Sunday, January 14, 2024

2024 Dunkley By-Election

DUNKLEY (VIC, ALP, 6.27%)  By-election March 2
Jodie Belyea (ALP) vs Nathan Conroy (Lib) and others
Cause of by-election: Death of previous incumbent Peta Murphy
Outlook: interesting; seat margin is just above average swing for government vacancies

Early this year we'll get the first electoral test for the Albanese Government on its own turf when the division of Dunkley goes to the polls in sad circumstances after the death of popular previous MP Peta Murphy.  Last year Labor sensationally captured Aston from the Liberals during a period of honeymoon polling, while the Coalition had a pretty good swing result when it retained the uncompetitive seat of Fadden in Peter Dutton's home state.  By-elections are more random and a lot less predictive than politics junkies tend to think they are, but an outer-suburban seat, on a loseable margin, with the honeymoon gone, seems much more significant.  

The by-election has been announced for March 2.  The writ will be issued Jan 29 with close of nominations Feb 8.


Dunkley, towards Melbourne's south-eastern fringe, has generally been a marginal seat, changing hands from Labor to Liberal in 1990 then back again in 1993, 1996 and 2019.  Liberal Bruce Billson, at one stage irrepressible Small Business Minister, was its longest-serving incumbent from 1996 to 2016.  A redistribution prior to the 2019 election made Dunkley more Labor-friendly and in 2022 it slipped just outside AEC "marginal" status, but no-one should be tricked by that, especially not for a by-election.

At state level Dunkley includes the Division of Frankston and parts of Carrum, Hastings and Mornington.  These are all now Labor seats except for Mornington, which Chris Crewther narrowly won against a teal independent after losing Dunkley federally in 2019.  Frankston and Carrum were part of the "sandbelt" battle that determined the 2014 state election in Labor's favour, but have since become much more solidly Labor.  In the context of a general state election pattern of swings against Labor in many outer suburban seats but swings to Labor in the south-east, what happened in the Dunkley area was pretty much on the state average with the exception of seats where there were personal vote factors (like star candidate Paul Mercurio picking up Hastings).  

For federal elections since its creation, the graph below shows about how Labor did on 2PP in Dunkley compared to the national average, as adjusted for later redistributions to make it more comparable to today.

So for instance a lean to Labor of 4 points on the graph means that in a general election where the national 2PP was tied 50-50, Labor would expect to win present-day Dunkley about 54-46.   The colours of the dots show the actual winning party (thus the Liberals won Dunkley in 2010 and 2016 but would probably not have won in either year on the current boundaries based on Labor's performance at the time.)  

The seat as it currently stands would have been initially Labor-leaning compared to the national average when Bob Chynoweth was the incumbent, but he lost in 1990 off the back of the Liberals' success in Victoria generally under Andrew Peacock.  In 1996 with Chynoweth's personal vote back on board it was again slightly Labor-leaning (indeed probably only Chynoweth's personal vote came anywhere near saving it after an adverse redistribution), but not enough to stop Bruce Billson winning it for the Liberals.  The seat then became a Liberal-leaning seat with Billson's personal vote until there was a move back to Labor in 2010 with Julia Gillard as Labor leader.   It might have shifted back towards the Liberals but in 2016 Billson retired.  

Rather than new incumbent Chris Crewther building a personal vote in 2019 the seat became more Labor-friendly even after accounting for the redistribution.  Crewther was a somewhat contentious MP and there appeared to be a strategic decision to leave the notionally ALP seat to its fate.  Peta Murphy was hence elected pretty easily at the second attempt.  

Overall federal voting patterns in Dunkley seem to have had a lot to do with (i) the national voting pattern (ii) whether either major party leader is Victorian and (iii) candidate factors.  Local issues and demographics will have been in the mix but it is interesting how strongly candidate factors have stood out.  

To try to throw Dunkley into the same "outer suburban" box as western Sydney and north-western Melbourne would be completely mistaken; this seat has not behaved like those in recent elections and even swung to Labor in 2019.  (Dunkley also voted No in the Voice referendum but had a higher Yes vote, 44.2%, than the national average.)

