Chapter 9: How votes are counted

In House of Representatives and other lower house elections (except in Tasmania and the ACT), each electorate elects one representative. The three main ways used in Australia to determine the winning candidate in this situation are or have been:

  • first past the post
  • full preferential voting
  • optional preferential voting.

First past the post

In a first past the post system, voters simply indicate (usually with a cross) the candidate that they prefer over all others. The system is used in the United Kingdom, among other places. The winning candidate is the person who gets most votes, regardless of whether or not that candidate wins a majority of votes. Consider the following result, taken from the seat of Melbourne at the 2010 federal election.

Seat of Melbourne, 2010 federal election
CandidateVotes (per cent)
Georgina Pearson (Family First)1.6
Adam Bandt (Green)36.2
Joel Murray (Australian Sex Party)1.8
David Collyer (Australian Democrat)0.7
Penelope Green (Socialist Party of Australia)0.7
Cath Bowtell (Labor)38.1
Simon Olsen (Liberal)21.0

Under a first past the post system, the winner would have been Labor’s Cath Bowtell. Even though she only gained 38.1 percent of the total vote, much less than a majority, she won more votes than any other candidate.

Many commentators would argue that such an outcome would not be terribly democratic. A party winning electorates with 35 or even 40 percent of the vote could win government with much less than a majority of votes across the country. This happens regularly in the United Kingdom where first past the post counting is used. In 2015, for example, the British Conservative Party won a slim parliamentary majority and therefore government after winning just 37 percent of the nationwide vote.

Preferential voting

Preferential systems of voting deal with the problem of winning candidates not necessarily winning a majority of votes by making voters indicate a series of preferences among the candidates. Voters write ‘1’ in a box on the ballot paper next to the candidate they want to see elected, then ‘2’ against the candidate they would prefer after their first choice, and so on. In the case of Melbourne, voters would have to indicate seven preferences by numbering the boxes on the ballot paper 1, 2, 3, 4, 5, 6 and 7.

Counting the votes

Where no candidate wins more than 50 per cent of the first preference votes, the least popular candidate (the one with the fewest first preferences votes) is eliminated. The second preferences of voters who voted for that candidate are then distributed among the remaining candidates. If there is still no candidate with a majority of votes, the remaining candidate with the fewest votes is eliminated and the preferences of his or her voters distributed. This process is repeated until one candidate gets more than 50 per cent of the votes.

In Melbourne in 2010, no candidate won a majority of first preference votes. The Democrat candidate, with fewest votes, was eliminated and his preferences distributed (see table). No candidate had a majority of votes after this process. The Socialist candidate was then eliminated, followed by the Family First candidate and the Sex Party candidate. Labor’s Cath Bowtell still had more votes than the other two remaining candidates after the fifth count but was still well short of a majority. After the Liberal candidate’s preferences were distributed, the Greens candidate Adam Bandt won the seat of Melbourne with 56 per cent of the final vote.

Seat of Melbourne, 2010 federal election
CandidateVotes after first countVotes after second countVotes after third countVotes after fourth countVotes after fifth countVotes after sixth count
Georgina Pearson (Family First)1,3891,4451,522 Excluded   
Adam Bandt (Green)32,30832,48132,67633,37134,66450,059 Elected
Joel Murray (Australian Sex Party)1,6331,7391,9162,120 Excluded  
David Collyer (Australian Democrat)602 Excluded     
Penelope Green (Socialist Party of Australia)613671 Excluded    
Cath Bowtell (Labor)34,02234,13434,28734,54734,98239,268
Simon Olsen (Liberal)18,76018,85718,92619,28919,681 Excluded 

As the Melbourne example shows, preferential voting always produces a candidate with a majority of votes. It also shows that it may take quite a few preference distributions to achieve this result. In Melbourne, some voters may have helped to elect the candidate who was their fourth, fifth or even sixth preference. The Melbourne result in 2010 was unusual, since in most cases, the candidate with the largest first preference vote goes on to win the seat. It was particularly unusual in that minor parties almost never win lower house seats in Australian elections. The fact that a Green was elected as a result of Liberal preferences in Melbourne also shows that parties sometimes advise their voters to give preferences to candidates with whom they share little ideologically so as to inflict damage on another rival party, in this case Labor.

