PROPORTIONAL REPRESENTATION SOCIETY OF AUSTRALIA

The Proportional Representation Society of Australia made a submission in June 2000, summarised below, to the Review Panel on Elections to the Aboriginal and Torres Strait Islander Commission that is formed after each round of ATSIC Regional Council elections.

Summary

In this submission the Proportional Representation Society of Australia recognises the many good features of current Regional Council election procedures and makes practical suggestions for implementing improvements that increase vote effectiveness.

In particular, the Society draws to the attention of the Review Panel how most of the problems arising from large numbers of candidates being elected without a quota of votes can be fixed through simple alteration of the way the quota is struck and transfer values are defined. Under no circumstances should the transfer value of a ballot-paper rise in the course of a scrutiny.

This submission uses the fairly detailed analysis of thirteen disparate Regional Council scrutinies to provide empirical evidence and construct numerical examples to illustrate various effects and potential solutions to problems.

The Society would welcome the opportunity to elaborate upon the comments and suggestions made, or to be involved in designing materials that will help stimulate voter, candidate and Returning Officer awareness of both quota-preferential principles and the reasons why they are applied.

Key recommendations

The quota-preferential system of voting and counting should be retained and enhanced to promote vote effectiveness.

The current requirements for casting a formal vote should remain, and concerted efforts be made to improve awareness of how to make full use of the single transferable vote.

In the setting of ward boundaries, except where there are compelling reasons to the contrary, an odd number of vacancies, no fewer than seven, should be be the norm.

In order to minimise the number of votes that are wasted, procedures relating to distribution of surpluses should be altered as follows:

Provision should be made for the possible replacement by countback of a vacating candidate elected that way.

The Proportional Representation Society of Australia exists to promote public understanding, acceptance and use of quota-preferential methods of determining election outcomes. Quota-preferential methods empower voters by giving them a direct and effective say in the composition of the bodies elected to represent them, and keep wasted votes to a minimum.

As ATSIC Regional Council elections are conducted on quota-preferential principles, and involve substantial numbers of voters right around Australia, the Society has been keen to analyse the performance of the system at last year's elections, and to make suggestions for improving vote effectiveness.

Detailed study of thirteen elections

To complement its examination of the relevant legislation and statutory rules governing the conduct of the 1999 Regional Council elections, the Society made a detailed study of printed records from 13 of what were likely to be the most complex scrutinies on account of the numbers of candidates nominating.

After contacting the Australian Electoral Commission about the availability of counting records, the Society grouped wards as smaller (up to four vacancies), medium (5-6 vacancies) and larger (7-10 vacancies), and aimed to achieve a reasonably representative cross-section in terms of number of vacancies and location in urban or non-urban areas.

Among the elections studied closely, 92 candidates were vying for 34 vacancies in the larger wards of:

Overall summary of candidates and vacancies

1,044 candidates stood for 387 vacancies in 124 wards at the 1999 Regional Council elections, an average of 2.69 candidates per vacancy. The maximum number of candidates was 27, for 10 Alice Springs vacancies.

As there were insufficient candidates in three three-member wards, and the number of nominations equalled the number of vacancies in four one-member wards, one two-member ward and one five-member ward, voters had an opportunity to participate in 115 separate elections on 9 October.

Table 1 below shows that the ratio of candidates to vacancies declined slightly as the number of vacancies rose. However, what is of particular importance to voters is that as the number of vacancies rose, they had the benefit of many more candidates from whom to choose without their ballot-papers ever becoming unwieldy.

Informal voting and related matters

The Proportional Representation Society of Australia applauds the fact that, as in the admirable Hare-Clark system of the Australian Capital Territory, marking a single first preference is sufficient for a ballot-paper to be accepted as formal. Provided that the need to use numbers has been widely advertised, this should keep the level of informal voting modest.

Nevertheless it is also important for voters to know that they cannot harm the prospects of those they most wish to see elected by marking further preferences for other candidates. Once they understand that the marking of preferences is simply the expression of an order in which candidates are to have access to whatever remains of their single transferable vote, voters will appreciate that it is in their interest to mark as many preferences as what they truly hold.

Unless resources can be devoted to educating voters about how to make the most of their vote, a liberal formality regime inevitably means that the level of exhausted votes will escalate in some circumstances, and that those elected at the end of scrutinies may fall considerably short of a quota of votes.

