- Date published:
4:00 pm, April 18th, 2018 - 78 comments
Categories: greens, labour, national, nz first, polls - Tags: colmar brunton, labour-green government, labour-green-NZF government, pollwatch
Colmar Brunton has, to much fanfare, released its latest poll. As Mickysavage noted recently while I’ve been on internet hiatus, (thanks for covering!) this result is a bit of a correction from a post-government high, with a 5% drop for Labour relative to their February poll, and a 2% improvement for NZF and 1% improvement for the Greens. (the latter being a bigger deal than you might expect, as internals between these two polls apparently had the Greens back under threshold) It is still an overwhelmingly strong endorsement of the current government, but it’s possibly also reflecting a bit of frustration that the government is not moving as fast as all of its fans would like on all of their priorities. I won’t cover the exact same ground as him, but calling this anything more than a setback for Labour is trying too hard. They’ve just lost some of the additional ground they gained post-election, probably due to overpolling in that February poll.
Statistically, it’s worth noting that the drop in this one poll, either for Labour or for government-vs-opposition on the Party Vote, is not significant enough to rule out it simply being a combination of under- or over-polling for the parties involved. While the margins of error in this poll were lower than the change for both New Zealand First and Labour, you need to add together the margins of error from both polls before checking for statistical significance, so until a second poll shows up confirming this dip, it may literally mean nothing, as per my prior disclaimers on polling. From an MMP perspective, in fact, this poll is the better of the two, as I’ll discuss below regarding threshold maths.
I’ve currently got my seat calculations for all three scenarios (expected, left-wing MoE, right-wing MoE) up on Datawrapper, which I am currently trying out instead of doing two of my three excel graphs. Its political chart is pretty great for doing seat allocations, in my opinion.
Fieldwork for this poll was conducted from the 7-11th of April, so it captures the very beginning of Marama Davidson’s co-leadership of the Greens, and also included the day before the announcement, when some people may have been assuming Julie Anne Genter could win. Make of that what you will. Also worth noting, as below, that Simon Bridges’ first ever preferred PM result was returned with this poll, and while it’s a relatively junk stat, it’s also abysmally low.
1 NEWS Colmar Brunton poll (3/3): Preferred PM: Ardern 37%(-4), Bridges 10%(+9), Peters 5%(+1) #nzpol
— Colmar Brunton (@ColmarBruntonNZ) April 16, 2018
How low? Looks like someone else has handily compared for us:
Simon Bridges' debut Colmar Brunton Preferred PM rating (10%) is worse than Andrew Little (12%), David Cunliffe (12%), and David Shearer (11%). Wow.
— Finn Jackson (@_Finn____) April 16, 2018
(Additionally: Only three actual party leaders have polled lower than this: Bolger, English, and Little, and there is absolutely no precedent for a right-wing leader with such low preferred PM ratings)
This is with the known phenomenon that right-wing or conservative politicians enjoying a much more instant boost to their popularity among their own party than left-wing or liberal ones do. Bridges is basically a space-filler, and his own party is implicitly admitting it here, rating him even lower than Labour’s succession of Davids, or their misadventure with Little as Leader. That’s got to hurt for Simon, but I wouldn’t expect National to be sharpening their knives just yet: everyone knows that sticking with this leadership team and looking unified even if they’re not is the only way they have a shot in the 2020 election, and pushing over Bridges only makes things even worse. This is where people like me really wish someone were polling comparative approval ratings for the Leader of the Opposition and Prime Minister, so we could do some more solid analysis on whether this is simply a failure of Bridges’ name recognition, or to what extent it’s an approval gap or simply disapproval of both our most prominent political figures. (Ardern’s 37% is actually not terrible when compared to Clark’s rise in popularity under her own government- Labour PMs traditionally do better once they’ve settled into government for a year or more, so she’s still ahead of the curve there)
Microparty news tends to be the same: There is still small but significant support for the Māori Party, and next to none for ACT.
This also shows that perhaps NZF’s dip under threshold was part of a brief period of discontent. This poll notably does capture the new Transport GPS, which was likely a win for all three parties to some degree, but the biggest win for the Greens, and just misses the new oil exploration policy announcement, which is likely to be seen as a loss with at least some NZF voters, and at least a moderate win for the Greens and climate activists within the Labour Party, despite it actually being a very “first steps” sort of policy that doesn’t roll back any existing fossil fuel exploration, and still allows for exploration on land in Taranaki.
I’ve once again run my probability model with 2,000 elections simulated for likely coalitions within this poll’s margin of error, and the results are pictured to the right. There is also two invisible too-small-to-count single outcomes: one where there was an outright Labour Government, which likely occurred by my random number generator simultaneously spitting out that National overpolled, Labour underpolled, and both NZ First and the Greens fell under threshold, and one where Parliament was hung and either ACT, NZF, or the Greens would have had to cross the house or commit to abstaining on confidence and supply to form a government. Excel doesn’t count half-per mille results on pie charts, so those two outliers are removed from the graph automatically.
While I give the threshold results to a rough 5% rounding above, the Greens only came under threshold 1.6% of the time in this run of my model, while NZ First came in under threshold 52.9%. While it’s even probabilities to come over threshold at 5%, in reality every party that’s consistently polled at or above 5% before an election has managed to pull off coming in over in our brief MMP history, so I wouldn’t count NZF out just yet, so if we do assume NZF is getting over threshold or winning an electorate at the next election, (essentially removing all the under-threshold results for NZF) it’s roughly 60%/40% odds that they’ll be needed for the next government if the election result looks roughly like this poll. (Which it almost certainly won’t, things are always changing, but it lets us know where the electorate is sitting now)
At this time Colmar Brunton haven’t posted their detailed results, but their response rate is falling in their last poll, and I expect it will have done so again this time, or at best regained a fraction of its lost ground from February- something to keep in mind for polls outside of the election season. This does mean that these polls are more likely to represent relatively decided or motivated voters, and less likely to represent swing or demotivated voters. It’s quite possible that this will hurt National more than Labour, as swing voters tend to largely vote based on a perception of unity and competence within governments and oppositions, and right now National is looking very shambolic in those regards.