Saturday, December 5, 2015

North Sydney Votes Live Comments (and postcount if needed)

North Sydney (Lib 15.9% vs ALP)
Trent Zimmerman (Lib) vs Stephen Ruff (Ind) and Arthur Chesterfield-Evans (Green)
Assessment: CALLED: Zimmerman (Lib) will be elected after preferences. 
Estimated final margin 60:40 if Ruff finishes second, more otherwise.

This is a rather hastily posted North Sydney live comments thread (refresh now and then for new comments) as the count in progress has so far been a bit more interesting than expected, and after seven booths the contest for the seat was still alive!  If it's still alive at the end of the night that won't be a good result for the Liberals (and if they lose it will be a shocker) but let's see if it is or not.  My apologies for not doing more advance coverage of this one; I have simply been too busy with work.

Here are some important things about this count:

1. If the final 2PP contest is between Trent Zimmerman (Libs) and Arthur Chesterfield-Evans (Green) then Zimmerman will win easily.  The interest lies in a possible contest between Zimmerman and the independent Stephen Ruff, who after seven booths is second on primaries with 5% over the Greens (and hence likely to remain there).  I don't expect we'll get a Zimmerman/Ruff 2CP tonight.

2. Although Zimmerman is copping some very large swings in early booths, projecting him down to about a 45% primary based on booth matching, the early booths haven't quite been representative.  They've tended to be booths where Joe Hockey polled especially well.  The swing after seven booths wasn't uniform but was correlated to the Liberal primary vote so far.  My model based off the first seven booths had the projected swing dropping from 16.5% to about 14%, which would leave Zimmerman with about a 47% primary, an easy win.  Since I did that projection the AEC projection of the swing has come down to 14.6%.  Unless the swing goes up a few points, Zimmerman will win.  (It may be 45% would be plenty anyway as it would leave Ruff needing about 85% of preferences, but we don't know that for sure.)

I'll add more comments after dinner!

8:00 pm: The ABC and AEC models have now more or less converged on mine, which still projects a swing of 14.1% for a primary of about 47, leaving Ruff needing something like 92% of preferences (which will be impossible).  I think this is just about callable for Zimmerman, even without detailed knowledge of the preference flows to Ruff, but because of the usual high prepoll vote these days I'm inclined to leave just the tiniest window of token doubt open for the PPVCs later tonight in case something unusual happens.

8:05 pm: All projections are now kicking up towards 48%.  (Mine is 47.6 now.)

8:45 pm: Not much has changed here.  Zimmerman's projections have continued to improve; the ABC now has him just over 48.

8:50 pm: It looks like the AEC has ditched the Zimmerman vs Chesterfield-Evans 2CP.  Whether we will see a Zimmerman vs Ruff 2CP tonight I'm not sure of yet (I have no reliable evidence that we will).  Anyway I can no longer maintain any plausible doubt that Zimmerman will get this.

Sunday 6:40 pm: The AEC is now gradually reconstructing a Zimmerman vs Ruff 2CP.  After six polling booths, Zimmerman has a 61:39 lead.  Using multiple regression against the Zimmerman and Ruff primary votes, I project this on current primaries to come down to 59.7% for Zimmerman, but that doesn't factor in anything for postals, on which Zimmerman's primary position should improve.  This suggests he is headed for a 2CP vs Ruff of probably slightly better than 60:40, maybe 61:39.

Sunday 10:50 pm: The ABC is now projecting 62.4% 2CP for Zimmerman, but since this projection has the Greens on minus one I think I'll stick with what I've got.

Monday 10:30: I've been arguing about projections with Antony over at the Poll Bludger thread, but we're not exactly fighting over sheep stations here.  (My projection off seven booths and taking postals into account is now 61.2 - that's using the booth "swings" rather than multiple regression, which is a little prone to overfitting).

Monday 3:50: Some movement with another three booths added; the ABC projection of Zimmerman's 2CP (complete with "residual" error) is now 61.8 and mine is now 60.8.  Mine may prove conservative because independents don't do very well on postal voting.

