Saturday, December 16, 2017

Bennelong Live: Majority On The Line (Plus Post-Count)

Bennelong: John Alexander (Lib) vs Kristina Keneally (ALP)
2016 margin: Liberal 9.7%

CALLED: Alexander (Lib) retains - government retains its majority


Here's a graph of the 2016 vs 2017 booth swings in Bennelong.

This is quite a strong relationship with 47% of variation explained.  Some booths (see the Tally Room map to see where they were) swung big and some didn't move much at all.  However those that swung big were mostly those that had swung big to Alexander in 2016 and the swing was just returning to sender.  This is suggestive of a sitting member having successfully worked certain communities that Labor either neglected or failed to appeal to in 2016, perhaps because the seat wasn't in play, but targeted more competitively this time.  It would be worth looking at data from other seats to see if the Coalition performed strongly in booths with large Chinese populations in competitive seats in 2016.

Sunday 5:40: More postals have been added and the flow did indeed weaken (quite a bit actually) so that Alexander made no progress from 54.8% and will probably stay around there.

Sunday 2pm: The AEC has added nearly 6000 postals and those added broke 61-39 to Alexander.  This is consistent with my suspicion that postals might be better for him comparatively then last time, however trends seen in early postals quite often moderate in later postal counting.  Alexander is now on 54.8% 2PP (a swing of just under 5%) and might end up a little over 55.

Sunday 10:50: 

Update from the AEC: The AEC began counting postal votes today 9am for #Bennelong. Anticipate counting up to 9,500 postals today. These may not flow though the Tally Room until late today. Last night's results now available at polling place level.

Also with so much silly extrapolation about this result going on, it is useful to check William Bowe's graph of by-elections as predictors of the result at the next election.  For by-elections held after the first nine months of a government's term, the relationship is weak, explaining less than 7% of variation.  It's also shallow, with each point of by-election swing being worth only about 0.12 points at the next election, not half a point as conventionally thought.

Live Comments

9:45 End of night wrap: What we've seen here is a massive relief for Malcolm Turnbull and his government, but not much beyond that.  The swing is currently at about 5.6% and should come down when postals are added, making it a run-of-the-mill by-election swing.  The government is unpopular, so the swing should have been larger, but the sitting member recontested, so it should have been smaller. These things may well have cancelled out. The outcome therefore lands currently bang in the middle of my "inconclusive" range.  There were a number of wildcards from a campaign riddled with blunders and questionable strategies on both sides, but either the voters have ignored most of that or they have taken notice of both sides' errors equally (it's hard to know how these things will pan out, which is why I gave Keneally a fair chance.)

Labor are spinning the result as a triumph, but will be disappointed not to get closer.  Trying to argue the consequences of a by-election swing against the government being extrapolated to a general election ignores the fact that such swings historically aren't.  Should Labor now proceed to come a cropper to the Greens in a by-election for Batman (if there is one) they could be looking at a rocky start to 2018.

The disgruntled right within the Coalition - Christensen, Sukkar, Andrews, Abbott et al have not been vindicated by this result.  The passage of same-sex marriage has not harmed the Coalition, and nor has having a moderate candidate who himself strongly supported it (being one of the few to back putting the Smith bill directly to Parliament).  The result has strengthened Malcolm Turnbull's grip on the leadership, which was very shaky just weeks ago.

The Australian Conservatives have outpolled the Christian Democrats, but have nonetheless done nothing remarkable here, barely getting their deposit back.  These parties will need to look at merging or running a joint ticket lest they cannibalise each other's vote under the new Senate system with its weak preference flows.  The Greens have suffered from relevance deprivation.

9:03 With the booths all in, turnout is 73.25%.  There's about another 15% to come when the postals are added, so we may make it to the high 80s.  The informal vote is currently running higher than I thought at 7.4%, though based on the New England example this will drop somewhat with checking and with postals.

8:47 This is a bad result for conservative rebels in the Coalition. They will not get any oxygen from this.  They didn't really get any from Queensland either, largely because it took so long for the result to be clear.

