Pages

Wednesday, March 3, 2010

Marginal mayhem- Instant runoff voting results baffle UVA students

University of Virginia students were confused by the results of recent instant runoff voting election for student body. Strangely enough, the candidate with the most 1st and 2nd choice votes lost. Even the winner of the UVA's IRV election didn't understand the results. Consider that UVA "has ranked ...among the top 25 nationally since the first U.S. News rankings came out in 1988." What more do we need to know to tell us that instant runoff voting is bad for voters?

Marginal mayhem
The University Board of Elections should educate the student body further about its voting methodology Lead Editorial / Opinion March 3, 2010

Though polls are closed for the University-wide student elections, questions remain about the University Board of Election’s preferential voting procedures. For those unfamiliar with this system, the UBE uses a technique called instant-runoff voting.

UBE released a statement yesterday to address the confusion surrounding the results of the Third Year Council vice-presidency election. “If no candidate receives the absolute majority of first-rank votes, the candidate who has the least amount of first-rank votes is eliminated,” UBE Chair Jennifer Kim said about the system, “and the second-rank votes associated with the eliminated candidate are distributed to the remaining candidates in the race.” Essentially the system breaks down into a series of instantaneous rounds of eliminations. After a candidate is eliminated, UBE’s system distributes the votes of those students who chose that candidate to their subsequent preferences.

Abebe Kebede, the runner-up in the aforementioned election, received the greatest number of votes in both the first and second rounds, but neither amounted to an absolute majority. Thus, in the third and final round, which was decided by the thinnest of margins — one vote — Natalia Mercado was declared the winner.

This system, complicated as it may be, is designed to select the candidate that best
represents the student body’s interest. Imagine if a candidate participating in a five-way race was supported by a narrow segment of the student population but was overwhelmingly opposed by the majority of University students. Such a candidate might realistically receive the greatest number of votes — for the sake of this hypothetical, say about 25 percent. In an election decided by simple plurality, this candidate would be the winner. In a system of instant-runoff voting, however, the elected candidate would be the one who holistically represented the greater range of student opinion and consistently
received votes during each round. The candidate with the more polarizing campaign, however, would have minimal votes during the following rounds and would be eliminated.

In an e-mail to Kim, Kebede said it was “troubling that all four candidates after looking at the results could not understand the process.” Though candidates are required to know this information prior to entering the race, it is difficult to assume the voters understand the process.

UBE should recognize that if this process had been better communicated to students, much of this controversy could have been avoided. Kim acknowledged that “the average student voter … may not be aware of the intricacies of the IRV process.” UBE’s voting website provided an explanation about how to choose candidates based on preferences. No information, however, was included about how the winner would be determined.

Ultimately, it seemed as if most students were unaware about how winners were chosen. If this had been a more conspicuous election, such as the one for Student Council president, the controversy assuredly would have been much greater. Because instant-runoff voting is not the norm, UBE must be more open about its voting system.


Visit this link to sign up for email updates:
http://feedburner.google.com/fb/a/mailverify?uri=ProtectUsElections-StopInstantRunoffVoting&loc=en_US

2 comments:

rbj said...

Also in Burlington VT, the candidate with the most 1st and 2nd choice votes also lost. But the measure of a candidate's broad support isn't just measured by the number of 1st and 2nd choice votes, anymore than the measure is from just the 1st choice votes. The reason why these surprizes happen in IRV is because IRV doesn't ask the right question, which is which candidate beats every other candidate in voter preference? In other words, which candidate beats every other candidate in a runoff between the two.

Both IRV and the "traditional" election method does not deal with that question when there are more than two credible candidates.

Kathy said...

IRV does *not* solve the spoiler problem of a nonwinning candidate causing the most popular majority favorite candidate to lose the election. This is what happened in Burlington, VT in the last mayoral contest when the pairwise favorite candidate of all voters was eliminated before the final counting round.

Can we get a copy of what all the voters' choices were at UVA so we can analyze the election? If so, please send an e-copy of the voters' choices to kathy.dopp at gmail.com.

Sometimes IRV also exhibits nonmonotonicity where increasing a candidate's vote share causes that candidate to lose the election, whereas the candidate otherwise would have won! IRV's a crazy method of counting votes. Burlington, VT, Pierce County, WA, and Aspen, CO voters recently voted to eliminate it.