You and Your Vote
An Analysis of the 2008 Election Results
A description of the Proportional Representation voting system used in NSW Local Government elections is provided in the Counting the votes section on the NSW Electoral Commission Web site.
What follows is an attempt to explain the practical application of this voting system in determining the make-up of our new Council. Central to this analysis is the accompanying spreadsheet summarising the ballot counting process. The figures used to prepare this spreadsheet were extracted directly from the results section of the NSWEC Web site. Candidates are listed in the spreadsheet in the order in which they were elected (or the reverse order of exclusion from the Count). The below-the-line (BTL) vote for each candidate is listed first, followed by the above-the-line (ATL) vote for each Group (against the names of the leaders of the respective Groups). Count 1, then, is the summary of the total first preference votes for all candidates. This allows us to determine the Quota for this election, calculated as follows:
In the 2008 Palerang election:
Under normal circumstances, a candidate must obtain a Quota of votes to be elected. The Quota remains unchanged for the whole Count. If there is a large number of exhausted ballot papers (those with no preferences for any continuing candidates in the Count), it is possible for candidates to be elected without having reached the Quota.
If a candidate obtains the Quota on their first preference vote count, they are immediately declared elected. In the 2008 election, Catherine Moore, Ian Marjason, Richard Graham and Judith Miller were all elected in this fashion.
Votes accumulated by a candidate in excess of a Quota are known as surplus votes. When a candidate achieves Quota, any surplus votes are distributed according to the preferences on the individual ballot papers in the parcel that led the candidate to achieve Quota. When Quota is achieved on first preference votes alone, the parcel will comprise all the ballot papers currently held by that candidate.
To see how this works, let’s look at a specific example. Note that an above-the-line vote is the same as a below-the-line vote that simply allocates preferences down the nominated Group (i.e. first preference to the first member of the Group, second preference to the second, and so on). In the first Count of the Palerang election, Count 1, Catherine Moore received a total of 1342 first preference votes. The Quota for this election was 731 votes, so there will be 611 surplus votes to pass on as preferences. But which ballot papers should we look at to determine where these 611 preferences should flow? In the present case, with 35 candidates, there could be as many as 34 different second preference choices. This problem is overcome by considering the second preference nominated on all 1342 ballot papers, but allocating each preference as only a fraction of a vote. The fractional multiplier, called the Transfer Value, is calculated as:
where a continuing ballot paper is one indicating a preference for at least one candidate still in the Count (i.e a ballot that has not exhausted). In the present case then, the Transfer Value for Catherine’s surplus votes is:
Thus, each of the second preference votes on Catherine’s 1342 ballots will be passed on as 0.4552 of a vote. If your maths is up to it, you will see that with 1342 ballot papers, we will then be allocating a total of:
which is exactly the number of surplus votes that we have to pass on.
Since all the above-the-line votes will implicitly nominate Judith Turley as the second preference, Judith will be the primary recipient of Catherine’s surplus votes. But Catherine also received 408 below-the-line votes, and these votes could have nominated any other candidate on the ballot paper to receive the second preference. You can see how these preferences were distributed by scanning the column below Count 2 in the spreadsheet. Note that only whole vote increments are recorded. Partial votes are held over until they add up to a whole vote, and are then added to the tally of the respective candidate. Judith Turley did indeed receive the majority of the preferences, a total of 513. The other 98 were distributed to other candidates.
Counts 3 – 5 are the distribution of surplus votes from the other candidates who achieved Quota on their primary vote. Note that surplus votes for Group candidates will flow mostly on to the second member of the respective Group, since most of the votes will have been above-the-line votes that implicitly allocate preferences down the Group. Surplus votes from Richard Graham, an ungrouped candidate, flow more widely because the preferences are all individually specified (below-the-line).
The next phase of the Count is the exclusion of candidates, starting with the candidate who received the least number of primary votes. In the present election, the first candidate excluded was Allan Schmidt. In Count 6, the five ballot papers that nominated Allan as their first preference are then distributed to the candidates listed as second preference. Note that these preferences are distributed with the same value that they were received. In this case, they are all whole votes, and not subject to a Transfer Value. By scanning up the relevant column of the spreadsheet, you will be able to see how Allan’s five preferences were distributed—two to Nerrida Hart, one to Mike McColl, one to Kev Fiebig and one to Terry Bransdon.
This same process is followed for each Count through to Count 22, the candidate with the least number of votes being excluded at each round. Before proceeding to Count 23, note that in Count 21 we see our first exhausted ballots. An exhausted ballot is one that contains no preferences that can be allocated to continuing candidates, that is candidates still in the Count. This may be because all the remaining preferences on the ballot are for candidates who have already been excluded, or it may be because no more preferences have been nominated. Although there is no way to tell for sure, the exhausted ballots in the present case are most likely below-the-line ballots that have nominated the minimum number of candidates (5), all of whom have now been excluded.