8:40 The two big PPVCs are in and haven't made much difference to the primary count.  They might not have such a strong preference flow for the Liberals though.

8:26 No real change with the projected 2PP sitting around 55-45.  Prepolls and postals may shift this, but probably not by more than a couple of points.

7:51 As the ABC is pointing out, the Coalition can now, if it wants, engage in partisan referrals to the Court of Disputed Returns.  Alexander's 2PP is improving, and I wouldn't rule out that he could get into what I call the "good" result zone, which is 56-58.

7:46 Nothing moving here, I think this can safely be called.  The result is currently in the "meh" zone where it can be spun whatever way you like but does not tell us anything we don't already know.

7:40 Tony Burke suggests this by-election is different because we've never had one where a government could lose its majority.  Federally that's true, but Queensland had one of those at state level and threw the majority government out!

7:37 The Conservatives are on 4.7% to the CDP on 3.0%, which is not as decisive as polls predicted.

7:32 The projected swing at the AEC is running ahead of the ABC and has increased to 5.2% with 9% counted.  That's still not enough, but it's getting closer.

7:29 This might be a good night for ReachTEL and a bad night for Galaxy/Newspoll.  That's a change if so.

7:23 As far as I can tell, the ABC two-party projection is not booth matched, but they are treating it as if they are.  I don't have a by-booth feed yet, but it looks like things are going unexpectedly well for Alexander because of a strong flow of preferences from the CDP and Conservatives.

7:10 After four booths the AEC has a seven point primary swing, but also some swing away from the Greens and to the Christian right parties on primaries.  So this is not a great start for Keneally in terms of winning chances when we get to preferences.

7:06 The ABC is showing booth differences that seem to now be from the total for the last election (ie not booth matched).  The AEC swings appear to be booth matched.  The swing in the first few booths wasn't large enough for Keneally, but Tony Burke has reported that at the Carlingford booth they have a double-digit swing on primaries.

6:55 We have votes! However it's rather confusing because this is a new booth location for the Ryde booth, and the ABC is showing swing figures that don't match the previous Ryde figures.

6:30 It's worth keeping an eye on certain clusters of booths based on their patterns last time.  Labor actually got a swing in 2016 in the Ryde booths (other than Middle Ryde), but the Eastwood booth had a double-digit swing to Alexander. Trent Zimmerman (Lib) has just mentioned that the Eastwood booth's turnout has roughly halved.  This might be a sign of poor turnout generally.

6:10: Not expecting anything too substantial before about 7 pm.  Most of the booths attracted more than 1000 votes last election.

6 pm: Well, here we go, though it will be a while before we get the first booth in.  Something else to mention about the stakes of this one: if Labor wins, it gets the numbers to act with the crossbench to bulk-refer MPs to the High Court over Section 44.  As the Coalition is much more likely to lose any by-elections thus caused, that could be the slow-burn path to a mid-term change of government.  But this is all several hypotheticals away, for now.


Welcome to my live comments page for today's Bennelong by-election, including the post-count for as long as it remains of interest.  As usual this is intended as a more forensic companion to the coverage on ABC News Live, and there will be similar live comment threads at Poll Bludger, The Tally Room and other places.  Comments here will start from 6 pm and will scroll with the most recent on top of the page, refresh now and then for the most recent comments.

If this seat falls, the Turnbull Government loses its majority, although it would be unlikely to collapse by reason of that happening alone.  The government very easily saw off a similar situation in New England two weeks ago but that was against weak opposition.  Polling and historic evidence both suggest Alexander will probably be narrowly returned (see my preview comments, including poll results) and the betting markets have him favourite at around $1.33.  In my view, that's a bit short. There are a lot of strange factors at work making this one a singular contest.  I can't find any objective reason to say Keneally will win, but I can see lots of subjective ones why she might.

