How to Manage the Eden-Monaro By-election in a time of Covid-19

(Update 25 May – the by-election date has been set for 4 July. You can check my guide to the Eden-Monaro by-election over at the ABC Elections website.)

With the decision of Labor’s Mike Kelly to resign from Parliament, there will need to be a by-election held in his marginal seat of Eden-Monaro.

With a margin of just 0.9%, and with NSW Deputy Premier and well-known local state MP John Barilaro tipped to contest the by-election, Labor will have a fight on its hands to retain the seat. This despite the fact that no government has taken a seat from the opposition in a century, not since the special circumstances of the Kalgoorlie by-election in 1920.

With Covid-19 restrictions still in place, it may also be an unusual by-election. The Speaker has asked the Australian Electoral Commission (AEC) for advice on special procedures that may be needed to protect voters and staff.  The by-election may be delayed until after the toughest of the social distancing regulations have been eased.

A useful guide for Eden-Monaro will be the Queensland local government elections, held on 28 March as Covid-19 restrictions mounted. There were significant changes to the conduct of polling by both Electoral Commission Queensland (ECQ) staff and by party campaign workers. It changed the way electors voted, how scrutineers observed the count and how results were reported.

At the time there were calls for the elections to be deferred as a public health risk. The elections went ahead and there has been no spike of Covid-19 cases in the aftermath.

But the expected surge in pre-poll voting has the potential to delay the release of Eden-Monaro election results. Is Eden-Monaro an opportunity to test procedures for counting pre-poll votes under secure conditions before 6pm on polling day?

When can the By-election be held?

There is plenty of scope to defer the by-election if necessary until after the toughest of the social distancing rules have been lifted.

When a vacancy occurs in the House of Representatives, the Constitution states that the Speaker shall issue a writ to hold a by-election. If the vacancy occurs close to when a general election will be held, the Speaker may choose not to issue a writ.

As the Speaker has already done, (see below), he consults on appropriate dates with the Australian Electoral Commission. He may also discuss arrangement with the government and the opposition.


There is nothing in the Constitution or the Commonwealth Electoral Act that specifies a minimum period within which the writ must be issued. Once issued, the writ must conform with the election timetable set out in the Commonwealth Electoral Act.

According to the Parliamentary Library, from 1949 to 2017, the average time between a vacancies occurring and a by-election being held was just over seven weeks. The variability in length is largely determined by the timing of writ issue.  The Christmas-New Year period can shorten or extend the period between vacancy and election day. Writs were issued at once for pre-Christmas by-elections in North Sydney (2015), New England (2017) and Bennelong (2017). The Griffith by-election (February 2014) was delayed by there being no time to hold the by-election before Christmas.

The five July 2018 by-elections (Braddon, Fremantle, Longman, Mayo, Perth) followed an extended delay of 11 weeks. The vacancies occurred around 10 May, the Speaker announced on 24 May that the by-elections would on 28 July, but writs were not issued until 15 June for a six week formal campaign. This was one of the longest periods between vacancies and by-elections.

After the issue of a writ, the by-election timetable is defined in legislation. The writ will specify –

  • The date for close of rolls, which must be between 7 and 17 days after writ issue.
  • The date for close of nominations, which must be between 10 and 27 days after writ issue. The declaration of nominations and draw for ballot position takes place the next day.
  • Early voting begins five days after the declaration of nominations.
  • Polling day must be a Saturday between 23 and 31 days after the close of nominations.
  • The minimum time from issue of writ to polling days is 33 days, the maximum 58 days.

How Social Distancing Rules Changed Voting at the Queensland Local Government Elections.

The timetable for the March 28 Queensland local government elections was set in legislation. Council terms ended, executive orders were issued and arrangements for polling were announced in mid-February, well before concerns over Covid-19 emerged in March.

The elections went ahead but with a number of late changes to procedures, issued by regulations under electoral and public health legislation. Measures included –

  • Health concerns for elderly residents meant that Electoral Commission Queensland (ECQ) mobile polling teams were prevented from entering nursing homes and  hospitals.
  • The ECQ encouraged the use of postal voting and pre-poll voting.
  • The ECQ expanded access to telephone voting. Previously it had only been available to blind and low vision voters, but was expanded to include people who were in isolation or had health concerns about attending a polling place. The number of telephone votes rise from 151 in 2016 to 8,428 in 2020.
  • The distribution of how-to-vote material outside polling places was banned.
  • Hand sanitiser was made available, queue control kept voters apart, voters were encouraged to bring their own pens or pencils. There was more on the day cleaning in polling places.
  • Some polling places were abandoned and voting concentrated in venues with more space for voters to spread out.
  • For social distancing reasons, scrutineers were not permitted to observe the count after the close of voting. Scrutineers were permitted to observe the check counts after polling day, but the number of counting tables in rooms was reduced.
  • Social distancing interfered with all face-to-face aspects of campaigning including door-knocking, public canvassing and campaign rallies.

