Written By: - Date published: 8:30 am, February 20th, 2018 - 47 comments
Categories: greens, labour, national, nz first, polls - Tags: coalitions, colmar brunton, MMP, polling, pollwatch, threshold, trends, TV One
Kia ora koutou and welcome to the second edition of Pollwatch, where I, your host, copy polling into spreadsheets and analyse what its margins of error mean so that you don’t have to, and try to share some dispassionate and hopefully objective analysis on trends. (although, relying on the human brain as it does, I concede it may be less reliable within its limitations than the numbers) Today, we analyse the Colmar Brunton poll sweeping in to the capital just ahead of Cyclone Gita, and like a cyclone, there seems to be a certain feature dominating the forecasts. Colmar Brunton’s detailed report will be available here, but as of the drafting of this post, that page still shows December’s poll, so I have made do with One News’ brief analysis instead.
First, a little disclaimer- my seat totals differ slightly from the official One News ones, because they have access to unrounded party vote figures and I don’t. I have left these deviations in to accurately show the limits of reading this stuff off public polling information, but thought you all deserved to know. In the official seat counts, Labour has 59 seats, and National 54, suggesting Labour has been rounded up to 48% and that ACT, despite not registering in the results as provided on the TV, has at least 0.4% support, as that’s the only way for Labour to lose a seat without National or the Greens gaining one on this poll.
On either result, however, Labour would require the Greens to govern, meaning any snap election held anytime soon would likely result in a coalition government without New Zealand First.
How likely, you may ask? Well, after copy-pasting a new formula to 50 different spreadsheet tabs, and completely rejigging my totals columns, I have for you a new colour in my projection graph, because this new poll made considering the possibility of an outright Labour government a thing we should take seriously. Those words sound scary, but remember, at several times during the Key government, he was polling at a high enough level to do the same, so this is no guarantee this trend will continue forward to 2020.
Those of you paying close attention will note that not only were there 2 National victories in 2,000 random simulations, but that excel sometimes sums totals to over 100% in its enthusiasm for rounding. This small chance of a National victory even with Jacinda Ardern at the height of her powers should hopefully temper our rhetoric around the National Party leadership selection: there is still a chance whoever is chosen by their caucus (still no moves to let their members help choose…) could be Prime Minister, especially once the potential of a comeback closer to the election is considered, so for those of us who weigh in, we should do so with our genuine opinions of who would be the best pick for New Zealand, not for our hopes of sabotaging their party, and of course, any National supporters listening can then take them with a rather large pinch of salt due to our different perspectives, if they wish to listen to our ideas at all, of course.
Speaking of English’s successor, this poll largely completed fieldwork before he announced his resignation on the 13th, but contained at least one full day after it too, given its finish on the 14th. Uncertainty around this news may at least partly explain the giant bump for Labour, or may simply represent the lack of effective opposition his party have provided so far- my position so far is that this is probably an accurate representation of the state Bill English’s (and to a lesser extent, Steven Joyce’s) post-Key holding actions have left his party in: they put everything they had into holding out for the election, and couldn’t quite do it, and now the decline has, perhaps, set in.
Back to more left-wing matters, it’s worth talking about the government’s coalition and support partners, New Zealand First and the Greens. New Zealand First, polling at 3%, is absent from all of these graphs because they are under threshold, and didn’t win any electorates in the previous election, so I have assumed for now that they would not were any snap election to be called right now with similar levels of support to this poll. Even considering the margin of error, there is no chance for New Zealand First to make it above 5%. These simulation results also assume a 95% chance David Seymour retains Epsom, and factors him into the National Government chance, as his presence or absence does not really seem to significantly alter its character.
The Greens, however, at roughly 5%, have about a 50/50 chance of being above the threshold. Those are dangerous waters, and suggest that action to remind the public of their wins, and accrue more, are needed to make progress if Labour is to continue to have a medium-term ally in government, and an effective left-wing check on their tendencies to compromise with corporate interests. Like New Zealand First, this is likely an adjustment to their role as being mostly-in-government, although a friendlier one in this case, and to the reality of the dead rats one needs to eat in coalition, and arguably reflects the mistake in process made by Labour that made them think the Greens wouldn’t have issues supporting the Electoral Integrity Amendment (Waka Jumping) Bill. More on that subject in another post, as that subject is worth an entire post.
And finally, you can see here the extremes of the margin of error in a nested donut graph, with National just able to squeak into government when the margin of error goes almost entirely their way in every respect, hence why it only cropped up twice in two thousand simulations. (This should also reinforce how rare these two extreme scenarios are to everyone, the reality is that scenarios only a little different from the first graph posted are the most likely)
Now we’ve gone through what this specific poll tells us, some commentary on trends. New Zealand First seems to be at or near the genuine bottom of the barrel in their support at the moment, having fallen to a comparable level to the prior Reid Research poll. This suggests that they’re at their natural core support of around 3-4%, and they had made soft gains of 5-6% in the election, some who immediately deserted on their choosing Labour, and other who are not impressed with their performance in the coalition government thus far. There is enough time for them to pull back and survive the 2020 election, but they’ll need to get some big wins before then, and reconnect with the left-wing popular base that seems to be really connecting with Ardern right now.
Secondly, the Greens are still falling. This is once again danger territory for them, and it will be a hard slog to even hold ground against an ascendant Labour party that is eating into some of their strongest policy areas despite arguing not delivering as well, scooping up large amounts of the Māori and queer vote. Some of this may be a lack of credit for their own wins, some may be getting the blame for decisions Labour has forced their ministers or Party into as a result of coalition talks when the reality is a bit more nuanced, and part of it may just be that the party is suffering from a news cycle very dominated by Labour ministers and a National opposition eating up large amounts of oxygen between its extra time to talk to media and its current leadership contest going on in smoky back rooms.
National’s fall here isn’t yet a hugely significant trend, but it could suggest a slow leak of support. This is something to monitor in the future. When you’re above 40%, you actually need a large 4% drop for a loss of support to be guaranteed to be significant in any one poll, but smaller drops over time can add up to a trend, even if any individual poll’s drop could simply be down to the margin of error.
Lastly, Labour has a huge bump in this poll, one that would be significant so long as the result isn’t rogue, and we’ll need to wait for the next one to see if this is just Colmar Brunton’s recent tendancy to overhype Labour’s performance, or if it represent a genuine and large shift to Labour as the most popular party in the country right now, or merely an overestimation of a slow push in their direction. (a shift not unwelcome, although perhaps some of us were hoping it would come more at the expense of National than of other parties, especially as there’s a chance that eliminating the Greens next election could be disastrous for them) The bump wasn’t so dramatic in previous Reid Research or Roy Morgan polls, but this does reflect general indications that a Labour(-Green) government without New Zealand First is currently very likely across all three companies that release their polling data publicly.
We’re also in the unenviable situation that Roy Morgan seem to have taken a break from polling New Zealand politics, which is unfortunate, as they were previously the most reliable polling company in terms of releasing public data on a roughly monthly turnaround. Whatever you think of their lean and accuracy, (my opinion was their accuracy was good, but that they leaned a bit obviously leftward in their results, which was not a bad thing back when Colmar Brunton leaned right and Reid Research tended to lean slightly less right) that regular information was a really useful reality check, and I personally hope they will return to releasing poll information soon, and are just taking a well-earned break.