Pollwatch: 11/02/2019

Written By: - Date published: 10:12 pm, February 12th, 2019 - 18 comments
Categories: act, Amy Adams, greens, labour, maori party, national, new conservatives, Nikki Kaye, nz first, polls, Simon Bridges - Tags: , , , ,

Kia ora koutou readers, and welcome back to the first Pollwatch of 2019, and as our official Worst News™ will tell you, (okay, maybe that’s my opinion of their recent coverage rather than an official thing) it’s a doozy.

Part of the reason I started modelling elections probabilistically myself is that looking at just the most likely result of a poll confuses you. This poll is definitely one of those- you would miss its biggest, most worrying implications (for both left and right!) if you simply looked at the number of seats each party gets in the most likely scenario, and ignored the chances of various governing arrangements and the implications of the approval ratings. Assume my usual cautions about waiting for a second poll to confirm dramatic results are in effect here, please.Roy morgan 11/02/2019: 63.2% chance outright labour govt, 36.8% chance labour-green coalition

This is, on its face, an amazing result on the left: no conservative nationalist party you have to compromise with, National in utter defeat even if they don’t eat ACT like they’ve promised to try to, and the Greens still around… or are they? In actuality, this simulation run had them knocked out of parliament 45.4% of the time. (The astute among you will note that means that 17.8% of the results had Labour winning so many seats that they wouldn’t even need the Greens to pass laws!)

This means that once we consider this poll result probabilistically, however, we see Labour aren’t actually behind the governing threshold. The most likely outcome is that they can govern alone, when you consider the Greens being under threshold and the Greens not being needed as essentially the same scenario. (They’re materially different in some ways, but not in terms of setting broad government policy)

This is the second result showing the possibility of an outright Labour government since Ardern became Prime Minister, and the third where National’s chance of governing relied entirely on the whole poll being rogue. It is however the first showing it as the most likely outcome.

We on the left should be worried about this- we’ve already killed off two of our seven serious parties last election. Even though I’m personally quite happy to see NZF gone, (they are definitively under threshold here) I would still like a more credible centrist party to replace them, and push Labour and the Greens back a bit more to the left.

We should also be very wary of results near the threshold of governing alone. Parties that have been in that position before have found that it tends to make people reconsider voting for the opposition- no election has ever delivered such a result, and with good reason. Kiwis voted for MMP to avoid having only two choices. Not having friendly critics to the left or the centre will merely embolden Labour to make choices that may not be right for all of us. Smaller parties are an important conscience in our system, and we need to either have stronger results for the ones we do have, or have a few more, for our democracy to remain healthy. I hope people seriously consider that when they vote in 2020.

It’s also worth noting that Labour should want either NZF or the Greens to be polling at the 7% level before the election. 7% is high enough to be completely out of risk of falling under the threshold on a statistical level.

Now, while I have plenty of criticism of Newshub in general and their political section in particular, they have done something really, really good in this poll: they are publicly releasing their approval/disapproval figures for Prime Minister and Leader of the Opposition, an actually reliable way of tracking how our two nominally most senior politicians are doing. Ardern is of course smashing it, with a rocking 52% net approval. (net approval being the percentage of people who approve of her performance less the percentage that disapprove, basically her margin of awesomeness) Despite his name, Bridges is underwater with an abysmal -28.9% net disapproval, news that would, if National has any intention of trying to win the 2020 election, have him rolled as leader immediately.

Let’s put this in context for a moment- Donald Trump, the President who has apologized for white supremacists, seperated underage children from their families to persecute them for crossing the border, had an immigration ban struck down as racist, put a credibly alleged rapist on the Supreme Court, and basically failed to do anything substantive except tax cuts, controversy, and discrimination, and is basically objectively the worst US president ever, currently has a net disapproval rating of -11.9%.

This is because, unlike Bridges, Trump commands the support of a whopping 88% of Republicans. (to be fair, identification as Republican is going down, too, so it’s likely that this remains high because people are tending to quit his party and claim they’re independents rather than protest him from the inside) If we assume absolutely nobody who voted for ACT, the Conservatives, TOP, or even NZF, Labour, and the Greens supports Bridges’ performance as Leader of the Opposition, (a ridiculously charitable assumption) Bridges would command 53.4% approval among National voters. That’s also abysmal, and let’s look at two controversial US Presidents to show you why- both of them had impeachment hearings.

Bill Clinton, right after his impeachment began, had a net approval rating of about 30%. That’s right, he was above water when Republicans decided to take a shot at him. At his very, very lowest, he still had an 11% net approval. And at that point, he still commanded the approval of 82% of his party, when he was revealed to be a cheater, a liar, and had tried to discredit his own aide to cover up for himself. This is a man whose public supported him more than Simon Bridges.