This suggests several questions for the Eden-Monaro by-election

  • Will NSW public health regulations permit AEC staff to visit nursing homes and hospitals?
  • Should how-to-votes be banned as a public health measure?
  • Will scrutineers be allowed in polling places on election night.
  • Will the AEC encourage greater use of postal and pre-poll voting?

The current provisions for postal voting should allow any voter with concerns about attending a polling place the right to apply for a postal vote.

How Did Covid-19 Change Voting Patterns in Queensland?

Data from the Brisbane City Council election held as part of the state-wide local government elections, and from the Currumbin and Bundamba state by-elections held on the same day, are a guide as to what might happen at the Eden-Monaro by-election.

Fear of Covid-19 saw a massive decline in the number of people who attended polling places on election day.

  • In Brisbane City Council, the proportion of ordinary polling day votes fell from 66.0% in 2016 to 26.5%. Postal voting rose from 12.2% to 23.9%, and pre-poll voting rose from 13.2% to 28.7%. There was also a big jump in absent voting from 7.4% to 18.3%, this increase largely due to pre-poll absent votes. Overall turnout dropped from 84% in 2016 to 79% in 2020.
  • At the Bundamba by-election, polling day votes fell from 68.6% of all votes at the 2017 state election to 38.8% at the by-election, postal votes fell from 8.7% to 5.1%, and pre-polls rose from 16.1% to 54.7%.
  • At the Currumbin by-election, polling day votes fell from 61.3% of votes in 2017 to 28.8%, while pre-poll votes rose from 24.9% to 58.7% and postal votes from 9.6% to 11.6%.

In Eden-Monaro at the 2019 election, just over 50% of votes were cast on polling day, 41% were pre-poll votes and 5.8% were postal votes. The Queensland results suggest the proportion of both pre-poll and postal votes will increase at the Eden-Monaro by-election.

Can the By-Election be Conducted Entirely by Post?

While several states use full postal voting for local government elections, it has never been used for state or federal elections.

There are two ways to access a postal votes, by being registered under special conditions as a permanent postal voter, or by applying for a postal vote at a single election.

Permanent postal voters are registered to automatically be sent postal votes after the close of nominations. All other voters who want a postal vote must complete an application before a postal vote pack is dispatched. Voters can apply for a postal vote once the date of the election is known.

Before a postal vote can be admitted to the count, the details on the outside of the postal vote envelope must be checked against details provided by the voter when becoming a permanent postal voter, or provided by the voter on their application form.

As the law currently stands, these procedures couldn’t be applied to a full postal vote election. The mass mail-out of postal vote packs to all voters would not have enough security to ensure postal vote packs aren’t being hijacked on the way to voters, someone other than the designated voter completing the vote and envelope.

You also wouldn’t believe the number of invalidating mistakes made with postal votes, such as families deciding to put all their ballot papers in the one return envelope.

In addition, Australia Post has recently cut the delivery time on letters. It would be highly likely that Australia Post could increase delivery times within Eden-Monaro, but it is a regional seat. Gambling on delivery time for all votes seems unnecessary when the Queensland local government elections suggest that in-person voting can still be conducted safely.

In addition, postal votes are the slowest category of votes to count. Going from 5.8% of votes as postal votes to 100% would certainly slow down counting and knowledge of the result.

Postal voting will likely have a greater role in the Eden-Monaro by-election, but should be conducted by registration or application as normal.

Can How-to-Votes be Banned?

The ban on how-to-votes at the Queensland election was done to comply with public health law. If a similar ban was introduced for the Eden-Monaro by-election, it would be done under NSW public health law, but may require a change to polling arrangements by the AEC.

The Queensland local government elections were conducted under optional preferential voting. The absence of how-to-vote material seemed to have had little impact on the rate of informal voting.

It was a different matter at the Bundamba and Currumbin by-elections. Both were conducted under state electoral law which requires full preferential voting. In Bundamba the rate of informal voting rose from 8.2% in 2017 (6 candidates) to 10.9% at the by-election (4 candidates). In Currumbin it rose from 4.7% in 2017 (4 candidates) to 8.1% at the by-election (4 candidates).

These are high rates of informal voting for a by-election involving only 4 candidates. The rates were even higher on election day when the how-to-vote bans applied, 11.5% in Currumbin and 16.1% in Bundamba.

