Wednesday, May 9, 2018

The Expected "Super Saturday" By-Elections

Today's four resignations from the House of Representatives following the Section 44 disqualification of Labor Senator Katy Gallagher is expected to trigger a day of at least five by-elections, or at least a cluster of by-elections close to each other.  The following seats are affected:

Braddon, TAS (ALP, 2.2%)
Fremantle, WA (ALP, 7.5%)
Longman, QLD (ALP, 0.8%)
Mayo, SA (Centre Alliance vs Lib, 5.0%; Lib vs ALP 5.4%)
Perth, WA (ALP, 3.3%) 

See The Tally Room for detailed histories of these seats.  Also see the Poll Bludger thread for Perth.  All seats will be contested on the old boundaries, irrespective of redistributions.

It's possible that given the strictness of the High Court's ruling, other MPs may come under pressure to resign or be referred to the High Court (note: as of Friday the media are suddenly all over the Anne Aly story, which has been known via Jeremy Gans' Twitter comments for months), though the Coalition may not be in any great hurry to hunt down any more and invite more scrutiny of its own remaining unclear cases.  The by-elections are not just a nuisance for Labor, but also for the Coalition, which must either throw resources into contesting them seriously or else chicken out and leave voters wondering what all the fuss was about.

Australia has never had a day with five federal by-elections before, so it would be quite a novelty.  Three were held on the same day in 1981 and 1984.  In 1994 four were held across three weekends following a cluster of resignations, but the resignations came on different days.  At state level, NSW has often held multiple by-elections on the same day.

The by-elections might be avoided if an election is imminent, as there is no requirement to hold them at any specific time.  However delaying the by-elections on the grounds of an imminent federal election would give the game away that an early federal election was intended, kickstarting a tediously long (again!) de facto campaign three months out.  It would be a very brave government to commit itself to an August election in this manner without even knowing whether its budget measures are working (let alone whether any impact of them doing so is durable) and with a long history of trailing in the polls.  Leaving the seats vacant for most of a year (in the case of a potential May 2019 election) would be unacceptable so it's easy to see why a rush of mid to late June by-elections is expected.

Assuming that's the case, what do historic patterns say about these sorts of by-elections?

The General Picture For Opposition By-Elections

Firstly, the history of by-elections in Opposition-occupied seats.  By my count there have been 61 of these since Federation (this includes those held by the junior Coalition partner at the time).  Two of those are dubious inclusions since the original election had been voided for irregularities, meaning the Opposition may or may not have really won had the original vote gone smoothly.  In two more cases the Opposition won the seat back unopposed.  In 26 the election was contested, but not by the government, with the Opposition of the day retaining 23 of those and losing three.  The losses were: Maranoa 1921 Nationalists to Country before there was a coalition, Cunningham 2002 Labor to Green and Lyne 2008 National to Independent.

Of the 31 cases where the Government contested a seat the Opposition had won fair and square at the previous election, the Opposition retained the seat 30 times.  Kalgoorlie 1920 is the famous exception - Labor's Hugh Mahon had been expelled for sedition, attempted to win his seat back and was defeated.

A 2PP swing either exists or can be reasonably estimated for 27 of these 31 Government-contested by-elections.  On this basis, there were 16 swings to the Opposition and 11 swings to the Government of the day, with a mean and median swing of 1.1% to the Opposition.  However, the swings have been quite oddly distributed, with 17 of below 4% one way or the other, six in the range 6-8% and four in the range 12-14%.   The standard deviation of the swing is six points.

The swing to Oppositions is therefore about five points less than it usually is in government-held seats that go to by-elections.  This would be largely caused by the fact that it is usually an Opposition MP who is resigning and taking their personal vote with them, rather than a government one.  There might be a small amount of backlash against the Opposition for causing the by-election in some cases, and governments might be more likely to not bother contesting by-elections if they expect an unflattering result.

For Government seat by-elections there is a relationship between by-election swing and how a government is going in polling at the time.  One might expect the same for Opposition seat vacancies, and indeed it is the case - with a warning that 17 of the 20 data points come from old Morgans (for the three in the Newspoll era I've used rolling averages):

It's not a bad relationship by the standards of such things, with the swing to the Opposition in polling explaining 31% of the variation in by-election swings.  Labor currently has about a 2.3% swing to it in polling (based on my One Nation adjusted aggregated figure) and on that basis, all else being equal, would expect about a 2.5% swing in its favour at a by-election for a Labor seat held right now. (All else is not necessarily equal as discussed below!)

