Redistribution of Excess Votes

jordainius

So in constituency A, the quota is 15000 and John Doe get's 20000 first preferences and goes through on the first count. Fairly straightforward.

But in the next count his "surplus" votes are redistributed so in this case 5000 votes are counted and second preferences win these votes.

My question is; how is decided what counts as a surplus vote? Are 5000 random votes taken from John Doe's 20000 and redistributed accordingly? (what about the 15000 non surplus votes; do they effectively count for nothing?)

Or am I completely missing something here? I've looked up stv online and on wikipedia and it doesn't make it clear what constitutes a surplus vote which is why I ask here! I'm getting the impression that a "non surplus" vote for a candidate who passes the quota is effectively binned? Surely I have this wrong?

the groutch

the 5,000 are chosen at random.
not the fairest system ever, but alot quicker than the having go through 20,000 votes to get it in exact proportion.
there are plenty of tallymen watching the process, to make sure it is completely random, and in no way biased

davef1000

I was talking about this with some people the other day. Is there a resource somewhere that describes the process? Even better, is there something that specifically relates to the Irish system? I'm very interested to learn more about the way votes are transferred/eliminated/etc.

loldog

the groutch wrote: »

the 5,000 are chosen at random.
not the fairest system ever

If you're picking that many at random, it's a pretty good sample of public opinion... however if the surplus is relatively small, say a few hundred then you could get some bad skewing of the results.

.

stop

the groutch wrote: »

the 5,000 are chosen at random.
not the fairest system ever, but alot quicker than the having go through 20,000 votes to get it in exact proportion.
there are plenty of tallymen watching the process, to make sure it is completely random, and in no way biased

Never liked this. While I accept the tallmen probably do a good job at ensuring no bias, I don't like my vote, or anyone elses for that matter, being discarded "at random". I'd prefer a full recount, do it right not do it quick.

dilallio

jordainius wrote: »

So in constituency A, the quota is 15000 and John Doe get's 20000 first preferences and goes through on the first count. Fairly straightforward.

But in the next count his "surplus" votes are redistributed so in this case 5000 votes are counted and second preferences win these votes.

My question is; how is decided what counts as a surplus vote? Are 5000 random votes taken from John Doe's 20000 and redistributed accordingly? (what about the 15000 non surplus votes; do they effectively count for nothing?)

Or am I completely missing something here? I've looked up stv online and on wikipedia and it doesn't make it clear what constitutes a surplus vote which is why I ask here! I'm getting the impression that a "non surplus" vote for a candidate who passes the quota is effectively binned? Surely I have this wrong?

Not exactly.

John Doe's 20,000 votes are placed in his section or pidgeon-hole in the count centre.

The counters will sort these in piles of 50, in order of the no 2's. Some of JD's 20,000 will not have a no 2 indicated, so these votes are deemed to be non-transferable.

Say in this case just 15,000 of JD's 20,000 have a no 2 indicated, this means that JD has 15,000 transferable votes. This works out as 3 times JD's surplus.

The counters will then remove one third of JD's no 2's (transferrable votes / surplus), ensuring that they take one third from the top of each & every pile. This ensures the random element and is extremely important. An e-voting system would count all of them and apportion the exact number of no 2's correctly, but by taking the same number from all piles ensures that this difference is very small.

Each of the no 2's removed from JD's pile, are then moved into the pidgeon-hole for the candidate who received the no 2.


There is a very good podcast (5 parts) which explains the full count process here.

ninty9er

Dilallio has it most correct.

It is NOT a random sample and only the surplus is distributed.

So of 20,000 votes, 15,000 is the quota portion.

All 20,000 votes are counted for transfer purposes.

If 10,000 show a preference for Ken Eljono, then KE will receive a transfer of 2,500 votes (or 50% of the preferences available for transfer representing 50% of the no.2s on John Doe's ballots)

Willie O'Dea is an example of massive transfer operation from 2007. His transfers elected Peter Power and Michael Noonan on count 2. Count 3 was Power's surplus. In this case it's only the votes transferred from O'Dea that are counted (i.e the Power 2 bundle of O'Dea's 1s is separated into the 3rd preferences from O'Dea and distributed accordingly)

In the case of surplus the non-transferable votes are excluded from the percentage. In this way it is impossible to determine how many people actually put a no.2 for Power or Noonan, only that 47% of preferences were for Power and 9% for Noonan.

