At 12:06 PM 4/26/2010, Kathy Dopp wrote:
Peter, I cannot answer all your questions in detail, even if I had the time, but this is what I've learned thus far re. proportional multi-winner methods.
Kathy, I don't think you have realized the context. These are elections within a political party, and they are small. I assume that the ballots will be centrally tabulated, thus precint summability is irrelevant. A good STV method is reasonable for them.
1. Stay as far as possible away from methods like STV that are not precinct-summable and do not treat all voters' votes equally. (STV degenerates to IRV and has all the same flaws including nonmonotonicity, tendency to elect extreme right or left candidates and eliminate more popular centrists, the spoiler effect, not precinct summable, removes voters' rights to participate in the final counting round (or alternative requires that the voter figure out how to rank all the candidates), etc.
Yes, there are problems. But if proportional represnetation is desired, STV is pretty good. Asset Voting, indeed, is a tweak on STV-PR, and is one which allows candidates to complete the process even if voters bullet vote, which solves that problem. It's also possible, as Quinn points out, to use a predeclared candidate list; perhaps this list starts out and is used, but then the candidate personally takes over the votes only if there is an impasse that isn't properly resolvable by the original ballots.
(STV with the Hare quota is not at all the same as IRV, it's more like IRV with a majority requirement, which is what Robert's Rules actually recommends. The Hare quota will fail, generally, to elect all the seats, but that's where the candidates can take over. You can't do it from the ballots themselves because that requires the voters to be able to rank all the candidates.... It's a big discussion, actually.)
Alternatives to consider include 2. forms of approval voting that are multi-winner and proportional (choose one that is also precinct-summable, they may be all fair as far as I know) There are two recent working as-yet-unpublished papers by two political scientists on this topic of proportional multi-winner approval systems that you may be able to search for on the MPSA web site from the conference that just ended in Chicago. Authors are Steven Brams and Marc Kilgour. There are at least a half dozen different multi-winner election methods using approval ballots but not all of them are proportional and some of them are not precinct summable. A working paper coauthored by Brams and Kilgour describes a new? proportional approval method that is supposed to be precinct-summable but I haven't read it yet.
Precinct summability isn't an issue for these people.
3. the party list system is tried and true in many countries. Often one is allowed to vote either for a party or for a particular candidate within a party list of candidates, so that the voters can determine the order of which candidates in a party are elected first, second, third, etc. depending on what proportion of votes that party receives. There may be more than one counting procedure for party list systems.
Missed point, again: this is for a political party. How do you use party list with a political party, unless you create sub-parties. Asset in effect does this but without the party, per se. It is as if each candidate has his or her own party, which he or she controls.....
---- Election-Methods mailing list - see http://electorama.com/em for list info