Preferential voting gives voters the opportunity to register support for a minor party or Independent candidate that they like, but know has no chance of winning, while using their second and later preferences to help determine who does win the electorate. For most commentators, this feature of preferential voting provides distinct advantages over first-past-the-post voting. Critics of preferential voting argue that they should not be forced to indicate preferences, even if they are low preferences, for candidates whose views they find offensive.

Optional preferential voting

Optional preferential voting in electorates each choosing one representative provides a middle path between first past the post and preferential voting. It has been used in New South Wales since the 1981 state election and was used in Queensland from 1992 to 2015. In both cases, Labor governments introduced optional preferential voting.

Optional preferential voting requires only that voters register a first preference on their ballot papers for their votes to be counted. Voters who wish to register one or more additional preferences among the remaining candidates are able to do so in the normal way, using consecutive numbers. Unlike full preferential voting, these voters do not have to indicate preferences among all the candidates. They can, for example, just indicate their first preference with a ‘1’ and leave the rest of the ballot blank.

Counting the votes

If no candidate gains more than 50 per cent of the first preference votes, then the least popular candidate is eliminated, as in full preferential vote counting. Any second preferences of voters who voted for that candidate are then distributed among the remaining candidates.

The ballot papers of voters who have indicated only a first preference for the least popular candidate are set aside as ‘exhausted’. These exhausted ballots no longer count as part of the total number of votes among which a winning candidate has to gain a majority. The process is repeated until one candidate gets more than 50 per cent of the remaining votes.

The following example, taken from the electorate of Kogarah at the 2011 New South Wales state election, shows how an optional preferential vote count works.

Electorate of Kogarah, 2011 NSW state election
CandidateVotes after first countVotes after second countVotes after third count
Joseph Abdel (Christian Democrats)2,507 Eliminated  
Miray Hindi (Liberal)18,36019,15519,665
Cherie Burton (Labor)19,66819,92221,207 Elected
Simone Francis (Greens)3,9524,062 Eliminated 
Total votes in count44,48743,13940,872
Total exhausted votes 1,3483,615

Labor candidate Cherie Burton won the seat of Kogarah after the distribution of preferences from Christian Democrat and Greens voters. Over half of the Christian Democrat and Greens voters (3,615) did not express a preference for Burton or Hindi. While Burton won 51.9 per cent of the 40,872 votes that remained in the count after the distribution of preferences, she was elected with a little under 48 per cent of the 44,487 valid votes cast in the Kogarah electorate.

As this example illustrates, optional preferential voting always produces a winning candidate with a majority among voters who have expressed a preference between the winner and his or her main rival. It does not, however, always produce a winning candidate supported by a majority of all voters who have cast a valid vote.

Proportional representation

Preferential voting systems discussed above are generally used to elect one representative. Where more than one representative is to be elected from each electorate, proportional representation systems tend to be used.

Proportional representation has been used for federal Senate elections since 1949.  Versions of proportional representation are now used in  state upper house elections in New South Wales, Victoria, South Australia and Western Australia; lower house elections in Tasmania; Legislative Assembly elections in the Australian Capital Territory; and some local council elections.

The most common forms of voting and counting votes used in these elections are variations on the ‘Hare-Clark’ system. The system is named after the nineteenth century London barrister Thomas Hare and the Tasmanian Attorney-General Andrew Inglis Clark, who altered Hare’s system and campaigned successfully for its introduction in colonial Tasmania. The Hare-Clark system is a ‘quota-preferential’ system. This means that successful candidates must win a quota of all the valid votes cast and that preferences from other candidates can be included in a winning candidate’s quota.

Preferences in Senate ballots

Hare-Clark systems are complicated both for the voters who have to use them and for the officials who have to count the ballots cast. In Senate ballots between 1949 and 1984, voters had to indicate preferences for all of the candidates on the Senate ballot paper. When there were only a few Senate candidates, this was relatively easy. By the late 1970s, however, dozens of candidates were contesting Senate elections. The requirement to fill in large numbers of preferences meant that almost one in 10 voters made mistakes on their Senate ballot papers, resulting in their votes not being counted.

The Hawke Government dealt with this problem in 1983 by introducing reforms to allow voters the option of indicating by a ‘1’ the party or group they wanted to vote for in Senate elections. This ‘group voting ticket’ method of voting is often called ‘above the line’ voting, because the registered parties’ names appear on the ballot paper above a thick line separating them from the names of the individual candidates, which appear ‘below the line’. Voters could now vote with a ‘1’ in a party box above the line, or else fill out preferences in all the boxes ‘below the line’. Voting above the line quickly became extremely popular in Senate elections, with around 95 per cent of voters using this method and only five per cent preferring the older ‘below the line’ method. Variations of these ‘above’ and ‘below’ the line options are used in state upper house elections in New South Wales, Victoria, South Australia and Western Australia.

