Newsletter of the Proportional Representation Society of Australia





March 2002



PR Analysis of the 2001 Federal Polls

The Proportional Representation Society of Australia presents, as an insert to this issue of Quota Notes, an analysis and commentary on how the 2001 Federal polls for the House of Representatives could have resulted if a Hare-Clark form of quota-preferential proportional representation had been used instead of the unrepresentative system of single-member electorates still in use. The Analysis will also be shown at www.prsa.org.au where it will show the outcome in graphical form.

The analysis shows particularly marked discrepancies between voter support and the party complexion of the resulting MHRs in certain States. 

In South Australia the Coalition gained 75% of the seats for 45.9% of the first preference votes. In Tasmania the ALP gained 100% of the seats for 47.2% of the first preference votes. 

Another striking feature of the election was the fact that the ALP won only 7 of Queensland's 27 federal seats (26% of the seats). Under a Hare-Clark system it would have won 11 seats (just over 40% of the seats), which compares better with its first preference vote there of almost 35%.

Cleaning up Australian Politics

In January, On Line Opinion invited the PRSA to provide an article on the feature topic "How to clean up Australian politics". See the Web site at (www.onlineopinion.com.au/2002/Feb02/default.htm). 

Others that contributed to the February edition were Independent MHR Peter Andren, Labor MHRs Mark Latham, Carmen Lawrence and Wayne Swan, Australian Democrat leader Senator Natasha Stott Despoja, and State Lower House Independents Dorothy Pratt and Philip Pendal, who were originally elected as party candidates. 

PRSA President Bogey Musidlak took the opportunity to set out in an article "What's wrong with single-member electorates" the reasons why, with the excessive campaign concentration on a relative handful of marginal electorates, and manoeuvring for preselection in safe seats, the job cannot be done using single-member electorates. He started out with the observation, "The only way of improving the image of politics and politicians is to give voters a real say in who gets elected - something that the winner-take-all single-member-electorate system cannot achieve." 

Despite earlier bluster to the contrary when electorate enrolment tolerances were being reduced, governments with minority two-party support clung to power in the 1980s and 1990s. Oppositions continue to be wiped out and, shortly after the feature was published, the circumstances in which South Australian government was decided made it clear that instructions to draw boundaries that will translate majority two-party-preferred support into government are only window-dressing for an intractable problem. 

He pointed out how ineffectual various suggestions for reform would be, noting  "Whatever tinkering is done at the edges, the major distortions inherent in the winner-take-all nature of single-member electorates remain. While voter involvement can be increased somewhat, patterns of geographic dominance are still widely present, and a handful of marginal seats will always attract an unhealthy preponderance of attention." 

Primary elections would make for better conduct in the dominant party in a particular area, but do nothing to foster broader public debate. Voluntary voting would leader to greater attention to the demographics likely to vote in the most marginal 
 electorates, while optional preferential voting would remove some authoritarian strictures, but do little to expand the number of electorates seriously contested. 

In a second article "How to involve voters effectively", Mr Musidlak went through the problems of hybrid systems that tried to lessen massive distortions between votes and seats, and of list systems that often encouraged splintering of parties, or could result in the wastage of enormous numbers of votes. 

Japan's money politics scandals show the need for preferential voting in multi-member electorates while the lop-sided Senates prior to the introduction of proportional representation demonstrate that this is not enough. 

Quota-preferential counting, setting out to minimize ineffective votes, achieves fairness of outcomes. Nevertheless Senate experience with six vacancies in each State since 1990 shows the importance of having an odd number of positions to fill, as then a majority of votes translates into a majority of seats. 

Instead of making it simple for everyone to register a formal vote, party boxes were introduced, in 1983, and have since spread to the Legislative Council elections in Western Australia, South Australia and New South Wales

Noting that "Voters marking party boxes usually have no idea what is to be made of their indicated support", the article gives the NSW Senate outcome of 1990 and the Western Australian Legislative Council elections of 2001 as examples where most of those on whose votes the last place turned would have been greatly surprised at the consequences. 