By-elections generally

The first thing to understand when considering by-elections is that government and opposition vacancies are different.  In a government seat vacancy the government is (with rare exceptions) losing the personal vote of its incumbent, while in an opposition seat vacancy the opposition is generally doing so.  There is also a tendency for governments to cherry-pick which opposition vacancies to contest, while oppositions nearly always contest government vacancies.  As a result of this, the average 2PP swing in government vacancies is about five points higher than in opposition vacancies: I make it about 6.1% since Federation, and 6.3% (Dunkley's current margin) in the last 40 years.  However there's a lot of variation (in my 2015 Canning preview I found the standard deviation to be about 4.7%.)  A lot is read into by-elections but they are very "noisy" events in terms of the overall federal picture.  (Note: differences between my averages and some in Antony Green's article here arise because I include anything where a 2PP swing can be calculated or estimated, not just those where the major parties were the final pair.)

Labor has quoted a statistic of 7.1% average swing in by-elections since 1983 that has been described in various forms by media.  I finally managed to replicate this as correct (actually I think it's 7.2) if one limits it to 13 government by-elections with a vacating (not recontesting) incumbent and where the seat finishes as a 2PP contest.  Peter Dutton has been floating the stat that no first-term government has lost a by-election since World War II, however there have only been four of those.  Three were in seats on double-digit margins (one of these the opposition did not even contest) and the fourth had a recontesting incumbent and was held while the government was in honeymoon phase.  One of these, Canning, had a swing large enough to take down Dunkley if repeated.  This is a particularly useless piece of electoral history from Dutton.  

Murphy's personal vote

Much has been said about Murphy being a particularly popular MP.  Measuring personal votes is challenging and commentators tend to overestimate them.  One measure that is used is Senate vs House of Reps primary differences, but these are affected by the Reps fields in particular seats and also by some seats having a higher rate of voters voting differently in the two houses than others.

On this front Murphy did rather well.  She outperformed the Liberals by 7.46 points more in the Reps than Labor did in Dunkley in the Senate, compared to a state average difference of 0.61 points (an implied 3.4% 2PP swing effect, which is high - Clare O'Neill in Hotham was even higher).  This gap, however, was probably boosted by a strong ballot draw for Murphy in the Reps, and also by Murphy taking about 1% in primary votes from the Greens that would then return as preferences. This suggests that something like 2-2.5% 2PP (still well above average) is a more reasonable estimate.  On Senate 2PP Murphy's implied personal vote comes out a bit lower but I now think Dunkley is a misleading division for this because the flows to Coalition from the Liberal Democrats and United Australia in the Reps were stronger than average; in the Senate a lot of votes from UAP especially exhausted.  

Bereavement vacancies

Something I covered a lot re Canning (the last time an incumbent MHR died in office) was the lack of historic evidence of much if any of a sympathy vote in by-elections caused by bereavement.  That might seem surprising but it seems that voters think more about who will be their best MP going forward and aren't inclined to simply rubber-stamp their former MP's replacement.  Also the timing of retirement by-elections can often be massaged to limit the damage for the governing party.  It might even be that some replacement candidates suffer by comparison in cases where a particularly loved MP for their party passes away.  I did note in the Canning previews that with the rate of government bereavements falling sharply from the 1970s on, there was some weak evidence of muted swing in these sorts of by-elections since the 1960s.  That said Canning itself did not add anything to that, and if the swing from Canning is repeated, Dunkley falls. 

This one might be different, or not.  This is actually the first time that a federal by-election has been triggered by a female MHR dying in office (four female Senators have done so).  Breast cancer is a high profile issue and Murphy's history with the disease recurring just around the time she was sworn in as an MP in 2019 is particularly poignant.  The few state-level by-elections where a female MP has died in office don't provide a lot of useful evidence about whether there is any gender difference in whether there is a sympathy vote.  

How To Vote Cards

The Animal Justice Party initially recommended preferences to the Coalition ahead of Labor.  It seems this is mainly a protest against the state Labor government over duck-hunting (though federal live export issues are also cited).  Very few voters will follow the AJP card as minor party cards tend to be not well distributed and minor party voters tend to make up their own minds; the AJP will probably not poll much here anyway having got 2.07% at the last Senate election.  As of 19 Feb the AJP had changed its card and recommended preferences to Labor ahead of both the Greens and Coalitions.  The AJP has said:

"Over the last 48 hours Labor and the AJP has reached an agreement on how we will cooperate to pursue sensible and important policies. 