In contrast, let’s look at one Richard Nixon. At the start of his impeachment, he was underwater, too, at about -36%. (American figures are rounded to the nearest percent for this, sorry) At his worst polling, he had -42% net disapproval, a figure we should regard as an automatic resignation threshold for a leader, as we can hopefully agree Nixon held on for way too long. New Zealanders, for all our faults, are at least very intolerant of the appearance of corruption, and between backing Sarah Dowie who’s being investigated for harmful digital communications, thinking Jamie Lee Ross would make a good whip, going Trump-lite when New Zealanders hate that type of politics, and making the Labour government look competent by comparison despite its scandals, it’s definitely time for Bridges to go by any rational standard.

If Bridges’ approval is in decline, he’s just two or three months away from being in post-impeachment Nixon levels of popularity. That’s an absolute indictment on his stale, by-the-numbers attempt to keep National as a credible opposition-in-one-party. National voters have hung on because they felt cheated by coalition politics and because they wanted another crack at the election. Other leadership hopefuls have stayed united under Bridges because they have recognized 2020 as what it is: a hill so steep that any sane leader would refuse to climb it. So they backed stand-in Simon, and kept on the deputy nobody really wanted. They’re saying they don’t want to constantly change leaders like Labour did in opposition, but the reality is that if you can’t deliver, you need to try something new. National hasn’t managed it yet.

I can see two credible roads from here. One is to admit defeat soon, roll Simon as leader after the party vote continues to fall as he drags it down with him, and elect Collins as replacement. This represents a continuation of their existing strategy, but with the additional risk of alienating their centrist, socially moderate or socially liberal supporters, who won’t have a bar of Collins as their leader, although if there is a ready and credible political vehicle for such supporters to desert to that’s not aligned with the government, this option starts to make more sense.

The other is that someone like Adams, or Kaye steps forward and contests the vote, and wins. This person would still likely fail to win the 2020 election, so they would likely try to delay the vote late enough that they would not be held fully accountable for losing, and get another shot in as Leader of the Opposition. The problem with this route is that none of the people with limited degree of appeal and talent in National really want to try their hand as Leader of the Opposition until it looks more likely that they’ll win, as being opposition leader essentially starts a countdown timer on your career as party leader that might prevent you from becoming Prime Minister.

So either way you look at it, this is a poll that shows continued consolidation of the vote for Labour and National, something that should be viewed as bad news for both parties under MMP. They each need to think about how they make more friends, say, by lowering the threshold, deliberately splitting off parties with genuinely different ideologies with their blessing, or by other types of electoral reform.

I would note that, if we changed the law to allow any party that won their first seat outright, and assume that the Māori Party’s 0.8% is higher than 0.83%, then we would have 1 MP each for ACT, (from their electorate) the Conservatives, and the Māori party. We’d also retain four MPs from New Zealand First. This would be a very easy way to set a natural threshold that keeps out joke parties even on current results, allows small parties to die natural deaths when they become genuinely unpopular rather than simply take a little dip into micro-party territory, and makes it easier to clear that first hurdle into Parliament for new ideas to be debated on the national stage. I also scales well when next we need to increase the size of Parliament, and gives larger parties a chance to eat their friends a little without actually eating all of them.

(We actually have less MPs per person now than we did just before we added another 21 MPs to Parliament, so we might want to think about allowing it to grow sometime, and while we’re at it, changing the South Island quota (it currently just adds electorate seats as the North Island’s relative population to the South Island grows. A better way to handle it would be a minimum number of South or North Island seats, and a minimum number of list seats that the Electoral Comission can adjust without new laws) and the ratio of electorate seats to list seats to a set proportion, so that we still have enough list seats left for a proportional result if National or Labour ever get a result near 20% in a general election)

18 comments on “Pollwatch: 11/02/2019 ”

  • Comments are now closed

  1. swordfish 1
    12 February 2019 at 11:05 pm

    Newshub Reid Research also asked another (non-standard) question on leadership:

    Who should lead the National Party ?

