Saturday, May 2, 2015

Legislative Council 2015: Windermere, Mersey and Derwent Live And Post-Count

Derwent: CALLED (6:45 pm): Craig Farrell (ALP) re-elected
Mersey: CALLED (6:47 pm): Mike Gaffney (IND) re-elected
Windermere: CALLED (9 pm): Ivan Dean (IND) re-elected on preferences


Tuesday night: There was quite a lot of silly spinning on election night from various political staffers and so on on Twitter, but not much has made it into wider circulation.  One example that has is this Greens' on-the-night press release, which really has me scratching my head as I try to find a single claim in it that's true.

The first problem with it is the sheer number of errors in the claimed swings, which are actually meant to be from the 2014 state election, although in one place it says 2010.  The press release claims gains in 15 booths and swings against in 1 (I make it actually 11 gains and 5 losses).

Of the five booths where they claim a swing from "equal" to 0.5 points to them, they were down on the 2014 election at Dilston (-1%), Hillwood (-2.5), Inveresk (-0.6), Waverley (-0.3).

Of the five where they claim a 0.5-1.5% increase, they were up 1.6 in Invermay.

Then I get different figures for all the following where they claim a precise swing - George Town "1.9" (I get 2.1), Mowbray "2" (2.3), Norwood "4.6" (4.8), St Leonards "2.9" (3.3), Weymouth "9" (11.2) and Elphin -"0.5" (-0.7).  There have been some small changes to booth figures since the night, but they don't explain any of this. There seems to be some irregular method at play in their calculations that is flattening out the actual swings, but not consistently.

There is also the claim that Bleyer "managed to force the incumbent of 12 years to preferences".  As it turned out, after Bleyer's preferences were distributed, Dean still hadn't crossed 50%, so even had Bleyer not run he would still have needed some preferences from McLean to cross the line.

Even the press release's prediction that the seat would be decided on Green preferences turned out to be false.  848 Green preferences went to McLean or Dean.  Even if every one that went to McLean then went to Dean, then the claim would still not be true, since Dean is presently winning by 1958, a margin which won't change much.  Switching all 848 from Dean to Houston would still leave Dean the winner by 262.  Had the flow from McLean to Dean been weaker then Green preferences would have co-determined the winner, but it wasn't, so they didn't.

The claim that the same votes for a Lower House seat would see Bleyer elected is also dubious.  It is true the Green vote in the electorate increased slightly on the 2014 vote that saw Kim Booth re-elected, but the fact is that Booth was somewhat lucky to win because of the distribution of votes between other parties.  A one-point swing back to the Greens will not necessarily keep their seat in Bass next election if the swing back to Labor is greater.

Tuesday: As usual check out Ben Raue's excellent electorate maps.

Monday: Not much more to add here as postals continue to trickle in with the exact margins to be known next week.  However it has been revealed that Labor made 10,000 robocalls in Windermere - almost one for every two enrolled voters.  They don't seem to have had a lot of impact.  There is a lot of talk about the turnout - note that there are still votes coming in so the current totals of 77.7% in Windermere, 78.6% in Derwent and 82.3% in Mersey will go up by maybe a few points each.

Sunday 5:15 pm: Daniel McCullough (Examiner) has reported Dean has won with a 2PP of 55.7%, a swing to him at present of 0.7% on the 2009 result (this margin may go up or down by a few tenths as remaining votes come in).  That sounds like the split of the McLean preferences was about 51-49 to Labor, which was exactly what the scrutineering sample passed to me last night said (so well done those responsible).   In the end a comfortable win for Ivan Dean but an acceptable result for Labor who have no cause to regret running against him.  Update: confirmed, it's 55.70% 2PP for Dean for now.

Another six years for Ivan Dean who I greatly doubt will continue past 2021, so this should be an open contest next time.  Next year: Tania Rattray will be up for a third term in Apsley after being re-elected unopposed last time.  Adriana Taylor is eligible for a second term in Elwick after beating Labor fairly easily in 2010.