In Count 22, Howard Crozier achieves Quota, and Count 23 is the distribution of his surplus votes. Note now that only the parcel of votes that caused the candidate to reach Quota is used in the distribution of, and to calculate the Transfer Value for, surplus votes. In the present case, there are 30 surplus votes, although 18 of them exhaust. While again there is no way of knowing for sure, these are most likely to be above-the-line votes [for Group A] that did not specify any second preference [above-the-line]. No other Group A member is still in the Count, so the only way an above-the-line vote can be passed on is if there had been a second above-the-line preference nominated. Most of the remaining preferences are passed on to Kev Fiebig, at the top of Group E, since Group A had recommended nominating Groupe E as a second preference. Note also that the eight votes passed from Group A to Group E at this point are the only preferences actually exchanged between Groups in the entire Count, an interesting observation in light of all the fuss over preference deals in the lead up to the election.
In Counts 24 – 28 we continue the exclusion process. We are now seeing a steady trickle of exhausted ballots. Remember that a ballot will exhaust when it has no preferences for candidates that are still in the Count. By Count 28, we’ve excluded 22 candidates, so there’s plenty of scope now to have actually nominated quite a few candidates but to still exhaust.
In Count 28, Terry Bransdon achieves Quota, with 731 votes. The fact that this is exactly a Quota is simply a coincidence. In fact, we know that Terry received some fractional votes in the surplus distributions from Catherine Moore, Ian Marjason and Judith Miller, and it is probably in the attempt to allocate the fractional surplus vote that we find 18 more exhausted ballots. The jump in the number of exhausted ballots is once again probably due to [Group B] above-the-line votes that did not specify a second preference.
Now we get to the business end of the Count, and the point at which we find that the rural residential candidates in Groups D and E, and Anne Goonan, have snookered themselves. When John van der Straaten is excluded in Count 29, 212 of his votes exhaust. This spells disaster for the remaining rural residential candidates. Up to this point, the three of them, John, Kev Fiebig and Anne, were sitting, collectively, on 1022 votes, enough to have any of their supporters feeling comfortable about success for at least one of them. Their rivals at this point, Mark Horan and Walter Raynolds had only 695 votes between them. The exhaustion of so many of John’s ballots was not the end of the road, but it was the sign of things to come.
When Mark Horan is eliminated, in Count 30, most of his preferences go on to Walter, taking him from some 50 votes behind Anne to more than 100 votes in front of her. Only 43 of Mark’s votes exhaust. As an ungrouped candidate, all of his votes will have been cast below-the-line, and the preferences that he has inherited will be less likely to have passed through many of the (predominantly Group) candidates who have already been excluded.
Now comes the crunch. Kev Fiebig’s exclusion, in Count 31, follows the same pattern as John van der Straaten’s before him—301 of his votes exhaust and Anne Goonan picks up only a handful of preferences, leaving her in 10th place as the axe falls for the last time.
With Anne's elimination in Count 32, there are only nine candidates left, and Judith Turley, Paul Cockram and Walter Raynolds are elected without having reached a Quota.
What happened here?! Well, there were actually two ‘problems’. The first was simply that too many voters didn’t nominate enough preferences. In the case of John’s and Kev’s exclusions, it was primarily above-the-line votes that fell short, and most likely not because they didn’t nominate a second preference but because they didn’t nominate a third, fourth or fifth preference. The second problem was that there was no way that Anne, as an ungrouped candidate (without any above-the-line voting square), could ever receive any above-the-line preferences from anywhere, regardless of how many a voter nominated. That meant that as the Count drew to a close, even though the three ‘rural residents’ had 1022 votes between them, Ann could never accumulate more than 628 of them, since the remaining 394 originated above-the-line. That would still have been enough, but in the end, she picked up only 24 (below-the-line) preferences from the 659 votes held by the other two.
A large enough proportion of Anne’s preferences may have gone to John and pushed him over the line, had Anne been excluded first, but she wasn’t, and the rest is history, as they say.
According to the NSWEC Counting Sheets, 1065 ballot papers in the Palerang election exhausted i.e. they ran out of preferences. That’s more than 14% of the formal vote, and a significant number of voters who, although their votes were entirely valid, ultimately made no practical contribution to the election process. Much of the problem here is simply a lack of understanding of how the voting system used in Local Government elections works, and I hope that this analysis has gone some way towards rectifying that problem. The short of it is that, whether voting above- or below-the-line, you should allocate a preference to every Group or candidate that you consider worthy of consideration. This is all the more important if you vote below-the-line and include any number of the candidates who appear towards the bottom of the list on a Group ticket, those most likely to be excluded early in the Count. The alternative is simply that you will have to live with a Council that other voters elect.
I trust that, if you’ve read this far, you’ve found this exercise worthwhile. The election result may not have been the one many of us were hoping for, but it has certainly been an education. Rest assured that I’ll be dusting off this account in just under four years time, and reminding everyone of what they need to do to avoid this sort of problem at the 2012 election.