Across Utah elections are being held using a voting system called instant runoff voting (IRV), ranked-choice voting, alternative vote, or preferential voting.
This system is a fairly standard, reasonable method for selecting one winner from a field of many candidates. It is, in my opinion, much better than the plurality system that is most commonly used in elections in the United States. But when electing more than one member at once, there are different variants of the ranked-choice system, and the choice of which one you use has significant consequences.
The standard, traditional method of using ranked-choice voting for multimember elections is called proportional ranked-choice voting. In an election with k seats proportional ranked-choice awards a seat to each candidate that receives a fraction of the vote more than 1⁄k. So in an election for only one seat, anyone who wins more than ½ of the vote wins. In an election of two seats, anyone who wins more than ⅓ of the vote wins a seat, and so on. If not enough candidates win 1⁄k of the vote, then the ranked ballots are used to eliminate unpopular candidates and redistribute votes until all the seats are filled with 1⁄k of the vote. The details of how this works are explained nicely at the FairVote website.
As I understand it, the current method being used in some Utah multimember elections is what is sometimes called preferential block voting. This method requires a candidate to win at least ½ of the vote to get a seat. If no candidate passes ½ of the vote initiall, then candidates who have the fewest first-choice votes are eliminated and the second-choice on those ballots is counted as a first choice. This process is repeated until a candidate exceeds ½ of the votes, and that candidate is elected. Then the winner's name is struck from the ballots, and the previous losers are restored to their ballots, and the process is repeated.
This method is much simpler to understand than the traditional method, but it gives very different outcomes. Specifically, the Utah system is equivalent to holding a single-member election for the first seat, and then holding a separate single-member election for the second seat, and so on. This effectively guarantees that the majority will choose the winner for every seat. If a multimember district is filling nine seats, all nine will go to the majority party, even if that majority consists of only 51% of the voters.
Under the proportional ranked-choice system, minorities are represented approximately proportionately. So a minority consisting of 34% of the electorate could win one seat in a two-member election, and a minority consisting of 11% of the electorate could win one seat in a nine-member election.
Finally, there are two alternative methods that I and some of my mathematical colleagues noticed being used in some recent GOP caucases. I believe these methods are used by people intending to use preferential block voting but hoping to save time by skipping some steps. I call these two methods lazy methods.
It is important to note that all voting methods have the potential to produce some paradoxical or counter-intuitive results, but methods like the lazy block are much more likely to produce unfair or paradoxical results than either the traditional ranked choice or the Utah/preferential block method. I assume that the lazy block method is only being used in the local precinct caucuses, but since the counting under any ranked-choice method is trickier than in a traditional plurality, we really need some checks in place (say an auditor) to ensure that that each multimember election is indeed tallied according to the accepted rules, and not according to a random computer programmer's interpretation of those rules.
It is also important to be aware of the potential problems in our local caucuses. I have found my precinct to be very open to accepting my suggestions about voting rules. I encourage you to talk to your precinct about making sure that they are not using lazy block ranked choice.
# of votes | Preference order |
1 | ABCDE |
2 | BACDE |
2 | CABDE |
2 | DABCE |
3 | EABCD |
Under the preferential block method, the first runoff gives B as one winner, exactly as before. For the second runoff, return all the candidates to the pool who were previously eliminated.
Since B has already won the first seat, then B is eliminated, so ballots that had B first now go to A. This gives 3 votes to A, 2 to C and D, and 3 to E. Eliminate C or D (by coin flip, since they are tied for last)--say C. So C's votes go to A, who now has 5--still not a majority. D has fewest, so is eliminated, and D's votes go to A, who wins.
So the winners in this method are A and B. The choice of A seems reasonable, since A is almost everyone's second choice. Many would argue that B is also not a completely unreasonable choice here.
B and D are elected, despite the fact that A is everyone's first or second choice and 80% of voters prefer A and C both more than D.
B is now removed and B's votes go to D (since A and C are out), this makes B and D the winners—the same result as proportional ranked choice in this case.
In the Utah county caucuses in 2016 each precinct chose from one of the following voting systems:
The plurality voting system has the advantage of being fast and easy to implement, but it will often elect a candidate that most of the voters dislike. The more candidates that run, the weirder the results of a plurality election are likely to be.
As an example, in 1992 the presidential election was split three ways: between Bill Clinton, George H. W. Bush, and Ross Perot. Clinton won the election with 43% of the vote, Bush had 38%, and Perot had 19%.
Given what we know of Perot supporters, it seems very possible that if the candidate with the fewest votes (Perot) had been eliminated and a runoff election held between Bush and Clinton, Bush could have been the majority winner. So was the plurality system really fair? Did it reflect the will of the people?
The multiple-round runoff system involves eliminating the least popular candidates and having a runoff among the most popular candidates. Sometimes additional runoff votes must be held until only one candidate receives more than 50% of the vote. This system is much less likely to give a result that voters feel is unfair, but it takes a long time to hold and a lot of work to count the ballots.
Instant Runoff Voting (IRV) has the advantage of giving a result similar to the multiple-runoff system, and it is usually considered more fair than the plurality method. It takes a little more time than the plurailty system but a lot less time than the multiple-runoff system. It only requires one ballot and the main cost is in the additional counting. IRV is used in Australian national elections, and in several US cities and states. Overall this seems like a good compromise---more fair than plurality and easier to implement than multiple runoff.
The main problem with IRV is that many people implement it incorrectly (lazily), and the incorrect implementation can give very surprising (some would say unfair) results. The difference between the different methods may seem minor, but these little differences lead to very different results.
Copyright © 2021. Tyler Jarvis.