Monday, October 30, 2017

New England By-Election: Prospects And Polls

Seat: New England (Nat vs Ind 8.5%, Nat vs ALP 16.5%)
By-Election 2 Dec 2017
Incumbent: Barnaby Joyce (Nat)
Main opponents likely to include Rob Taber (IND) and ALP candidate
Former incumbent and 2016 opponent Tony Windsor (IND) not contesting
Outlook: High chance of comfortable Nat retain

With the date of the New England by-election set it's time for a general prospects and polls post that I will update through the campaign as opportunities arise.  I may be pretty slack about this as Queensland will be a higher priority and I have a lot of other stuff on at the moment.  I won't be doing live coverage on the night because of previously booked air travel, but there will be a postcount thread if it is needed.

Circumstances and history

The New England by-election follows the disqualification from office of incumbent Deputy Prime Minister and National Party leader Barnaby Joyce, who was found to be in breach of Section 44 of the Constitution because he was a dual citizen of New Zealand by descent, despite having been born in the electorate of New England.

New England has had a mostly conservative history.  It was last won by Labor (then spelled Labour, see comments) in 1910 and Labor has not been competitive in it since the 1980s.  Beyond that, in the last 104 years the rule of conservative parties was interrupted only by the 2001-2013 tenure of independent Tony Windsor, and also briefly by the 1922 expulsion of the sitting member from the Country Party.  Windsor himself was considered fairly conservative until he backed Labor to form government in 2010.  

In the 2013 election, Barnaby Joyce contested New England to shift from the Senate to the House of Representatives and Windsor did not stand. Windsor contested in 2016 and a close contest was generally expected but a mutually vicious campaign did not go well for him and in the end Joyce won convincingly.

The contest is of special interest because the government will lose its majority if it loses the seat.  (It has not, contra to what you may read in many papers and hear ad nauseum from Labor, lost it yet.)

Typical swings in government by-elections

I noted much of the history of by-election swings in my preview of the 2015 Canning contest. The following is a summary:

1. By-elections in government-held seats tend to produce two-party swings against the government.
2. These swings are somewhat larger on average if the government is polling poorly at the time.
3. There is little difference between by-elections caused by resignation and by-elections caused by death.
4. By-elections caused by disqualifications may be different because they tend to occur early in the term. There may be voter backlash against officialdom over the voiding of the result, and against being sent back to the polls.

In fact, we have very few examples of by-elections caused by voiding at federal level, and few recent ones even at state level.  But the widely quoted case of Jackie Kelly (Lindsay 1996) should be treated with some caution as a precedent.  This by-election, which saw a five-point swing to the incumbent, was held just seven months after the Howard government took office.  The Howard government was in honeymoon phase and polling at around 57-43 leads nationally around this time.  It had also been in office for such a short time that it had not done much to annoy voters.  Being sent back to the ballot box annoyed them though, and they responded by increasing Kelly's margin.

The New England case is unique in that the voiding comes sixteen months after the original election.  I cannot find any example of nearly such a long period between the original election and its voiding in Australian federal or state by-election history back to at least 1900.  (There were two state-level cases with a one-year gap in WA in 1906, and both of those incumbents lost.)  However even if this means it is not the same previous cases of voiding, there is one regard in which it is: the incumbent is a contestant, and therefore does not lose his personal vote.  Loss of personal vote is a major factor in the swings that normally occur in by-elections.

Can Anyone Possibly Beat Barnaby?

Things seem to be playing out in Joyce's favour so far with a weak line-up of opponents set to run against him.

There is not much history of non-classic (eg National vs Independent) federal by-election swings, compared to the history of typical 2PP swings that I looked at for Canning.  If we treat it as a National vs Labor contest then a 16.5% swing is required.  There has only been one larger 2PP swing (Wakefield 1938, 20.1%) ignoring artificial examples, though Canberra 1995 (16.1%) was very close.

Based on current government polling (showing about a 4% swing against the Government since the last election), a 2PP swing on average around 6.5-8 points to Labor would be expected for a vacancy caused by death or resignation, depending on how it's modelled.  Government polling could well get worse before the by-election comes, but that won't make a huge difference.  Given that there is no loss of personal vote, the swing should not even be that large and Labor would be doing well to reach 40% 2PP.

The Shooters, Fishers + Farmers and One Nation tend to poll well in New England, with 6.2% and 7.7% in the Senate respectively at the last election.  The Shooters have won a state by-election in a NSW rural seat recently and been competitive in another. However these owed a lot to state and local factors, and it has been hard to generate strong flows from Labor to the party.  With One Nation, the issue with obtaining Labor preferences is even more acute.  The Shooters have been reported as not contesting because they cannot find a good enough candidate.  One Nation have been indirectly reported as unlikely to contest but have not publicly said anything.  It appears doubtful that Joyce will face a serious challenge on the right.