Sunday 5 pm: We've had a provisional exclusion of Bleyer and her preferences split 54.6% to Houston, 23.6% to McLean and 21.8% to Dean - slightly stronger than the sample passed on to me yesterday, but that isn't surprising.  Dean now needs a mere 19.2% of McLean's preferences to get over the line.  (These figures will vary slightly with extra postals and so on still to be added - for instance if Dean only just gets the required amount now then we will have to wait for postals to trickle in to be sure of the result.  But he should get the required flow rather comfortably.)

Sunday 3:22 pm: A new set of rechecked primary figures were just posted, with 35 provisional and 96 out of division votes added.

Sunday 2:45 pm: Primaries for Windermere are being rechecked, with trivial changes.  I will be out a bit this afternoon so updates will be slow. Official results are here (refresh now and then to see if there's been any preference throwing.)

Sunday 1pm: I expect the primary votes received so far are being checked and we could have a provisional preference distribution later today or tomorrow.  With large gaps between the candidates it should be possible to determine the election without having to wait for all the postals.

Sunday: Windermere Analysis Windermere is possibly my favourite Legislative Council electorate to have a close look at.  Here is a lot more detail on it:

This spreadsheet gives the results for 2009 and 2015 and the swings by booth:

There were very large swings to Dean at Dilston and Hillwood. These were booths that were in the proposed Gunns pulp mill "impact zone" and that in 2009 voted very strongly for Peter Whish-Wilson and the anti-mill independent Ted Sands.  Both the Greens and others went backwards bigtime in these small booths at this election.  Mostly, these were not Green booths to begin with - they were conservative booths that were annoyed about the pulp mill.

Swings against Dean occurred in some of the working-class booths like George Town and Waverley. (Dean did well in blue-collar booths in 2009 apparently because of the pulp mill, so this makes sense.) Ravenswood is an exception only because it was Kathryn Hay's home base, which would have depleted Dean's vote there in 2009.  Labor got swings to them compared with Hay's result in all but four booths - Ravenswood being one of the exceptions (for the reason just mentioned) and two of the others being quite conservative booths where an endorsed and very working-class-sounding ALP candidate wouldn't have played too well.

A big question here is how much damage McLean did to the Labor vote.  If we look at the comparison with Hay's vote as the "Independent Labor" candidate in 2009 the pattern is very obvious:

There is a strong linear relationship between the two:

McLean% = .397*Hay%(2009)+5.91 

This explains 51% of the variation in the McLean vote.  The obvious outlier is, again, Ravenswood, where Hay's vote was inflated by it being her own booth.  In comparison there is no relationship between the McLean vote and the 2009 Dean vote, and a weak negative relationship between the McLean vote and the other 2009 results.

Nobody at all was fooled by Kathryn Hay's "independent" status in 2009 (her booth results explain 88% of variation in Labor's 2010 state election results and 92% for 2014) so the appearance of the above may be that McLean's vote came mostly at Labor's expense.

But if we compare Labor's 2014 state vote with McLean's result the pattern is slightly different:

It's still a strong relationship with 50% explained, but the slope of it is weaker:

McLean% = .2266*Labor%(2014) + 9.55

The higher constant term quite likely means that many of the votes McLean took from the 2009 Labor vote were already lost to the party by 2014.

The suggestion is that Labor lost about seven points from its 2014 election vote to McLean, so that had he not run the Labor vote would have been about 36%.  That would be a swing of about 5% from the state election.  On the downside, close to half that would have been reclaimed from PUP.  On the upside, running as a major-party-endorsed candidate in an upper house election is a disadvantage, and Labor's performance relative to their state vote here is much better than the Liberals' in Rosevears and Huon last year.


Wrapup (8:45) I have had some scrutineering figures based on moderate sample sizes over several booths that indicate that the split from the Greens to Labor is not very strong (63%) and there is very little happening off McLean.  No great surprises there; scrutineering sampling is hard and the estimates might be out, but they are enough to say Dean won't be caught.

Well that was another wonderfully fast count by the TEC. This isn't necessarily the end of my comments for the night, but it is the end of tonight's primary counting, and I'm not sure there will be too much more to add this evening.  So what have we seen?

