Batman: A Unique Federal By-Election

Batman (Vic): ALP vs Green 1.0%
Ged Kearney (ALP) vs Alex Bhathal (Green)
Incumbent David Feeney (ALP) resigned over eligibility issues
Outlook: No reliable data basis for predicting this seat.  

A by-election will be held for the Victorian seat of Batman in the near future after David Feeney became the first confirmed Labor casualty of the Section 44 citizenship fiasco.  Feeney threw in the towel when he was unable to find any positive evidence that he had renounced his UK citizenship circa 2007. Some Labor insiders believe the seat is now unsaveable while some are more upbeat that they may just hold it.

This could be the last time we'll be referring to the seat by the name "Batman".  There's a significant campaign to rename it after Simon Wonga, but that won't be decided until the redistribution process concludes later this year.

The heavy lifting by way of preview has already been done at Tally Room and Poll Bludger with their excellent by-election guides.  The seat's dramatically split voting pattern was laid out by Michael McCarthy in his pieces (here's the latest) on the "hipster-proof fence" (aka Tofu Curtain, Great Wall of Quinoa, Corduroy Line) around Bell Street, which divides the Green-friendlier and Labor-friendlier sides of the electorate.  Also of interest may be Kosmos Samaras' analysis of why Labor is getting trashed by the Greens in inner-city seats like the state seat of Northcote (the southern half of Batman) and (see also Tim Colebatch on this) what they can try to do about it.  A convenient step-up in attacks on the proposed Adani coal mine in Queensland probably isn't it (at least, not by itself).



All that really remains for me to do is to put this by-election into its proper historic context, by telling you it really doesn't have one.  It's been said that all by-elections are unique, but this one is especially so. And while that might sound like an open invitation to pay slavish attention to the polling that is bound to appear, I don't recommend doing that either.  Seat polling in Australia is a struggling enterprise at the best of times, but inner-city seats with high enrolment churn are especially hard to poll and we've seen a chain of severe (at times, double-digit) polling failures in them.  In 2016 in this seat, Newspoll had Labor winning 53-47 (reasonably accurate given the 51-49 margin) while a Greens-commissioned Lonergan that had the Greens winning 55-45 was well wide of the mark.

Why Batman Is Different

Batman is the latest example of a rare species of federal by-election caused by the former MP either having their place in the parliament taken away from them, or resigning because that was about to occur.  In all there have been eight previous cases of this - four election voidings between 1904 and 1920, Hugh Mahon's expulsion for sedition in 1920, the disqualifications of Jackie Kelly in 1996 and Barnaby Joyce last year, and the resignation of John Alexander pending the same fate also last year.

What sets the Batman by-election apart from all of these, firstly, is that the incumbent isn't recontesting.  There's been a view that these kinds of by-elections suit the party whose MP has been disqualified because voters resent being sent back to the polls because of legal technicalities.  However, it's been difficult to separate this claimed effect from factors like the recency of the election (in Kelly's case) and the personal vote of the incumbent.  Now we have a by-election where the incumbent is resigning in expectation of disqualification, but isn't recontesting, so if there is a sympathy vote, does it flow to a different candidate from the same party?  It seems unlikely that it would, given that parties do not even get sympathy votes at by-elections when the incumbent MP has died.

The next aspect that is unusual is that Batman is a Labor versus Green contest. The history of swings at federal by-elections mostly concerns two-party preferred swings between Labor and the Coalition (which can then be divided into government and opposition seat by-elections and so on).  It is hard to apply that history to a serious contest between an opposition party and a minor party on its own side of the fence.  The seat will be up for grabs again within 15 months, and the outcome does not affect the government's majority.  So for anyone who feels like voting Green to send a message to Labor about - well, anything really - this is basically a free hit.

 The only federal by-election to have previously finished as a close Labor-Greens contest was the Cunningham by-election won by the Greens' Michael Organ in 2002, but Cunningham was not an established Labor vs Greens seat at that time, and didn't become one thereafter.  The Greens had only scored 6.6% there at the previous election (with another 7.2% for the Democrats).

