Sunday, September 8, 2013

2013 Federal Election Late Night Wrap

This is my quick late night summary of where we stand after a more interesting election night than I was actually expecting!  The Coalition has won the election decisively, but it has been a middling win in 2PP terms and only a moderately heavy one in seat terms.  Note that the exact 2PP vote will change from now to the final tally in late counting and may end up being slightly more Coalition-friendly than what we have now.  It was not the rout that many were expecting following the 54:46 Newspoll and Nielsen and the astonishing seat betting plunge in the final days.

As usual a number of seats are in doubt and we will have to see how late counting goes in these to see exactly where Labor finishes up.  Currently 11 specific seats are gone (Bass, Corangamite, LaTrobe, Deakin, Braddon, Robertson, Page, Hindmarsh, Lindsay, Banks and Lyons).  Labor trails but is within striking distance in Dobell, Eden-Monaro, Reid and Petrie.  Labor leads but is not yet home and hosed in Barton, Capricornia, McEwen, Lingiari and Parramatta.  In 52 of its own seats Labor leads by more than 51:49, and indeed the first seat up from there is Coalition-held Solomon (NT), where Labor's internal poll that had them narrowly winning doesn't seem to have actually been far out.  So at the moment it looks like a loss of around 15 seats, but we will need to wait to see how late counting pans out and whether these close seats tend to fall on one side of the ledger or the other.  Usually they'll split fairly evenly.

Adam Bandt has retained Melbourne very comfortably (very late in the piece I marginally suggested him as the most likely crossbencher to win) while in Indi and Fairfax we will have to wait for the AEC to reconstruct the count between the Coalition and the third-party challengers.  Both seats are projected by the ABC as close, but I believe that is based on estimates, not exact figures.   When watching the count, be aware that last time in Denison, booths were reconstructed to the correct two-party order in alphabetical order, which resulted in wild swaying through the count.  (One of the odd features of Denison is that most of the Labor-leaning booths fall in the first half of the alphabet.)   Depending on when these seats are done, I may be able to do some modelling based on what we learned in Denison.  (It's quite likely scrutineers will have good information on preference flow.)

The model I constructed for predicting the outcome expected 52 Labor seats in its final version.  Labor would be ultra-unlucky from here to not win more than that, and won't win less, but we will have to wait a while to see how many more.  The model certainly missed on Greenway (refreshingly, the Western Sydney effect was not enough to cause the election of a clearly clueless and terrible candidate), Kingsford-Smith, Moreton (where it had no direct polling data but its owner made an incorrect judgement call based on Newspoll aggregates) and Lilley (where the JWS poll was way out).   There is doubt about seven other seats projected as Labor losses (with Labor currently ahead in four) and two other seats that were not projected as Labor losses (with Labor currently trailing in one). 

I've deleted a sentence here about 2PP because the 2PP nationally is currently running at over 53.2 and past experience suggests it may rise further and may even go over my final aggregated 53.5.  Whatever it should not be too far from it in either direction.  Models that modelled exclusively off the last election 2PP (and hence had nearly all pollsters leaning to Labor) have overestimated the 2PP, and I suspect this is because the shift from late polling to the election result is a variable aspect of each particular election, and not of the body of pollsters as a whole.  It looks like the seat loss for the given 2PP will be closer to the pendulum than my model expected but how large the difference is will depend on how the final results pan out.  The use of seat poll data doesn't seem to have harmed the model's strike rate (it changed relatively few seats anyway) but it's not clear that it was a big advantage. Probably I should have spent more time on it, especially to check whether the totals panned out better for Labor if I converted the projected margins in each seat to percentage chances. 


ReachTEL polling successfully predicted the winner of all the Tasmanian seats, but was a few points too generous to the Coalition in each case, only picking Franklin by a whisker.  At least in the case of Bass, it doesn't look like this was just late swing.   Labor's plight in Tasmania is extremely unlucky; at present the Liberal Party leads Labor just 50.5:49.5 in the state but Labor has only one seat to show for it.  Worse, the only swing in the state big enough to actually win Lyons fell in Lyons. 

The result was what I expected but I did not expect such a huge result for the Palmer United Party which could now have a significant chance to win a seat or two at state level if it manages to keep its act somewhere near together until then.  That said, voters may be more reluctant to vote for it when they are genuinely threatening majority government by doing so. 

The Senate

The Senate count is a diabolical mess and Tony Abbott could end up working with a crossbench that made the one Julia Gillard had to deal with look tame.  However, it is his fortune that his minority will not be in the house that forms government.  He may also find that he can work with most of the minority players (and if they veto his paid parental leave scheme, that will be a bonus).