In Mersey a popular incumbent was attacked over his voting record but again this sort of attack had no impact, in part because the opponent was a last-minute entry from well outside the electorate.  Gaffney won more than two-thirds of the vote in every single booth.  The margin was about to be expected and there is really nothing to learn here that we did not know already.

In Derwent the ALP easily retained a long-term safe seat, with the results showing that the percentage of voters who will not vote ALP under any circumstances is not that high, not even in the very worst rural booths, so there are certainly good signs of recovery.  All the same the result is nothing special given the low-profile opposition, and given that in its glory days Labor used to get well into the seventies in this electorate fairly often, so there is still a long way to go.

I should also say a word for the good result for Farrell's opponent Alan Baker: we have seen low-profile challengers do much worse in these elections.  While very inexperienced I think he's impressed people with the honesty of his comments.  He even won a booth!

In Windermere it looks like 70 is at least the new 64, with Ivan Dean on current figures (subject to checking) having apparently withstood the intense but sometimes wayward ALP attacks on him to be the oldest LegCo winner since 1965. Absolute confirmation should follow over coming days as preferences are distributed.  Ivan Dean has picked up a 4.8% swing to him on primaries, with a 5.5% swing away from the Greens. The latter was to be expected since they no longer had the Tamar Valley mill as a live issue, had a lower profile candidate, and spent much of the campaign fishing unsuccessfully for the worried-about-drugs vote that would surely never stray from Ivan Dean. (It's still a marginally better result for them than the state election.) Overall the result has come out rather similar to 2009, but there are some interesting booth-by-booth patterns in the wake of the demise of the pulp mill as an issue.  I may cover these more closely over coming days.

We're yet to see what the final 2PP will be - based on patterns from 2009 it would be about 56:44 to Dean, but maybe it will be a bit less than that (it was 55:45 last time).  I think Labor would consider a 2% swing to them from Hay's primary vote in 2009 (especially with an ex-Labor independent in the mix, and Hay higher profile and not a party candidate) to be at least an acceptable result.  Given the difficulties for Labor especially with voters liking to vote for independents in the upper house, this result seems to support polling that suggests that Labor is rebuilding well from the 2014 state election hiding.

All up,  I think these three elections tell us nothing that is translatable to the Lower House picture and that we did not already know from polling - Labor are rebuilding, the Greens aren't up to much, and the Liberals sat this one out after last year.   The usual takeaway is the difficulty of beating Upper House incumbents, with two of the three failing to attract major opposition.

8:14 Final for Mersey for the night with Gaffney on 75.23%.  Final for Derwent too with Farrell on 64.45%. Wrapup comments for the night to follow soon.

8:12 Postals in in Windermere and we are "FINAL FOR SATURDAY" at 43.99% Dean, 28.6% Houston, 16.62% McLean, 10.79% Bleyer.  Dean has done very well on those and he's now needing 21.9% of preferences.  There will be rechecking tomorrow and I'm probably going to hold off absolutely calling it for now while we are sure that there are not any significant errors in the count, or until someone feeds me some samples.  But it does look a lot like Dean has been returned.

8:03 Newnham is in and there was a small swing away from Dean in that one.  All booths are now in, still waiting for postals.  Dean is on 43.45% to Houston's 28.7%, McLean has come up to 16.71% and Bleyer is on 11.15% (some improvement on earlier, and OK in the circumstances).  Dean currently needs 23.5% of preferences.  I am all but certain he will get that given that he got 41% of Greens' preferences against Hay last time, even with Hay not being a party candidate, and even with the dreaded pulp mill.  But I'd like to see some scrutineering figures to be absolutely sure.

8:00 Gaffney now over 75% in Mersey with all booths counted, prepolls not added yet.

7:53 Waiting for Windermere's largest booth Newnham, and also postals.

7:44 Just about done for Derwent for the night with the Gretna booth apparently abducted by aliens but everything else counted and Farrell sitting just below 65.  I don't think that's a great score for Labor, but it's hardly disastrous either.

7:38 A bad booth for Dean in George Town with a 3.8% swing away from him and 6.7% to Labor.  However, we only have a few booths plus postals/prepolls to go to see what kind of preference flow Dean needs.  Whatever, it's going to be the sort of level he would be very likely to get.

