Candidates, Informal Voting and Optional Preferential Voting

In the final week of parliamentary sittings, Queensland LNP Senator James McGrath tabled the report of the Joint Standing Committee on Electoral Matters (JSCEM’s) inquiry into the conduct of the 2019 Federal election. (You can find the report here.)

One of the controversial proposals in the report, and one opposed by Labor and Green members of JSCEM, was a recommendation to move from full or compulsory preferential voting (CPV) to optional preferential voting (OPV) for House of Representatives elections.

The report provides few arguments in favour of OPV. The recommendation for OPV appears suddenly at the end of a brief discussion on informal voting, in particular noting the impact OPV’s use at NSW elections has on informal voting at Federal elections. Senate McGrath presented a different argument for OPV when tabling the JSCEM report in the Senate, referring to OPV as a measure “to maximise voter choice”.

I’ve written on OPV many time previously, and I would suggest reading this post I wrote in 2013 after a previous suggestion from the Coalition about introducing OPV.

Informal voting will always be higher under compulsory preferential voting because ballot papers with faulty numbering have few options for salvation available. This compares with OPV where ballot papers with incomplete, duplicate or missing numbers, or completed with a single tick or cross, can count for candidates to the extent that a voter’s intent is clear.

The JSCEM recommendation follows a rise in informal voting at the 2019 Federal election, largely caused by an increase in NSW. The federal election was held just eight weeks after the NSW state election. Where NSW uses OPV for state elections, CPV applies for Federal elections. As is usually the case, informal voting was highest in NSW at the 2019 Federal election, was the highest ever recorded in the state at 7.0%, and was up 0.84% on 2016.

Ten of the 12 House districts where informal voting was above 9% were in NSW, the highest being 13.3% in Blaxland. Those 10 districts were all in western Sydney, and all were districts where a large proportion of the population was born overseas and/or has poor skills in written English. The other two in the top twelve were Calwell in Melbourne’s north-west, another high migrant seat, and the Victorian rural seat of Mallee, which saw a field of 13 candidates and a resultant high rate of informal voting.

Research on informal voting is voluminous, reports having been conducted after every Federal election since 1984. Informal voting research is also a regular feature of state election reports.

All this research has pointed to a consistent set of causes for and correlates with informal voting. Among the consistent findings are –

  • Confusion with OPV at state elections, which has been used for NSW state elections since 1981, and was used for Queensland elections from 1992 to 2015.
  • A clear correlation between informal voting and electorates where a high proportion of voters were born overseas, speak a language other than English at home, and have poor English literacy rates. English literacy rates also impact in outback electorates with large indigenous populations.
  • The rate of informal voting increases as the number of candidates increases, especially in the category of informal votes caused by defective numbering.
  • Confusion with instructions on Senate ballot papers. The former instruction to mark one square above the line induced 1-only voting on House ballots. House votes with only a single ‘1’ were reduced by the 2016 Senate reforms, though there was a slight increase in voters stopping at 6 preferences. 1-only voting has also been much lower at by-elections where there is no Senate ballot paper.

That many migrants complete their ballot papers incorrectly is unsurprising as voting with numbers is incredibly rare outside of Australia. The overwhelming majority of countries around the world complete ballot papers by marking a box, usually with a cross. The rare countries that do use numbers, mainly Ireland and Malta, do not require all boxes to be numbered.

The table below is drawn from the AEC’s Informal Voting research on the 2016 House of Representatives election. You can find the report here.

Percentage of Informal Vote by Category – House Elections 2001-2016

Category 2001 2004 2007 2010 2013 2016
Blank 21.4 21.2 20.0 28.9 20.9 24.9
1-only 33.6 32.8 30.1 27.8 29.4 14.6
Incomplete .. 4.4 4.5 2.6 6.8 10.8
Ticks/crosses 12.4 9.3 9.9 11.8 10.5 7.6
Non-sequential 19.9 15.4 17.9 9.2 14.4 15.4
Scribble/Slogans 6.4 14.3 14.2 16.9 14.5 19.8
Other 6.2 2.6 3.3 2.9 3.5 6.8

At the 2016 election, a quarter of voters left their House ballot paper blank, and one in five put scribbles, slogans and other protests on their ballot paper without completing a formal vote. (Ballot papers with slogans and scribble as well as a correct sequence of numbers are formal.)