Material published on the Internet site of the Australian Electoral Commission suggests that the overall rate of informal voting was 4.95% in 1993 and 2.85% in 1996. Comparable figures for 1999 are not available at that location.

In 1993, informal voting rose as the number of vacancies increased, from around 4.3% when there were seven or fewer vacancies, to 5.0% in the range to thirteen, and 5.7% where fourteen or more candidates were to be elected. The highest overall rates were in Queensland, Victoria and the Northern Territory, with disturbingly high levels between 15% and 25% in some places.

By 1999, the situation had improved markedly. In the 13 scrutinies studied, the informality rate ranged from nil to 5.43% with a median value of 2.85%. Perhaps somewhat surprisingly, the rate was lowest for larger wards (7-10 vacancies, median 2.02%), with smaller wards next (2-4 vacancies, median 2.20%) and medium-sized wards highest (5-6 vacancies, median 3.69%).

The Society concludes that local factors must be important, and suggests that there be an examination of educational efforts in areas with rather high nominations and notably low informal voting to see if the good practice involved can be documented and translated to areas where problems were experienced. Perhaps the development of simple visual material about what happens to votes would help in sparking voter interest and result in lower rates of vote wastage.

While clearly no-one should contemplate making it harder for people to lodge a formal vote, the difficult question arises of whether single ticks or crosses ought to be accepted as formal votes. Ideally one would like as many people as possible to participate effectively in an election.

However the quota-preferential system is geared to a numerical statement of what is to happen to each single transferable vote, and most other public elections also require numbering of squares. The acceptance of as many papers as possible in Regional Council elections may lead to those voters becoming permanently disfranchised when they try to repeat these markings elsewhere.

On balance, the Proportional Representation Society of Australia therefore believes that the current formality provisions should be retained and that further effort be put into promoting the need to number squares.

What can be said about elected candidates?

Just as voters were not confronted with any strategic minefield of considerations, candidates were also assured of a fair go.

For candidates, there are just two possibilities: election (with or without a quota) and exclusion. No more votes are required of them once they were mathematically certain of election. The quota is calculated with exactly that in mind and any surpluses of elected candidates are transferred in accordance with the wishes of voters contributing to the election of those individuals.

The scrutiny operates as a continuous process revolving around whether anyone has sufficient votes to be declared elected and have a surplus distributed. If no-one is in that position, the candidate with fewest votes is excluded and those votes are transferred to continuing candidates where possible.

Where a candidate has been excluded, unless a ballot-paper indicates a further preference for continuing candidates, it can only be set aside as exhausted. The only remedy against this type of vote wastage is therefore a vigorous education campaign pointing out the benefits of marking as many preferences as are held, and emphasising that taking such a course cannot harm the prospects of election of those supported most strongly.

Except where there are organised groups successfully urging voters to mark squares in a particular order, a useful rule of thumb has been that candidates with half a quota of first preferences can expect to fare quite well, and that those with less than a quarter of a quota stand little prospect of success.

Table 2 below classifies first preferences received by candidates according to how close they were to the quota in their ward.

The three distributions were broadly similar with a slightly greater proportion receiving less than 0.3 of a quota in the smaller and medium wards. Compared with six years earlier, the greatest change was the steep fall, by around two-thirds, in the frequency with which candidates received very few votes (less than 0.1 of a quota).

First preferences for elected candidates, shown in Table 3 below, revealed more differences according to the number of vacancies.

As the number of vacancies rose, the distribution of first preferences obtained by elected candidates spread out a little more widely. Whereas three-quarters of successful candidates in smaller and medium-sized wards started with more than half a quota, two-thirds did so in larger wards. In addition, scrutinies in all smaller wards began with several exclusions as no candidates were able to obtain a quota of first preferences.

Six years earlier, wards with more than a dozen vacancies had more than 40% of successful candidates starting with less than half a quota, and a few candidates at all levels were able to secure election from a starting point of less than 0.3 of a quota. In 1999, apart from the absence of such wards, the biggest change was that, compared with wards of equivalent size in 1993, a greater proportion of elected candidates in larger wards started with between 0.3 and 0.5 of a quota.

Three of the four elections in smaller wards saw the election of the same group of candidates as would have occurred under first-past-the-post rules. This phenomenon occurred in two of the five elections in medium wards but not at all in larger wards. The greater spread of candidates’ first preferences in larger wards was a factor in one or more with moderate initial support building up their progress totals as exclusions continued.