It's worth noting the circumstances of the two by-elections where there was a meaningful swing against an Opposition although it was polling above its previous election result.  These were the Griffith by-election in 2014 (for the seat of former Prime Minister Kevin Rudd) and the Cunningham by-election in 1977 (following the death of 14-year incumbent and former Whitlam Minister Rex Connor.)  The dot-point more or less on the line near those is the 0.1% swing against Labor in the Bendigo 1960 by-election on the death of 10-year Labor incumbent Percy Clarey.

Given that the average swing to Oppositions in Opposition-held by-elections is small, it does make sense for governments to contest non-safe seats held by Oppositions.  The trend of governments usually not doing so really took off in the last thirty years, in which only two of 18 Opposition by-elections have been contested by the ruling party.  Prior to that, governments contested about 70% of the time.

Specific Factors

With the exception of Perth, the Labor-held by-elections are unusual in that they would be recontested by disqualified incumbents (assuming Susan Lamb actually manages to renounce thisa time!)  This creates a difference with most by-elections because there is no loss of personal vote, and that should also count in Labor's favour compared with the typical Opposition-seat by-election.  The new MPs have only been in parliament for just under two years themselves, and have probably not had much time to build up a personal following.  However Justine Keay (Braddon) and Susan Lamb (Longman) both defeated Coalition incumbents - in Lamb's case a two-term incumbent and in Keay's a one-term incumbent who was also a former State MP.  Whatever personal vote Wyatt Roy and Brett Whiteley may have had would have been factored into their 2016 result and should therefore be added to the Labor's starting line for the by-elections.  Thus we have two examples of something very strange indeed - by-elections where the "vacating" party has not an in-theory loss of personal vote, but a double sophomore effect (albeit a weakened one) on its side!

Aside from any anger at Labor for causing the by-elections through its own sloppy screening (which was only as sloppy as anyone else's) the Braddon and Longman by-elections therefore have personal vote dynamics more typical of a government vacancy, while Fremantle is somewhere between the two.  And on that basis, Labor should be expected, barring seat-specific factors, to win them all easily.  And that is even before considering the (slender) evidence on whether voters view disqualification by-elections differently from other by-elections.  [Edit: Former Liberal incumbent Brett Whiteley looks set to be the Liberal canddiate in Braddon, so this actually means that Labor will have a partial sophomore effect while the Liberals will not have a loss of personal vote, beyond any caused by Whiteley being out of the job for two years.]

There are, however, some seat-specific factors that are not in Labor's favour.  In Longman, Labor owed its 2016 win to a decision by One Nation to get rid of Wyatt Roy.  The Coalition will presumably learn its lesson and run a candidate more palatable to One Nation supporters.  One Nation polled 9.4% in 2016 and could well poll considerably more in a by-election now.  The preferencing behaviour of One Nation supporters has also changed since the 2016 election, and I think that is not just because the 2016 contests involving the party were aberrant, but also because the party has become more overtly Coalition-friendly.  A One Nation candidate polling in, say, the mid-teens and splitting preferences a la the Queensland state election could in theory do as much as six points of damage to Labor's 2PP in Longman.  Which sounds like enough to account for all of Labor's advantage and then some, except that any increase in the One Nation vote in the seat is likely to draw more from the Coalition than Labor.  Moreover, One Nation voters are not very directable when it comes to how-to-vote cards, as the Queensland election showed.

In Braddon, the Tasmanian Liberals  had a very bad campaign in 2016.  They were dogged by recriminations over the dumping of Senator Richard Colbeck to a position he eventually lost (but has since recovered) and also by perceptions that the party's state ticket was too male-dominated.  More significantly than these things, it appears the "Mediscare" campaign resonated in a state that feels perpetually vulnerable, and that the themes that polled well for Malcolm Turnbull in the inner cities fell flat in the regions.  The party may have learned some things since then, and plenty of goodies for Tasmania were rolled out in the Budget.  There could also be spillover from the Liberals' convincing state election win, but Tasmanian voters have a history of strongly separating state and federal politics, and two factors that helped them at state level (fear of minority government and state Labor's poker machines policy) will not apply here.

Overall while most of the Labor seats are loseable, the current balance of evidence suggests each one, considered by itself, should really be retained.  Longman is the most uncertain, but I am not sure it is as lineball as people are saying.  Perhaps the Coalition's best hope could be if the by-elections somehow become a referendum on Bill Shorten (especially if the Coalition doesn't have too much to do with this being the case.)  The by-elections don't affect who is the government and are probably only for a year anyway, so just maybe voters could see it as a free hit at the Labor leader (as Cunningham 2002 was against Simon Crean). I'm not that convinced this will happen though.

