Vote splitting

From electowiki
This video is a simple, fun, and effective explanation of how "vote splitting" sometimes prevents the election of a popular candidate. This video was published by the Center for Election Science in 2013, and advocates for approval voting.
Wikipedia has an article on:

Vote-splitting (sometimes referred to as a "split vote") occurs in an election when a voter supports more than one candidate but is either unable to express their endorsement on the ballot, or doing so worsens the likely outcome from the voter's perspective. Vote splitting generally isn't a problem when two candidates are on the ballot competing for the same seat, but vote splitting can be a problem when three or more "serious" candidates compete for a single seat. The "vote splitting" problem is sometimes referred to as the "spoiler effect", since the third (or fourth, or fifth) candidate is often accused of spoiling the election for one of the two perceived frontrunners.

The first kind of vote splitting is a consequence of the ballot type. This means more expressive ballots have less or no vote splitting, if paired with a method that safeguards against the second type of vote splitting..

The second type of vote splitting is normally defined by either unwanted consequences or strategic voting but is hard to define rigorously. There are a number of criteria related to vote-splitting which are more rigorous.

Single-mark ballot

See also
Single-mark ballot

The biggest issue with a single-mark ballot is that it can cause a high amount of vote splitting. It is particularly problematic in single member plurality elections. However, other systems that still use a single-mark ballot such as Runoff voting still have vote splitting.

A standard example of vote-splitting is when two candidates are similar, so they each get half the votes they would if the other were not running. This is a good example of failing the Independence of clone alternatives criteria. Another issue is that the voter's preferred candidate may be highly unpopular in the constituency. This means that using their only vote on that candidate has no influence on the result and could be better served on a second or third choice. This is a problem of voter impact which can be thought of how clearly a voter's true choice is translated into the election of a candidate. This is often referred to as the Wasted vote problem if they still vote for their favorite or the favorite betrayal problem if they vote for another candidate. This system rewards voters for not voting for whom they really want and as such promotes strategic voting. Furthermore, it implies that the voter's choice will be heavily dependent on their estimates of how others will vote, not on their preference. The use of fake polls or deceptive reporting can have large effects on election outcomes where strategic voting is emphasized. With all systems of voting, the second-order effects of people trying to vote strategically must be considered.

Ranked ballots

See also
Preferential voting

The major advantage of ranked voting is to eliminate the standard forms of vote splitting which are present in Single-mark ballot systems. This can largely eliminate issues that give rise to the wasted vote, though some ranked methods still fail independence of clones and most ranked methods fail the favorite betrayal criterion.

Unfortunately, it can be proven mathematically that no non-random ranked ballot method can be entirely free of spoilers due to Arrow's impossibility theorem. In particular, independence of irrelevant alternatives is incompatible with majority rule. Ranked ballots are still proposed as a solution to vote splitting for two reasons. First, the types of election where spoilers exist are greatly reduced; and second, for any deterministic method to be entirely free of spoilers requires some assumptions that are often considered unrealistic - see below.

Cardinal ballots

See also
Cardinal voting systems

Some cardinal voting systems pass independence of irrelevant alternatives, meaning that the removal of a candidate who didn't win will have no effect on who won. For instance, both range voting and Majority Judgement do so.

This fact, however, only implies a lack of a spoiler effect if the voters don't normalize or otherwise adjust their ballots based on the candidates who are running. In a ranked ballot format (absent truncation and equal rank), adapting one's ballot to who's running would be considered tactical voting, but as cardinal systems allow for multiple honest ballots, the line between honest voting and tactical voting is blurred.

For Majority Judgement in particular, Michel Balinski and Rida Laraki argue that the categories have common definitions, which implies that the voters' evaluation of the candidates will be independent of who is running, and hence that its IIA compliance eliminates the spoiler effect.

Relation to proportional representation

See also
Proportional representation

Vote splitting is often conflated with Proportional Representation but they are completely distinct concepts. Vote splitting is related to strategic or expression issues at the time of filling out a ballot by a voter. Proportional Representation is a measure of the outcome of an election. The relationship is that vote splitting is a major cause of reduced Proportional Representation. Systems designed to achieve high Proportional Representation but that still use single-mark ballots often do not reduce the amount of vote splitting but instead mask its effects at the partisan level. A mixed electoral system, for example, still has a single member plurality component with all the vote splitting issues of a full single member plurality system.

Another confusing point is that Proportional Representation is most clearly defined for Single-mark ballots but Single-mark ballots have the largest problem with vote splitting. Using another ballot format makes it much less clear what proportional representation means. Some concepts include proportionality for solid coalitions, justified representation and stable winner sets.


In the United States, a famous example of a split vote occurred in the 2000 U.S. Presidential Election when Green Party candidate Ralph Nader attracted voters who might otherwise have voted for Democratic Party candidate Al Gore because of the similar left-wing platforms of both candidates. Because of the very narrow margin of victory of Republican Party candidate George W. Bush over Gore, many blamed Nader's candidacy for causing his loss and thus being a spoiler (although the votes that went to the eighth-place candidate in the contested state of Florida could also have potentially covered the split).

In Canada, the Progressive Conservative Party had held power under Brian Mulroney throughout much of the late 80's thanks to a loose coalition of conservative voters in the western provinces and nationalist voters in Quebec. The coalition collapsed, though, and in the 1993 election the right-wing split the votes between the Reform Party and the PCs, with nationalist voters in Quebec flocking to the newly founded separatist party Bloc Quebecois. The result was over a decade of Liberal rule in Canada.

In modern-day Canadian politics, the Reform Party and the Progressive Conservatives have now merged into the United Conservative Party of Canada, and have been in power since 2006. The left-wing now faces the same vote-splitting that plagued the right, with the four left-wing parties being the Liberals, New Democrats, The Bloc Quebecois, and the Green Party. In the past two election cycles (2008, 2011) there have been efforts among left-wing voters to vote strategically to defeat conservative candidates.

See also