Only three candidates with more than half a quota of first preferences failed to be elected, two in circumstances in medium-sized wards where there were more candidates in this category than there were vacancies.

As happened six years ago, about half of the successful candidates at all levels were elected with less than a quota. Table 3 below illustrates how the proportion of candidates for whom less than 0.8 of a quota sufficed (at the point when the last unsuccessful candidate was excluded) increased noticeably with the number of vacancies.

Margins between the last unsuccessful candidate and the last one elected were generally not very narrow, averaging nearly 10 across the thirteen scrutinies studied. As the number of vacancies rose, these margins shrank markedly, starting with an average of 16 in smaller wards, and dropping to averages of 9 and 5 in medium-sized and larger wards respectively.

Final margins as low as 1 in Kaurna and 3 in Alice Springs highlighted the importance of having scrutiny procedures that are completely above reproach and capable of being explained without embarrassment even to excluded candidates who initially believe they somehow got a raw deal. A good starting point is to have procedures that seek to make the most of every voter’s efforts.

In all of the four large wards, more than three-quarters of a quota of votes was exhausted, and over a quota in two of them. Three of the five medium-sized wards had exhaustion of virtually a quota, or more, while in the four smaller wards, this level of vote wastage was under half a quota. Median levels were respectively 0.34, 0.97 and 1.04 of a quota as the number of vacancies rose.

Losses by fractions were all less than 0.1 of a quota in smaller wards whereas two of those in medium-sized wards were greater, and all of those in larger wards, getting as high as 0.44 of a quota in one ward.

The combined loss through fractions and exhaustion exceeded a quota on six occasions, three each in medium-sized and larger wards.

Table 5 sets out the distribution of formal votes into those for successful candidates, those for the last unsuccessful candidate, those exhausted and those lost by fractions.

The table illustrates how voter satisfaction, as measured by votes for elected candidates, increases with number of vacancies, and how quickly votes for the last unsuccessful candidate trail off. It also makes it clear that the level of exhausted votes is rather high under current procedures and progressively assumes greater prominence vis-à-vis votes for the last unsuccessful candidate. Finally, it shows how loss by fractions grows with number of vacancies.

Boundaries

The Review Panel is specifically charged with the task of reviewing current electoral boundaries and making recommendations about any new ones. The Society’s comments on these matters are brief and based on general principles, in part as the reasons for the current mix of arrangements are not always clear.

Following their analysis of local government elections from the point of view of choice available to voters and the extent to which they exercise an effective vote, constituent Branches of the Society have found empirical evidence pointing to the conclusion that at-large elections serve voters better rather than does division into wards. For instance, in April 1984 the New South Wales Branch commented in its published booklet on the previous year's council elections, "Except in extraordinary circumstances, elections should be at large rather than in ridings or wards, as this is the arrangement most likely to give the representation that the voters want."

Tables 1 and 5 set out some of the benefits to voters from having more vacancies in wards. The proportion of votes for elected candidates would have been higher in 1993 because of the numerous wards of very large size that existed then. At the same time, informality rates were also higher.

If nine people are to be elected, the quota is just over 10% (one-tenth), so those left without direct representation need be no more than 10%. Taken as three 3-member wards, there would be a 25% (one-quarter) quota in each area, and hence up to two-and-a-half times as many people could legitimately feel disgruntled about the lack of an effective vote.

In a 5-member ward, 16.6% (one-sixth) of voters could be left dissatisfied and in a 7-member ward, 12.5% (one-eighth). Once the number of vacancies increases beyond 11, the gain in accurate reflection of voters' wishes is much more modest for each further increment, and the number of candidates on the ballot-paper may start to become an irritant, particularly if several squares have to be marked to record a formal vote.

Consequently the Proportional Representation Society of Australia recommends that seven vacancies be seen as the desirable minimum for any ward, unless there are extremely good reasons for smaller numbers to be used. We believe that there are currently too many wards returning just one, two or three representatives. The uncontested elections and general level of nominations in these circumstances appear to reflect dissatisfaction on the part of the community also.

The Society also suggests that there be an odd number of vacancies in each ward unless there are specific compelling reasons to the contrary. This is because whenever there is an odd number of vacancies, 50% of the votes determines a majority of the places.

However, if the number of vacancies is even, this need no longer be so. For instance, in a 6-person ward, the quota would be just under 14.3% (one-seventh). If candidates expressing opposing views on a particular issue obtained 44% and 56% combined support respectively, three espousing each view would be elected to the Council.