In the case of Fremantle, there might be some temptation to either not contest it or to run dead in the faint hope of throwing it to the Greens.  Both Perth and Fremantle saw Green votes around 17% in 2016 and the party is possibly in better condition in WA at present than in many other states.

The Opposition has already started spinning the results by claiming the by-elections will be a referendum on the government's tax policy.  This sets the scene for what will presumably be a claim of any swing to Labor as a triumph, but in fact the dice are loaded in Labor's favour and only swings above, say, 5% to Labor should be considered interesting.  Any swing to the government would be a good result for it.


The Mayo by-election is so unusual that there really is no basis for predicting it.  Australia hasn't had a federal by-election caused by a crossbencher resignation since 1946. The party Rebekha Sharkie was elected under has had its leader resign to contest state politics, had another Senator disqualified and consequently lost a Senate seat, and changed its name, all since 2016.  It also underwent a rollercoaster ride in public support in the leadup to the SA state election (ending up a lot lower than it would have liked) and the feeling that the bubble has burst could well spill over into this one.  Another advantage that the Liberal Party has in trying to evict Rebekha Sharkie is that the 2016 result factors in the troubled tenure of Jamie Briggs, who Sharkie defeated.  The Liberals can be expected to run a much better candidate this time.  It should come down to what voters think of Sharkie herself and whether she has been a popular incumbent.  It is unlikely Labor will threaten, as the closeness of the 2016 Liberal-Labor 2PP owes a lot to Briggs.

If there are any more by-elections I will add them in here.  I will probably also do a piece dealing with each seat once the dates and at least the main contenders are known.  Comments on remaining Section 44 issues as well as the by-elections are welcome on this thread.

Meanwhile, just noting some early betting odds to see how they go: Perth 1.15 vs 5.00, Fremantle same as Perth, Braddon 1.35 vs 3.00, Longman 1.85 vs 1.90 (all Labor first).  Mayo Sharkie 1.80 Lib 2.40 Labor 5.50.


  1. sorry a bit off topic but I am looking for a breakdown of age to party vote.. found something dated 2013..... but looking for something 2016 or 2o17 esp .... 49 to 64 and 65 plus.... cannot find this on the web

    1. Probably be in some summary of the Australian Election Study (which is rather "bouncy" in its data anyway). We don't have rich exit polling in this country as in the US.

    2. can you help if you have time

  2. think the betting odds have the probable results correct

  3. How is it that Braddon (and Bass) can get such enormous Liberal primary votes at the state election (even when Labor is doing relatively well), but be competitive marginals federally. I really can't comprehend how there are that many voters out there that are comfortable putting a Liberal in the number one spot in one election but preferencing them below Labor in another in such a short time frame.

    1. It's mostly because of the majority government dynamic at state level. About 20-25% of Tasmanian voters don't really give a fig between Labor and Liberal at state level provided that the Greens are kept at bay. So they swing backwards and forwards to vote for whoever can win a majority - at some stages in the cycle Labor gets 50% and the Liberals 25% at state elections, at others it's the other way around.

  4. Your anti-Coalition views continue to shine through in everything you write and all your commentary on political issues. How you can pretend to be an independent, academic commentator is amazing

    1. Oh dear oh dear oh dear oh dear oh dear.

      Firstly, I am not an "academic commentator" and I do not pretend to be one. I have no current relationship with any academic institution, and I have never had any such relationship as it relates to my political commentary. I am however independent - I have never been a member of any party, and there is no party that I support with any reliability at all, or feel particularly close to.

      Secondly, you clearly haven't read enough of my work. For example you obviously haven't read the grief I gave Labor over its completely stupid position on Senate reform (now tell me how my "anti-Coalition views shone through there). You obviously haven't been following the stuff I've been saying on Twitter about the Tasmanian Greens' xenophobia this week. You haven't seen my attacks on the Tasmanian left parties for their attempt to destroy free speech in Tasmania in 2012 (which I continue to remind them of whenever they claim to be pro-free-speech). Those are just a couple of examples; there are so many others.

      If it comes to predictions you've ignored everything I've said about how Labor's lead in the polls does not mean they are shoo-ins for the next election. You've also ignored my successful prediction that the Liberals would win 13 seats in the Tasmanian election, among many other such examples.

      How you can pretend to have a clue is what is really amazing here!