Saturday, May 18, 2019

Election night arrangements and election watching tips (2019)

Donations welcome!

If you find my coverage useful please consider donating to support the large amount of time I spend working on this site.  Donations can be made by the Paypal button in the sidebar or email me via the address in my profile for my account details.  Please only donate if you are sure you can afford to do so.


My coverage tonight and to come

I will be doing live blogging for the Mercury from 6 pm.  The link to the live coverage is here.  The link to the tweet that links to the live coverage is here.  I am not sure yet how long the coverage will go or whether I will need to take any breaks to write articles.  There is an email address on the live coverage for people to ask me questions during the coverage.

I will probably not be checking emails, tweets or for comments on this site regularly during this time, and I ask journalists not from the Mercury not to call me until the live blog has finished.  However if you have interesting scrutineering samples from Tasmania - especially of the rate of below-the-line voting for Lisa Singh in a Senate booth (please say which booth!) - you're extremely welcome to SMS them to me (0421428775) or email them to me (  I probably won't be able to reply immediately.

The live blog is paywalled and I don't know if there will be a paywall free link or not. Subscription options are available starting from $12.  (In a kind of reverse Antony Green situation, a public media source did try to poach me for this election but negotiations collapsed from the point where I mentioned the concept of being paid.)

Once the live blog has finished I will go home and late night coverage will continue on this site for as long as I can manage it.  Postcount threads will be unrolled tomorrow through the day.  At the moment I intend the order of business to be Tasmania Senate, followed by non-classic/interesting/messy Reps contests, followed by one thread for classic 2PP Reps contests, and I do not expect to look at the Senate in other states until at least Monday.

When will we know?

We should get exit polls right after 6 pm, which are still a bit of a vague science in this country.  The Galaxy exit poll in the recent NSW state election was wrong by 2% 2PP.  Votes will build up from maybe 6:30 and if the result isn't close we could know who has won the Reps within a couple of hours.  This is especially likely if Labor does better on two-party preferred than the very herded polling we have seen, which is distinctly possible.  However, it's possible based on current polling that the result could still be in doubt at the end of the night, or at least should Labor win there may still be doubt about whether they have a majority.

This year's count will be slowed by the massive rate of pre-poll voting.  Many seats will not get to 50% counted before the within-electorate prepoll booths are counted, many of them late tonight.  In the Victorian state election, swings differed markedly between on-the-day and prepoll voting so there may be a late night shift in whatever swing is going on.  The experience of that election suggests we should be really careful about calling anything projecting to less than about 54-46 off less than 40% of the vote, and seats with high Jewish votes such as Macnamara and Wentworth may require more care still.

Misleading Signs

A big theme in discussion today has been the use of signs that mimic AEC sign colours, especially Chinese-language signs in Chisholm and some other Victorian seats.  Unfortunately electoral laws are insufficiently clear to spell out beyond doubt that these signs should be illegal, and it is therefore difficult for the AEC to respond to them.  It is illegal to mislead an elector in relation to the mechanics of voting, but it isn't illegal under electoral law to put out signs that look like official information but are not.  (Whether it is illegal under laws regarding impersonating a Commonwealth body is outside my expertise).  There needs to be attention to this following the election as this has been a problem in many recent elections.

Seats with the "wrong candidates"

The AEC has procedures for deciding which two candidates to conduct the on-the-night two-candidate-preferred count between, usually based on the result of the last election in that seat.  (A challenge against the practice of revealing these choices from 6 pm by Clive Palmer failed for reasons not yet released.)  However when there is a change in the final two, the AEC will be counting a two-candidate result that is irrelevant to the final result.

It's hard to say yet how many seats will be affected this year.  Three-cornered contests (Liberal vs Labor+Green, eg Macnamara and Higgins, or Labor vs Liberal/National, eg Gilmore and because of the Liberal disendorsement, Lyons) are obvious candidates.  Mallee is another possibility.

When the wrong 2CP candidates are selected, any 2CP figure for the right candidates that the ABC put up will be their estimate only - not real numbers.  The ABC has improved its practices for being clear about this.  This may not prevent the ABC on both TV and website from calling some of these seats for candidates, but in the past such calls have often been premature and in some cases wrong.

