Surplus

Our man in Havana

How do they decide which of a canidates votes are the surplus?

Do they just take the top 100 or so and distribute them? Say you took 100 votes from another part of the stack surely this would give a different result?

I bet Nicky Kelly is hoping for this to happen!

sceptre

Hopefully this will just answer all the questions anyone might have about how PR-STV works:

Officially it all works as the following:

Let's assume the following: (you'll notice I'm making the figures easy to divide)
3 seater constituency
Registered voters 100000
Votes: 62345
Spoiled: 2345
Valid votes: 60000
Quota is (60000/4)+1 = 20001
(note quota is number of valid votes divided by (seats plus one) plus one)

Results:(1st count)


Barlow, Ken 12000
Bear, Teddy 25000
Keane, Terry 9000
Lawlor, Liam 250
Oddball, Iman 7000
Zig 6750
(Total 60000)

Teddy Bear is elected as he has exceeded the quota of 20001. Teddy bear has a surplus of 4999. This has to be distributed.

And here's where it gets complicated. We can't just use the last 4999 votes that Mr Bear got. So what we do is count all of Bear's votes and divide them in proportion. Essentially we have to turn 25000 votes into 4999.

So each surplus vote that is distributed from Teddy Bear's first preferences has to be multiplied by 4999 and divided by 25000.

So let's say that all the second preferences for the other candidates amount to:
Barlow 10000
Keane 3000
Lawlor 500
Oddball 5000
Zig 6500

When they're divided, they amount to
Barlow 2000
Keane 600
Lawlor 99
Oddball 1000
Zig 1300

Adding them to first count gives us:
Barlow 14000
Bear 20001 (elected)
Keane 9600
Lawlor 349
Oddball 8000
Zig 8050

Now we eliminate Lawlor and distribute all of his votes (no division here)
Lawlor's next available preferences are:
Barlow 250
Keane 80
Oddball 10
Zig 9

Giving a new result (after count 3) of:
Barlow 14250
Bear 20001 (elected)
Keane 9680
Oddball 8010
Zig 8059

Now Oddball gets eliminated
His next available preferences are:
Barlow 1000
Keane 1809
Zig 5250

Giving us:
Barlow 15250
Bear 20001 (elected)
Keane 11489
Zig 13309

We could eliminate Keane and distribute the surplus at this point but we don't bother. We just declare Barlow and Zig elected without reaching the quota (it would be a waste of time to distribute Keane's votes as we'd already be left with just three candidates - plus as mentioned below not all votes will be transferrable so all candidates may not reach the quota).

What makes it even more complicated is that not all of TBear's votes (or Oddball's) will be transferrable - some people will just put Bear down as 1 and not put a number 2 - very few people (as most will do the top 3 or 4) but this has to be allowed for.

jd

Originally posted by sceptre

Teddy Bear is elected as he has exceeded the quota of 20001. Teddy bear has a surplus of 4999. This has to be distributed.

And here's where it gets complicated. We can't just use the last 4999 votes that Mr Bear got. So what we do is count all of Bear's votes and divide them in proportion. Essentially we have to turn 25000 votes into 4999.

So each surplus vote that is distributed from Teddy Bear's first preferences has to be multiplied by 4999 and divided by 25000.

This only true for the for the first count. Afterwards, its the votes transferred from other candidates from the previous count that are used to calculate what other candidatees get from the surplus...

Electronic voting is more accurate, as all he votes are always used in calculating the distribution of the surplus

simple eh

Victor

Well actually surpluses are distributed at random, as it would take too long to calculate otherwise. The electronic system had to have this 'random' factor added this election to ensure the method of each calculation was the same through each constituency.

Say the quota is 20001 and the candidate has 21001 votes after the last count. 1000 votes are then picked at random.

Victor

sceptre's post edited (80000 changed to 60000 and so on).

jd

Originally posted by Victor
Well actually surpluses are distributed at random, as it would take too long to calculate otherwise. The electronic system had to have this 'random' factor added this election to ensure the method of each calculation was the same through each constituency.

Say the quota is 20001 and the candidate has 21001 votes after the last count. 1000 votes are then picked at random.

But at random from the last votes the candidate with the surplas received! ie not his poriginal number 1's

Victor

Originally posted by jd
But at random from the last votes the candidate with the surplas received! ie not his poriginal number 1's

I understand (but have not studied) that it is throughout the total (original plus transfers) - it is fairer this way.

sceptre

Originally posted by Victor
sceptre's post edited (80000 changed to 60000 and so on).

Looking closely at it, the maths are skewed up anyway. It'll give anyone who's interested some idea though (if you see the total maths error and figure out why it makes a balls of the whole example you understand how it works) - blame my typing it late at night.

Clearer example is at

http://www.wikipedia.com/wiki/Single+transferable+vote

Victor

I give up

dahamsta

There are pretty good explanations here and here.

(No offense to sceptre's explanation. )

adam