Analysis: IRV and the Ferndale Mayor’s Race
(Kevin Deegan-Krause, Nov. 7, 2013)
Editor’s Note: Prior to the election, The Oakland County 115 News invited readers to share how they would vote in an election where the four candidates were ranked. We kept the respondents anonymous and sent only the resulting votes to Political Science Professor Kevin Deegan-Krause to get his take on how they compared to the actual election results. The following is his response:
Ferndale’s recent mayoral election results give us a hint of why voters in 2004 were smart to pass a ballot initiative to establish a voting method called Instant Runoff Voting. They also show why it is unfortunate that bureaucracy at the state level keeps the city from actually using the method that its citizens voted into law.
In simplest terms, Instant Runoff Voting is a system that asks voters to rank candidates instead of just picking one. If nobody wins a majority, it looks at the second-choice of the least popular candidate and moves the vote to the second choice and keeps doing this until some candidate receives more than half of the votes. This voting system ensures that the winner is always a candidate with a majority. It also prevents “spoiler” candidates who keep other, similar candidates from winning. (The most famous example, arguably, is the effect that Ralph Nader’s campaign in Florida had on Al Gore in 2000).
Instant Runoff Voting is especially important places like Ferndale which do not have party primaries and simply give victory to the candidate with the most votes. This system for electing mayors and council members is not unusual in Michigan, but it has the potential for big problems because there is no way to narrow down candidates to a clear choice. The problem is even bigger for cities where the candidates do not carry party labels. In practice it means that many candidates from the same side are all competing with one another in the only round of the election. This means that if many candidates from one side split the vote, a lone candidate from the other side might win even if his or her side is much less popular. One way to solve this might be to introduce party primaries, but the creators of our city charter seem to have thought this might be too divisive, and it would not resolve the problem of third parties, especially if two or more parties were on the same side of major political issues.
The answer that voters picked in 2004 is an answer that is unique in Michigan but not that unusual in other places. Ireland, Australia and Malta all use a version of this system as do cities including San Francisco, Oakland, Minneapolis, and many other small cities and special voting districts. Many student councils also use this approach and many corporations and membership organizations have adopted it for their board elections.
Ferndale’s 2013 elections give us some insight into how the system would work and why it might solve some problems. The problem in this year’s vote is that no candidate for mayor got a majority of the vote. Dave Coulter came close with 47.7% of the vote, but that means, at least in theory, that 52.3% of the voters preferred some other candidate, and that Coulter might theoretically have lost a one-on-one race against a single candidate, so it is worth looking at how Instant Runoff Voting offers a better arrangement.
In order to do that, of course, we’d have to know not only the vote totals but how voters would have ranked the candidates. Voting results don’t give us that information but Crystal Proxmire at the Oakland County 115 conducted an informal poll that helps. While it is by no means a scientific poll, she got a reasonable number of responses that corresponded fairly well to the overall results and so we can use her numbers for an example. Her 34 respondents sent in lists of all four candidates in the preferred order. There were 14 different rankings in all, the most common of which was Coulter>Covey>Wells>Parton which was sent in by 10 people.
The great thing about having this ranking information is that we can figure out who voters prefer in head to head elections. To do that, I took the 34 responses and weighted them slightly to reflect the actual first choices of voters in the elections (doing this did not produce that much change), and then looked to see for each pair of candidates who the majority of voters preferred. The results are in the table below. In the actual election, Coulter led Covey 47% to 33%. In a head-to-head election with only these two candidates, the Oakland County 115 poll suggests that Coulter would have won 61% to 39%. Likewise Coulter would have won against the other two candidates: 81% to 19% against Wells and 90% to 10% against Parton. If Coulter had not run, the polls show that Covey would have beaten Wells 60% to 40% and Parton 87% to 13%. In a Wells-Parton race, Wells would have won with 79% to 21%.
If these two candidates ran head-to-head: | 2. …against this candidate | ||||
Coulter | Covey | Wells | Parton | ||
1. This candidate would receive this % of the vote… | Coulter | – | 61% | 81% | 90% |
Covey | 39% | – | 60% | 87% | |
Wells | 19% | 40% | – | 79% | |
Parton | 10% | 13% | 21% | – |
From this we can see that Coulter’s 47.7% was not a sign that a majority preferred another side and merely a sign that voters on the same side tended to split their vote a bit among other candidates who were in some ways similar. But the important thing is that we could not have known this without knowing voters’ second, third and fourth preferences.