Monday 4:30: The ABC is now posting two projections, one based on booth swings and one on preference flows.  After 11 booths these have Zimmerman on 62.1 and 59.7 (I still have 60.8).

Monday 6: With 14 booths plus nearly 2000 postals realigned the ABC projections on 60.6 and 60.2 have pretty much converged with mine which is currently on 61.  Anyway we're not likely to see much change from here so I'll only update if things move significantly.

Tuesday 7: A point isn't really significant but for what it's worth after 26 booths all the projections have gone just below 60 now - the ABC has 59.8 and 59.7 and mine has 59.6.


So, What Does It All Mean?

Unfortunately I didn't post much by way of goalposts for this by-election in advance of it, beyond saying on Twitter and elsewhere that a clearcut result (meaning outcome not in doubt) by the end of the night was a good result for the government while anything else wasn't.  Referring to a "swing" against the 2PP from last time (65.9%) doesn't make a lot of sense given that we are comparing an ALP opponent to a more dangerous Independent opponent in a seat Labor has never won, but if it does come out around that range, then it will probably end up around or slightly lower than the 6% swings against governments that are normal in by-elections where the same parties contest.  This said, Ruff was probably not the most formidable possible independent opponent - yes, he was endorsed by Ted Mack, but Mack's time was a while ago, and Ruff did project a rather scruffy (pun intended) image.

The primary vote swing is attracting most attention, but it is mostly explicable by the loss of Joe Hockey's personal vote and the presence of Ruff as an opponent instead of Labor.  What's most likely happened is that the Liberals have lost votes to Ruff, the Greens have lost votes to Ruff, and Labor voters have scattered between Ruff, the Greens and the gaggle of micro-parties.  The micros have done rather well, with the exception of Palmer United, whose 0.43% and dead last is dire even by the standards of inner-city electorates, where PUP always struggle, and is the sort of vote the Monster Raving Loony Party would be embarrassed by.  Comparing primary votes in an election with 13 candidates to one with 6 candidates is always difficult.

Some may say this result sits uneasily with the Turnbull honeymoon, but the Coalition's current federal polling is not much different to their 2013 election result and so no boost off that should be expected.  There is also a history of large primary vote swings against governments that lose very high-profile members.  Comparisons to cases like the Bradfield by-election when Brendan Nelson retired are misleading as the Coalition were in Opposition at the time.  (As it happens, this primary vote swing is the largest against a government since Labor lost the Canberra by-election in 1995, the largest against the Liberals since Malcolm Fraser quit Wannon in 1983, and to find a larger PV swing against a Liberal government that didn't involve a three-cornered contest with the Nationals we need to go back to Wentworth 1956, when the endorsed Liberal candidate was roughed up and nearly defeated by a bunch of "Independent Liberals").  By the way this seems to be the first time since Bradfield 1952 that an Opposition has piked from a Government seat by-election (and wisely so.  Had Labor contested the whole thing may have become a kick-Shorten exercise and they might have finished third.  Or even fourth.)

We can see why the Liberals saturated the electorate with great expense as they were very much afraid of a local-issues backlash, and the result suggests with good reason.  The superficially bad nature of the primary-vote swing will make for some headlines the government could do without, at a time when the revolution is looking scruffy about the edges thanks to Macfarlane and Brough.  Objectively though, I think the result is actually quite good, but also that it tells us nothing positive about how the government's travelling that we didn't already know.


  1. Nice analysis as usual. Minor correction: that would be Wentworth 1956, not 1960.

  2. I think the absence of Labor in this by election was relatively, as you pointed out their was the chance of them coming third or fourth. Even in the Future the chance of them falling into third, especially behind the Greens in safe Inner-City Liberal electorates [such as North Sydney, Warringah, MacKellar, Bradfield, Wenthworth (NSW), Higgins, Kooying, Goldstein (Vic) and Curtin (WA)] will be heightened and not out of the question.

  3. I was following all the commentary quite closely that night since I was on the evening shift over at my self storage unit. Nothing like listening to politics when you're trying to kill time eh? Haha!