Tue 9 Jan 2018 11:27AM

Multi-Winner Plurality PR?

SW Sara Wolf Public Seen by 376

WONK TALK: @adamzielinski had an idea the other day in the middle of our meeting. It was a response to a non-PR multi winner approval system. Here it is if I have it right:

What if you did a multi-winner system where everyone voted for one candidate. Plurality style. If there are 5 seats the top five get elected. Isn't this sort of a crude PR?

It's not expressive or really honest, and it maintains the spoiler effect, but it seems pretty damn equal and I wonder how accurate it might be?


Kristin Eberhard Tue 9 Jan 2018 7:52PM

Sounds like you are describing Limited Voting or Single Non-Transferable Vote. It is used in 147 local jurisdictions in the US as well as in several other countries. It can achieve more proportional results than plurality, but is vulnerable to vote splitting.




Alan Zundel Wed 10 Jan 2018 12:50AM

Vote splitting is one problem. Another is that if there is one or more front runners, they may suck up most of the votes and then a minority of voters are selecting other winners with many fewer votes. In other words, one candidate wins with 60% of the votes and another with 10%. They each get one seat but one represents a lot more voters than the other.


Adam Zielinski Mon 22 Jan 2018 5:40AM

The reason we were discussing this was because apparently the Independent Party of Oregon is in favor of this kind of voting system, or something like it. They wanted to know if we were open to amending Star voting now or in the future to allow multiple winners. At first I thought they were talking about proportional represnetation such as STV or Reweighted Range voting, but no, they were just talking about multiple winners for a given legislative seat or district. It was unclear whether they supported each voter getting one vote, or each voter getting the same number of votes as there are winners.

Lake Oswego city council has a unique voting system where there are three winners and each voter gets three votes. In their last election, four candidates ran, and three people won. And each voter got to pick three candidates each.

The Sierra Club elections are like this as well. You have to have at least one more candidate on the ballot than there are seats. So someone has to lose. But otherwise, if there are three seats available, voters get three votes and the top three candidates win.

I tried to ask why the IPO liked this voting method more than STV or RRV, and they said it was simpler and that STV and RRV were way too complicated. Also they said they use approval voting for all their internal votes and it works fine and they see no reason to consider or try any other methods.

It seems to me that if you had a multiple winner election, but with each voter only getting one vote, this could actually work as a quick and dirty way to achieve proportional representation. But not if each voter gets the same number of votes are there are winners.


Alan Zundel Wed 24 Jan 2018 4:59PM

Not really like PR. If it were a partisan election either there would be only one candidate from each party or there would be multiple candidates from each party. In the first case it can't be proportional because no matter how many votes your party gets you can only get one seat. In the second case party voters have the split vote problem, or you may end up filling all the seats from the majority party.

I can't see why they like such a system either. It is similar to single-seat plurality in that voting is simple but outcomes are terrible.


Clay Shentrup Sun 25 Feb 2018 8:56PM

Seems like they should just use multi-winner Approval Voting then. Or RRV Approval Voting, which is VERY simple. You just weight each ballot by the number of winners it approved, plus one.

Even simpler is Asset Voting. It needs more empirical research but has some incredibly nice properties.


Dag Arneson Thu 1 Mar 2018 6:38PM

I had an idea for something similar to this, with two rounds, first a single vote quota round, then the remaining seats filled with a score voting round, conducted simultaneously on one ballot.

Another idea for a system like this, similar to asset voting, would be to have the candidates each pick an ordered list of the other candidates, and then as the candidates are elected or eliminated, surplus votes are allocated according to their list.


Sara Wolf Fri 2 Mar 2018 7:25AM

What do you mean by a single vote quota round? Can you draw the ballot?


William WAUGH Sat 3 Mar 2018 3:37AM

Assuming a fixed or limited count of people who can sit in the legislature and assuming two rounds of voting, here's how I'd spend those two rounds. First round is about who gets seats, and goes by RRV. Second round, citizens assign proxy to whichever legislators they want to. In votes on legislation, the members vote with what proxies they have.


Dag Arneson Tue 13 Mar 2018 7:48PM

The quota round works like in STV, a candidate is elected if they get more than 1/(N seats + 1) of the vote.


Dag Arneson Tue 13 Mar 2018 7:54PM

RRV is good. I think the biggest problem with systems with multiple rounds and reweighted ballots, like STV and RRV, is that the results are very difficult for a layperson to interpret. This isn't a huge disadvantage where there is high trust, like within a relatively homogenous political organization, but for public elections presents a problem. The disadvantage of these systems for intra-organizational elections is that they require a software implementation to count effectively. The system I have proposed above could be conducted with separate plurality and range votes, which are relatively easy to come by.


Clay Shentrup Wed 14 Mar 2018 6:32AM

This isn't inherently proportional any more than at-Large Plurality.


Sara Wolf Sat 17 Mar 2018 4:07AM

None of us were recommending it, just contrasting with Lake Oswego's even less proportional system.