In the case of eliminations all ballots become a "number one" for the next preference in the count, therefore non-transferable votes come into play in an elimination as the first preference has not been elected.
It's much more complicated than people give it credit for. It's extremely democratic, that's not to say it's the best form of democracy.

http://www.electionsireland.org/counts.cfm?election=2007&cons=159&ref=

Up de Barrs

ninty9er wrote: »

Dilallio has it most correct.

It is NOT a random sample and only the surplus is distributed.

So of 20,000 votes, 15,000 is the quota portion.

All 20,000 votes are counted for transfer purposes.

If 10,000 show a preference for Ken Eljono, then KE will receive a transfer of 2,500 votes (or 50% of the preferences available for transfer representing 50% of the no.2s on John Doe's ballots)

Willie O'Dea is an example of massive transfer operation from 2007. His transfers elected Peter Power and Michael Noonan on count 2. Count 3 was Power's surplus. In this case it's only the votes transferred from O'Dea that are counted (i.e the Power 2 bundle of O'Dea's 1s is separated into the 3rd preferences from O'Dea and distributed accordingly)

In the case of surplus the non-transferable votes are excluded from the percentage. In this way it is impossible to determine how many people actually put a no.2 for Power or Noonan, only that 47% of preferences were for Power and 9% for Noonan.

In the case of eliminations all ballots become a "number one" for the next preference in the count, therefore non-transferable votes come into play in an elimination as the first preference has not been elected.
It's much more complicated than people give it credit for. It's extremely democratic, that's not to say it's the best form of democracy.

http://www.electionsireland.org/counts.cfm?election=2007&cons=159&ref=

This highlights a flaw in the system, as you say when a surplus is being distributed only the bundle of votes that took the candiate over the quota is examined. It may arise that this bundle does not have enought continuing preferences in order to distribute the surplus entirely.

In the case of Brian Crowley in the European election he had a surplus of 7,874 but only 5,765 were distribute as this was the total number of votes in the bundle that put him oveet the quota which had continuing preferences.
This is done for efficiency reasons as it would be too time consuming to go back and look at all the candidates votes to decide on the distribution of the surplus.
http://www.electionsireland.org/counts.cfm?election=2009E&cons=242&ref=

In the Seanad election they apply the PR STV in its purest form right down to distributing fractions of votes.

_Bella_

Goes through it here, starts at 22:30

http://www.rte.ie/player/#v=1092345

ninty9er

Up de Barrs wrote: »

In the case of Brian Crowley in the European election he had a surplus of 7,874 but only 5,765 were distribute as this was the total number of votes in the bundle that put him oveet the quota which had continuing preferences.
This is done for efficiency reasons as it would be too time consuming to go back and look at all the candidates votes to decide on the distribution of the surplus.
http://www.electionsireland.org/counts.cfm?election=2009E&cons=242&ref=

Nope. This is done as Crowley didn't have enough votes on his own, therefore his second preferences don't transfer as he used them all.

Up de Barrs

ninty9er wrote: »

Nope. This is done as Crowley didn't have enough votes on his own, therefore his second preferences don't transfer as he used them all.

Sorry if I'm not explaining this properly for you, as it is applied in Dail elections, where a candidate has a surplus it may not all be transferred if there are not enough continuing preferences in the bundle of votes that took him over the quota. In a Seanad election this would not arise as the entire vote of the elected candidate would be examined to distribute the surplus not just the bundle of votes that took him over the line.

Seve OB

So here is the legals on how the voters are transferred

http://www.environ.ie/en/LocalGovernment/Voting/PublicationsDocuments/FileDownLoad,1895,en.pdf


if you can understand it, you are a better man than I am, cause it all looks like legal waffle to me which never makes sense in plain aul english!

Dymo

What about the case of a full recount like in Laois/Offaly, will the sample be random again and could this change the result.

irishgeo

What happens your number 2 vote if the person you gave it too is elected already?

I did give my No 3 to someone who was elimated later on, so how does it work?

Zynks

This may be a silly question, but I just can't figure it out...

Say candidate X has 14k votes, 2k votes over the quota. This excess is then redistributed. And here comes the question, WHICH of his 14k votes are re-distributed? I can't think of a fair and precise way to do that, where the next preferences of the 14k votes are equally and proportionally reflected.

