Talk:Proportional approval voting

This article was nominated for deletion. Please review the prior discussions if you are considering re-nomination:

keep, 19 June 2015, see discussion.
delete, 14 September 2009 (UTC), see discussion.
delete, 14 October 2007 (UTC), see discussion.

More examples, and comparison of quotas in different Proportional Representation schemes

More examples, both here and elsewhere, would be helpful. E.g. showing that in a 2-winner contest, out of 100 voters, 34 supporters of only candidate C could get represented by C, with the other voters choosing between A and B. Comparing different PR schemes with the same scenarios would be instructional. ★NealMcB★ (talk) 14:07, 10 March 2017 (UTC)[reply]

In particular, an example in which the ballots demonstrate the ability to vote for more than the number of winners, and/or in which PAV winners are different than block plurality, would be good. ★NealMcB★ (talk) 17:40, 17 September 2018 (UTC)[reply]

Sounds great to me! Closed Limelike Curves (talk) 04:04, 16 February 2024 (UTC)[reply]

Merger proposal

The following discussion is closed. Please do not modify it. Subsequent comments should be made in a new section. A summary of the conclusions reached follows.

To not merge, given that the methods are sufficiently different to sometimes yield different results; it is accepted that they are very similar, and opinion differed as to whether they were similar enough to be discussed on one page; no consensus with stale discussion and maintaining status quo. Klbrain (talk) 16:43, 6 October 2024 (UTC)[reply]

I propose that Sequential proportional approval voting be merged into Proportional approval voting. The two articles describe the same election method. They describe the method using different names, but the logic itself is mathematically identical. Dhalsim2 (talk) 23:48, 29 June 2018 (UTC)[reply]

Thank you for noticing the similarity, but the articles should not be merged. As the references in both articles clearly show, these are two different topics. Someone incorrectly added the sentence to Proportional approval voting saying the particular method was invented by Thiele and rediscovered by Simmons. The references do not support this. The table on page 122 of Kilgore's 2010 paper shows the two methods giving different results. StarryGrandma (talk) 00:59, 1 July 2018 (UTC)[reply]

Thiele invented both methods. Sources: 1. (here the method is called Thiele's optimization method) http://www2.math.uu.se/~svante/papers/sjV9.pdf, 2. Math PhD Warren D. Smith acknowledges Warren D. Smith of discovering a special case of his harmonic method for when approval ballots are used. That special case is PAV https://rangevoting.org/QualityMulti.html#acknow — Preceding unsigned comment added by ParkerFriedland (talk • contribs) 06:45, 6 June 2019 (UTC)[reply]

I read Kilgore's paper in its entirety and worked through each of his examples by hand and by computer. I do now acknowledge that PAV and SPAV can yield different results. I was previously confused due to the example in the Wikipedia PAV article. The example had wrong logic in it. I have now corrected the example. Though PAV and SPAV do occasionally yield different result due to ties in small sample sets, most of the time, it empirically seems that the results are the same, so I do think that there could be a legitimate argument to merge the two articles, but I acknowledge that due to the possibility of a difference, there is certainly less importance to a merge. Dhalsim2 (talk) 01:52, 12 July 2018 (UTC)[reply]

I'm glad we agree they are different, and I think should not be merged. I'd love to see more research on the question of when the results are the same vs different, but that should be published somewhere outside of Wikipedia. ★NealMcB★ (talk) 03:36, 21 January 2019 (UTC)[reply]

The two don't agree, regardless of ties. SPAV is a greedy approximation algorithm for PAV that elects candidates one at a time. The greedy strategy elects a potentially very different slate of candidates from PAV.

That being said, I think the two systems are procedurally similar enough to be described in one article. SPAV is a strategy for calculating PAV, which suggests the two articles should be merged, probably with a section that explains SPAV. Otherwise, we'd be forced to spin off a dozen different articles, one for every integer linear programming algorithm you could use to approximate PAV.

As precedent, Kemeny–Young doesn't have separate articles describing every possible approximation method. Closed Limelike Curves (talk) 04:03, 16 February 2024 (UTC)[reply]

The articles should not be merged. The way the algorithms work is so different that one of them (SPAV) probably shares the massive problems that STV has when it comes to Risk-limiting audits, because both are sequential with an exponential set of possible elimination sequences, while PAV is linear in a sense making it very easy to calculate the margin of victory and the necessary sample size. That is a critical difference for states that require RLAs, and merging the articles would confuse the discussion. And yes, we need appropriate coverage and citations for these issues.... ★NealMcB★ (talk) 18:24, 3 August 2024 (UTC)[reply]

The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.