The by-election is also an important test for Cory Bernardi's Australian Conservatives, both in terms of the potential for the party to succeed in New South Wales (where it is anecdotally attractive to disgruntled Abbott fans) and in terms of its potential to supplant or absorb Fred Nile's Christian Democrats.  Bennelong is an excellent test seat for this as the Christian Democrats polled very well (over 6%) at the 2016 election.  Bernardi's party is eyeing off One Nation's Senate seat in NSW and we may find out whether this is a realistic threat.

In terms of some basic statistics for the election, enrolment is 106,582.  However, recent urban by-elections have rarely had turnouts much above 80% even when both major parties have contested.  Perhaps the intense focus on this by-election will drive a higher turnout, but I'm expecting at least 10,000 and possibly 15,000 or more to not show up.  I'm also expecting about 5% of votes cast to be informal.

13,387 voters have pre-polled.  Nearly all of these were at Epping and West Ryde Prepoll Voting Centres (6848 and 6275).  There have been 16,210 postal vote applications of which 439 have been withdrawn as duplicates.  Of the postals, 45% are Liberal "party postals" and 7% are Labor's, compared to the 2016 election at which 19% of Bennelong postals were Liberal "party postals" and insignificant numbers were anyone else's.  A "party postal" is an application submitted on material sent out by a party (for instance a flier for a candidate might include a postal vote application form).  The voter who uses a party postal to apply won't necessarily vote for that party, but is believed to be more likely to.  In 2016, postals in Bennelong were 4.3 points better for the Liberals than booth votes.  As that gap might in theory increase, and as there are no on-the-day absent votes in by-elections, Labor will need a solid lead on the night for the seat to be callable this evening.


  1. I reckon Malcolm Turnbull making this by-election *mean* something (the comment regarding it being a reflection of HIS government) might have nullified, to a degree, the usual by-election backlash against the Government.

    To my knowledge, that comes about the feeling (of the electorate) that they can punish the Government without destroying it. Turnbull raising the stakes might nullify that.

  2. The Bennelong result doesn’t really give us many pointers to the next federal election. It’s already bleedingly obvious that the LNP has a hercuelean task ahead of it to retain government – both electoral history and electoral mathematics are stacked against them. Inasmuch as this result tells us anything though, it’s that the governments position may not be completely irretrievable. If they hang in and don’t self destruct, with a bit of good fortune coming their way, it indicates there is just a possibility we could witness a rerun of 2004.

    The ALP should have achieved an 8-10% swing in the circumstances. A government that changes leaders, then just scraped in at the subsequent election and has a leader with Turnbull’s netstat, should be swinging in the breeze, just waiting to be cut loose. That they didn’t exposes the flaws in Shorten’s leadership and judgement.

    The ALP needed just needed an everyday candidate that could say, “Hi I’m a hardwoking member of the local community, I feel your frustrations etc, but hey, don’t focus on me, take the opportunity to send a message to the government”.

    But no, what do they do? Dredge up a fly-in candidate from the other side of the city whose husband was the mayor of Botany until just last year. Not only that, but she was associated and intimately connected with one of the most ugly periods in NSW political history and led the ALP to its worst defeat in history. Just as NSW Labour was tentatively showing signs of green shoots, they bring in a high profile candidate to remind everyone of the history-and in an electorate that showed one of the highest swings in the state in 2011.

    Aside from that, there is the the issue of the governments precarious majority. By ratcheting up the importance of the result with a high-profile candidate, they forced people to think more carefully about their vote. More carefully about lodging a normal by-election protest vote.

    The smart strategy would have been to play it down, keep it as low key as possible, then have a shocking upset on the night.

  3. Minor point but the TPP in Epping Heights showing on Sunday (62-38), that is a swing to the Liberals, simply looks miscalculated by about 6%.

    1. Looks like it has been corrected by the amount you expected, now showing 55.6-44.4.

  4. As I said on Tallyroom, the biggest swings tended to be in the booths along the edges of the seat (in all directions).

    In 2016, the Absent vote was significantly better for Labor than the Ordinary votes. There's no absents at by-elections, so perhaps the big swings in these booths is simply due to the Absent vote being absorbed into the booth vote?

    1. It would be interesting if that geographic pattern happened consistently at other by-elections.