[EM] [Election-Methods] [english 94%] PRfavoringracialminorities

[Raph Frank]

On Wed, Aug 20, 2008 at 10:09 PM, Juho <juho4880 at yahoo.co.uk> wrote:
> Rules for registering candidates may be different in different countries and
> may also be method independent in many cases. Parties may often have a
> formal role, but I don't know what the typical rules in STV-PR countries
> are.

In Ireland, they are pretty open.  However, they have been complaints that
they are annoying.

It is something like you have to be nominated by 20 registered voters, but
your nominees have to actually be physically present when you register.

Also, you have to pay a deposit and if you get 25% (I think) of a quota of
votes, you get your deposit back.

> There are however ways to fix this in the list based methods even if the
> districts sizes differ, but in STV this is probably trickier to do (having
> roughly equal size districts would be easy).

It isn't that hard.  Also, the number of seats per district can be
varied to achieve better seats to voter matching.

Also, if districts are at least 5, then the max imbalance is only
around +/- 10%.

> I think it should be pretty much the same thing. Roughly, if n% of the
> voters rank "X style" candidates first then "X style" candidates should get
> about n% of the seats.

The formal rule is called proportionality for solid coalitions.

Basically, if a party (coalition) is supported by n% of the voters
and all those voters rank all coalition candidates before any other
candidate, then the party should get n% of the seats (subject to
rounding error).

However, in real life, voters don't always vote for their party like that.


> STV is in principle good at this (but multiple small size districts may
> favour large parties).

Yeah.

> That may be a reasonably close approximation of party support but does not
> handle properly horizontal votes that are like
> MyCommunistFriend>MyRightWingPartyCandidate1>

Basically, 'personal' votes mess up the measure.  If voters tend to vote
for candidates, then you can't just assume first preference vote is a
measure of party support.

> What is a better way to handle the bias that e.g. small districts may cause
> in STV?

Increase the district size :).

> Maybe an explicit party vote?

I think the ability to transfer to the a candidates list is the best
way to handle it, if increasing district size is not feasible.

This doesn't fix the party support proportionality.  However, I
don't think that is important, what's important is achieving
proportionality of what the voters want.