As already mentioned, it is only through education about the meaning of preferences that the level of exhaustion of votes when candidates are excluded can be prevented from spiralling.

However, the scrutiny rules for the treatment of surpluses of elected candidates have a direct bearing on the level of votes wasted through not helping to elect candidates. The current definition of the transfer value causes votes to be unnecessarily written off as exhausted or lost by fractions when they could readily be treated as being fully effective.

In its study of the thirteen scrutinies, the Society found that in the transfer of surpluses amounting to 327 votes, the extraordinarily high number of 166 votes was deemed exhausted or lost by fractions. This is completely unsatisfactory, the more so as practical options exist to ensure that such wastage of votes is kept to a minimum and consequently the wishes of all those voting are reflected more accurately in the final outcome.

First, instead of just dividing a surplus by the total number of papers contributing to a candidate's election, there should be an assessment of how many of those papers are not transferable, and an attempt made to incorporate those non-transferable papers fully within the elected candidate's quota. Success will mean that the voters concerned have the whole of their vote contributing to the election of candidates.

Voters have been given the choice of how many preferences they mark. Attempts to realise the full value of their vote should not stop just because they have made particular choices.

Second, if each vote is given a value of 10 or 100 at the start of the scrutiny, the quota comes down a little (thereby avoiding elected candidates having to obtain more votes than strictly necessary) and, because we are effectively operating to one or two decimal places, losses by fractions can often be whittled down by a factor of ten or a hundred.

Among the scrutinies studied by the Society, unnecessary wastage of votes was best exemplified in the Alice Springs election where 27 candidates contested 10 vacancies. Two candidates (Hoosan and Hunt) each had a quota of first preferences, and a third (Shaw) was elected when the first surplus was distributed.

The three surpluses totalled 76 votes and in their attempted distribution, 18 were exhausted and 26 were lost by fractions – that is, continuing candidates were credited with just 42% of the excess votes that the three candidates received. These largely-avoidable losses are all the more disturbing because at the end of the scrutiny, five candidates were bunched within ten votes and the one who had to be excluded was at that point just three votes behind the last of those to be elected.

Taken together, the two remedial techniques proposed by the Society would have reduced vote wastage when the three surpluses were distributed to as little as 2.9 or even 0.25 votes! The 18 votes currently deemed to be lost by exhaustion would all have fitted into the quotas of the three elected candidates.

Better ways of distributing surpluses in practice

In the Alice Springs ward, there were 785 formal votes for the ten vacancies, so the quota, the first whole number greater than one-eleventh of this total, was 72. Candidates Hoosan and Hunt were elected on first preferences, with 108 and 101 votes respectively. The first task was to distribute Hoosan’s surplus of 36 votes.

From the scrutiny sheet, it is possible to ascertain that (disregarding Hunt) 58 votes for Hoosan had Shaw next, between 15 and 17 could not be transferred to any continuing candidate, between 6 and 8 had H Furber (the last of the unsuccessful candidates) next and between 3 and 5 placed Kenny next. Other candidates had the next available preference on either zero, one or two ballot-papers.

For the sake of illustrative concreteness, let us assume that 16 papers were non-transferable, seven and four respectively had Furber and Kenny next, and that six continuing candidates came next twice each and eleven were next just once each. Any distribution consistent with the detail on the scrutiny sheets, including of course the one that actually occurred, will produce essentially the same results as can be found in Table 6 below.

In the actual scrutiny, the transfer value was 1/3, resulting from the surplus of 36 being divided by 108. Multiplying out by the numbers of ballot-papers involved, Shaw was credited with 19 extra votes (taking him beyond the quota), five votes were deemed to be exhausted, Furber’s progress total was increased by two and Kenny’s by one, and everything else was written off as lost by fractions.

An alternative approach is to start by saying that those voting for Hoosan but not having a further preference for a continuing candidate should still get a full vote’s worth of influence if possible. This is achieved by arranging for the whole of the 16 non-transferable votes to be incorporated within Hoosan’s quota, leaving the surplus of 36 to be shared among the 92 transferable papers, at a transfer value of approximately 0.3913.

In these circumstances, Shaw receives 22 additional votes instead of 19, and the only other change is that losses by fractions increase from 9 to 11 (instead of everything occurring in lots of 0.3333, multiples of 0.3913 can be lost when continuing candidates attract at least one next available preference). The net effect is that votes wasted are reduced from 14 to 11.

