Saturday, March 13, 2021

WA 2021 Live

WA 2021: Start position Labor 40 Liberal 13 National 6

(Includes Darling Range, notionally Labor occupied by Liberal via by-election, and Geraldton, notionally Liberal but incumbent switched to Nationals).

Labor has won the election overwhelmingly.  Nationals likely to win more seats than Liberals.  

Labor on track to win upper house majority.  Some prospect of micro-party wins.

Expected Labor gains from Liberal (9): 

Bateman, Carine, Darling Range,  Dawesville,  Riverton,  Kalgoorlie,  Hillarys,  Scarborough, South Perth

Expected Labor gains from National (1): Geraldton

Liberal seats at risk to Labor (2): Churchlands, Nedlands - both currently likely to fall

Nationals seats at risk to Labor (1): NW Central -  currently likely to fall

Latest updates appear below the line scrolling to the top.  Refresh every 10 mins or so once counting is well underway for the most recent comments.


------------------------------------------------------------------------------
Donations welcome!

If you find my coverage useful please consider donating to support the large amount of time I spend working on this site.  Donations can be made by the Paypal button in the sidebar or email me via the address in my profile for my account details.  Please only donate if you are sure you can afford to do so.
------------------------------------------------------------------------------

-------------------------------------------------------------------------------------------------------------------


Brief note Sunday 2 pm: I haven't been able to get postcount threads up yet because of being busy preparing for work trip, but hope to do so tonight.  What is notable is that in Mining and Pastoral, the Daylight Savings Party are currently extremely close to winning via preference snowball from only 0.17% of the vote (almost last), and seem a very real chance to get a seat.  

End of night wrap: I could go on but have been up for over 20 hours on 2 hours sleep, so I think I will leave it here for tonight.  What we have seen is a victory as massive as the most lopsided polling predicted, and perhaps even then some, with the PollBludger swingometer currently projecting a ridiculous 69.2% 2PP for the McGowan government.  The Liberal Party is clinging to a faint hope of tying the Nationals' seat count, but might even be reduced to just Cottesloe and Vasse.    Opposing a popular first term government while one's own party is in power federally is a tough gig for any Opposition, and a swing to the government would not have surprised in any case.  However, the WA Liberals have been unable to find adequate leadership and have made terrible blunders, most critically on COVID-19 border-related issues and by trying to compete for Greens votes.  Their overrated strategy of trying to discourage voters from giving the government "total control" appears to have failed, in my view because people currently want clear leadership that will take command of the COVID issue and the post-COVID economy.  Even so, to actually see the extent of the predicted mayhem has been fairly amazing.  At one point the Liberal seats were becoming callable as lost so quickly I could not keep up with them.  

Credit should go to YouGov for issuing some brave polls that have proved to be pretty accurate pointers to the scale of the bloodbath that has been inflicted on the conservative parties.  This is especially noteworthy given that so few pollsters polled at this election at all.  

The Upper House picture is still developing, and will continue to do so over coming weeks, but for the time being it appears that Labor should win a majority.  No micro-parties are currently on track to win off tiny vote shares, but at times Legalise Cannabis and Australian Christians have been notionally on for seats off a few percent of the vote.  

Thanks all for the interest, there will be more coverage on Sunday.

---------------------------------------------------------------------------------------------------------------------

10:16 In East Metro the Greens have a second path to victory which is to get over Labor at the point where they are currently eliminated.  

9:54 The ABC has results calculators up now.  Remember these are only a model and are also relying on live numbers instead of attempting to project final vote totals.  The calculator gap between the Greens and Legalise Cannabis in East Metro has already closed to 0.4%; I suspect the Greens would actually win on current numbers because of under-counted BTL votes.  The other LC calculator win for Sophia Moermond in South-West is not guaranteed either, but the lead there over the Greens still exceeds 1%.

9:39 The Aus Christians seat that was possible in South Metro earlier has now dropped off, to the benefit of the Greens, who are now on for two seats in the calculator.  This is looking like a rerun of the 2017 election where there would be the odd undeserved win by a party polling a few percent, but no completely ridiculous wins off fractions of 1%.