Cunningham shared a feature with Green state by-election victories over Labor in Fremantle (WA 2009) and Northcote (Vic 2017), which is that in each case the Liberal Party didn't contest.  However, Labor narrowly retained the Melbourne state by-election (2012) under the same circumstances, in a seat that is now solidly Green.

One might expect the Liberals to also sit out Batman, since their normal result in the seat (third place at the last three elections, less than 20% primary in 2016) is embarrassing enough even without the normal by-election anti-government reaction factored in.  That said, Michael Kroger has suggested the Liberals might run someone to try to spoil Greens candidates if they decide that they don't like Alex Bhathal's position on Palestinian issues.  I would imagine that running to get, say, 15% of the vote in the hope of making a three-point difference on preferences wouldn't be a lot of fun for any Liberal candidate.  Presumably the intended payoff of such a miserable exercise would be bragging rights in the three-cornered contest for the seat of Melbourne Ports, with its high Jewish vote.  Whether Kroger is serious about all this or whether it is a warning shot to future Greens candidates remains to be seen.

In any case, Kroger doesn't have the final say - my suspicion is the party will pike, lest the momentum of having Labor on the rack over Section 44 issues be blunted by a shabby result.  The possibility of a loss to the Greens placing pressure on Bill Shorten's leadership would also be a juicy temptation.

If the Liberals don't run (as widely expected), that becomes trouble for Labor when we look at the preference distribution from 2016.  The Greens' regular candidate for this seat, Alex Bhathal, led by 0.9% on primaries, but had expanded this to 3.7% on the preferences of a gaggle of left-wing micro-parties. However, Liberal primary votes then split 63.6% to 36.4%, because the Liberal Party had decided to preference Labor over the Greens, and Liberal Party voters have rather high how-to-vote-card follow rates.

The evidence of Northcote (the south end of Batman) in particular is that once Liberal voters no longer have a how-to-vote card to follow, they behave much more evenly between Labor and the Greens.  Thus an overall preference flow (mostly from Liberal voters) in Northcote of 66% to Labor in 2014 became only 46% to Labor at the 2017 Northcote by-election.  At the 2016 federal election the flow of preferences to Labor in Batman was 55%. This may have been because of David Feeney's unpopularity (see below) but the number of left-wing micro-parties contesting would also have been a factor.

Candidate Factors

We don't have to pay much attention to candidate factors in the case of Bhathal, since she has been the Greens candidate for this seat five times since 2001.  She will have a high degree of name recognition, but that would also have been the case in 2016, so mostly that is already factored in.  In the lead-up to this by-election there have been some minor but ineffective rustlings against her within Greens ranks.

However in the case of Labor, candidate factors are very relevant.  David Feeney had an utter shambles of a campaign in 2016 and as a result his personal vote was net-negative.  Assessing the size of a candidate's personal vote is messy but one measure that can work well (all else being equal) is to compare the difference between the party's House of Representatives vote and the party's Senate vote, with the same difference for other candidates of the same party in seats in the same state, especially surrounding seats.  This debate came up recently when defenders of Melbourne Ports MHR Michael Danby claimed Danby had a good personal vote because he had polled five points higher than Labor's Victorian Senate vote.  In fact, Stephen Murray showed that Danby had the third worst personal vote by this measure of any successful Victorian ALP candidate.  The second worst score belonged to Peter Khalil (Wills) - who as a new MP was not comparable - and the worst score was Feeney, with a House minus Senate difference of just 0.4%, nearly eight points below the median for Victorian Labor incumbents.  Given that incumbency is typically considered to be worth at most two points on average (and probably less), this suggests Feeney's personal vote was somewhere around -6%.  I doubt it's really that bad though; it could be though that Bhathal has a personal vote through her repeated attempts at the seat and that that made Feeney look a bit worse in this comparison than he was.

It would stand to reason that where a bad incumbent is ditched, a party's base vote should improve.  However there are not that many obvious examples to compare with, and many of these (eg Robertson and Bass in 2010) involve first-term incumbents only.  When a first-term incumbent does not recontest, their party still has a personal vote advantage, since at the previous election the other side had an incumbent, and now nobody does.