    (1) All Respondents

    Collins 17.3%
    Bridges 13.4%
    Bennett 6.2%
    Adams 4.0%
    Mitchell 1.5%
    Other 8.0%
    Don’t Know / Don’t Care 49.9%

    (2) Intending National Voters only

    Collins 27.1%
    Bridges 19.6%

    https://www.newshub.co.nz/home/politics/2019/02/revealed-national-voters-prefer-judith-collins-to-simon-bridges.html

    .
    Here are one or two comparisons I’ve done on Preferred PM / Leader Performance:

    https://thestandard.org.nz/latest-reid-research-poll-result-disastrous-for-national-and-bridges/#comment-1582297

    https://thestandard.org.nz/latest-reid-research-poll-result-disastrous-for-national-and-bridges/#comment-1582604

    • Sacha 1.1
      12 February 2019 at 11:30 pm

      This linking to whole posts rather than specific comments thing is not a good feature, @lprent – any news on a fix?

      • lprent 1.1.1
        13 February 2019 at 7:59 am

        Subject to time. I have found that after spending 5 months working offshore last year that whenever I take a 5 day weekend, mostly what I want to do is read, sleep, do delayed apartment maintenance and play with the kitten.

        However I did start programming on the site last wèek. Hopefully that will continue this weekend.

        The bad news is that I have a new work project and the next round of booked vacation is Easter.

        The good news is that Dermot Nottingham now has a bankruptcy as well as a ankle bracelet. So he isn’t wasting my vacation time with observing court – so I have a lot more of it available.

    • veutoviper 1.2
      13 February 2019 at 8:24 am

      Swordfish made four main contributions on the “Latest Reid Research poll result …” post (plus several short thank you responses related to the four main ones.)

      The four were at comments numbered:

      2.2.1.1 – comparisions of the Newshub poll with election day results

      2.2.1.1.1.2 – relating to NZF polling

      44 – includes some comparisons on preferred PM/Leader Performance

      44.1.1.1 – more on comparisons of preferrec PM/Leader Performance

      So, presumably Swordfish was attempting to link to the two 44 thread comments.

      Well worth looking at these two comments by scrolling one of the general links provided in Swordfish’s 1 above.

  2. Robert Guyton 2
    13 February 2019 at 8:03 am

    “Despite his name, Bridges is underwater”
    Nice.

  3. patricia bremner 3
    13 February 2019 at 8:05 am

    This is a pendulum swing back to Labour and the Greens after Key, and a punishment by those who voted for Winston seeing him as a handbrake on Socialism. The refusal to vote isn’t counted, so it is hard to know if this group has increased since the election.

    Winston has mellowed, he has aimed to encourage a younger demographic to join NZ First, and Fletcher Tabuteau is an example of the new blood. Well educated resourceful and showing a good grasp of political smarts, he is a possible successor.

    So the lull in voter support could be misleading, as NZ First has always performed better than polls suggest. There are many who remember Winston as one of the few to query the behaviour of banks and the one% towards tax during the “Wine Box Inquiry”. and more have come to value the “Gold Card”

    It is pleasing to see the stirling efforts made across a wide range of problems being tackled by the Coalition under Jacinder Ardern’s Leadership being recognised by the electorate.

    The exposure of the seedy underside of the rightwing contrasted sharply, and some will not vote rather than be associated with the tenor of scandal and funding by Chinese power brokers. Had Jacinda Ardern not been even handed, they could have rallied, but she is working for the regions as well as Auckland, so a pause?

    The “Delivery” in 2019 will decide where polls go next.

    • greywarshark 3.1
      13 February 2019 at 9:04 am

      I think the next thing we need is a four year fixed term parliament and I would like to see the general election in October, no later. It gives a more reasonable time for a government to display its policies and implement them, so we can see them working, and know they are delivering what the country needs in the right fashion.

      I have always held back from having this longer period on the premise that it gives another year of backwardness and pain when the RWs get to wreak their malign gobbling-goblin mindset, but events have convinced me that the election cycle enters the consciousness too early with the three year period.

  4. greywarshark 4
    13 February 2019 at 9:25 am

    That’s a great post Matthew and thoroughly thought through and explained. I note you think
    that Labour getting too strong a position, not balanced by diverging views from smaller mates, would not be good.

    It made me think of the USA ,which since Russia capitulated and lost its position as leader of the communist phalanx, has become overblown with hubris and posturing and is OTT.

    That sage thought – Power tends to corrupt etc. and we wouldn’t want too much of that heady stuff getting into the NZ Labour Party, getting drunk with power and exciting new ideas that would be world-leading, and lead to knighthoods even. While we end up with hoodie nights killing dairy owners for cigarettes. Brave new world Old Labour. We want the New to stick to your knitting for da workers, and da small business, capisce!