That voters at both by-elections received three ballot papers, two local government using optional preferential, and one by-election for full preferential, is without doubt a cause of the increase in informal voting.

But the absence of how-to-votes would also have played a part. The Labor Party and LNP did not have identified candidates for the Gold Coast or Ipswich councils, but did for the by-elections. Had candidates been allowed to issue how-to-votes, most by-election voters would have received how-to-votes with full preferences to take with them into the polling place. This would have cut the rate of informal voting.

Without how-to-votes, political parties resorted to other tactics. Parties had big signs with sample how to votes that people could take pictures of with their smart phones. Signs also had qcodes allowing voters to download a how-to-vote.

At the two by-elections, the flow of preferences was slightly weaker than at an election where how-to-votes were distributed. In Currumbin, the actual Green preference flow to Labor was 72.8% compared to a 75.6% distribution at the last state election. That 75.6% included some fourth party preferences so would have been a higher Green preferences flow. One Nation preferences split 35.6% to Labor and 64.4% to the LNP, not much different to the last state and federal elections in Queensland.

In Bundamba, which finished as a Labor vs One Nation contest, Green preference split 66.7% to Labor and 33.3% to One Nation, a very high leakage of preferences to One Nation. The LNP how-to-vote recommended a preference for One Nation, but they split 50:50, well down on the 70% to One Nation where counted out at the last state election.

In Brisbane City Council under optional preferential, Green preferences flowed to Labor at about the same rate as in 2016 despite the lack of how-to-votes. Countering this, the Greens recommended preferences in more wards in 2020. There was a weaker flow than in 2016 of Labor preferences to the Greens where Labor candidates were excluded, perhaps an indication of the lack of how-to-votes. There was also a higher rate of exhausted preferences in the 9-candidate Lord Mayoral race than in the wards where there were generally only three candidates.

What does this mean for Eden-Monaro? As a Federal by-election, full preferential voting applies. Past by-elections have shown that informal voting is reduced without the complication of a Senate ballot paper. But the by-election could attract a dozen or more candidates, and if how-to-votes are banned, the rate of informal voting would be certain to rise as voters make mistakes trying to invent lower order preferences between candidates they’ve never heard of or find equally unattractive.

A confusion people make is to think how-to-votes are all about preferences. They aren’t. The major parties distribute how-to-votes for two fundamental reasons –

  1. to attract first preference votes from late deciders, and
  2. having attracted a first preference, to make sure the voter completes a formal vote

Preferences will be important at the Eden-Monaro by-election. The Labor Party will want as many Green preferences at it can get. The Liberal and National candidates will want to maximise intra-Coalition flows. Green and National preferences both flowed at 87% in 2019, a rate that would be hard to match without how-to-votes.

Can how-to-votes be distributed without hand-to-hand contact? In recent years parties have become keen on re-cycling how-to-votes, but can that be allowed in the current climate?

The Electoral Act does not allow the display of campaign material inside a polling place, so AEC staff can’t have a supply of how-to-votes. Even then, how could a display of how-to-votes be of use if a voter can’t take them to the voting partition. South Australia displays registered how-to-votes on voting partitions, but it seems too late (and to me would be undesirable) to set up a similar system under Commonwealth law.

Love them or hate them, how-to-votes might stay for the by-election. But party workers would have to obey social distancing laws. Workers would have to stay further apart, allow voters to approach them rather than the reverse, and making voters “run the gauntlet” to enter the polling place has to be stopped. (In my opinion, having to run the gauntlet should be stopped Covid-19 or not.)

Should Scrutineers be Banned from the Election Night Count?

This was by the far the most contentious measure adopted by the ECQ. Scrutineers could enter the polling place to see the ballot boxes opened at 6pm. They then had to leave, but could enter once the count was completed to observe the sealing of material for dispatch to the Returning Officer. Scrutineers could have been given the count at this point, but it appears scrutineers either left or they weren’t told the tallies.

On 28 March, indicative preference counts were only done for the Bundamba and Currumbin by-elections. Scrutineers could not observe these on election night, and with various ECQ computer problems delaying the result, knowing what was going on was impossible.

In Currumbin, it was impossible to know the result with only 28.8% of votes cast in polling places. It required the giant Tugun pre-poll centre to be counted. It took more than 11,000 votes, more than a third of all votes in the electorate. This was counted for first preferences on the night, but staff ran out of time to conduct an indicative preference count. As scrutineers weren’t present, it wasn’t even possible to have an approximation of the preference count.

