Monday, December 1, 2014

Victorian Election Postcount: Lower House

Seat Total ALP 47 L-NP 38 Green 2 Ind 1
(Green win of Prahran subject to recount)

This post (work in progress) will follow the remaining Lower House seats that are of interest.  In all of them the lead is currently substantial and I do not expect any of the leads to be overturned on late counting, but maybe one or two will.  The most interesting seat is the three-cornered contest in Prahran which has its own post (Postcount: Prahran).

Note that Pollbludger has some figures on number of remaining votes.

In the other seats there is probably a perception that large leads on Saturday night could be pulled down on post-poll counting given the vast number of prepolls and the tendency for postal voting to favour the conservatives.  The flaw in this logic is that many of the postals are already counted so on average the postcounts should be more similar to the main game than usual.  But we are already seeing some surprising differences to 2010, and also that the differences are varying by seat.

I'm sequentially writing up seats as time allows through Monday, so when finished this article will include Melbourne, Frankston, Bentleigh,  and Brunswick.  Brunswick isn't in any doubt, but it might do something interesting anyway.

Morwell has been removed from the list as the Nationals have clearly retained it.
Shepparton has been removed as there were some concerns Suzanna Sheed (Ind) might do spectacularly badly on prepolls but that hasn't happened. She has come down from 53.5% 2PP to 52.9% but with such a mass of early votes (13,000) out of the way there is not enough left to throw to shift that.
South Barwon has been removed as the Liberals have increased their lead on prepolls and clearly won it.
Each seat will have updates posted down the page.  As seats are considered of no further interest they will be moved to the bottom of the pile, which is initially sorted alphabetically.

Frankston (ALP defeats Liberal)

You might think the Coalition should have been slaughtered in Frankston because of disgust with previous sitting member Geoff Shaw.  But it's often turned out it isn't like that: a lot of voters vote on party lines, and even losing a contentious first-term incumbent still means you don't have to worry about the personal vote of an incumbent for the opposite party.  On the night Labor had about a 51.3% 2PP lead in Frankston, normally easily enough, but this has come down to 50.3.

In 2010 the 2PPs for the Coalition were 52.2% booth, 55% postal, 51.4% early, 41% provisional and 45% absent.  What has changed this time is that the early votes, instead of favouring Labor slightly compared with the booth primaries (36.2% Labor, 34% Liberal, 13.4% Shaw, 8% Green) have instead favoured the Liberals (33.3% Labor, 36.5% Liberal, 14.8% Shaw, 8% Green) by about three 2PP points.

As a result Paul Edbrooke's lead over Sean Armistead has come down to a mere 192 votes.  Not panic stations yet for Labor by any means as they would be expected to build the lead up again when absent votes are counted, probably by more than enough than the Liberals would gain on the remaining postals (of which there are less than a thousand)  However, if the absent votes do not deliver, there'll be trouble.  At the moment my projection has Labor by 400 but that's too close to be too confident.

Tuesday 3 pm: Figures for this seat have been frustratingly offline for a while during rechecking but are now partially online.  Armistead has moved to within 60 votes of Edbrooke but this seems to be on the addition of further postals, with no absent votes included yet.  In other words, to be expected - but Edbrooke needs the absents to do what they normally do.

Wednesday 4 pm: Some figures that I assume to be out of electorate prepolls have been added and appear on the ABC site but not yet the VEC site and show Labor's lead out to 161. That makes Labor's position more comfortable as the absents now have to not only favour the Coalition but do so by a larger margin than before. Still not a foregone conclusion but it would be a large surprise for the seat to not fall.

Wednesday 5:30 pm: Labor gains three votes on a batch of a few hundred absents.  I don't know where they are from or whether they're representative; all else being equal Labor should do a bit better than that but at this stage any gain is a good gain.

Thursday 5:20 pm: The lead has narrowed to 135 votes on either postals or prepolls. If 2010 absent vote levels are any guide there's still a four-figure sum of absents remaining, but we don't know the final turnout.

Thursday 10:30 pm: More absents have been counted  and Paul Edbrooke's lead has increased on them as expected.  He is now up by 255 and there cannot be anything significant left to throw here now with formal turnout at 90% counted.  It's been a really good effort by Armistead to get this close to holding this seat but not close enough.  Barring substantial miscounting (which is extremely unlikely), this is now beyond doubt.

Friday 1:09 pm: Further gains to Edbrooke on provisionals, he leads by 331.

Monday 1:05 pm: The distribution of preferences is underway and I'm advised that with only Armistead (Lib), Edbrooke (ALP), Shaw (Ind) and Swain (Grn) left, Armistead is still ahead by 223 votes.  Green preferences will very heavily favour Edbrooke while Shaw preferences won't so heavily favour Armistead, judging from the initial 2PP throw.  It should be noted that while the initial 2PP throw showed an Edbrooke win by 336 votes, there are very rare cases of these initial 2PP throws being wrong by more than that once the full distribution is counted, so it isn't quite all over yet.

Monday 2:50 pm: Preferences held by the Greens (which are not necessarily 1 Green) went 72.4% Labor, 15.5% Liberal and the rest to Shaw.  Edbrooke leads Armistead by 1883 with 5547 Shaw preferences (not all 1 Shaw) to throw, meaning Armistead needs 67% of the Shaw preferences to win.   The 2PP quick throw said he won't quite get there but we wait to find out for sure.