9:17 With 13.9% counted in East Metro I now get 4-1 and Legalise Cannabis so that's the second seat for them.  On current numbers they are getting over the Greens by 1%.  However, the Greens have a swing approaching 3% against them in the upper house but only 1.2% in the lower in the same electorate.  So this suggests the Greens will rise and may get up in this seat.  

9:11 The ABC are now projecting two each of Australian Christians and Legalise Cannabis to my one, presumably off the back of changing figures.  They are not posting live result calculators yet.  They are projecting 22 Labor seats but I am unsure what the basis for this is.

9:02 The Liberals have got off the canvas in Nedlands and are back in that one, so it is still possible they can manage four.  

8:54 William Bowe has provided booth-matched swings that show that in the lower house Labor has swings of close to or above an Upper House quota everywhere, and way over 1.6 quotas for Mining and Pastoral.  It looks to me like if anything like this occurs in the Upper House, Labor wins a majority with at least 20 seats, probably more.  

8:44 South West on an initial attempt off 5% counted I got 3 Labor 1 Liberal 1 National and 1 Legalize Cannabis, though the latter was by less than 1% over the Greens.  This could yet fall over but the dope party is on 2.7%.  For Agricultural the Health Australia spiral is currently failing. For Mining and Pastoral I thought a result with Labor on 4, Liberal 1 and the Shooters eliminating the Nationals was unrepresentative but it is actually fully consistent with the booth matched swings (see above).  

8:25 S Met initial run off 2.3% counted - 4 Labor 1 Liberal 1 Australian Christians, undeservedly elected by preference snowball off 2.6%.  Reasonable margin on the calculator so keep an eye on how this goes later.  

8:17 E Met initial run off 1.9% counted - 4 Labor 1 Liberal 1 Green

8:06 Having a very preliminary look at the upper house - if the button is pressed on North Metro with 0.72% counted and ignoring BTLs then "Liberals for Climate" (aka Flux) win.  Extremely early and heaps of time for this to fall over but we are not yet seeing that micros are hopelessly uncompetitive.  (Update: with 3.9% counted, Flux still notionally win but their lead over the Greens at the end is too small and would fall over on BTLs).   So currently more likely 4 Labor 1 Liberal 1 Green

7:46 Didn't realise I'd gone 40 minutes without a comment, just updating the totals at the top.  Anyway all but four Liberal seats already seem to have fallen.  Carine and Churchlands are both dicey; the Nationals are quite likely to at least win as many seats as the Liberals, if not more than them.

7:06 It's got worse: Riverton appears to be toast.  It is hard for the Liberals to get more than 5 at present.  

6:57 This is a very grim start for the Liberal Party.  They are projecting ahead in Vasse and Cottesloe.  There are four seats with no votes counted (one of which has an adverse scrutineering report for them).  One of those is Hillarys which is very likely to fall on the swings we're seeing so far.  We are already projecting towards the Liberals winning only 4-5 seats, but some of the seats where they are behind on small samples such as Scarborough and Riverton are still competitive.

6:52 Dawesville projecting strongly to Labor though not so much as in the seat poll.

6:51 The trend in Geraldton has turned around with Poll Bludger projecting it to ALP.

6:45 For the rest of the night we're going to play a game called Ten Blue Bottles, which consists of watching the Liberal Party lose seats and seeing which ones it manages to save.  Interestingly one green bottle showed some sign of not falling in the first booth: Geraldton, where Lib-turned-Nat incumbent Ian Blayney gained a large swing in the first booth off the Liberals and One Nation.  

6:37 More booths from various seats coming in and we are so far seeing the kind of massive Liberal to Labor swing implied by Newspoll.  

6:28 Colossal swing (over 20%) in the first booth in ALP "marginal" Burns Beach.  These swings may be a little misleading because of shifting vote forms and redistributions.  Also a massive swing approaching the same size in Thornlie hospital votes.  

