In 2019 two cities in Utah began using a type of ranked-choice voting for elections. And while ranked-choice voting is more effective in increasing minority representation than first past the post voting, Utah’s version may be harmful to minorities, argue Jack Santucci and Benjamin Reilly. As votes under this system can ‘cascade’ downwards from the first winner to others from the same party or group, they write that ‘block-preferential’ voting could lead to unfair outcomes if adopted more widely.
Americans seeking an end to polarization and genuinely multi-party politics are pinning their hopes on ranked-choice voting (RCV). The success of RCV in Maine and its uptake in cities and towns across the country marks an important step in this process.
However, a new kind of ranked-choice voting has emerged in some multi-member town councils in Utah. It looks a lot like the proportional version that multi-party advocates want, but works against the interests of fair representation. This “block preferential” system is a mistake which could undermine the broader push for RCV. To understand why, we need to think about how electoral systems are designed.
Normal multi-winner RCV — what it is and how it works
In a multi-seat district, ranked-choice voting is meant to increase minority representation. That can be racial, ideological, partisan, or whatever. The key to this result is called a quota, and the quota falls as the number of seats rises. In a four-seat district, for example, the quota is 20 percent of votes. To figure out a rough quota – the proportion of the vote needed to win a single seat — divide 100 percent by the number of seats in the district, plus one.
This version of multi-winner RCV delivers fair outcomes, producing broad multi-party assemblies in countries like Ireland and Australia. Its use in the US has mostly been in cities, where it came with increased racial and gender diversity. This is due to the quota —– the more seats elected per district, the easier it is to win one seat. And this is the type of RCV in the Fair Representation Act, which Lee Drutman and The New York Times have argued would help more parties to win seats.
Block-preferential voting is not normal RCV
Last year, for the first time, two Utah cities used a seemingly similar system. It looks like multi-seat RCV — there are multiple seats per district, and voters rank candidates – but its counting process works very differently
Unlike regular RCV, which works well in single-member electorates, or its proportional version designed for multi-member electorates, the Utah system applies a formula designed for single-seat contests to the election of multiple seats. In so doing, it treats an elected member as though they had instead been the lowest polling candidate, passing on all their second-choice votes at full value. This means that voters who choose the first-elected candidate then get their votes counted towards the next candidate too. And the next.
Consider the following example, which shows how the same voters fare under three different systems: single-seat RCV, “normal” multi-seat RCV (the single transferable vote), and the block-preferential system now used in Utah. (We thank Andy Eggers for this example.)
Assume that 51 percent of voters have ranked their ballots A through Z. And 49 percent of voters have ranked their ballots in the opposite direction, Z through A. In other words, voting is polarized, as we have two groups of voters who support different sets of candidates.
In single-seat RCV, candidate A wins. They have 51 percent of votes.
Now, let’s switch to a two-seat district.
In “normal” multi-seat RCV, the victory quota is 33 percent. Candidates A and Z are elected, both having gotten more than a quota of first-choice votes.
But in the block-preferential system, candidate A wins the first seat, then is eliminated from the next count. All their ballots flow to B, who gets the second seat. The minority Z-through-A voters have won no representation.
This is fundamentally against the spirit of fair representation and can have very pernicious consequences. If voters tend to rank candidates from a single party or group, the sequence of instant runoffs will see that group win every seat. Votes “cascade downward” from the first winner to the others.
Australia, which has used RCV for over a century, once used this system to elect its national Senate. Following the US example, states doubled as districts, with three seats to be elected. Of 60 total state-based elections, 55 produced single-party delegations.
The system was widely derided and seen as producing illegitimate results. Reformers called it “a mongrel” and “a blockhead system.” Some even noticed that the downward cascade effectively let some voters vote twice. In 1948, the government switched to normal multi-seat RCV, and it has stuck since.