As there are ten vacancies, the quota would be 72 for any number of formal votes between 781 and 791 inclusive – the closer that this is to 781, the greater is the fraction that needs to be added on because the quota is the next highest whole number after division by 11. Those regularly conducting elections with smallish numbers of votes recognise that wastage from being forced to have the quota too high can be reduced by giving each vote a value of 10 or 100, that is, working to one or two decimal places. Of course, losses by fractions are also greatly diminished as soon as this is done.

If each vote is given a value of ten, the quota becomes 714 (71.4 of the ‘old’ votes) after dividing 7850 by 11. Similarly, the quota drops further to 7137 (equivalent to 71.37 of the "old" votes) if each vote is worth 100.

Table 6 sets out the impact on candidates, exhausted votes and losses by fractions of:

• first using the number of transferable votes as the divisor for the transfer value (combined wastage 11 votes instead of 14);
• then just calling each vote 10 (wastage down to 6.5 votes);
• then combining both of the changes immediately above (wastage now at 1.9 votes); and finally
• calling each vote 100 and adjusting the definition of the transfer value (combined wastage falls further to a negligible 0.14 votes).

Even greater levels of unnecessary vote wastage were found in the distribution of Hunt’s surplus that followed, primarily because between 42 and 45 of the votes for him either did not get beyond a first preference or had further preferences for just Hoosan and/or Shaw who had both already been elected. Extraordinarily, 22 votes of the surplus of 29 were wasted, 12 being deemed exhausted and a further 10 lost by fractions.

Again, as it was not possible to ascertain from available printed material either the exact number of non-transferable papers or the numbers with individuals as next available candidate, a distribution of the 101 papers that is consistent with the scrutiny particulars was constructed for the purposes of illustration in Table 7 below. In particular, 43 ballot-papers were taken to be non-transferable and 12 and 8 (both also one more than the minimum possible) were assumed to have placed next the candidates whose progress totals increased as a result of the distribution of the surplus.

Table 7 shows how wasted votes would have fallen from 22 to:

• just 6 if the transfer value became the surplus divided by the number of transferable papers;
• 14.2 if each vote was simply given a value of 10;
• 0.5 if both the preceding changes were made; and
• 0.04 if each vote were given a value of 100 and the number of transferable papers were the divisor in the transfer value.

Taken together, Tables 6 and 7 show how, by adopting a better definition of the transfer value, instead of 36 votes (half a quota!) being set aside as exhausted or lost by fractions at the start of the scrutiny in a closely-contested election, either 2.4 or 0.19 votes could have been lost by fractions. In this process, no candidate or voter would have grounds for complaining about changes that simply improve the level of vote effectiveness.

In addition, the further 8 votes wasted when Shaw’s surplus of 11 was then distributed could have been reduced to 0.5 or 0.05 votes respectively by adopting essentially the definitions in the two rightmost columns in the two tables.

Having over 40 extra votes credited to candidates after the distribution of the first three surpluses should be seen as an important potential improvement in scrutiny procedures. The Proportional Representation Society of Australia therefore suggests that each vote be called 100 for the purposes of the scrutiny and that the number of transferable papers become the divisor when a quota comprises ballot-papers all of the same value.

The two-pronged changes to the definition of the transfer value illustrated above are in accordance with the long-used rules for conducting small-scale quota-preferential elections of both the Proportional Representation Society of Australia and the Electoral Reform Society of Great Britain and Ireland.

Use of the number of transferable votes as the divisor in establishing transfer values is already quite widespread in public elections. It was adopted in Malta and Eire upon the introduction of proportional representation there during the 1920s, and has operated since the 1970s for both New South Wales local government and Legislative Council elections. It is also a feature of the Hare-Clark system introduced in the Australian Capital Territory in 1995 with a stated aim of deliberately minimising the number of votes set aside as exhausted.

Avoid transfer value anomalies in more complicated situations also

The Alice Springs scrutiny also highlighted a further defect in the definition of the transfer value when an elected candidate has ballot-papers of different value in his or her quota. An extremely crude procedure devised in the 1980s, presumably to simplify matters in the days when the possibility of computer counting was not remotely on the horizon, is specified in legislation: the surplus is divided by the total number of ballot-papers that constitute the progress total of the elected candidates, to establish a single transfer value.

