Saturday, March 17, 2018

2018 South Australian Election Night Live


From base of notional 25-19-3 seat distribution:
Apparent Liberal gain: King
Possible Labor gain: Mawson (in some doubt)
In doubt: Adelaide (Lib ahead), Newland (Lib ahead)
Outcome: Liberals have won, almost certainly in majority, most likely result 25-19-3 or 26-18-3.

Note: I will try to clear comments tonight but don't expect replies

Updates (Refresh for latest)

11:42: The ABC has turned off the projection, meaning that some seats that were projected as easy wins have dropped back into in-doubt status.  Of these Adelaide is the closest with a current real margin of just 67 votes.  Here in 2014, booth votes went 51.6% to Liberal but declaration votes went 54.5% to Liberal.  So while the seat can be projected to about 51-49 we have to wait and see if the pattern repeats.  Some doubt can also be entertained about Newland, where we have a 50.9-49.1 margin on booth votes, but in 2014, declaration votes were overall weaker for the Liberals than booth votes, and this time there will be a greater proportion of declaration votes that are prepolls.  Note that in each of these cases I am not necessarily matching exactly the same voting areas, which creates some further issues.

10:51: More serious doubt about Mawson now, dropping to 50.4% on the projection.

10:48: Heysen is clearly a Liberal win now whatever the exclusion order so that's 25 for the Liberals

10:30 Current upper house figures in quotas - Lib 3.79 ALP 3.53 SAB 2.32 GRN 0.68 CON 0.43.  Without much doubt that would be 4-4-2-1 if it stayed that way, but need to keep an eye on whether Labor drop back significantly below the Conservatives.

9:48 Labor's lead in Mawson has been chopped in half so that one is in doubt.  On the Liberal list, Adelaide with an actual count of 51.4% and a projected 50.6% hasn't been called into question but is worth keeping an eye on.

9:34 With the Liberal primary tracking towards 38% it is now reasonable to start asking why both major polls had the Liberals on only 34% (albeit in the ReachTEL case without redistributing undecided).  This is starting to look like a statistically significant failure.

9:32 Tom Kenyon has conceded Newland.

9:30 A quick look at the Legislative Council.  With 16.1% counted the Liberals are on 31.8%, Labor on 28.8, SA-BEST 19.2, Green 5.7, AC 3.6, LDP 2.5, AJP 2.2, Dignity 2.  On those numbers the Liberals would win four seats, Labor three, SA-BEST 2, the Greens 1 and there would be a drawn-out contest for the last one between Labor, SA-BEST and the Conservatives (who have polled poorly).  Based on Senate patterns Labor would have a good chance but perhaps this would be different this time. Let's see where the numbers land with a lot more counting.

9:17 SA-BEST have fallen behind on the two-party count in Heysen anyway.

9:14 At a very rough estimate based on the classic seats, I have the 2PP swing at about 1.1% to Labor.  This seems to have delivered no net seats because much of it is being wasted in ultra-safe seats like Croydon, Elizabeth, Kaurna and Playford.  It suggests strongly that preferences shifted to Labor's detriment and that the SA-BEST flow could well be below 50-50.

9:03 This is only the seventh time in the last 49 years that an Opposition has won government while the same party is in power in Canberra.  For the previous examples see What Kills State Governments: Age Or Canberra.

9:02 I have not called a Liberal majority yet because of doubt about Newland but I expect the Liberals will win it.

8:57 Looking at Heysen, currently the gap between SA-BEST and Labor is 5.3 points and the Green vote is 12.2%.  Ignoring other parties, that would require Labor to gain at .42 votes/Green preference.  Looking at the 2016 federal election results, there were no seats where Labor gained on NXT at quite that rate.  In a few they gained at 35-40%, in some at lower rates and in a couple NXT beat them.  (Greens voters are not big how-to-vote card followers.)  I suspect that SA-BEST will remain ahead of Labor on current figures.  A bigger issue for them could be the overall two-candidate situation which looks very close.

8:38 So looking at options for a Liberal majority.  I'm assuming that from their 27 notional seat list there are two won by the crossbench, and a swap of King (gain) for Mawson (loss).  Ignoring Dunstan where there are no figures there are four of the remaining 25 for which there has been doubt - Elder, Heysen, Newland and Waite.  There is still doubt about Heysen but at present they're holding the rest.

8:34 ABC now projecting Labor just ahead in Liberal-held Waite (unexpectedly) but this is off some very Liberal-friendly booths; we have to see if that holds up with more counting.

8:27 A bit of life in Elder which has come back below 51% to Liberal.

8:14 There is a large swing against the Liberals in Waite, but they are holding on there so far.
From the Liberals' notional 27, they're losing two to the crossbench (Mt Gambier and Frome) for 25.  They're currently ahead in King and currently losing Mawson.  There's an exclusion order question on Heysen.  (Labor have improved in Lee now.)  So currently the Liberals are hovering around 24-25, we have three crossbenchers and Labor 19.

8:06 ABC has Labor ahead in the notionally Liberal seat Mawson, so that's one Labor might yet pull back.

8:04 Liberals ahead now in another Labor marginal, Lee.  There was a big late betting plunge in Lee and King. 

Geoff Brock doing OK in Frome so far so it looks like the three expected crossbenchers are all holding.

7:59 Now we have more trouble for Labor - they're behind early in one of their marginals, King.

7:56 I agree with Antony that Xenophon is in trouble in Hartley and he's not out of the exclusion-order woods either.

7:54 Heysen has an exclusion order issue - SA-BEST are not far enough ahead of Labor currently to avoid being swamped on Greens preferences and knocked out.  We need to keep a close eye on the relative primaries here to see if SAB can even win this one.