Among the celebrated innovations in the Hare-Clark system, used continuously for nearly a century in Tasmania, and now entrenched in Australian Capital Territory law, are the Robson Rotation and countback features, which have eliminated safe seats and brought a wider choice of candidates into more contact with the voters they hope to represent. 

In terms of the most appropriate electorate size, the article observes "Once you have seven or nine vacancies in an electorate (12.5 per cent and 10 per cent respectively is enough to secure election), a shift of even 2 or 3 per cent in support is quite likely to change representation. No area can be taken for granted or written off as a lost cause, and there are incentives to build further support to win more seats."

Electronic Voting Theme of ACT Branch AGM

To mark the tenth anniversary of the historic PR plebiscite victory on 15th February 1992, the ACT Branch ensured that opinion-formers were aware of the significant moment. The Branch issued a media release celebrating the way in which the Hare-Clark system has delivered on supporters' expectations, and continues to be improved through collaborative efforts. 

The theme of the ACT Branch's AGM in March was the ACT's pioneering experiences with electronic voting and counting. Guest speakers were Carol Boughton and Matt Quinn from Software Improvements, the ACT-based company announced in April 2001 as the successful tenderer for the development of the electronic voting and counting systems used at the October 2001 general elections for the ACT Legislative Assembly. ACT Electoral Commissioner Phil Green also attended. 

Linuxcare Australia had earlier nominated Software Improvements as a major subcontractor in its tender for the electronic voting and vote counting system. Elections ACT sought by 1st February details of proposed solutions to: 

provide electronic voting within four pre-poll centres around Canberra, and at a number of polling places on polling day; 
be able to enter data of electors' preferences from paper ballots; and 
use a computer program combining electronic voting data and data entered from paper ballots to undertake the distribution of preferences under the Hare-Clark system. 

After the close of tenders, the USA-based parent company of Linuxcare merged with and into Turbo-Linux at which time it assigned all performance and financial obligations of the Linuxcare tender to Software Improvements. 

The fixed timing of the general elections on the third Saturday in October meant that many deadlines were immovable, a somewhat unusual situation within the software industry. Absolutely essential to the intensive task was the ability to clarify requirements and obtain continuous feedback from Elections ACT. In the early stages the essential elements were identified along with those, such as audio instructions for some users, that would be 'nice to have'. 

Contracted programmers began working first on the voting interfaces, then the process for entering preferences recorded on ordinary ballot-papers, and finally the systems for counting the votes and reporting on the scrutiny. 

A data dictionary was prepared and Institute of Electrical and Electronics Engineers standards for documentation of code and its alteration adhered to. Test procedures were written in parallel with the code modules. Because of the tight timing, formal testing was rolled into acceptance testing. For instance, minor late changes were required to the counting program to deal with gains by fraction that are possible because exhausted votes are always kept to a minimum in the ACT. 

Voters had a choice of reading instructions in a number of languages. Voting involved the swiping of an issued barcoded card at a terminal within an isolated local area network from which a zip disc was taken at the end of the day as a full record of voting activity. 

The computer confirmed that the barcode was valid (that is, it was issued at that particular polling place, and hadn't previously been used) and informed the ballot-box server using an HTTP-based protocol that it was starting a vote. The ballot-box server replied with the name of the polling place and the Robson rotation that the booth should display. The cursor always started at the top left place on the screen image of the ballot-paper. 

The keypad had only a few active keys, enabling voters to move up and down column entries, and in simple fashion between group columns. Numbering of preferences was sequential, starting with the first. Every keystroke and vote could be compared, and an error message generated if they did not agree. If no preferences were entered, voters saw a message saying that their vote would be informal if nothing changed and asking whether they wished to proceed. 

After lobbying from sight-impaired groups, having audio instructions available as part of the voting module became a priority relatively late, posing significant challenges that were met to the delight of voters no longer needing to rely on others to follow their voiced instructions. 

Pre-poll voting began on 2nd October, five days after nominations were declared and initial ballot-paper orders were determined. Because all the necessary levels of validation and acceptance could not quite be achieved by then, electronic voting was not available until a week later. 