That is where Instant Runoff Voting comes in. It formally collects those other preferences on the ballot itself and then uses an easy procedure to figure out the most-preferred candidate.
Here’s how it works using the results we found in Ferndale. On the ballot, voters rank all the candidates using as many numbers as there are candidates, in this case 1,2,3 and 4. Then the ballots are tallied as normal according to the number of first place votes received by the candidates. In this sense the system is exactly the same as what we do now, and if one candidate gets a majority of first-place votes, it stops there. The candidate with a majority wins.
The instant runoff comes in where one candidate does not win a majority, as in Ferndale in 2013. In this case the procedure is quite simple. We look at the ballots of the candidate with the least number of first place votes and see who those voters liked second-most. We then give the ballots to those candidates. In the case of the Ferndale election the example is actually rather simple. Dave Coulter got 1634 first-place votes but he would have needed 1714 to have an absolute majority so we look at the candidate with the smallest number of first-place votes, Sherry Wells. According to the Oakland 115 survey, all of Wells’ voters put Coulter in second place, so all of her 310 first place votes go to Coulter and he wins with a majority of 1944 votes out of 3428. End of story.
If the results for Parton and Wells had been reversed and Parton had received the fewest number of votes, the story would have been slightly more complicated, but not much. Parton’s supporters all preferred Wells second, so their votes would have gone to Wells. But adding 310 votes to 335 would not have given Wells a majority (in fact she would have remained in third place), and so we would need to go to another instant round. In this case, with Parton out, we now look at the next preference of Wells’ voters (some of whom initially came from Parton). From the results of the poll, just about three-fourths of Wells’ voters preferred Coulter to Covey, so 477 votes would have gone to Coulter, giving him 2111 votes.
In either case, Coulter wins a majority because the voters of the candidates with the fewest votes tended to prefer him over Covey. If Covey, Wells and Parton’s voters had all been on the same ‘side’ with their voters preferring one of those three candidates first, the transferred votes would have helped Covey more and resulted in his victory. In that case, Instant Runoff Voting would mean that Wells and Parton did not play a ‘spoiler’ role in keeping the candidate of the most popular side from getting elected.
So in these examples, Instant Runoff Voting serves the purpose it was intended—it ensures that the candidate who would have won head-to-head races against the others is the one who actually gets elected. It lets us keep our non-partisan election and means we don’t have to add a potentially expensive primary while still making sure that the most popular candidate wins.* Now if only we can get the state to make sure that the voting machines it is selecting will allow us to vote this way. But that is another story, one that is being covered regularly in the Oakland County 115.
Read our previous article on Instant Runoff Voting at https://oaklandcounty115.com/2013/09/30/irv-whats-missing-from-ferndales-election/
*As somebody who teaches political science for a living, I am obligated to point out that it is not quite that simple and that there are some other twists, but they are so minor that I do not want to confuse matters by introducing them here. I would be glad, however, to discuss them in more detail by email: kdk [at] wayne {dot} edu
Kevin Deegan-Krause is an Associate Professor of Political Science at Wayne State University. He teaches on democracy and elections and civic engagement. He was also elected in 2012 as a member of the Ferndale School Board. He sent his comments all the way from Bulgaria where he is studying political parties and elections.
Appendix:
Numbers I used to calculate the results above.
Ranking submitted | Number submitting | Total voters if extrapolated using actual results | First choice | Total first choice for candidate (equal to actual election results)** | |||
Coulter | Covey | Wells | Parton | 10 | 711 | Coulter | 1634 |
Coulter | Covey | Parton | Wells | 2 | 142 | ||
Coulter | Covey | – | – | 1 | 71 | ||
Coulter | Wells | Covey | Parton | 7 | 497 | ||
Coulter | Wells | Covey | – | 2 | 142 | ||
Coulter | Wells | Parton | Covey | 1 | 71 | ||
Covey | Coulter | Wells | Parton | 4 | 759 | Covey | 1137 |
Covey | Coulter | – | – | 1 | 189 | ||
Covey | Wells | Coulter | Parton | 1 | 189 | ||
Parton | Wells | Coulter | Covey | 1 | 167 | Parton | 334** |
Parton | Wells | Covey | Coulter | 1 | 167 | ||
Wells | Coulter | Covey | Parton | 2 | 207 | Wells | 310 |
Wells | Coulter | – | – | 1 | 103 | ||
**Reduced by one from the actual results to make calculation slightly more simple |