Other short-lived uses of the system at local-level elections saw minority parties learn to game the system. By tight transfers of preferences with allies, they were able to lock out majority-supported parties entirely. In all these Australian cases, the dysfunction of the block preferential system eventually saw it replaced with normal, multi-seat RCV envisaged in the Fair Representation Act, which has worked well.
Image credit: Peter Rukavina (Flickr, CC-NC-SA-2.0)
It’s not just confusing — it’s dangerous
How does something like this get adopted? The answer is by accident, without considering what it means for minority representation. Reformers come to incumbents with the idea of ranked-choice voting — the majoritarian form, meant for single-seat districts. And incumbents say, “can you make that work for our multi-seat elections?”
In a low-stakes election with weak party or group attachment, voters are not likely to cause much “downward cascade.” That’s because they are unlikely to keep rankings within their groups.
But if voting is polarized — in partisan, ideological, or racial terms — block preferential voting could lock out the opposition.
So far, the system has caught on in Republican locales. The two Utah towns, Payson and Vineyard, are in a county where Hillary Clinton got only 14 percent of votes in 2016. And back in the late 2000s, two more cities in North Carolina experimented with similar rules — similar in the sense of using ranked-choice to find the “majority slate” in a multi-seat district.
If block preferential voting is confined to small and overwhelmingly conservative places, it is not likely to find controversy. But if it gains broader popularity it could become a target for voting-rights litigation, undermining the broader push for RCV across the country. And it is far from the same “ranked choice” system in the Fair Representation Act.
Please read our comments policy before commenting.
Note: This article gives the views of the author, and not the position of USAPP– American Politics and Policy, nor of the London School of Economics.
Shortened URL for this post: http://bit.ly/37l3tAR
About the authors
Jack Santucci – Drexel University
Jack Santucci is Assistant Teaching Professor of Politics at Drexel University. He researches parties, elections, and electoral systems in the United States.
Benjamin Reilly – The University of Western Australia
Benjamin Reilly is a political scientist at The University of Western Australia. His work focuses on democratization and electoral reform in ethnically divided societies.
Your analysis of “block preferential” voting is completely on target and its “RCV” label has been a matter of concern to those of us at the Ranked Choice Voting Resource Center for some time. Not having an official name for this voting method, we have referred to it as “Sequential Multi-winner RCV.” We are convinced that this method would not pass voting rights muster in that its outcomes differ little from the usual at-large method.
Certainly, if Utah and others similarly situated wish to use this method, they are within their rights to do so, however, we would all be better served it it were labeled more accurately as “Block Preferential Voting” or “Sequential Multi-winner Ranked Voting.” This would clearly distinguish it from Ranked Choice Voting which, in its multi-winner form (single transferable vote), seeks to produce proportional representation.
Thank you for issuing this warning to the RCV support community and others.
Jack, please give the formula used for calculating block preferential voting results! This post has left me confused and frustrated. I want to understand how this works, but I can’t, and I won’t without the formula. Surely I’m not the only one.
I agree with Vivian.
At face value it seems to make voting more fair.
Let’s say I like everything about the Green party and nothing of the red. If I assumed my green vote was a throw away because of this opposition voting mentality,I would have to vote blue “no matter who”.
With ranked choice I could vote Green Party and then blue without my “preference” spoiling the chances of the “blue” ones chances.
Seems pretty straight forward to me.
To Vivian’s question, there is no real formula, in the sense of a Droop (or Hare, or whatever) quota. But the process looks like this. (1) Set the number of seats in the district at M. (2) Set the victory threshold at a majority of continuing ballots. (3) Allocate the first seat. Does anyone have a majority of first choices? If yes, they get the seat. If not, eliminate trailing candidates, then transfer those votes, as in IRV. (4) Allocate the second seat. Does anyone have a majority? If not, reallocate every ballot from the winner of the first seat. Does this help?
The core idea is that we’re using IRV to find a majority *slate*, not just a single majority winner. This brings me to Andrew’s question. The underlying theory of representation is that the majority should get every seat. If we apply that to a small city council, it follows that this system is designed to produce a legislature without an opposition party (or faction).