A total of 16,465 formal and just 94 informal votes (0.6%) were cast electronically out of nearly 200,000 votes for which the overall informal rate was just under 4%. Compared with levels of support received on paper, voting for the Australian Democrats (11.2%) and ACT Greens (10.3%) was up around 3.5% and 1% respectively, while that for Labor (38.2%) and Liberal (30.7%) fell by about 4% and 1% respectively. 

At the centralized counting centre, data entry from paper-based votes also occurred over an isolated local area network. Two operators entered all preferences written on each paper in a batch of fifty that was kept in the same order. The Deputy Electoral Commissioner supervised the resolution of apparent discrepancies, making any final corrections and committing the batch irrevocably to the electronic database. 

Records of progress that included an 'outcome' based only on the papers whose electronic processing was complete were regularly made available to the media and, as the ACT Branch had anticipated, resulted in numerous misleading reports. At the same time, candidates were alerted to the likelihood that very small numbers of votes would again determine some orders of exclusion. 

An extensive question-and-answer session in which the ACT Electoral Commissioner also participated brought out a number of interesting points. 

On the electronic side, informal voting was much lower than with paper ballots (where it continues to be largely deliberate), and the data entry completely correct. A large sample of written ballot-papers found that about 1 in 75 voters made mistakes in numbering. 

But for the demonstrable accuracy of all the electronic data procedures, a recount would probably have had to be granted in Ginninderra, where only 55 votes separated two Liberal candidates at one point. Candidates' awareness of the likely closeness of that exclusion decision because of the availability of full interim particulars meant they could not argue that they should have been given greater opportunity to organize a scrutiny presence more in keeping with that uncertainty. 

The scrutinies were concluded 12 and 13 days after polling, seven days again being set aside for return of postal votes needed before the quotas can be struck. This was 2-3 days longer than required for the original counts in 1995 and 1998. However, the Molonglo recount in 1998 that altered the order of exclusion of ALP candidates, and resulted in one change to the list of those successful, did not finish until 21 days after the scrutiny. 

The entry and checking of written preferences on the numerous Robson Rotation variations of the ballot-paper was more difficult to manage than expected, resulting in slower committal of processed batches. However, the development of an electronic record of all ballot-papers means that any casual vacancies occurring in the life of the current Assembly can be determined almost as soon as all the consenting defeated candidates are known.

When the matter of computer voting from home in the near future was raised, the major obstacles currently were found to be the reliability of identification, and the possibility of coercion where votes were received in quick succession from several members of a household. A report on the electronic voting and counting experience for the ACT Legislative Assembly should appear on the Elections ACT Web site (www.elections.act.gov.au) later this year. Improved barcode readers and mechanisms for faster data entry from votes marked on paper are two areas noted as needing more study. Last-minute unavailability of separate capacity for the Internet connection led to the election-night difficulties on the Web site.

All other Australian electoral authorities showed a keen interest in the pioneering ACT work that one described as 'courageous'. Software Improvements has subsequently made a presentation to the Electoral Council of Australia (www.eca.gov.au) and has begun productive discussions with some jurisdictions. Particular interest was expressed in the audio module that allowed unsighted voters the opportunity to mark a ballot-paper without anyone else knowing their wishes.

Also presented on that occasion was the Austrian-developed PowerVote equipment used in some constituencies in Eire on a trial basis in 2002. It came in a briefcase weighing 25 kilograms, and was reputed to be able to take 1000 votes a day, but was unable to incorporate Robson Rotation into its procedures.


2002 Proportional Representation Society of Australia

National President: Bogey Musidlak 14 Strzelecki Cr. NARRABUNDAH 2604

National Secretary: Deane Crabb 11 Yapinga St. SOUTH PLYMPTON 5038

Tel: (08) 8297 6441, (02) 6295 8137 info@prsa.org.au

Printed by Prestige Copying & Printing, 97 Pirie Street ADELAIDE SA 5000