Tuesday 3:10 pm: A scrutineer expects the quick throw to be pretty accurate.

Tuesday 4:00 pm: Big thanks to Jackson Whiteley for the figures,  Paul Edbrooke has won the seat of Frankston by 334 votes (a change of just two from the quick throw) after the preference flow subject to official confirmation.

Bentleigh (ALP defeats Liberal)

This seat delivered the Coalition government in 2010 when they won the post-count after trailing slightly on the day.  Based on a Galaxy 53-47 poll my seat model narrowly tipped it to be retained but this doesn't currently look like happening.  At the time of writing Labor's Nick Staikos leads the Liberal incumbent Elizabeth Miller by 298, with almost all the postals and all but 900 or so prepolls gone.  The main share of votes to come is about 2000 absents.  In 2010 absent votes in this electorate favoured Labor to a trivial degree and split almost exactly the same as booth votes.  In this instance the booth votes were more in Labor's favour and so the absent votes would have to swing substantially to the Liberals despite the booth votes swinging slightly to Labor.  This seems rather unlikely.  Miller is likely to need about 55% 2PP of the remaining votes and that's a very big ask.

Wednesday 4:15pm: Some figures that I assume (etc as for Frankston) have been added showing Labor's lead increasing slightly to 319 and making Miller's task even more difficult.

Thursday 1 pm: Labor's lead has come down to 284 with the addition of some further prepolls or postals.

Thursday 10:40 pm: Absents have continued to favour Labor here and a lead of 357 is easily enough to hold off whatever trickle of votes may be left.

Melbourne (Greens defeat ALP)

This seat was euphorically celebrated by the Greens on election night and there appeared no doubt to anyone (including me) that they had won it.  However they then did very badly on postals and after Sunday checking moved a whopping 111 two-party votes from Sandell (Green) to Kanis (ALP), Sandell's lead was cut to 323.  In 2010, absent votes for this electorate favoured the Greens by five points compared to their booth results, and disfavoured Labor by the same amount on the same basis, while early votes slightly favoured the Greens and Liberals at the expense of Labor.  Remaining postals and provisionals favour Labor but there will not be many of them.  So I expect the Green margin to increase, but we don't yet know this for sure.  (There was also a by-election in this seat in 2012 which I haven't looked at yet.  By-elections are "different" because of the lack of absent voting.)

As at 3:15pm, some early votes have been added and they are 40% Green, 26% Labor, 28% Liberal compared to the booth votes of 42.6%, 30%, 22.4%.  That's not very good for the Greens compared to the 2010 patterns and increases the uncertainty about the seat.  Later I may put up a projection table for it.

4:20 Monday: The early votes added have nonetheless improved Sandell's position slightly (probably because the Labor vote is down too) and she now leads by 506 votes with 50.94% two-party preferred.

1:00 Tuesday: Sandell has made further gains and now leads by 681.`

Later Tuesday: I thought an update on this article had gone through but it hadn't.  Kanis has conceded the seat on Facebook.  Conceding makes no difference to the count process even if the candidate who has conceded later wins (which happens in very rare cases).   It's just a nice thing to do when there is no longer any reason to hope you can win.

Brunswick (ALP defeats Greens)

You may think I'm taking leave of my senses to even mention a seat where Labor currently leads with 53.9% 2PP.  And indeed I do not doubt that they will win.  But I mention it because in 2010 there was a colossal swing in Brunswick from the booth vote (55.8% to Labor) to the final result (53.5%), based on early votes splitting 52-48 to the Greens and absent votes 54.7:45.3 to them.  Swings of this size (2.3%) from booth totals to final results are very rare.   The 2010 swing was noticed by Michael McCarthy on election night.

In 2014 there are a lot more prepolls, and the Labor booth vote was lower (I didn't get the exact figure before they added postals; think it was about 53.5).  If the differences between prepolls and booth votes, and between absents and booth votes are repeated in 2014, Labor's current 1900 vote margin projects to be cut to something like 100-200.  But almost certainly those patterns won't be repeated; because of the increased number of prepolls, the prepoll voters are likely to be more representative of the electorate as a whole and hence the swing on them lower.  (The Prahran count is an example of this.)  The seat may end up with a very modest Labor margin and I think is worth keeping half an eye on.

Late Tuesday: I thought an update had gone through on this seat but it hasn't.  As I expected the absent votes have shown a lesser swing compared to booth votes so far than in 2010; indeed, so far they have shown no swing compared to the booth votes  at all (53.8% to Labor).  That might come down with out of electorate prepolls but confirms that the Greens won't be able to make this super-close, though it may still come down to 52ish.


  1. TPP count in Melbourne has been updated, 0.3% shift in Greens favour.

  2. Something to note with Victorian elections, with their late November timing (after the end of uni and Year 12 exams, but not in school holidays) is that the demographics of early voters are skewed younger (and hence more favourable to Labor, and especially the Greens) than would be the case at a federal election, or state elections in other states. Of course a comparison with recent Victorian elections is valid as 2002-10 were also held at this time of year.