When the AEC selects the wrong 2CP candidates, a realignment is conducted in which the votes are distributed to the right candidates.  This is usually done alphabetically by booth.  Because different booths have very different voting patterns, the 2CP result swings around wildly during this process, often causing the seat to be projected wrongly by the ABC computer and the media.  To predict where such seats end up it is necessary to use regressions off the primary votes, and I hope to post these here where necessary.

ABC calling seats prematurely

The ABC computer system will often call classic-2PP seats as won once its projections off a certain percentage of the vote have a party ahead by a certain margin.  Sometimes these calls will be premature.  In cases in the past candidates have turned around 49:51 or worse margins on the night when postal votes are added, plus at this election there are an even greater number of within-electorate prepolls (PPVC votes) that will be counted late on the night but may display differences in swing pattern with booth voting.   In Victoria we saw some seats the ABC was projecting into the 53-54 range for Labor not actually fall.

The Coalition tends to improve its position once non-ordinary votes (out-of-electorate prepolls, postal votes, out-of-electorate votes on the day and provisional votes) are added.  In 2013 on average this was worth about 0.25% two-party preferred, and I haven't checked the 2016 figures but think they were similar.  Flynn is one seat where the swing-back in post counting tends to be huge, often around 2 points. The ABC will use past patterns from each specific seat to project changes from the on-the-night totals.  It may be that for some reason the pattern changes slightly.

The 2PP, and preference shift

One of the big issues in modelling this election has been the possibility that the Coalition will do much better on One Nation and United Australia Party preferences than in the past.  With these parties apparently set for a combined 8-9% of the vote, this could affect the final 2PP by a point or so, and will be a major factor in Queensland seats especially.

As the night goes on and two-party preferred counts start to appear, we may well get a feel for whether a substantial preference shift has happened and how big it is.  We will not know the exact 2PP vote on the night, and at the last election it took many weeks to be finalised.  We will probably know it to about half a point by the end of the night.


A reminder that political-party commentators who appear on TV panels will often give biased readings of how results are going and act as cheerleaders for their parties in any seat where there is any chance (or even some seats where there's none).  This isn't always the case and sometimes a party insider will try to be objective.


Expect the Senate count to be very long, very slow, confusing and in the case of close contests challenging to project.  Thanks to the last election we have more of an idea of how party preferences flow (basically, all over the place) but it will take a while before below-the-line votes for Lisa Singh and Jim Molan are unrolled and we have a clear idea where they stand

Do not attempt to use Senate how-to-vote cards to model Senate outcomes. Most voters do not follow them, especially for smaller parties.  Please note that if someone emails me a model that does this, I will probably not even reply.  

We will probably know a few things by the end of the night, perhaps including who is in the box seat for seat 5 in every state, but much will unravel slowly over the several coming weeks and some seats could be in doubt until the press of The Button.  That said, it may be for a half-Senate election that nearly all the seats are clear quickly.

We will only have official Senate counts dealing with party totals on the night, and there's some doubt how advanced these counts will be even at the end of counting tonight compared to the Reps counts.  A common misconception is that only above the line votes are counted on election night so that parties with high below-the-line rates will gain in coming days - this is completely wrong.  Both above and below the line votes will be counted but they will appear by party as an undivided total.

In Tasmania Lisa Singh was elected in 2016 by an organised below-the-line rebellion against her party's ticket order.  A reappearance of this could cause the Tasmanian race to become about candidate totals not party totals, and to be very complex to project.  The main scenario here is that it could be unclear whether or not Labor's #3 John Short will be able to overtake Lisa Singh's primary tally on party preferences before he is excluded.  If Singh is excluded first, her preferences will leak massively so if the race is close then that may affect Labor's competitiveness for their third seat. For this reason the Tasmanian race may be especially unclear even when most votes have been counted by party.  Races for the final spots in all states will also be sensitive to slight shifts in the vote totals in the post-count.

I hope you all enjoy the coverage!

1 comment:

  1. Thanks for all your work, you genuinely make voting less stressful and much more interesting! All the best for the big night tonight!


The comment system is unreliable. If you cannot submit comments you can email me a comment (via email link in profile) - email must be entitled: Comment for publication, followed by the name of the article you wish to comment on. Comments are accepted in full or not at all. Comments will be published under the name the email is sent from unless an alias is clearly requested and stated. If you submit a comment which is not accepted within a few days you can also email me and I will check if it has been received.