- Date published:
10:00 am, May 19th, 2020 - 53 comments
Categories: campaigning, greens, journalism, Judith Collins, labour, national, nz first, polls, Simon Bridges - Tags: 2020, anne tolley, campaign, electorates, judith collins, list rankings, newshub, poll, pollwatch, reid research, simon bridges, watch
I’m back with a new edition of Pollwatch for those of you who like some statistical analysis with your speculation!
This was a doozy, definitely proving that UMR wasn’t off-base with the leaked results and that National is giving a disastrous performance, worthy of 2002 Bill English- in fact, literally only 0.1% better than that final 1999 election result for National. If this were the 2020 election, it would be the second-worst National result ever, and the actual worst election under MMP for the Right ever, as in 2002 ACT was at the height of its powers with a 7% result, and its 2020 “bounce” still has it under 2% and only just managing to get an actual caucus back. (It’s worth noting that national polled even worse than this in the lead-up to 2002, though, and election campaigns tend to narrow the result no matter who leads)
Unfortunately most of the graphs here are so boring they’re barely worth looking at- there’s a 100% chance of a Labour-only government if this were the election, something I’ve mentioned before that would be very dangerous in practice and that I hope voters would narrow away from in the election, either by softer Labour supporters considering the Greens, or by natural closing of the gap between parties. That said, the trend still shows that polling is quite volatile since late 2019, likely at least partially due to long periods between polls now that Roy Morgan is out of the New Zealand polling arena.
Newshub also hasn’t given their net approval numbers which actually mean something, so I won’t comment on Preferred Prime Minister polling, other than to note it is as disastrous as expected, and that Bridges and Collins are in a statistical tie for second place, so knives are clearly being sharpened, and perhaps not just by Collins.
That said, I have an entirely new addition to my overall model for you all, as I was working on something to let us know how close we are to National hitting overhang seats, so for the very first time I’ve adapted the Strong Transition Model from the UK’s Electoral Calculus site to allow me to make predictions based on the national vote for which electorates will change hands. This sort of model tends to be the best way to predict the overall number of electorate victories, (1 to ACT, 0 to NZF, 35 to National, 35 to Labour) so please take individual calls with a grain of salt, but those numbers will likely be good predictors as we get closer to election date, if the number of electorate wins becomes relevant. I’ve also corrected one National prediction to a Labour one, as to my knowledge the Greens don’t plan to reprise the very strong campaign in Nelson, so it is very likely to go to the Labour MP factoring that in, and given that nobody is calling Takanini yet, I’m predicting it will go to Labour with a result this strong, which seems fair if we assume it’s a possible bellwether seat like its demographics tentatively suggest it might be.
You’ll also note my model doesn’t predict NZF will win Northland. They’re out of parliament on these results, but they actually can win Northland with a stronger party vote result than in this poll, according to my model, as the exodus of National support might make Peters look like the most attractive option, so put a pin in that electorate race. With some actual way to predict electorates, I’ve updated my model for the assumption that ACT gets one, NZF get none, and as far as we can tell, minor parties also get none.