Indies (especially Windsor) have been the most significant opponents to the Nationals in this seat for some time, but without Windsor it is doubtful anyone has the profile.  Windsor would not have been likely to win anyway as he is now considered too close to the left in the electorate.  A lower profile independent could have a good chance of running second instead of Labor, but taking down Joyce will be a tall order.  Because of Windsor's high profile as a former member, it seems his 2016 result is not a realistic baseline.

Rob Taber, who finished a distant second to Joyce in 2013, is reported to be running and is reasonably prominent.  Taber commended Windsor highly when running in 2013 so may well be seen as a Windsor-ish independent.

I don't expect blowback over Joyce's ineligibility to damage his chances.  Labor will attack Joyce for continuing to act as a minister while ineligible, for causing the by-election and for not resigning from the ministry earlier, given that he says he had a gut feel he was going to lose the court case.  I expect this to all be seen as Canberra beltway stuff by the voters of New England and ignored. There will be some resentment that Joyce has even been unseated on a "technicality" at all, and this will work in his favour.

Peter Hartcher has an interesting rundown of possible issues as seen by Joyce and his critics.  These together with general assessments of the government are typical mid-term by-election fodder and in some respects the campaign issues may not be that different to in the 2016 one. Labor intends to run on similar issues to those Windsor would have run on: the NBN, Gonski, energy and university cuts.

A possible wild card is any further coverage concerning Joyce's personal life.  Sharri Markson of The Daily Telegraph on October 21 claimed the then Deputy PM was facing "a deeply personal crisis" involving "issues that have affected his marriage of 24 years" and referred to "staff leaving at the height of the situation".  The report was very vague on details.  At the same time a report in the Herald-Sun claimed that a high-profile unnamed pro-"family values" federal politician had been in an affair with a staffer for several months.  It was easily assumed that these two reports were connected.

The Daily Telegraph also reported Tony Windsor as having published "vicious innuendo about" Joyce's personal life on social media, and Andrew Bolt has made similar claims.  Windsor and others on social media have indeed been making comments about alleged sexual harassment issues from years ago (unrelated to the current media reports), said to involve one or more unnamed politicians.  Many observers have assumed that these claims, unsupported by any evidence that I have seen, are aimed at Joyce.  However unlike the celebrity Weinstein-type harassment scandals now breaking on a seemingly daily basis there has been no support from anyone claiming to be a victim of any Australian politician, so barring such support it's hard to see these whispers causing damage at this stage.

Can The Election Be Declared Quickly?

The by-election comes four days into an eight-day spell of House of Representatives sittings which will be difficult for the government because of the absence of Joyce (which means the government has to rely on the Speaker's vote to win motions opposed by Labor and the crossbench, for instance to gag members). There has been some speculation that if Joyce wins with a big outright majority, his win might be declared early, allowing him to resume his place in time to restore the Coalition's "floor majority" before the second sitting week starts.

I am very sceptical of this, given that a result can only be declared once it is mathematically certain, based on sufficiently checked votes, who the winner is.  The counting on the night will need to be rechecked before a declaration can be made.  If Joyce is not well over 50% on primaries it becomes more difficult still, because even if he thrashes his AEC-selected two-candidate-preferred opponent on the night, it may require a full distribution of preferences to establish who is actually second.  The by-election for Higgins in 2009 took nine days to declare despite the winner polling 54.6% on primaries.  Perhaps a declaration could be possible towards the end of the sitting week but the idea that he'll pop back into his seat on December 4 seems extremely unlikely to me.  More likely the government is set for two weeks down a seat.


Polling seen at the time of the election being called was not very useful as it included Windsor.  An especially useless but amusing piece of polling was a fake live phone poll on Nationals letterhead peddled by anti-Joyce forces.  Further giveaways included the incorrect margin of error and the male/female percentages all lining up perfectly without any rounding issues.

An Australia Institute ReachTEL taken in September found a 57-43 lead to Joyce over Windsor after preferences, off primaries of 44.6 to Joyce and 26.5 to Windsor with 5% undecided (so say, really something like 47% to Joyce).  Nationals internal polls have been reported with even stronger figures for Joyce against Windsor.  Seat polling is facing an accuracy crisis in Australia, and polling of this seat in the leadup to the 2016 election found it to be much closer than it actually was.

More polling will be added when I see it, but will be most useful once the field has settled a bit.


  1. The seat was won by Labour in 1910, not Labor.

  2. Since there is nothing called the "Beltway" around Canberra, can we please stop using this term to describe insider gossip? We are not yet a US colony...