In addition, unless there is a clear system in place, on a recount you are likely to get a different redistribution every time...

Am I missing something here?

flahavaj

Zynks wrote: »

This may be a silly question, but I just can't figure it out...

Say candidate X has 14k votes, 2k votes over the quota. This excess is then redistributed. And here comes the question, WHICH of his 14k votes are re-distributed? I can't think of a fair and precise way to do that, where the next preferences of the 14k votes are equally and proportionally reflected.

In addition, unless there is a clear system in place, on a recount you are likely to get a different redistribution every time...

Am I missing something here?

My thoughts exactly! Very odd way of doing things.

paul71

flahavaj wrote: »

My thoughts exactly! Very odd way of doing things.

All the 2nd preference votes are counted and the surplus is divided in the same proportion.

Capt'n Midnight

Zynks wrote: »

In addition, unless there is a clear system in place, on a recount you are likely to get a different redistribution every time...

Am I missing something here?

Votes are mixed up from all ballot boxes when arrive in count centre , so that's the random bit


not sure how exactl ythe pick the transfers, but once picked they are put in a bundles and labelled so the case of a recount it will be the same bundle counted each time

Zynks

paul71 wrote: »

All the 2nd preference votes are counted and the surplus is divided in the same proportion.

That would be fair at the 2nd preference level, but it would degrade immediately at the next level of redistribution - which seems to happen a lot.

For redistribution to be fair, votes would have to be limited to 1st and 2nd preferences only (which is not a bad idea IMO).

Zynks

Capt'n Midnight wrote: »

Votes are mixed up from all ballot boxes when arrive in count centre , so that's the random bit

not sure how exactl ythe pick the transfers, but once picked they are put in a bundles and labelled so the case of a recount it will be the same bundle counted each time

That seems to put in question the validity of the recount system since the source of the bundle itself could be questionable (if the case of "beyond 2nd preferences")

paul71

Zynks wrote: »

That would be fair at the 2nd preference level, but it would degrade immediately at the next level of redistribution - which seems to happen a lot.

For redistribution to be fair, votes would have to be limited to 1st and 2nd preferences only (which is not a bad idea IMO).

Now you have it, the proportional basis only works on the first distribution. But lets be honest in the absense of a computerised system if you applied to the subsequent eliminations the counts would take weeks.

K-9

Zynks wrote: »

That would be fair at the 2nd preference level, but it would degrade immediately at the next level of redistribution - which seems to happen a lot.

For redistribution to be fair, votes would have to be limited to 1st and 2nd preferences only (which is not a bad idea IMO).

I don't know the ins and outs but my best way of explaining it is that the cubby holes you see on the TV pics are used to distribute the votes. So they set them out based on 2nd preferences etc.

Though if you've voted for say the bottom 3 candidates in order of elimination, your 4th vote is of equal value to the 2nd preference of the 3rd last eliminated candidate!

If you go on the RTE site and their election site they have a decent guide, based on practical examples.

Zynks

K-9 wrote: »

I don't know the ins and outs but my best way of explaining it is that the cubby holes you see on the TV pics are used to distribute the votes. So they set them out based on 2nd preferences etc.

Though if you've voted for say the bottom 3 candidates in order of elimination, your 4th vote is of equal value to the 2nd preference of the 3rd last eliminated candidate!

If you go on the RTE site and their election site they have a decent guide, based on practical examples.

Thanks, I found it:http://www.rte.ie/news/election2011/mechanicsprstv.html

It confirms my suspicion that there is an element of 'faith':

This way of doing things assumes that, because the ballot papers have been mixed at the very start of the count, their order is random and so any slice taken from the top of a bundle of votes will be a random and therefore representative selection of all the votes in the bundle. An alternative way of dealing with this matter is to pass all the ballot papers on to the continuing candidates but at a fraction of their value (this is known as the 'Gregory method').

sceptre

Multiple threads merged - all of them were basically asking the same question.

Always worth checking the page (and the previous one) to see if the query you've got has already been answered once or twice.

/mod

K-9

Zynks wrote: »

Thanks, I found it:http://www.rte.ie/news/election2011/mechanicsprstv.html

It confirms my suspicion that there is an element of 'faith':

Found that bit odd but you do have the famed tallymen keeping an eye on things.

Maybe favours party machines in that respect?