7:30 Ravenswood is in and Dean has topped this booth! Hay won this classic ALP booth last time but she was from the suburb and there has been a 3% swing against the "Labor" vote (counting Hay as Labor).  Dean's position has again improved on my projection and the expected share of preferences required is not going up (it's now at 18%).  He should get this.

7:27 Strong performance by Dean in the Hillwood booth has improved his outlook slightly - he has a 24% swing to him in this booth with an 18% swing away from the Greens.  The Greens won this booth in 2009 partly over the pulp mill and partly because their candidate lived thereabouts.  A bad night for the Greens who are struggling to break double figures.

7:23 Gaffney on the other hand is keeping his lead and it doesn't look like the three-to-one margin is going to change much.  It might even increase slightly.

7:19 Farrell is not getting quite the big swing in the urban booths that he was in the little ones and his projected score has now dropped back to 65, which is about what he is on now.  This might end up being only a so-so margin for Labor though admittedly there has been no need to try all that hard against low-profile opposition.

7:15 A big flurry of booths in and although Dean has won the Labor booth of Waverley, it is the first booth to show a swing against him from 2009.  He's currently on 44.4% and my projection reckons he will finish about there with Houston on about 30, and Dean needing 22% of preferences to win.   If it stays that low he should be OK - preferences usually don't flow strongly in these elections as there are no how-to-vote cards.  But 80% flows have sometimes been seen.

7:00 Well this is starting to look a bit more interesting in Windermere now.  The swing to Dean in the Rocherlea booth, another working-class booth, was very small.  His projected total dropped again to 45% and his projected required share of preferences rose to 20%.  That's still OK, but as we get more of these Labor booths that share might rise to something that we won't know the result of tonight.

6:57 A not-so-great result for Dean in the George Town South booth where the swing to him is only a few percent.  Although Dean's primary is currently on nearly 50% I project him to drop to 46% (which would still be enough as it would leave him needing only 17% of preferences to win.)  But this is not over yet and we have to keep an eye on where the gap between Dean and Houston ends up (seems no doubt they will be the final two.)

6:55 The remaining interest in Derwent and Mersey is bragging rights for the easiest win.  My projection has Gaffney just ahead of Farrell at the moment.

6:50 A couple of unrepresentatively strong Dean booths have just been added in Windermere.  He is currently tracking for 48% to Houston's 29%.  It's not over yet (we'll need to see if the swings stay constant into the Labor booths) but this is a really good start for Ivan Dean.

6:47 More booths in in Mersey and this is beyond any doubt with Gaffney tracking for 76% and could win every booth.

6:45 More booths have come in in Derwent and Farrell has over twice the required swing in every one and is tracking for low 70s.  I see no reason to doubt he has won the seat.

6:40 Gaffney wins the Forth booth with a 29% swing so he is now tracking for 72%.

6:36 Great news for Dean as he wins another booth (Pipers River) with a 10% swing.  If the swings in these small booths hold up he will barely be taken to preferences - but will they?  It is looking like the McLean challenge isn't amounting to anything and the Greens now have a swing against them.

6:34  Two more small booths in Derwent have Farrell still projecting at 75%.  I'd like to see a couple of booths that aren't small rural booths and then I think this one can be called.

6:32:  A smaller swing in the Glenora booth has knocked Farrell's projection down to 75%.

6:30 pm:  First booth in in Windermere and it's a great start for Ivan Dean with a swing of about 10% in the Weymouth booth.  If that swing holds in other booths he will be fine.  McLean polling poorly in this booth in last position in it.

6:29 pm:  Gaffney wins Mersey Vale booth 100-49.  The projection off this booth is 75% for Gaffney.

6:27 pm:  Two booths in from Derwent and Baker has won the Bronte booth by a vote (!) but more significantly Farrell has easily won Ellendale which has been Labor's single worst booth in the electorate lately.  The swings are 30% and 56%.  The projection off these booths is 85% for Farrell but that is not reliable yet.

6:25 pm: A pic from counting shows Farrell with a bigger pile in Derwent but presumably from one of the major booths.