As has consistently been the case for decades, around half of all errors were caused by the way voters completed their ballot paper. The pleasing feature of the 2016 report was that the incidence of 1-only informal votes halved, presumed to be due to the new Senate ballot paper instructions indicating voters should give at least six preferences. There was a slight rise in voters stopping at six preferences.
But other categories of numbering errors remained stubbornly high.

The incidence of 1-only voting has always been higher in NSW, pointing to the link with optional preferential voting. Ticks and crosses have also tended to be higher in NSW, and in high migrant population seats, and many of those seats are in NSW.

The link between the 1984 Senate changes and informal voting at House elections was clear from the 1984 informal vote research report. The 1984 informal vote stands out clearly in the graph below that shows the average candidates per seat, and the rate of informal voting ,at House elections since the introduction of full preferential voting in 1919.

South Australia learnt the lesson of the 1984 spike in informal voting. In 1985, South Australia dopted a savings provision for lower house votes in conjunction with adopting a Senate-style ballot paper for Legislative Council elections. Lower house candidates can lodge a registered ticket of preferences to save ballot papers with insufficient preferences, mainly those with only a first preference. Around 4% of ballot papers are saved from informality by these ticket votes.

The savings provision still applies for South Australian state elections. It results in there being a substantially lower rate of informal voting at state elections compared to Federal elections in South Australia.

Despite clear evidence that a savings provision can cut the informal voting rate, other jurisdictions have not followed the South Australian example.

Informal Voting and Candidate Numbers

When full preferential voting was introduced in 1919, it had only recently been adopted for state elections in in Victoria and Western Australia. NSW and South Australia did not move to full preferential voting until 1930, and Queensland did not adopt full preferential voting until 1963. Tasmania has never used full preferential voting for state elections.

The conflict between state and federal voting laws explains the high incidence of informal voting in the 1920s, as shown in the above graph. Some of the rise in informal voting in recent years has arisen with conflict between CPV and the use of OPV for state elections in NSW and Queensland.

After 1930, informal voting settled into a consistent rate of 2-3% for five decades until a rise that began in the 1980s. Over those decades, one of the recorded features of informal voting was its tendency to be higher in contests with only two-candidates. Such contests became much rarer after 1980.

But as the number of candidates has increased over time, so has the proportion of informal votes. The same relationship can be measured in comparing electorate results at single elections. Some of the increase since 1990 also reflects the ever larger size of Senate ballot papers.

The average number of candidates is not always a real measure of the task faced by voters. The stacked bar chart below shows a different measure of the task voters face by plotting the proportion of contests with a given number of candidates.

From 1949 to 1966, over half of all districts had between one and three candidates. Between a quarter and a third of all electorates had only two of three candidates as late as 1987, but since then, two and three candidate contests have disappeared. Four to six candidates became the most common number until 1996.

At six of the eight elections since 1998, half of all electorates have had seven or more candidates.

So for most of Australian political history, full preferential applied at elections where voters only had to provide three or four preferences. Now voters regularly face twice that number, required to give preferences to candidates they either don’t know or may find equally objectionable.

In its report on informal voting at the 1987 election, the AEC noted that roughly 85% of all informal ballot papers with an identifiable first preferences could have been admitted to the count with different formality rules. All had first preferences for a candidate who would not be excluded from the count so preferences on those votes did not need to be counted.

To protect full preferential voting, 85% of ballot papers with defective number that could count were rejected to protect full preferential voting from the much smaller number that had incomplete preferences.

Given the decline in support for major parties since 1987, that 85% figure may have slipped to about 70%, the figure where major party voting sits these days. We don’t know the actual figure because the AEC has not conducted research on identifiable first preferences since 1987.

The above finding backs a point I’ve always made about how compulsory preferential voting is implemented in Australia. We have too many formality rules defining what can’t count, and not enough defining what can count.

As the savings provision in South Australia amply demonstrates, there are ways to cut high rates of informal votes simply by coming up with a method that allows more incomplete ballot papers to count.

At the JSCEM inquiry into the 2010 election, I outlined the South Australian provision in detail, and also suggested adopting what I called “progressive” informality, where a ballot paper remained formal as long as its missing preferences did not have to be counted. The committee seemed more interested in the South Australian method but nothing ever came of the inquiry.

As I suspect will be the result of the current JSCEM recommendation on OPV. The report lacks any coherent argument in favour of OPV, which makes it look like a wish list from an LNP Senator. And the LNP is strongly in favour of OPV from its experience in Queensland.