6:18 Here we go - 14% swing in the first small booth in the Nationals ultra-safe seat of Central Wheatbelt. And also some hospital votes in Riverton (the Liberals are actually winning those, and the swing to Labor from them is modest but there's no guarantee they've taken votes in the same places.)

6:15 A link is showing on the WAEC page but at present it goes to a file not found page.  However (thanks to @GhostWhoVotes) a working link is here: https://www.elections.wa.gov.au/elections/state/sgelection#/sg2021

6:11: I understand that only above-the-line votes for the Legislative Council will be counted tonight, so if there are close contests expect the Greens vote to go up a little in post-counting while major parties may decline.  

6:08 (WA time): Scanning the horizon for votes or even a results link on the WAEC website; nothing yet.

Intro (Saturday morning)

Just a quick intro post for tonight's count.  I will be offline most of the day today but will have what may be brief or may be detailed live comments here from 6 pm WA time.  The scale of the comments will depend on how much is going on; if things are very settled and counting in the upper house is very slow then I may not do more than a few hours.  I will try to do a fair amount of washup and postcount posting on Sunday, and then after that I expect to wind back to an hour per day in the late afternoons and evenings WA time, as I will have work commitments during the day from March 15 to 24.  Whatever I offer will probably be but a footnote to the coverage available on PollBludger.  

The final Newspoll came in at 66-34, which in my seat model on average gives the Liberals and Nationals four seats each and Labor the rest if accurate.  However there is a lot of possible variation in the Liberal seat outcomes on that figure - they might win only one or two, or they might get into the high single figures.  There is also a YouGov poll of Zak Kirkup's seat of Dawesville which finds him losing 40-60.  It's very unlikely this poll will be wrong enough for him to save the seat; four Queensland polls by the same method had an average 2PP error of just 3.2%.  

The Newspoll might be wrong (the new polling methods have only been tried in one previous election and lopsided elections are hard to get right) but I'd be surprised if it was way, way out - I'm expecting the Labor 2PP to have a 6 in front of it.  This has been an almost perfect storm in terms of the conditions for an election to be as lopsided as possible.  Incidentally one of the most lopsided elections in vote share terms that I've ever seen, Tasmania 2002, comes out at 66-34 on a 2PP estimate I did this week (but in that case bolstered by semi-optional preferencing and the drive to avoid minority government.)

As for the Upper House if the swings in the Newspoll are replicated there (a huge swing to Labor with all of Liberals, Nationals and One Nation smashed) that might be bad news for micro-parties (which is good news for democracy if so); I have added some new simulation results to my Upper House page.  

10 comments:

  1. This comment has been removed by the author.

    ReplyDelete
  2. This looks like a polling success!

    ReplyDelete
    Replies
    1. I would hold off on celebrating until the final tally comes in. If Poll Bludger's 69-31 projection is correct, the error on the 2pp in the final Newspoll will have been as large as it was in 2019 (3%), albeit in the opposite direction. Obviously it didn't change the result this time (which is what Kevin and other psephs have been saying all along - with leads this big, polling error can't save you), but it still doesn't imply that the polling in this election was necessarily a success.

      For comparison to other recent 60+ 2pp landslides: in NSW 2011 (64.2%, the previous record for largest landslide), the Newspoll was off by just 0.1%, while in QLD 2012 (62.8% 2pp), the polls were off by between 0.8% to 2.8%. I'm not sure that the idea that "polling in a lopsided election is difficult" is necessarily true - plotting 2pp margin against size of polling error basically produces no correlation whatsoever.

      (of course, all of the above is invalidated if the final 2pp ends up closer to 67 or 68 than the 69 projected by Poll Bludger)

      Still, knowing the media, "predicted winner wins by quite a bit more than expected" isn't going to show up on their radar as any kind of polling error, so I expect plenty of celebration of the polls (the head of polling at YouGov has already started talking about how this shows they don't herd: https://twitter.com/DrCampbellWhite/status/1370708067105468417). Any thoughts on your end, Kevin?