We are now satisfied that directing preferences to Labor after negotiations delivers good policy outcomes and demonstrates that major parties cannot take the AJP and our voters for granted, or ignore us."  

What were those outcomes?

Other factors

Again as noted in the Canning previews there is a relationship between how a government is polling and how well it does in by-elections, with some of the swing against governments being caused by them tending to be less popular at by-election time than at the previous election.  As of early 2024 the Albanese Government was running just about at its 2022 vote in aggregated polling, which would be consistent with a slightly lower than average swing for a given type of by-election.  2PP polling had been shifting fast in the second half of 2023 but has so far stabilised in early 2024.  That said it is fairly common for government polling to decline during February-March, so we should keep an eye on the national polling picture.    

This is a first-term by-election.  During Canning there was discussion of governments tending to do better in first-term by-elections, but this was just a distraction, because most of the first-term vacancies have been in opposition seats.  Indeed after eliminating North Sydney 2015 (Labor did not contest) and Lindsay 1996 (recontesting incumbent in voided seat) there are only two precedents left in the last 90 years, Canning 2015 (with a mid-campaign change of Prime Minister) and Hughes 1984 (while Hawke government was polling through the roof).  In the early 20th century first-term government vacancies as often as not saw large swings.  

The by-election will be another test for whether the Coalition can perform under Peter Dutton's leadership in Victoria, or while the Victorian Liberals are such a mess.  The Coalition failed the first test of whatever this was in Aston (it was not so much the result there but the margin) and another disappointing performance would raise serious questions about problems with the Liberal brand in Victoria, after a lousy third place after preferences in the Mulgrave state by-election.  

Cost of living is the issue in Australian politics at the moment and will feature heavily in the campaign.  Local infrastructure issues, however, often feature in by-elections and this one is no different with the Liberals keen to use the shelving of the Frankston-Baxter rail upgrade to their advantage.  However the latter has drawn an unflattering headline for Peter Dutton after a cool reception from state politics.  

In late January the government announced that it was revising the promised Stage 3 tax cuts, with changes that comparatively advantage lower and middle incomes and disadvantage the higher income brackets.  This debate will feature strongly in Dunkley with the government arguing that it is responding correctly to changed circumstances and the need for cost of living relief, and the Coalition arguing that the government has broken trust with voters and is failing to address bracket creep.  In general polls have found the stage 3 tax cut changes to be well received, though many of the polls have had long and sometimes problematic wording.  

The Advance organisation is spending up big on extremely unsubtle attack ads against Albanese, which will test whether its methods have any staying power or whether any major credit it was given for the Voice success was undeserved. 


Labor has announced the preselection of Jodie Belyea, a well-credentialled women's services and community worker who was recruited into the ALP relatively recently, in part through Murphy herself.  We will have to see how Belyea goes during the campaign but on paper at least Labor seem to have found another candidate who can make a genuine claim to come from outside politics, someone who in another world could have been an independent.

The Liberal Party has chosen Frankston Mayor Nathan Conroy, beating a field of other hopefuls including ex Carrum MP Donna Hope (formerly Bauer).  Conroy bolted onto Frankston Council North-East ward in 2020, being first elected with 26% of the vote (over 8000 primaries) at the first attempt. He was immediately made Deputy Mayor and a year later elected Mayor around the table (he has now been elected for three years in a row in that position, which is apparently unprecedented; Victorian councils often share the top job around).  Prior to Council politics Conroy was at one stage "general manager of a multimillion-dollar business" - which turns out to be a bowlo, with the claim being both criticised as hyperbole by opponents and defended by those who work with him.

The Age reported that there was a possible Section 44 issue surrounding Conroy, which was later revealed as him having been born in Ireland (he speaks with a strong Irish accent). He has renounced Irish citizenship.  

I expect Dunkley to be a two-party contest and other candidates to be a sideshow.  

A controversial candidate in 2022, who is running again, is MyPlace founder Darren Bergwerf (IND), a sovereign citizen, conspiracy theorist and Holocaust "questioner".  In 2022 Bergwerf very narrowly missed getting his deposit back and qualifying for public funding.   I am not convinced Bergwerf is eligible to be elected on account of a likely automatically inherited Dutch citizenship.