For the Eden-Monaro by-election, there is at least the advantage that the AEC has long established procedures for doing indicative preference counts. Unlike the ECQ, the AEC also has a established and fully tested virtual tallyroom to publishes results. The AEC does not publish polling place results in the virtual tallyroom on election night, but it does make its detailed datafeed file, including polling place results, readily available for download.

If scrutineers are banned from polling places, they might still be permitted to watch the pre-poll count under more controlled circumstances.

Which raises another thorny problem. If there is going to be a surge in pre-poll voting, how can the AEC count the increased number of pre-poll votes in a timely manner on election night.

Anyone who remembers the Wentworth by-election will know this is a problem. In Wentworth, all polling places had reported their results by 9pm and it looked like a clear win for Independent Kerryn Phelps.

Then at around 11pm, well after everyone had stopped following the results, the four huge pre-poll centre counts were reported, and they significantly narrowed the result.

If something similar happened at the Eden-Monaro by-election, the minority of votes cast on the day may be counted early, but we may be waiting hours to know the result because of the delays in counting pre-polls.

Should Pre-poll Vote Counting Start before 6pm in Eden-Monaro as a Trial of the Procedure?

Two weeks ago I wrote a post (you can find it here) where I set out the case for allowing pre-poll votes to be counted in secured counting centres on the afternoon of polling day.

In 2019, 41% of votes in Eden-Monaro were pre-polls compared to just over 50% cast at polling places. Based on the Queensland case, on the day votes could fall to 30% and pre-poll votes reach 60%.

In 2019 there were only two polling places in Eden-Monaro that took more than 2,000 votes, only one taking more than 3,000. There were seven pre-poll centres that took more than 3,000 votes including just under 12,000 taken at the Queanbeyan pre-poll voting centre.

It is possible that all polling places might be counted by 8:00 or 8:30 on election night, and the result will then depend on how long it takes to count pre-poll centres that have taken more than 5,000 or 10,000 votes. Note that pre-poll votes are usually counted centrally with several parallel counts undertaken.

Should we use the Eden-Monaro by-election to test the viability of allowing pre-poll votes to be counted on the afternoon of election day? As I point out in my post on the subject, New Zealand counts all its pre-poll votes on election day under secure conditions, with the results released after the close of polls.

Counting pre-polls before 6pm would require a change to the Electoral Act. The act currently prevents any ballot box from being opened before 6pm.

I run through all the pros and cons of early counting in my blog post, so I won’t go through them again here. I’ll just make it clear the early counts must be done under secure conditions with communication devices impounded, and results can’t be released until after 6pm.

Knowing the result of an election in a timely manner is a key feature of a open and accountable electoral process. If for public health reasons there are restrictions placed on scrutineer access to counting on election day, and this result in delays in the release of results, then there are reasons to adopt other procedures to aid transparency.

The difficulty of counting pre-poll votes on election night has already been raised as a major problem by the AEC. That problem can only get worse at the Eden-Monaro by-election. Should we amend the Electoral Act for the Eden-Monaro by-election as a trial measure allowing the early counting of pre-poll votes to ensure there are no unnecessary delays in the result becoming known?

By-elections have regularly been used by Electoral Commissions to test procedures for later use at general elections.

Parliament will sit before the by-election, so there is time for the Electoral Act to be changed to allow counting of pre-poll votes before 6pm, if only on a trial basis.

With all the other issues likely to interfere with the conduct of the by-election, it seems worthwhile trying to reduce confusion by preventing unnecessary delays in the result becoming known as soon as possible after the close of polls.


8 thoughts on “How to Manage the Eden-Monaro By-election in a time of Covid-19”

  1. Banning how-to-vote cards should be permanent. Voters should be capable of constructing their own how-to-vote cards. Worst case they could copy them from media advertising.

    COMMENT: Tasmania and the ACT do, but neither jurisdiction uses full preferential voting. The NT’s restriction of them also began under optional preferential voting. Under full preferential voting every ballot paper must have a valid sequence of preferences to be formal. Yet 75% of ballot papers never have their preferences counted, requiring 75% of voters to complete a meaningless set of preferences just to allow their first preference to count. It is why the major parties distribute so many how-to-votes, to capture first preferences and ensure those votes are formal.

    1. “Yet 75% of ballot papers never have their preferences counted, requiring 75% of voters to complete a meaningless set of preferences just to allow their first preference to count. It is why the major parties distribute so many how-to-votes, to capture first preferences and ensure those votes are formal.”

      Do 75% of ballot papers not have their preference counted because that’s the top 2 in a preference distribution? If that’s a problem we should be counting the preferences of the winner where the preference is never distributed under any system? Refer to Jordan Ellenberg ‘How not to be Wrong’.

