[EM] STV+AV

Jameson Quinn jameson.quinn at gmail.com
Thu Feb 2 04:41:05 PST 2012


Yes, this works. One downside is that, unlike STV, a hand count becomes
quite untractable.

Jameson

2012/2/2 Raph Frank <raphfrk at gmail.com>

> One possible way of combining AV + STV is to allow equal ranks.  This
> method becomes a method that is very similar to approval in the single
> winner case.
>
> When determining if a candidate is elected, all candidates at the rank
> share the remaining vote strength, but when determining if a candidate
> should be eliminated, the candidates at the rank get the full strength
> of the ballot, like in approval.
>
> A feature of STV is that the proportionality for solid coalitions
> criterion isn't affected by the elimination ordering, so you can use
> any rule that you want.  This assumes that candidates who exceed the
> quota are elected (and elected candidates are immune from
> elimination).
>
> Assuming there was a ballot:
>
> A: 1
> B: 1
> C: 2
> D: 2
> E: 3
> F: 3
>
> In round 1, the ballot would count as
>
> For electing:
> A: 0.5
> B: 0.5
> Others: 0
>
> For eliminating
> A: 1
> B: 1
> Others: 0
>
> So, the vote is shared when determining if someone has reached the
> quota, but is full strength when determining who to eliminate.
>
> If a candidate has more than a quota of "electing" votes, he is deemed
> elected.   Otherwise, the candidate with the lowest "eliminating"
> votes is eliminated.
>
> If B was eliminated and A was elected with 133.3% of the quota (so 75%
> of the voting strength consumed), then the ballot would count as:
>
> For electing:
> A: 0.75
> B: 0 (eliminated)
> C: 0.125
> D: 0.125
>
> To eliminate
> A: 0.75 (irrelevant since already elected)
> B: 0.75 (irrelevant since already eliminated)
> C: 0.25
> D: 0.25
> Others: 0
>
> This means that if all party supporters vote for party members as
> ranks 1 and 2, then the party is guaranteed to gets its share of the
> seats.  This follows from the solid coalition criterion.
>
> However, when determining the intra-party ordering, rank 1 candidates
> are considered approved and rank 2 candidates are disapproved.  The
> most approved party candidates (up to the limited number of seats the
> party gets) are the last to be eliminated, so they will be the ones
> elected.
>
> Standard PR-STV becomes IRV in the single seat case.  However, this
> system becomes a method that is very similar to approval in the single
> seat case.
>
> This means that for parties that get 1 seat, the intra-party decision
> is made by IRV, but with this method, the intra-party decision is made
> by approval (or at least an approval-like method).
>
> Assuming Meek's method for the transfers, then the rules are
>
> For both vote types (for election and for elimination)
>
> Vote strength passed to the next rank
>
> - the amount that would be passed to the next rank if all candidates
> at the current rank were ranked in order
> - This is the product of (1 - keep factor) for all candidates at the
> current rank times the vote passed into this rank
> - Nothing is passed to the next rank unless all candidates at the
> current rank are elected or eliminated
>
> Sharing (for election)
>
> - vote strength not passed to the next rank is shared between
> candidates at the current rank in proportion to their keep factors
>
> Sharing (for elimination)
>
> - vote strength not passed to the next rank is given at full strength
> to all candidates at the current rank
> ----
> Election-Methods mailing list - see http://electorama.com/em for list info
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20120202/e47f4c2e/attachment-0004.htm>


More information about the Election-Methods mailing list