The new Labor candidate, Ged Kearney, is a very high-profile name as President of the ACTU since 2010, and has been seeking a political career for some time.  The ACTU is not quite as influential as it was in the Bob Hawke days (Hawke's approval as leader of the ACTU was being routinely polled for years before he entered parliament) but the post has still produced some heavy hitters since, such as former Labor leader Simon Crean and former Minister Martin Ferguson (the Batman MHR prior to Feeney).

Overall

One possible approach to a very foolhardy attempt to model Batman is to use Northcote as a base for the kind of swing that might occur.  In Northcote Labor copped an 11.6% swing; in Batman anything over 1% will mean the Greens pick up the seat.

The primary difference between Northcote and Batman is the Labor candidate factor.  In Northcote, Labor lost a popular 11-year incumbent and Minister while in Batman they are losing a dud.  In Batman, they also have a much stronger replacement candidate than in Northcote.  The second important difference is that Northcote was a state government seat while Batman is a federal Opposition seat.  The not-especially-popular Andrews government may still have some drag effect on the result, but it shouldn't be as much.

These differences should be worth several points.  But are they worth as much 10.6 points combined?  They could be (indeed there might even be a swing to Labor after the Feeney 2016 disaster), but I'm mildly sceptical.  However, if another difference is added by assuming that the Liberals run and preference Labor, there could be enough there for Labor to hold on.

Another approach is to just ignore 2016 and ask, if we use 2013 as a base, is there enough reason for a 10.6% swing to the Greens?  That line makes things look more positive for Labor, because while the impact of the Liberals not running a candidate could be as high as five points, if there was no reason for an Opposition to be affected by the usual swing factors at by-elections, then maybe Labor should hold.  However, that ignores the portion of the 2013 to 2016 swing that had nothing to do with David Feeney's terrible campaign.  The Green vote in the Senate in Batman increased by 4.25 points from 2013 to 2016 in the context of a status-quo state result.  That demographic change has probably continued since, so underlying shifts plus the Liberals not running would be about enough to wipe out the 2013 margin by themselves.  So Kearney will have to campaign very well to counteract any by-election effect.

I don't think it is hopeless for Labor but it does appear to be quite difficult.

Poll (25 Feb)

A poll-shaped object has been detected in the form of a landline-only Lonergan of 693 voters.  It is not reported (and should have been) whether this poll was commissioned by anyone or just done by the company off their own bat.

Aside from the modest sample size, causing the 53-47 to Labor result to be within the notional margin of error, the poll has numerous red flags.  No-one in their right mind would poll landline-only in an inner-city seat anymore, after the numerous robo-polling failures in such seats in recent times (it's risky enough in any seat!)  The report acknowledges that respondents were heavily skewed to voters aged 35+, meaning that the scaling from younger voters is going to be from a very small sample, increasing the poll's unreliability and notional margin of error.  That's even assuming that the young voters who were reached were representative, which they very likely weren't.  The poll also appears to have used a forced-choice issues question on whether voters were more likely or less likely to vote for a party that supported the Adani coal mine - isn't a voter in Victoria (even in Batman) allowed to not care about a hypothetical hole in the ground in Queensland?

If this poll is correct it will be a very, very, very lucky hit!

Update (1 March)

Lonergan's chances of being lucky have increased as there has been a significant attack on Bhathal from within. While rumblings of such internal dissent had been present in the lead-up to the by-election, they seemed to have been smoothed over, but the leaking of excerpts from a detailed complaint (albeit a dismissed one) can't be helpful.

In recent days betting odds as lopsided as Greens 1.10 Labor 4.50 have been seen but I suspect it was closer than that even before this latest incident. 

Update (13 March): Fake Poll Alert

There hasn't been any new polling data.  A claimed Labor internal poll dated 9-11 March has appeared on Twitter but I consider it to be a fake.  It has the same 2PP as Lonergan's poll but different primaries, together with a screed about the success of Labor's campaign that refers to "media allies" and attempts to distract from issues like Adani or refugees.