The 2016 changes to the Senate ballot

What happened to the preferences of voters who voted above the line? In elections between 1984 and 2013, each political party registered the way that it wanted the preferences of voters who voted for it above the line to be distributed among all the candidates on the Senate ballot paper. This gave the parties control over which other parties would receive votes that could not be used to elect one of their own candidates (see below). At the 2013 election, a group of small parties used this feature of the Senate voting system to channel preferences to each other. This strategy resulted in some candidates with small numbers of initial votes to win seats in the Senate. Ricky Muir of the Australian Motoring Enthusiast Party, for example, won 0.5 per cent of the first preference votes in Victoria but gained a Senate seat through accumulating preferences from other parties.

After much debate about this development, the Coalition Government, with the support of the Greens, passed amendments to the Commonwealth Electoral Act 1918 that removed the ability of parties to direct Senate preferences to other parties. A ‘1’ vote above the line for a party could now only be used to elect candidates from that party. The amendments stipulated that voters who voted above the line had to indicate preferences for at least six parties, by numbering at least six party or group boxes above the line. Voters who wanted to vote for individual candidates below the line could still do so by numbering at least 12 candidate boxes. Slightly confusingly, the amendments allowed Senate votes to be counted as valid if a voter numbered fewer than six boxes but numbered at least one box above the line, or numbered fewer than 12 boxes but at least six boxes below the line.

The specific effects of this move from full preferential voting to optional preferential voting in half-Senate elections are still difficult to predict. The 2016 federal election involved a full Senate election, so the quota for electing senators was lower than for normal half-Senate elections (see below). As a result, some minor parties won Senate seats that they would be unlikely to win in a half-Senate election. The changes will reduce the chances of very small parties with little electoral support achieving Senate representation. Whether it will favour any of the more established parties over others is unclear. It is almost certain to result in some candidates being elected to the Senate without achieving a full quota (see below). 

Quotas in Senate ballots

Hare-Clark quotas are calculated by dividing the number of valid votes by one more than the number of places to be filled, and then adding one vote to the result. The size of the quota is thus determined by the number of places to be filled and the number of valid votes. In the 2013 half-Senate election, the quota that candidates in New South Wales had to reach to be elected was as follows:

(4,526,382 valid votes  / (1+6 Senate places)) +1 = a quota of 646,627 votes

In the 2011 New South Wales Legislative Council elections, the quota to be reached by successful candidates was as follows:

(3,948,985 valid votes / ( 1+21  Legislative Council places)) + 1 = a quota of 179,501 votes

Most of the difference in these New South Wales quotas is explained by the difference in the number of places to be filled in each election (six for the Senate, 21 for the Legislative Council). In percentage terms, the quota for a half-Senate election is approximately 14.3 percent of the overall vote, while for the New South Wales Legislative Council it is approximately 4.5 percent.

For double dissolution federal elections such as 2016, the quota in each state is a little over half that for normal half-Senate elections (roughly 7.7 per cent rather than 14.3 per cent), since 12 rather than six  Senators are elected.

Counting the votes

Vote-counting for Hare-Clark systems is complicated. It has three main stages. The first stage is to determine which candidates have a quota or more of first preference votes. The first candidate in the Senate groups put forward by the Labor and Coalition parties always get more than a full quota in each state at Senate elections. In 2010, for example, the first candidate in the Coalition group, Concetta Fierravanti-Wells, won 1,610,385 first preference votes, or 2.71 quotas. Most of these votes were cast for Coalition above the line. John Faulkner, the first candidate in the Labor group for New South Wales, won 1,515,446 votes, or 2.55 quotas. Fierravanti-Wells and Faulkner were declared elected.

The second stage in the count is to distribute the surplus votes won by candidates like Fierravanti-Wells and Faulkner; that is, any votes over a quota that they do not need to get them elected. These surplus votes are transferred according to second preferences using a ‘transfer value’ to allocate them to other candidates. The transfer value is the total number of votes won by a candidate, minus the quota, divided by the original total. For Fierravanti-Wells, this transfer value was:

(1,610,385 – 593,218) / 1,610,385 = 0.632

All the second preferences of voters who gave their first preference to Faulkner were then multiplied by 0.608 to determine how many votes should go to which other candidates. As a result, the second candidate on the Labor group, Matthew Thistlethwaite, was elected. The same process was carried out with Coalition votes, resulting in the election of the second Coalition candidate, Bill Heffernan.