      Delete
    2. Looks to me that if the Newspoll 2PP does end up 3% or so out the main cause is likely to be preference shifting. On current numbers they have Labor a bit low, the Greens a bit high by a similar amount, the Nats a bit low, the Liberals a bit high, so all that should be close to a wash in 2PP terms and most of the difference between them and the PollBludger 2PP estimate looks like preferences.

      Delete
    3. Interestingly if I use the last-election preference flow estimates by Antony Green (to Labor: 0.832 Green, 0.394 One Nation, 0.466 Others), I actually get a 2pp estimate of 67.2% off the final Newspoll primary votes. Really should have thought of incorporating last-election preference-flow-based 2pp estimates into Meridiem (my model at Armarium Interreta) - I even pointed out how they're on average slightly more accurate than the published 2pp to you in an email but forgot to follow up on it!

      I will also note that if I apply the above pref flow estimates to the current projected primary vote totals, I get a 2pp of 69.1%. All of this suggests that, if there is a 3% error, it might be more due to primary vote error than preference shifting.

      (Note to fellow readers: I'm not saying that the Newspoll did poorly - we have to wait for the final result to pass any kind of judgement on that - but I do want to push back on the idea that the polls did well because they "called a Labor landslide" without any analysis of the difference between the final polling and the election result.

      If Labor ends up winning 67-33 in the 2pp, then yes, the Newspoll did do really well, but if the final result is 69-31 or 70-30, that's an error as big as, if not larger than, the 2019 polling failure.)

      Delete
    4. It seems a significant part of it is the 1.2 point discrepancy between the published 2PP of 66 for the poll and the expected last-election figures of 67.2 (I got this too). This is unlikely to have been caused by rounding and more likely to have been caused by Newspoll's last-election formula differing from ours (it doesn't strictly use the last election for that state). I get the impact of the primary votes being wrong if the preference flows were the same as last time at around 1.4 points (I think the Labor primary is up most of a point since I wrote my comment on March 14). These two combined do not leave much for preference-shifting, about half a point.

      Delete
    5. Based on the data here: https://antonygreen.com.au/final-two-party-preferred-result-for-2021-western-australian-election/

      It looks like the final 2pp will be 69.7% to Labor; thus the final Newspoll will have been 3.7% out on the 2pp which is worse than the 2019 Newspoll (3%).

      On the primary vote, I calculated a combined 10.3% error:
      -Labor overperformed by 2.95%
      -Liberals underperformed by 1.72%
      -Nationals overperformed by 1%
      -Greens underperformed by 2.08%
      -One Nation underperformed by 0.74%
      -Others overperformed by 1.8%

      That's actually worse than the final Newspoll of 2019 (9.9% error on primaries), though in 2019 the polling error afflicted every single pollster in very similar ways (which shouldn't happen unless some pollsters are adjusting their methods to remove outliers, i.e. herding) making it worse from a statistical standpoint.

      Delete
  3. Well, back on Feb 22nd (http://kevinbonham.blogspot.com/2021/02/wa-2021-how-lopsided-do-you-want-it.html ) I said the Libs would fit in a motorbike and sidecar and the Nats in a Barina. Almost coming true, though the Libs might need something bigger if they catch up in Carine.

    ReplyDelete
    Replies
    1. Hi Jack,

      You were spot on with your prediction on Feb22, so I think that Kevin owes you the reverse of a porcupine fish for your spot on predition.

      https://www.abc.net.au/news/2021-03-19/labor-wins-churchlands-as-liberals-reduced-to-two-in-lower-house/100016418

      dedwards

      Delete
  4. There's a path for 5th Labor to beat the Greens in SMET. If the Liberals are excluded before the Liberal Democrats (gap is 0.24%) their prefs push the Lib Dem over both the Greens and Labor. Then Labor only needs to overturn that 0.3% margin and the Greens will be excluded, electing Labor.

    ReplyDelete

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.