      COMMENT: It’s not a problem in the count. It’s a problem that people cast a first preference for a candidate but have their votes made invalid because of errors in preferences for other candidates when those preference don’t have to be counted.

    2. To clarify the previous post and reference to Jordan Ellenberg. It is the loser on 2CP preferences who never has their preferences distributed? This could have changed two divisions in the last Queensland election with the winner going to the more extreme candidate on each side of divide.

      COMMENT: In single member contests, candidate are excluded in ascending order of first preference vote. At the end of the count, there are two candidates left in the count, one is the winner, one the loser. No ballot paper cast as a first preference for either candidate ever has their further preferences counted out, but any ballot paper with a first preference for either candidate and an error in further preferences is informal. I see no reason to exclude ballot papers for errors in preferences that never need to be examined for counting.

    3. The two are the Greens win in Brisbane and the ON win upper Queensland. Neither would have won under a different concept of preference.

      COMMENT: Candidates are excluded in ascending order of vote. Preferential voting as used in Australia rewards candidates with the most first preferences. Almost all electoral systems used around the world reward candidates with the most votes. If you’d like to advocate something like Condorcet voting you can, but no one attempts such methods of voting except in small committees.

  2. A minor quibble. The three different ballot papers voters at the Bundamba by-election received had three different voting systems. The ballot for Mayor used optional preferential, the one for the state by-election was full preferential, and the one to elect local area Councillors used a multi member first past the post system to elect two people to the local Division – so people had to mark two boxes.

    COMMENT: In my background pages on the by-elections I went into detail about that. I thought it was a little too specific to single out as part of a broader topic.

  3. Scrutineers are a vital element in ensuring both integrity and confidence in the electoral system. Polling clerks, particularly at smaller polling places, often benefit from having experienced scrutineers present. While indicative two-candidate preferred counts are conducted at each polling place, negating some of the art of preference estimating, there are occasions when casual polling clerks make mistakes that are challenged by scrutineers and addressed by the officer-in-charge. Scrutineers also buttress confidence in the system, particularly for partisan supporters who sometimes erroneously believe that polling clerks are somehow finding a way to manipulate election results. Scrutineers are the simplest antidote to conspiracy theories about our electoral system. Managed properly, perhaps with some or face masks, COVID-19 should not be used as an excuse to exclude scrutineers.

    COMMENT: I agree with you, but would point out that the election night count is only preliminary. The official count is the check count conducted after ballot papers are returned to the Returning Officer, and that is always done by more experienced staff with more experienced scrutineers watching. At the Wentworth by-election, there were three polling places where the two-candidate preferred count had been mucked up, and that was fixed the next day in the check count.

  4. Antony, with Andrew Constance now running for Eden-Monaro, are there precedents for state and federal by-elections being held on the same day, in geographically overlapping areas? It makes sense for cost reasons as well as COVID reasons to do this in Bega/Eden-Monaro. But might be confusing for voters and perhaps not easy for poll workers – eg some polling places would have to hand out a state and a federal ballot paper; separate teams needed for counting if results not to be delayed etc; two sets of How to Vote cards etc.

    COMMENT: Stephen, even if the state government wanted to hold the Bega by-election the same day, something I doubt they want to do, there are a series of logistical problems.

    There can’t be a writ for a state by-election until there is a vacancy. Constance does not have to resign until the close of nominations, which is three weeks before polling day, and three weeks is short for a state by-election campaign. Constance would have to resign earlier than required to have the by-elections on the same day. Back in 1975, Wagga Wagga had a by-election one week before the Federal election, Wal Fife having resigned three weeks before the dismissal brought on the Federal election.

    Then there are the logistical problems. I’m not aware of precedents of overlapping state and federal by-elections being held on the same day. Queensland has held state by-elections on the same day as Council elections and it highlights the problems that would occur in Eden-Monaro. There had to be separate rolls and ballot paper issue as there is no software to allow double mark-off. As would be the case in Bega, optional preferential voting at council elections clashed with compulsory preferential at Qld state by-elections, creating a surge in informal voting.

    I doubt that the by-elections will be held the same day. What usually happens is that the Eden-Monaro by-election will be completed before writs are issued for Bega. That was the case with Lyne/Port Macquarie in 2008. The government will give Constance a just-in-case opportunity to re-run in Bega if he loses Eden-Monaro, as the NSW Labor government gave to Rob Oakeshott in 2008. This did happen with the Country Party’s Ron Brewer in 1974, resigning as MP for Goulburn, defeated contesting Eden-Monaro at the 1974 Federal election, and then winning back his old seat in July 1974.

Leave a Reply