The Libertarian Party's AEC registration has been approved and it the party will have a first run under its new name with candidate Chrysten Abraham, a 2022 state and federal candidate for the former LDP.  

Also running: Alex Breskin (Greens)Reem Yunis (Victorian Socialists), Heath McKenzie (Australian Democrats who are currently in danger of deregistration), Bronwyn Currie (Animal Justice)  

The ballot order is pretty much perfect for the Liberals - they have drawn first and Labor last (in full it is Liberal, AJP, Libertarian, VicSoc, Bergwerf, Greens, Democrats, Labor).  This could be worth at least 1% as while the donkey vote is at most half a percent, it is also a big disadvantage to be way down a relatively long list.  Labor were ahead last time in this seat.  

There has been some nonsense in the Australian about Labor preferencing the Greens with regard to the statements of some Greens on Gaza.  The Labor preferences will never be distributed so what Labor does with its preference recommendations is immaterial to the outcome, whatever people think of the symbolism.  


As polls or poll-shaped objects appear I will note them here.  Note that seat polling is not very reliable in general.  

1. Australia Institute uComms 5-6 Feb:  A mixed robo/SMS poll with a sample of 626.  After distribution of undecided, Labor 40.1% Liberal 39.3% Green 8.2% (Green candidate not named) Libertarian 1.6% IND/other 10.8%.  (It's possible the four others who are running could get 10.8% but many voters would have been envisaging a teal-type independent and there isn't one.) 2PP on respondent preferences is 52-48 to Belyea, but I would expect at least 53-47 off these primaries.  Automated polls struggle to get a fair sample so while these numbers look reasonable overall there are issues with the young voter breakdown (looks too conservative) and also some hilarious crosstabs for the Greens who are supposedly getting 1.7% in 51-65 but 27.9% in 65+.  The poll includes an issue question about the stage 3 tax cuts that while similar to other results includes skewed wording that leads the respondent to agree with the Government - this question should be ignored completely.  It also includes other issue questions that are not very relevant.  

2. YouGov: 51-49 to Liberals off primaries of  Lib 40, Labor 33, Greens 9, Bergwerf 7 and the rest 2-3% each.  Sample size only 394 (effective 256) which is adventurously low given the issues with seat polls.  

3. Painted Dog: n=278 and 26% undecided with no report of full primaries by The Nightly (whatever that is), moving right along then ...

666. Alleged Labor Internal Poll, good luck finding any details: a supposedly "leaked" (which means given away) Labor internal has appeared on Sky on 11 Feb.  Labor are said to have the same lead as in the TAI poll, but the main purpose is to claim that voters strongly disapproved of Frankston Council's performance.  The point of this is to tie in with a Labor campaign attacking Conroy over council rate rises, suggesting that while he was strongly elected his performance might be another thing.  No details are available based on which to judge whether voters were assisted to reach this view. 


This is a by-election that sits pretty much bang on the standard swing line for a government vacancy. That doesn't mean that it will be super-close, but arguments as to why the swing should be more or less than average can be made in either direction.   Possibly the fact that the Coalition did so badly in  Aston skews things in Labor's favour, but there were candidate factors as well as honeymoon polling factors at play there.  Here the Coalition has picked a suitable local candidate, albeit one whose council performance is controversial.  

For Labor I think the equation is pretty simple: a win is a win is a win.  Their overriding objective should be to win the seat by any margin at all and get a new MP in who can start building a personal vote (especially as the opponent looks to be no mug at vote-building and might not be easy to get rid of if he wins).  To win comfortably would be nice but it really doesn't matter.  A loss by any margin is bad but it's not the end of the universe either.  

For the Liberals it's less straightforward.  A win would be terrific but I think there's a big difference between a close loss (inside say 53-47) and a heavier one.  If the Liberals do not get a decent swing here then this will be further evidence of a general Victorian problem that would be a big headache going into the next election.

I intend to have live coverage of this by-election on the night, depending upon when it is.  A sidebar Not-A-Poll has been added for those who want a go at picking the result.  

No comments:

Post a Comment

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.