This unweighted average results in some voters being required to contribute proportionately more than others to the candidate’s election. It always creates anomalous situations, and sometimes affects actual outcomes. The most glaring defect occurs when the transfer value of some ballot-papers (typically of very small value when received by the elected candidate) rises as a result of a surplus being distributed. In that event, the voters whose ballot-papers experience this phenomenon actually get more than one vote!

The problem can be visualised by seeing what happened to the 83 votes of Shaw upon election, 64 being first preferences and 58 other ballot-papers having been transferred at value 1/3 from Hoosan. The definition of transfer value applying in Regional Council elections has each of the 122 ballot-papers moving on at value 11/122 or approximately 0.09.

Voters giving Shaw their first preference were therefore finding themselves with 91% of their ballot-paper’s value helping to elect Shaw and 9% available for someone else. Voters giving Hoosan their first preference already had two-thirds of its value contributing to his election and one-third flowing to Shaw, and now had 0.09 moving on to someone else: in other words over 0.24 of their vote’s value (nearly 75% of what was left when it came to Shaw) contributed to Shaw’s election and the rest was available for others.

Put simply, the procedure specified in the legislation always gives papers of least value in the composition of a quota undue weight in determining how the flow of further preferences is translated into increments to progress totals of continuing candidates.

There is no justification for such a disparity of treatment. A more principled approach would involve starting with the assumption that the same proportion of each vote contributing to a candidate’s quota should be taken to be used in securing that election and the rest made available to continuing candidates.

Table 8 illustrates this difference and related information in circumstances where all papers are transferable. To maintain the focus on this particular issue, adjustments have not been made to reflect the effects of an improved definition at the point when Hoosan’s surplus is distributed.

In practice (and this was the case with Shaw’s votes), not all papers will be transferable. The first step should always be to calculate continuing vote weight by multiplying numbers of transferable papers in each batch of a particular value by that value and summing these products.

If this aggregate continuing vote weight is less than the surplus, each paper should be transferred at the same value as that at which it was received: in these circumstances, there will be an unavoidable loss through exhaustion. Otherwise, where ballot-papers for the elected candidate are transferable, each previous transfer value should simply be multiplied by the surplus and divided by the (aggregate) continuing vote weight.

From the scrutiny sheets themselves, it is only possible to say that there were between 12 and 22 non-transferable papers among those helping to elect Shaw and leave him with a surplus of 11 votes. For the sake of concrete illustration, let us assume 18 papers were non-transferable, 10 of them coming from among the first preferences. In that case, the continuing vote weight would be 54x1 + 50/3 = 212/3 and so all previous transfer values would be multiplied by 33/212 or roughly 0.156, in the case of transferable papers, while the non-transferable papers, of combined weight 10 + 8/3 =12.66, would contribute all of their remaining value to Shaw’s quota.

Consequently, 54 ballot-papers with first preference for Shaw would move on at value 33/212 while 10 would have their entire value incorporated within Shaw’s quota. Of the 58 ballot-papers with Hoosan as first preference, 50 would move on at value 11/212 and the other 8 would contribute their entire remaining 1/3 to Shaw’s quota.

A scrutiny where transfer values rose twice

Remarkably, the scrutiny in Tumbukka, where there were seven vacancies and the quota was again 72, provided two examples of the transfer value of some ballot-papers rising in mid-stream.

After the distribution of one surplus (both votes were unnecessarily lost by fractions!) and seven exclusions, candidate Johnson was elected, with 78 votes, to which contributed 78 ballot-papers of full value and four of value 0.02702702. The new transfer value was 6/82 or roughly 0.07317073, meaning that the four papers of least value had more than doubled in worth to new candidates as a result of having just helped elect Johnson! On this occasion, two of the six votes in the surplus were deemed to be exhausted and three lost by fractions.

The next exclusion saw the election of candidate Clark with a progress total of 88, made up of 88 ballot-papers of full value and a further 11, some of transfer value 0.02702702 and the rest of transfer value 0.07317073 (the actual numbers cannot be inferred from the available scrutiny sheets). Clark’s surplus of 16 was divided by his total number of 99 ballot-papers and the transfer value 0.16161616 subsequently attached to each of them.

This may have caused some of the 11 ballot-papers that had already helped elect another candidate to experience virtually a six-fold increase in value! If seven or more of these 11 papers had a particular continuing candidate next, the increases would have made a difference to progress totals.

