV. Nor is there anything in IRV as a voting system that gives an advantage to less-established parties over more-established parties. Less-established parties would have a better chance at winning than they do under our current system, but IRV does not give them any special advantage. In fact, wherever there are single-member districts two dominant parties emerge and have the advantage, whether under IRV or under our current plurality system.
Several serious problems in the U.S. political process are related to our single-member plurality voting system. This review of the research and arguments related to Instant Runoff Voting supports the contention that IRV will eliminate or mitigate these problems, and that overall it would be worth switching from our current voting system to IRV.