  5. rata 5
    13 February 2019 at 9:40 am

    “Part of the reason I started modelling elections probabilistically” :).
    I should have stopped there but breezed the whole article.
    What a load of waffle…
    The article precise is basically that “any things possible”.
    Meaning the writer has no idea what will happen.
    To be fair only the last 6 weeks before the election really count
    and nothing before then means anything it’s just filler.
    And even then a maverick like Peters could upset the whole apple cart.
    May as well let Oliver the octopus decide.
    If nothing else it was enthusiastically presented waffle and
    if you are getting paid to write this waffle good on ya.
    I’m staying with Oliver as his accuracy rate is higher.

    • Robert Guyton 5.1
      13 February 2019 at 9:45 am

      If having your say about electoral possibilities this far out from the election is pointless, as you claim, why are you having your say about electoral possibilities this far out from the election?

    • Matthew Whitehead 5.2
      14 February 2019 at 6:49 pm

      Nobody gets paid to post here, FYI.

      I don’t analyse polls to tell you what will happen, at least not prior to the last week before election day, that’s a fool’s errand.

      I analyze them to tell you what the poll actually tells you about the political environment now. Of course it’s a lot of “anything could happen,” because we’re nowhere fucking near election day. I tend to be bigger on implications and try to be extra careful with pollwatch to actually inform people rather than give them political commentary, although I do a little of that.

      The time before the election campaign counts, but it doesn’t count for a lot because most people aren’t watching carefully, but they will be reminded of some of the more relevant facts.

      Yes, all my posts are going to be super nerdy. If you don’t like it, this blog has other authors.

  6. DJ Ward 6
    13 February 2019 at 10:05 am

    There is always the possibility that National gets a slight margin in party votes with the Greens and NZF below the threshold. You then get a National government ruling on its own.
    Support for National is not absolutely linked to support for Bridges. Few would abandon voting National just because the leader is a dud. It’s not like they have any real voting options. Voting conservative or blue green ends up weakening the Right vs Left vote. So voting will likely be strategic towards National despite the disapproval of Bridges.

    • Matthew Whitehead 6.1
      14 February 2019 at 6:50 pm

      5% is not a “slight margin.” That’s a lot of ground to make up.

      If this were a week before election day I’d say they would be enjoying another term in opposition and Simon would be forced to resign.

  7. AB 7
    13 February 2019 at 3:15 pm

    Matthew – how much volatility in voting behaviour does your model recognise?
    There have been a few events in NZ politics that shift a lot of votes quickly – most recently the Ardern leadership ascension and the Brash Orewa speech.
    There have been other smaller ones – the OPEC oil crisis and the Muldoon ascension to leadership that wiped out the Kirk/Rowling government. And Key’s leadership that seemed to turbocharge what had been the Clark governments slow loss of support.
    My feeling (without any real evidence) is that we may be entering a time of greater volatility that could confound probabilistic models.

    • Matthew Whitehead 7.1
      14 February 2019 at 7:03 pm

      Other than assuming parties that won at least one electorate last term will win one again, most of my assumptions are statistical. I also look at who various parties declare they’ll work with before the election, and in the lieu of such announcements, I assume parties in government will stay in government, (although I position centrist parties in the centre of any relevant graphs) and that the biggest party of the government will only invite into coalition those parties that are absolutely necessary to govern, and will pick the party closest to its own values. (This doesn’t stop them from working with others, of course)

      Essentially my model just takes the party votes from a poll, and randomly distorts them according to the probability curve of their margin of error a couple hundred times. (that is, smaller distortions are more likely, larger distortions less) I then have two hundred totalling sheets that look up those numbers and run an MMP seat allocation based on them, and any assumed electorate seats. (I can actually plug in poll numbers from electorate polling, too, seeing we got some of that last election, and have a random number generator determine if a candidate wins or not)

      Each MMP seat allocation check the total number of MPs and figures out how many are required to pass a law, then totals the seats for various groups of parties and determines who is kingmaker, or whether there’s a given type of government that’s natural. The main sheet then re-totals how many elections within the simulation fell into each scenario, and whether NZ First or the Greens fell under threshold in that election. I then re-run the simulation ten times by pasting the numbers into a table, so I have 2,000 results, and graph the results for you, putting each simulation’s results into its own table to graph the trend.

      The volatility of voter behaviour is evident to the same degree it is in the raw data- sadly with regular polling outside of election campaigns now ended with Roy Morgan out of the picture, we don’t really have any good check on the volatility of voter opinion during non-campaign times. (The continuing polls tend to perform a little bit better during campaign time, but have their own accuracy issues)

      This approach is statistically valid for any poll that isn’t a rogue, not that we always know a rogue when we see one, at least not until we have hindsight on it.