Unless a change is made to the unsophisticated procedure for dealing with ballot-papers of different value that contribute to a quota and surplus, it is only a matter of time before a material injustice occurs to some candidate.

With Regional Council scrutinies now conducted electronically, there is simply no excuse for retaining deficient procedures and risking outcomes that are unfair. Changes should be made now to the definition of the transfer value, first to minimise exhaustion and loss by fractions, and second to avoid anomalies arising if no account is taken of the different values at which an elected candidate has received particular batches of votes.

The Proportional Representation Society of Australia therefore asks the Review Panel to take up the principle of giving every voter maximum possible effect for his or her single transferable vote, through the continuing vote weight concept in more complex transfer value settings. Those currently losing part of their vote unnecessarily in distributions of surpluses will be grateful, and others really have no business even thinking of complaining that people are getting full value for their vote. That is what the quota-preferential method is meant to be all about!

Deferring the distribution of a surplus

If the definition of the quota is changed in line with what is proposed above, it will be possible for candidates to be elected with surpluses of less than one vote. Should the progress totals for continuing candidates lowest in the poll be several votes apart, in manual counting it would be preferable to delay the distribution of any small surplus(es) until one or more exclusions have been carried out. If transfer value definitions have been properly sorted out, it becomes fairly much a matter of taste whether to include possible deferment of transfers of surpluses in the writing of a program to conduct a scrutiny electronically.

As a general rule, the fewer the number of continuing candidates, the more straightforward will any distribution of a surplus become. Hence as long as no candidate is extremely close to a quota (no closer than the total of undistributed surpluses) and the difference between the two candidates lowest in the poll is less than the total of undistributed surpluses, exclusions should take precedence, because neither the order of election or exclusion can be altered.

A respectable case can be mounted for ignoring situations where a candidate is rather close to a quota, even though as a result of deferment the order of election of candidates (and therefore who is elected and who not) could change in circumstances that are not completely far-fetched.

Adequate provisions relating to the deferment of surplus distributions are to be found in the procedures governing New South Wales local government elections, and were in place in the Senate scrutiny rules before the extensive changes made to the Commonwealth Electoral Act in 1983.

Bulk exclusions

In its examination of scrutinies, the Society noted that some bulk exclusions had occurred and that there had been a deferment of a distribution of a surplus in Barra in order that a bulk exclusion take place. However, in Broome Outer Areas, Wanga Pulka North, Gascoyne, Gattang and Barra, either opportunities for bulk exclusions were missed or the bulk exclusion was not extended as far as logically possible.

An examination of the extremely convoluted language governing these provisions showed that they are incapable of totally picking up situations where the sum of the votes for a group of candidates is equal to the progress total of the one standing next highest in the poll (and therefore each of those candidates will necessarily be excluded before the one standing next highest in the poll).

This is not the only problem with the current bulk exclusion provisions. To put the matter succinctly, they take the key elements of the hunt for a bulk-exclusion opportunity out of sequence. The first thing to be certain of in practice is that there remain at least as many candidates as there are unfilled vacancies - there is no point of even thinking theoretically about excluding beyond this number! Then one can check whether aggregate progress totals downwards from a particular candidate are less than the next highest progress total and perhaps can't put anyone past the quota either.

In fact, a reasonable case can again be made for not even worrying about the possibility of someone being elected during the course of a bulk exclusion, although it is true that in some cases the order of election could alter because a candidate is held to be elected as a result of a smaller or larger set of ballot-papers.

The current bulk exclusion procedures start looking for situations where the order of election or exclusion cannot alter even theoretically, and then incorrectly presume that it is all right to take the largest available subset if the biggest example found happens to end up electing too few candidates. That line of reasoning constitutes a fairly elementary logical lapse in regard to which the main consolation is that it is not likely to have a noticeable deleterious effect in large-scale elections with fairly disciplined or predictable voting patterns.

Casual vacancies

The Statutory Rules providing for casual vacancies to be filled by re-examination of the ballot-papers counted for the outgoing candidate are on a par with world best practice because they ensure that no more than a quota of votes can be parcelled out to see which consenting candidate those voters were most prepared to have as a replacement. It would be prudent to insert further clauses saying that in the event of a need to replace someone who filled a casual vacancy, the votes of the relevant person successful at the general election should again be re-examined (as in the Hare-Clark system used in Tasmania and the Australian Capital Territory), rather than voters being automatically invited to return to polling places for a mid-term by-election.