7:52 Colton has turned around with more booths in the 2PP so Labor is no longer winning it.  At this stage there is no seat that is changing on a 2PP basis.  Labor must start taking seats from the Liberals on a 2PP basis or the Liberals will win, probably outright.

7:51 In Giles, an exclusion order issue is looming, but on early figures Labor could be too far ahead for it to matter.

7:48 Frances Bedford bolting in in Florey, so that's one off Labor's pile to the crossbench.

7:47 SA-BEST doing quite well in Heysen which was always one of their most likely seats.

7:37 At this stage Labor aren't in trouble in any of their 20 notionally held seats, but given that the Liberals notionally hold 27, they need to win some notionally Liberal seats to hang on.  At the moment they are leading in Colton according to the ABC but that is a projection off a very pro-Liberal set of booths so I am suspicious of it.

For the Liberal part, we can knock off one for Mt Gambier where Bell is winning easily, but he can be treated as Liberal anyway.

7:30 I am especially interested in the notionally Liberal ALP-held seats (Newland, Mawson, Colton, Elder). Early ABC estimates are encouraging for Labor in Newland and Colton.

7:23 ABC projection suggests a pattern that the majors are both doing a few points better than the final polls and SAB and Others worse.  On that basis, a very close night ahead.

7:18 Troy Bell, ex-Liberal, doing very well early in Mt Gambier.

7:15 Mawson, another potential SAB target, distant third there too.

7:09 Davenport was a potential SAB target but for now they're a distant third.

6:56 SAB competitive in Finniss where they appear to be running second, but they need the Liberal primary to come down a few points.

6:46 The early ABC projections are positive for Labor in terms of the state vote but off 0.1% it's too early to take much notice of that - probably comes to a large degree from uncompetitive electorates.  By the way I recommend following the ABC site as it is easiest to see all the electorates at once - just with the usual concern about preference projections that are not real votes.

6:34 Some very early booths in, as usual rural ones that are strong for Liberals, so the state totals will take a while to settle down to anything vaguely meaningful.

6:16 Betting continuing to plunge on the Liberals, now down to $1.42 after not being favourites until today.

6:10 An exit poll has been reported by Nine with a 36-31 lead for the Liberals and a mid-teens figure for SA-BEST.  On these figures the Liberals would have good chances of winning.

Opening Post
Welcome to my attempt to post live comments on Batman and South Australia simultaneously.  Polls are just closing in SA.

Last night I chickened out of trying to pick a winner based on the public polls, which suggest a very close contest unless SA-BEST preferences split strongly one way or the other (something which there seems to have been very little real attempt to sample.)  Today the vibe from informed commentators seems to have been very much in favour of the Liberals and there was a big betting plunge in their favour in certain seats.  Against that Labor has claimed their internal polling points to a result of about 21-21-3-2, which more or less matches a reasonable read of the public polls and causes me to cynically suspect this read is therefore wrong.  But we will see!

This count is going to be messy and probably even worse than Queensland's.  There will probably be seats in which the final two are not known with certainty or even confidence tonight, and SA also has infuriatingly slow counting of postals and pre-polls, which don't get done til early in the working week.  Perhaps we will know the result tonight (more or less) or it might string out for several days or even until preferences are distributed in some seats.

I'll be keeping an eye on the ABC coverage but also keeping a watch for "bolter" seats with exclusion order issues that are easily overlooked.  Please keep in mind that figures on the ABC coverage are likely to include estimates that may not be real votes, especially as concerns flows from SA-Best.

I'll try to keep this page on SA time but will probably fail!


  1. "No seat changing on a 2PP basis"

    ABC has King as a Lib gain from Labor.

  2. Everyone seems pretty confident about Heysen now, but a week ago when pre-polls opened the polling had SAB 5 points higher. How much, if at all has that been factored in?

    1. I find so often that when there is a theory that prepolls will behave differently because of some campaign event or slide for a particular party, they don't. The classic case of this was the removal of Tony Abbott as PM during the Canning by-election. Prepoll voters have this uncanny tendency to anticipate the final result. But even if SAB are 5 points better in the remaining 20-odd% to be counted (because the count never gets to 100), that still wouldn't be enough - and I expect SAB to do badly against the Liberals on postals too.

    2. Well, if you're right about the foresight of the electorate then it's clearly game over. However, if not, 5 points on first preferences might well indicate a larger difference on 2nd and 3rd preferences as well. I guess we'll find out soon enough.

  3. You mentioned that the polls seem wrong. It's interesting that the Legislative Council result is closer to the polls than the House of Assembly results. Part of this is due to fact that SAB is 'contesting all seats' in the LC, whereas they are not in the House of Assembly, but the LC numbers indicate the problem pollsters have with the 'other' vote. A respondent may be intending to vote Aust Conservatives, Dignity, Liberal Democrats or Animal Justice, and will indicate that to the pollster, but when they get to the polls there are no candidates for those parties in their House of Assembly seat, so they vote for Liberal or Labor (or SAB or Green) instead. The error with 'Other' in the Newspoll is 1.7% which fits with this explanation. The other main error in the Newspoll is 3.3% with SAB. Newspoll 17%, Actual votes counted so far 13.7% for SAB. I think this indicates a late swing against SAB -probably even occurring in the polling booth. People think ' I don't like Labor or Liberal much, but these SAB clowns have unfortunately proved to be no better, so I will vote as I normally vote'. Reachtel gave 16% for SAB, so another indication that the downward trend for SAB was real.