Thistlethwaite and Heffernan both now had surplus votes. In cases like these, the surplus votes are again distributed using newly calculated transfer values. In 2010, this process added to the votes of the third Labor and Coalition candidates but did not give them enough votes for a quota. Two Senate places for New South Wales thus remained unfilled at this point in the count.

The third stage of the Senate count is to eliminate the most unpopular of the remaining candidates in turn and distribute their preferences to other candidates, as in House of Representatives counts. Many of the remaining candidates often only have a small number of votes, meaning that a large number have to be excluded before another candidate is elected. In 2010, after 259 counts, the third Coalition candidate Fiona Nash was elected on preferences. After 271 counts, the Greens Lee Rhiannon was elected to the final seat.

Proportional representation elections produce parliamentary houses that more accurately reflect the diversity of opinion in the electorate than houses based on first past the post or single member preferential or optional preferential systems. The greater proportionality of Hare-Clark systems can be seen by comparing votes and seats won for the House of Representatives and the Senate in 2013 (see the table below). The Greens won just one of the 150 House of Representatives seats with 8.9 percent of the vote, while the Coalition won 60 percent of the seats with 46 percent of the vote.

The Senate result better reflects the distribution of first preference votes. The over-representation of major parties is reduced and the Greens won as many Senate seats as their votes suggested they should. The same pattern occurs in state parliaments in which one house uses single member preferential voting and the other a version of proportional representation.

First preference votes and seats won in the 2013 federal election
 House of Representatives Senate 
PartyVotes %Seats %Votes %Seats %
Liberal-National Coalition45.660.037.742.5

Formal and informal votes

Ballot papers have to be marked according to the laws governing voting before they are accepted as ‘formal’ or valid votes. Ballot papers that are not marked according to the rules are deemed ‘informal’ and are not included in the count to see who is elected.

While some citizens deliberately cast informal votes - not marking the ballot paper or marking it incorrectly - most informal voting is caused by misunderstandings about the rules. Confusion among voters is undoubtedly heightened by the fact that they often have to vote according to different methods at Commonwealth, state or territory and local elections.

So-called ‘savings provisions’ in Australian electoral laws allow various degrees of latitude for ballot papers that are not filled in entirely correctly but where a voter’s intention is clear. While the Commonwealth Electoral Act specifies, for example, that House of Representatives ballots must include a full list of preferences and Senate ballots must now include six preferences above the line or twelve preferences below it, the Act also allows some ballots to be counted as valid even if a full set of preferences are not indicated, numbers are written twice, or other minor mistakes are made.

Laws regarding informal voting are sometimes controversial, particularly when differences between state and Commonwealth laws cause confusion among voters. In 1990, for example, the New South Wales Coalition Government of Nick Greiner legislated to make ticks and crosses on ballot papers informal. A single tick or cross on a ballot paper had previously been counted as a formal vote indicating one preference under New South Wales’s optional preferential voting system. The move to make ticks and crosses informal produced a higher than usual informal vote at the 1991 state election. This high level of informal voting was widely seen as disadvantaging the Labor Party. Labor amended the state legislation to re-allow tick and cross votes after it won office in 1995.

Legal disputes over elections

As the discussion throughout this Hot Topics indicates, the courts can become involved in elections at various stages, including the registration of parties, candidate preselections, voter enrolments and disputes between candidates during campaigns.

Legal challenges to the outcomes of Australian elections are heard by the Court of Disputed Returns, which is the High Court, the Federal Court or the relevant state or territory Supreme Court. Petitions to the Court of Disputed Returns have to be filed by a candidate or voter in the election within 40 days of the return of the writs officially announcing the election results. The Court is required to deal with the petitions speedily.

Most elections produce petitions to the Court of Disputed Returns on a range of matters that might have affected the result. These include the eligibility of successful candidates and irregularities in the handling and counting of votes. While most petitions are not upheld, those that are upheld may result in by-elections or casual vacancies.

Questions for discussion

  • Why has single member preferential voting generally been chosen over proportional representation voting for lower house elections in Australia?
  • Which of the voting methods outlined above do you think is best? Why?