6:20 pm: Nothing yet but it shouldn't be long now!

6:06 pm: TEC results page is here - there is currently an issue with the Mersey link going to results for Windermere but I'd expect that would be sorted out soon. The direct Mersey link is here.

6 pm: Polls are just closing now and small booths in Derwent could start coming in within 20-30 minutes based on past years.

5:15 pm: Polls close in 45 minutes, so if you haven't voted yet make sure you get inside the booth by 6 pm or you won't be allowed to vote.  A note on pre-election voting: the ABC here notes that "over 10,000" voters had already voted before today.  The total of all pre-polls and postals in these three electorates in previous elections (including informals) was 8167, and I'd assume the total excludes some postals yet to be received.  Based on last year's count we may well get pre-polls and already received postals counted on the night, but we'll see how this goes.

If there is a preference distribution for Windermere it could take place within the next few days, but there is also some possibility of it stalling partway through until enough postals are in to break a close exclusion point.

3:15pm: Just a note about Derwent booth-matching for tonight.  Some of the early booths to be counted should include small rural and remote booths like Bronte, Ellendale, Gretna, Hamilton, Tea Tree and Ouse.  These booths had extremely low Labor votes in 2011 and again at the 2014 state election.  So a few funny things could happen with them.  The first is that Labor might actually lose some and could even trail in early counting - do not adjust your set should this occur, since it won't last.  The second is that we could see extremely large swings (potentially over 50%) back to Farrell in some of these booths.  If this happens, it won't necessarily hold up through the rest of the count. I am using the 2011 results for booth-matching - I considered using 2014 state election results but they are messier in terms of matching the non-booth votes.

Opening Post (midday): Welcome to my live comments on tonight's three Legislative Council contests.  Three of the fifteen seats in Tasmania's upper house will be decided tonight.  You can see my previews and form guides here:

My in-depth statistical analysis of how Legislative Councillors vote can be seen here.

There will be live comments throughout the counts.  Polls are open until 6 pm, first figures may start appearing around 6:30 and based on past experience counting for the night could be done by about 9 pm.  Windermere will very likely go to preferences and I will follow that count over coming days until it is settled.  When I believe a candidate will definitely win the word CALLED will appear for that seat in the summary at the top of the page.

Craig Farrell (Derwent) and Mike Gaffney (Mersey) are facing only low-profile opposition and I very strongly expect them to both win easily, perhaps very easily.  

Ivan Dean's bid for a third term in Windermere is less straightforward - bidding to become the oldest LegCo winner since 1965, he's been heavily targeted by Labor's Jennifer Houston, while high-profile ex-Labor independent Scott McLean has kept out of controversy.  The Greens' Vanessa Bleyer is also running, but the Greens can't win this seat.  It is compulsory to number three spaces for a formal vote, so this is effectively a compulsory-preferencing contest.

As a contest, Windermere could be either a cracker or a fizzer, but either way it will be an interesting outcome, and one I've been looking forward to covering for some time.  If Dean is beaten it will make it rather more difficult for the Hodgman government to get contentious legislation passed.  Incumbents are usually re-elected easily unless they've done a lot wrong, and he might win easily, but there's been no polling, and it's a naturally close electorate.  

For Windermere, I'll be modelling the votes for Dean, Houston and Bleyer off the votes for Dean, Kathryn Hay (Ind Labor) and Whish-Wilson (Green) last time.  This might not work terribly well but it should at least mean the strong Green and strong Labor booths don't distort my impression of where we end up.  

Most of my attention will be on Windermere since I expect the result in the other two to be obvious quickly.  

Once the counts get underway, I'll be posting live comments, scrolling up.  Refresh frequently from 6:30 on for latest updates.

No comments:

Post a Comment

The comment system is unreliable. If you cannot submit comments you can email me a comment (via email link in profile) - email must be entitled: Comment for publication, followed by the name of the article you wish to comment on. Comments are accepted in full or not at all. Comments will be published under the name the email is sent from unless an alias is clearly requested and stated. If you submit a comment which is not accepted within a few days you can also email me and I will check if it has been received.