For Tasmania there is the usual 2-2 split, then the Greens on Labor preferences and the last seat is between the Liberals and the Palmer United Party.  There is currently a nexus point at which the Liberals, PUP and Liberal Democrats are very close together (although it is harder for the LDP than it looks because of below the line voting) and PUP need to survive that exclusion to win.  It will probably be a few days before it is clearer where this one is going but the incredibly high PUP primary has blown all the other micros out of the water and scuppered the Family First preference snowball.  I did a number of models where I gave PUP 4-4.5% and generally found that they had a lousy preference flow in those, but in this case they look like getting both Labor and Green preferences in surplus.  Clive Palmer said his party was going to surprise people with its performance.  In the Tasmanian Senate he is right; he surprised me.

For the ACT the Greens are currently showing as just losing to the Liberals, but bear in mind the ACT has high below-the-line rates which will hurt the Greens on their way up.  I don't think that the Greens will win.

For the NT the Australian First Nations preference snowball appears to have failed due to inadequate primary, and Labor is very close to quota anyway, so I think that is a 1-1 split.

For SA Nick Xenophon has polled nearly two quotas but it seems his running mate Stirling Griff is going to be high and dry, because on the current votes, Labor and Green preferences (grrrr!) go to Bob Day of Family First ahead of either the Xenophon group (no friends to the Greens of late) or the Liberals.  We'll have to keep an eye on this one to see if the Xenophon group can make more headway in late counting but the colourful calculator outcome of five parties taking six seats looks quite plausible. 

For WA the ABC calculator currently has an individual by the great name of Wayne Dropulich, of the Australian Sports Party, snowballing to win a seat from 0.22% of the vote. In Vic there is also a snowball involving the Motoring Enthusiasts Party from about half a percent. The modelling at the truth seeker blog (which I strongly recommend reading to follow these counts, rather than just relying on the ABC calculator's single scenario of the day) considers the ASP snowball uncertain but the MEP one just about a done deal.

In NSW the fortunate draw of column 1 on the ballot paper, plus confusion with the Liberal Party, has propelled the Liberal Democratic Party (essentially: US-style libertarians) to a big chance of a win off a huge primary vote.  Questions need to be asked about confusion between the party names but also whether the Liberals dropped the ball in terms of educating their followers where to find the Liberal box.  Personally, I think this is hilarious.  Readers would be well familiar with my pet rant on the subject of the Liberal Party's misuse of the term "liberal" to describe itself when it is often not; if a more liberal party is elected at its expense because of confusion, there's a certain irony at work.  But yes, I suppose I should take this seriously, a lot of voters have been fooled by the similarity in names.

In Qld PUP have very likely won a Senate seat off a huge vote.

So to Nick Xenephon and John Madigan it looks like we'll be adding at least one PUP Senator (quite likely two), a Liberal Democrat, possibly Family First, and perhaps two very obscure micros.  But we need to keep a very careful eye on changes in the primary vote

Stay tuned!   More comments over coming days.


Below is archived some strange fanmail I received recently:

Strange Fanmail

Permission to publish the following has been granted:

May I seek your kind permission to submit a pithy comment or two for possible inclusion to those already  in your profile [Ian Rist, etc] on your web page? [all of the below are plagarized--from Sydney Smith, Disraeli, Churchill and Marx-Groucho Marx-- for as aspiring pollies the Rhino Party has no ethics. More are to be found in the most excellent Penguin Book of Insults]






[this is Voltaire, badly paraphrased as I don't speak French]

In case you don't follow the fortunes of the Rhinos due to a vicious campaign by the Mudrok press  now need a swing of 108% to get office.However we will not abandon our principled policy of paying off Australia's foreign debt with an American Distress Card. The party will be re-launched as the Budgie Smuggler Party--seems to have paid dividends for Abbo--and we have a cunning plan to vote harvest using a mob of micro parties: the PALMER TITANIC II PARTY; the FREE BEER PARTY [aimed at the Ocker vote];the RIGHT OF THE SOUP SPOON party[aimed at the WA vote]; the RESETTLE REFFOS IN ANTARCTICA party [western Sydney vote] and the GOEBBELS party [aimed at Mudrok paper readers.] Eat your heart out Australian
Motoring Enthusiast Party

Thanks again for your most interesting page

      (email address removed - KB)
The two most abundant things in the universe:
 Hydrogen and stupidity. ~Harlan Ellison

And there was a follow-up:

 Do run it but no address--Rhinos are quite shy and I'd rather get into power through a coup
[As U may kno the Rhinos started in Canada. The Rhino was chosen as the name/symbol as its got a quick temper, tough hide and is prone to charge off in the wrong direction, thus the obvious symbol for a pollie. As an expat Canuk I was most happy to pinch most of their policies, though so far voters here have remained nonplussed].

If I may be indulged two more political insults:

Churchill-Attlee doesn't travel by train presumably as there is no fourth class

from Henry Clay a member of the American Senate on an opponent-He is brilliant but utterly unscrupulous. Like a dead mackerel at midnight he glistens and stinks in the moonlight.


I like that last one!  The quotes above have not been verified by me.


  1. Not sure if there are figures floting yet, but BTL dropped to about 8% locally.

    1. Ta; that's about what I expected, perhaps slightly at the lower end.

    2. It's a two booth sample, so not exactly definitive.