Median result if polls accurate is Labor government, not clear if minority or majority (approx 46 seats)
We're into the final days of the NSW campaign and after a cluster of polls in late February there was nothing useful statewide until the final week polls started. This thread will follow the major final week polls and also cover all the local polls (many of them disreputable) that I'm aware of. The aggregated estimate at the top will be changed as new polls are added.
So far this week we have had possibly final (or perhaps not) offerings from Morgan, Resolve and Freshwater. These have landed in a narrow band with 2PPs of 53.5, 52.5 and 53 to Labor, which is very similar ground to the late February cluster so there's not much more to explain at present about what that should do. My seat model's median projection if the 2PP is around 53 remains at 44-45 Labor seats (45 is just ahead of 44), which would be a comfortable Labor minority government with options of working with either the Greens or independents. However, it's well within the model's potential for error that a 53% 2PP would lead to a small Labor majority, or that it could lead to both majors having about, say, 41 seats and the crossbench deciding the winner.
It's also still possible that the polls are wrong and that we could see a 2PP of, say, 56 to Labor, which is most likely a majority if not necessarily a large one, or, say, 50-50, which would be likely to result in the Coalition having the most seats though even then not necessarily forming government. A Coalition majority is at rightly long odds since it appears to require both a very good distribution of seat swing and a significant polling failure.
I weight the Resolve (14-19 Mar) and Freshwater (Mar 19-21) polls higher than the Morgan (Mar 10-14), partly for recency and partly because Morgan's sample size was small among other Morgan foibles, but my aggregate after the first three polls is 53.0 not 52.9 because after reviewing detailed by-party 2PP breakdowns for 2019 recently published by Antony Green my estimates of some of the 2PPs for the February batch increased, causing me to revise my estimate of where things stood at that time from 53.0 to Labor to 53.4 to Labor.
Resolve's primary votes were 38-38 with Greens 8 (remarkably low), IND 8 and Others 8. I have examined whether 8% for independents could be plausible and I think it is ambitious but not impossible. Independents polled 4.8% last time including nine who polled more than 15%, three of whom won. At this election there are again significant independents in seven of those nine seats plus about another eleven with potential for independents to poll very well, though not all those independents will do so. Importantly those independents include four incumbents (three ex-Shooters and Gareth Ward) who were not independents last time. The 2PP was not directly published but implied to be 52.5 to Labor based on comments that it was a 4.5% swing.
Freshwater has a similar picture with 37-37 for the majors, the Greens on 10 and IND/others the same as Resolve on 16. (The AFR said both majors were on 33 in its article but that does not add up.) The 2PP was 53-47 (that's not especially generous to Labor; my last-election estimate is 52.9). It is interesting that Resolve's 2PP did not change despite Labor dropping 2% on primary votes and more transparency regarding preference allocation would be useful in explaining this.
Morgan also had the majors tied but on 34-34, with the Greens on 13, One Nation on 2 (now confined to seats they are actually running in) and independents and others on 17. This continues a pattern of Morgan having low primary votes for the majors.
Others, which will at least include the final Newspoll (I am not sure there will be anything else), will be added as they arrive.
Seat Polling
I mentioned a Pittwater seat poll by Freshwater (52-48 to Liberal vs IND) in the previous article. Others I am aware of are:
* A much-criticised Climate 200 Community Engagement poll of North Shore, supposed to show a 2PP of 50.7 to Liberal vs Independent, but there is no explanation of how the 2PP was calculated and 50.7 is not credible off the very incompletely reported primaries which suggest more like 55-45. Community Engagement is a non-transparent pollster which does not have a demonstrated record of successful predictions at elections. It's helpful that there was a very teal-like contest in North Shore in 2019 where it can be seen that the independent local mayor. Carolyn Corrigan, gained at .26 votes per preference. See for instance the Twitter thread by Yuan (Joseph) Lin that bolts down why the supposed flow to teal independent contender Helen Conway in Climate 200 's poll is implausible based on what we have seen before.
* Another Climate 200 poll of Pittwater with Liberal 34.3 Scruby (IND) 28.5 Labor 11.2 Green 6.4 and the rest others or "won't vote" and the teal candidate Jacqui Scruby ahead 50.7-49.3. This is presumably also Community Engagement and details are again inadequate. No polling methods details for these polls have been found. Incidentally the North Shore poll supposedly had 7% not voting, which seems unlikely unless non-citizens are included, since electors who don't vote are generally hard for pollsters to find.
* An inadequately reported Freshwater seat poll of Riverstone. The AFR, whose reporting of polls has been a rolling embarrassment to a supposedly upmarket paper, didn't see fit to tell us sampling dates, polling method, 2PP method or sample size. Labor led 54-46 but that was with One Nation polling 7%. One Nation aren't running, which would help the Coalition by cutting down on loss of potential preferences to voters voting 1 One Nation and stopping.
* Redbridge polls of Penrith and Parramatta, claiming to show Labor trailing 49-51 and leading 54-46 but Redbridge are not very good at estimating 2PPs; I get Labor ahead about 51-49 and 57-43. Both polls had very high One Nation votes (15% in Penrith, 13% in Parramatta) - high votes for One Nation and UAP are a common feature of Redbridge polls. By comparison in 2019 One Nation got 7.2% in Penrith and less than 4% in the upper house in Parramatta.