Not that lacking an argument in favour of OPV doesn’t mean Senator McGrath’s recommendation will not be adopted. After all, as I explained in a post from 2016, the Queensland Labor Party offered few justifications and minimal warning when reverting the state’s electoral laws back to full preferential voting in in 2016.

So Who Does Optional Preferential Voting Assist?

The non-partisan answer is that optional preferential voting always works in favour of the candidate in a contest that leads on first preferences. That’s because every ballot paper that exhausts its preferences is excluded from the tally of votes remaining in the count, lowering the finishing line for the candidate that starts the count in the lead.

If you remember this point, that OPV helps the candidate and party with the highest first preference vote, you understand why Australia’s two major parties have swapped their opinions on OPV over time.

It is not a swap based on principle, or supporting something that voters want, but rather on the basis of self-interest at the time.

In the days when DLP preferences helped the Coalition win seats, Labor supported OPV and the Coalition opposed it. One of the Labor Party’s icons, Gough Whitlam, was a supporter of OPV and tried to implement it while Prime Minister.

Another Labor hero, Neville Wran, implemented OPV in NSW in 1980. He also entrenched it in the state’s constitution, and many a Labor figure has since cursed OPV’s entrenchment as the rise of the Green’s has removed Labor’s former advantage over the Coalition on first preferences.

OPV disappeared from Labor’s policy platform in 1991, a year after compulsory preferential voting played a significant part in the Hawke government winning a narrow re-election.

Since 1990, it is overwhelmingly Labor and minor parties that have benefited from CPV, and the Coalition that has found itself on the losing side.

The same process followed the introduction of OPV in Queensland. When the Electoral and Administrative Review Commission (EARC) reviewed the state’s electoral system in the early 1990s, the Labor, Liberal and National Parties all lodged submissions opposing OPV. When EARC still recommended it, the Goss Labor government implemented OPV and benefited from it in three-cornered contests at the 1992 election. The Coalition virtually abandoned three-cornered contests after 1992.

After several elections of still recommending full preferences on how-to-votes, the Labor Party under Peter Beattie benefited greatly in 2001 from a switch to a “Just Vote 1” strategy, cutting off the flow of One Nation preferences to the Coalition.

By 2009, as One Nation declined and as the two Coalition parties merged to form the LNP, preference strategies changed. The LNP now urged “Just vote 1” and Labor, with the rise of the Greens, went back to recommending preferences.

So from the view of party politics, there are arguments for and against both optional and full preferential voting. Over the last five decades, both Labor and the Coalition have at different times supported both systems. And why they have done so has always been related to perceived electoral advantage at that time.

Are there Arguments for Optional Preferential Voting?

The best argument was the one put by the EARC inquiry into Queensland’s electoral system.

Nevertheless the Commission is concerned that electors are currently required to record views they may not have, by ranking in order of preferences all candidates offering in their electoral district. If they do not have a complete set of preferences they have either to invent preferences, or arbitrarily assign rankings to candidates about whom they know nothing or care less or accept that their ballot paper will be excluded from the scrutiny. The Commission believes that it is not unreasonable or oppressive to require every adult citizen to play a meaningful part in the choice of their government, and has set out its views on compulsory voting in Chapter Five. But having required that duty be discharged, it is inappropriate for the electoral system to corrall votes on behalf of candidates or parties who electors do not wish to support but merely consider less objectionable than the other on the ballot paper.

Electoral and Administrative Review Commission, Report on Queensland Legislative Assembly Electoral System, Volume 1 – The Report, p 59, para 6.24

It has always struck me that those who argue most strenuously in favour of compulsory preferential voting are objecting that people who vote for minor parties might have no interest in preferencing either major party. Such advocates ignore the problem that most votes declared informal for numbering errors are votes cast for the major parties.

Already I am hearing the argument put that voters can be “disenfranchised” by optional preferential voting, an argument that was also raised against Senate electoral system reform in 2016.

Saying voters are disenfranchised by optional preferential voting is rubbish. Voters are disenfranchised when their vote counts for no candidates, not when their vote only counts for some candidates.

If a voter chooses not to make a choice between certain candidates, and in the end their voter doesn’t get distributed to one of those candidates, the voter is not disenfranchised. Their vote still counts for the candidates they cared about and preferenced.

The folly of CPV without proper savings provisions was highlighted by the 2009 Bradfield by-election. There were 22 candidates, including nine Christian Democrats, and the Liberal candidate polled 56% of the first preference vote. Despite there being no need for any preferences to be distributed, and especially no need for any Liberal preferences to be distributed, all 73,000 ballot papers were checked for a valid sequence of 1 to 22. If you ever wonder why it takes so long to finalise election counts in Australia, this formality checking process is a significant factor.