What this really lets us do though, is predict how many National MPs will retain or lose their seats, and it’s an absolute bloodbath. With 39 seats from the party vote, (down from 56) they get 5 (down from 15) from the list once we assign Nelson to Labour. If we assume their list based on caucus rankings for now, (a reasonable starting point, but likely to have a few adjustments) this gives their List winners as:
2 – Paula Bennett (now list-only, to act as campaign manager)
3 – Paul Goldsmith (ACT to retain Epsom)
7 – Nikki Kaye (Labour to win Auckland Central, predicted 2.8% margin)
9 – Michael Woodhouse (Labour to retain Dunedin North)
11 – Alfred Ngaro (Labour to retain Te Atatu)
Paula Bennett and Paul Goldsmith are absolute necessities to retain, but this actually suggests their list ought to diverge by caucus rankings when you look into who they’ll lose. Let’s start with unranked incumbents, who are gone regardless of election result:
Moving on, here are those who’d need to find new employment if National can’t close the gap:
14 – Melissa Lee (Labour to retain Mt Albert)
15 – Chris Bishop (Labour to win Hutt South, predicted 5.9% margin)
17 – Anne Tolley (list-only, “to contest election as Speaker“)
20 – Nick Smith (I’m calling my model wrong here, as Labour could absolutely win this in a two-party race, and as far as I know there are no Green plans to try to win this election in 2020, which upsets how its maths works)
24 – Kanwaljit Singh Bakshi (Labour to retain Manukau East)
26 – Brett Hudson (Labour to retain Ōhāriu)
29 – Jian Yang (list-only)
30 – Parmjeet Parmar (Labour to retain Mt Roskill)
33 – Jo Hayes (Labour to retain Christchurch East)
36 – Harete Hipango (Labour to win Whanganui, predicted 4.4% margin)
38 – Denise Lee (Labour to win Maungakiekie, predicted 2.3% margin)
43 – Lawrence Yule (Labour to win Tukituki, predicted 2.3% margin)
44 – Maureen Pugh (Labour to retain West-Coast Tasman)
45 – Nicola Willis (Labour to retain Wellington Central)
47 – Agnes Loheni (list-only)
48 – Paulo Garcia (Labour to retain New Lynn)
…but with potential new blood in the following electorates:
Chris Luxon, candidate for Botany (expected to replace independent incumbent Jamie-Lee Ross)
Unannounced new National candidate for East Coast
Unannounced new National candidate for Invercargill
Tim Costley, candidate for Ōtaki
Simon Watts, candidate for North Shore
Nicola Grigg, candidate for Selwyn
Jake Bezzant, candidate for Upper Harbour
Once we factor in these losses and replacements, there’s some obvious considerations for tweaking National’s list.
A lot of their more senior talent has been shafted by the caucus reshuffle, and that’s very dangerous when your list is going to potentially be cut this close to the wire. If I were a senior National Party official, I’d be looking at bumping Tolley up to #4 or otherwise in single-digits on the list so that I didn’t get embarrassed by losing the assistant-speaker from caucus with my own hubris, and frankly I’d also consider bumping an additional woman up the list too, like Melissa Lee or Parmjeet Parmar, because if they went by caucus rankings alone, a result like this would leave them with eight women out of thirty nine MPs, and an overall very white caucus, too, which would open them up to further criticism that they aren’t even trying to look like New Zealand, and while I don’t like Melissa Lee, she sure as hell is a better pick as a list MP than Alfred Ngaro. They’ll naturally be hoping for another 3-4 list MPs minimum from the campaign, but they shouldn’t count on it- lists should both plan for the worst and offer for the best. The reshuffle has already thrown new blood high into the current caucus rankings, speaking cynically likely because they supported Bridges for leader, and the smokey room setting the list needs to counterbalance that.
On a wider view, if we look at the bloodbath to list-reliant MPs in opposition, it’s worth noting that Labour had similar issues in the 2014 election, and arguably are still recovering from them. It might make more sense for the competence of our parliament to set the size of parliament such that we have a list seat for every electorate seat, to limit the losses of senior MPs and institutional knowledge. That would, of course, give us 142 MPs this election, but that’s not completely ridiculous.
I hoped you all enjoyed the foray into the consequences of who wins and who loses- I’ll definitely come back to that later when the campaign is officially open and we have lists for National and Labour and I can tell you who might be in as well.
It’s also worth noting that as this is a statistical model, it’s likely in the election that Labour will lose a few electorates, and pick up additional ones from National, but if they’ve gained from 2017, which seems very likely right now, they’ll retain any electorate losers on the list, anyway, unless they’ve been badly demoted, so unfortunately I can’t tell you very much interesting about that until lists are released- and I’m certainly not going to go off the first draft of the Green list yet, given that voting is still in progress on that one, and it may be adjusted by the Executive afterwards if necessary.
Also, any candidates with a margin smaller than 3% are in fact very marginal calls by the model and could go either way- but it’s likely that most such upsets might be balanced out by losses of existing Labour electorates that this model can’t predict easily, or additional seats changing hands that it doesn’t accurately predict to. The idea is more to get at the right overall number of electorates for each party.