* An IPA poll by a hitherto unknown to me outlet called Insightfully and covering Dubbo, Bathurst, Orange and Upper Hunter and purporting to show very high votes for non-majors based off robopoll samples of less than 300 per seat. However the poll was taken before candidates were announced and there simply aren't non-major candidates capable of getting 55% in Dubbo (where there is no independent running this time) or 44% in Bathurst (where the only independent was banned from Lithgow Council for five years.) As such while the Upper Hunter numbers were good for the Nationals incumbent, it's hard to take any notice of this one.
* Some non-transparent nonsense on Sky about a poll by an unstated pollster for an unstated industry group supposed to show a 16% swing against Matt Kean on primaries in Hornsby with One Nation getting 15% (and that's raw with 5% undecided). Kean was still winning 58-42 but nobody is taking the claimed One Nation vote seriously. (They got 4.5% there last time.)
* A further batch of industry group polling on Sky (edited into the article) has further problems. Again there's no detail on pollster name, polling method, sample date (with one exception), sample size, 2PP method or the sponsor (though the sponsor of the first three seems to be an energy industry source that detests Matt Kean and might well have used skewed questions against him, since it's unlikely enough voters would have a view on Kean either way for some of the ratings given). We have:
- Leppington a supposed 52-48 to Labor off Labor 40, Liberal 32, One Nation 16, Greens 7 others 6. With the high propensity of One Nation voters to exhaust, this 2PP is unlikely; I would expect 56-44 by statewide preferences though probably a little lower in this seat. (It is credible there will be a high One Nation vote in Leppington).
- Goulburn a plausible 53-47 to Labor off Labor 35 Liberal 33 Shooters 13 Greens 9 One Nation (not running) 5 others 4. The combined Shooters and One Nation vote here is about what it was last time.
- Wakehurst a supposed 50-50 off primaries of Liberal 41 Regan (IND) 37 Labor 11 Greens 3 others 8. Very hard to credit that Michael Regan would have any trouble getting a gain rate of well above the .18 votes/preference needed to tie off these numbers. If these primaries are real Regan will win, but hard to believe the Liberal primary has gone down that much.
- North Shore a supposed 54-46 to Helen Conway off primaries of Liberal 34 IND 28 Green 12 according to narration (but 20 according to graphic) Labor 7 others 6 and who knows what the other 13 are. It seems suspicious that Labor would be down through the teal challenge but the Greens not at all and for that reason alone it's hard to believe that one.
- and the big one Kiama which is actually a fairly fresh sample taken March 13 - this might be useful information but we'll see. Ward (IND) 40, Labor 33, Lib 11, Green 8, other 7 and 53-47 to Ward (that seems about right since most of the Lib votes will exhaust and the Green votes may not split that strongly).
North Shore and Kiama may be separate batches. There's no evidence to indicate any of these should be reliable and they may be being put about by motivated actors.
Leaderships
The Resolve poll showed both leaders are very popular by recent standards with both on net +20 approval and Perrottet ahead 40-34 as preferred Premier (a modest and generally not sufficient lead as this skews to incumbents). Morgan had Minns ahead of Perrottet 52-48 among those who bothered to return the SMS on that question, which about 15% didn't. Essential has Perrottet ahead 36-33 but appears to have wimped out on voting intentions despite having a sample of 708, and Freshwater has Perrottet ahead 46-40.
Betting
Betting odds are not reliably predictive but it is interesting to keep an eye on how they go. Markets on result type currently imply Labor is about an 82% chance of winning with a 30% chance of a majority and a 52% chance of minority, to the Coalition with a 14% chance of minority and 4% chance of majority. Seat betting markets have been jammed on 42 Labor wins for ages and are now quite confident about Labor keeping Bega and gaining Heathcote, East Hills, Parramatta and Penrith with a little more doubt about Riverstone. In serious doubt on the Coalition side are Winston Hills, Ryde, Camden, Goulburn, Upper Hunter, Holsworthy and Tweed vs Labor and North Short, Pittwater, Wakehurst and Willougby vs teals. Two INDs are seen as in significant danger but still favourites - Gareth Ward in Kiama (vs both majors) and Helen Dalton in Murray.
The overall favourites tally is Labor 42 Coalition 41 Green 3 IND 7 (which I think would be a Labor government) but after adjusting for close seats a more realistic read of what the seat markets expect is Labor 44 Coalition 38 Greens 3 with IND 7 or 8. So the seat markets, the overall markets and the statewide polling are all pretty much consistent with each other.
Updates will follow with at least Newspoll to come. At this stage on the night I will have the usual live blog-style coverage here, with the usual rollout of close seats and the Legislative Council count during Sunday.
Updates: Wednesday
A fresh batch of Sky "industry group" polling comments have been added to the seat poll section above.
Shows even primary on average alp and coalition average 70 to 75% of primary vote.. green vote 10% .. still 15% to 20% everyone else assume 10% right which will have a large exhaust rate other mainly even major party preference.. this leads to the liberals being in trouble in any seat they are even or behind in.. I don't see how this can lead to anything but a alp majority govt. The size of this majority is the question
ReplyDelete