In an era when we are seeing more and more candidates on ballot papers, more and more voters are having to invent preferences for candidates they object to or don’t know, just to have their real preferences count.

As I have pointed out elsewhere, the overwhelming reason parties distribute how-to-vote material is about first preferences and formality. And the more candidates there are on a ballot paper, the more that how-to-votes become about formality.

The underlying assumption about vote counting under full preferential voting is that voters give equal weight to each preference on their ballot paper. Voters are assumed to care as much about their choice between a seventh and eighth preference as they do between their first and second. Every ballot paper is reduced to a choice between the two final candidates, and the constructed majority produced assumes the preference choice between the two was equal with every vote.

Optional preferential voting weights preferences by allowing voters to stop giving preferences once they no longer care about further choices. Compared to compulsory preferential voting, optional preferences effectively weight the count to the first nd higher choice preferences of a voter.

As the number of candidates increases, so does the number of voters inventing, transcribing or randomly allocating preferences. And that undermines the assumption that all preference choices made by a voter are equal.

That in the end is the advantage of optional preferential voting – it allows voters to concentrate on the choices they want to make rather than the choices they have to make.

And if you want to try and educate voters on how preferential voting works, isn’t it easier to explain by talking about who voters want to elect and how to number their ballot paper to express those views.

3 thoughts on “Candidates, Informal Voting and Optional Preferential Voting”

  1. My concern in respect of Optional Preferential Voting (OPV) is that the ability of ballots to exhaust would undermine the majoritarian principle of our electoral system.

    Although many candidates are elected under Full Preferential Voting (FPV) without receiving a majority of first preferences, the winner of each FPV contest will always have received a majority of preferences. That is, more than half of formal ballots would have expressed a preference for the winning candidate over all other candidates.

    So where, say, 100 000 formal votes are cast, the winner will always get to at least 50 001 votes under FPV, even if they did not receive 50 001 first preferences.

    Likewise, it is not possible for a candidate to be elected under FPV if a majority of voters prefer another candidate (notwithstanding that the former candidate may have received the most first preferences).

    However, under OPV, where voters choose to exhaust ballots – i.e. to leave some boxes blank – the number of ballots remaining in the last count may be less than the number for the first count. I.e. where 100 000 formal ballots are cast, and 5 000 ballots for excluded candidates do no express a preference amongst the final two candidates, only 95 000 ballots will be counted in determining the winner.

    This may lead to the situation where a winning candidate is preferred against all other candidates by only, say, 48 000 of the 100 000 formal voters. This situation can be compared with the blatantly undemocratic ‘first-past-the-post’ system, although the latter is obviously far worse.

    While there is an argument to be made that voters should not have to direct their preferences to candidates they do not wish to support (although giving a candidate, say, a seventh or eighth preference hardly amounts to supporting them), this must be balanced with the desirability of the majoritarian principle prevailing, that is, the winning candidate being preferred by a majority of formal voters.

    The fact that it appears that several seats would have been decided differently at the last election under OPV – i.e. several candidates who were preferred as against all other candidates by a majority of voters would not have won their seats – underscores for me the possibility of OPV occasionally leading to outcomes at odds with the will of the electorate.

    I also understand that any system which reduces informal voting cannot be all bad. I’m interested Antony in your thoughts as to whether OPV offends the majoritarian principle.

  2. Has an analysis ever been done for informal votes on who the first preference candidate was? Given your second bullet point at the top of your article, and given the “anti-establishment“ position of parties like One Nation, Palmer United, etc. I’m not convinced that Labor would necessarily be the loser under OPV.

    COMMENT: The informal vote research carried out in 1984 and 1987 assessed the first preference party of votes declared informal for incomplete numbering. It had a higher proportion of Labor votes but wouldn’t have changed results. There has been no analysis of preferred party on informal votes since 1987.

  3. Tom the first and best

    Partially optional/compulsory preferential voting could be a compromise option. It would deal with most long ballot paper informality issues, if the number of required preferences was 6 it would deal with Senate formality differences (excluding savings provisions) and still result is a low exhaustion rate.

    Increasing nominee requirements and/or minimum registered party size would be another option to reduce oversized ballot papers.

    Compulsory preferencing prevents voters unintentionally exhausting because they don`t think their candidate`s preferences will be distributed. Prahran 2014 could have been an example of this, if preferencing was optional.

Leave a Reply