The flat patch

Written By: - Date published: 10:18 am, July 5th, 2013 - 17 comments
Categories: election 2014, greens, labour, national, polls - Tags:

I’ve been having a look at the Roy Morgan trend. Oddly, I don’t see this Labour ‘flat patch’ that people are claiming is happening. What is happening is that the Greens are trending slightly down (as it usually does mid-term) and Labour isn’t rising fast enough to offset that. We’re still some way off the safe-zone for Labour+Greens, and the question is whether we’ll get there.

Labour averaged 33.1% in the past three months of Roy Morgans, that’s its highest quarterly average under Shearer. Labour has risen an average of 0.6 per quarter under Shearer, and kept up with that average last quarter. The graph below is 3-month rolling averages.

roy morgan labour green national

The combined Labour+Green figure is improving more slowly because of the Greens’ average 0.3% per quarter decline, giving Lab+Green a combined average 0.3% improvement per quarter and stalled last quarter.

In the graph below, I’ve got the rolling average of National vs Labour+Greens and shown the level Labour+Greens needs to beat to govern without other parties, and the level below which National won’t be able to cobble together a coalition. In between, New Zealand First is kingmaker (the conventional wisdom is that makes it odds on National – I wouldn’t put weight too much on the conventional wisdom when it comes to Winston).

roy morgan labourgreen national With 7-ish quarters to go until the next election, the trends would see Labour+Greens at 47.5% by election time, a percent short, that could potentially be made up by MANA, and National at 40.5%.

But, yeah, trends.

When was the last quarter that Labour averaged as high as it is now? December 2010. It had peaked at 33.6% the previous quarter. Less then a year later, Labour had plunged to its worst result in generations. You can’t bet on these trends, you have to make them happen.

Of course, the situation is a lot better now than then, because the Greens are averaging 12% now vs 8% back then and Nationa’s averaging 43%, not 51%. Then, the fear was that National might govern alone. Now, it’s that National might be able to govern with New Zealand First.

The situation is OK, not fantastic, but victory is in near reach for the Left. The fear is that Shearer will fluff election year like Goff did, and for much the same reasons, and that not enough of the lost votes will go to the Greens. I’m not worried about any current ‘flat patch’ as much as I am about Shearer coping when the acid goes on.

His inability to dismiss a couple of bad polls cleanly, the way that a seasoned politician like Key would, highlights the problem with the relatively raw Shearer.

The votes that have come Labour’s way seem to be more dissatisfaction with National than buying the Labour brand – and that’s because the Labour has yet to firmly brand itself and National, and Shearer and Key. The language and the lines of attack seem to shift each week. That’s self-defeating, and it suggests the same underlying problem that Labour had under Goff – it doesn’t know what it wants to stand for, so it doesn’t really end up standing for anything.

The nascent stuff is there, the party that will roll up its sleeves and get NZ working (jobs, fairness etc) vs the government that’s too busy cutting secret deals to look after ordinary people (SkyCity, job cuts etc) but its never articulated simply and over and over again.

Shearer’s challenge is to pull off that messaging while dealing competently with the predictable and unpredictable events that will arise, and convince New Zealanders that he has what it takes to be PM. The question for Labour is whether it believes he can do the job.

17 comments on “The flat patch ”

  1. Colonial Viper 1

    What ice field? Flank speed!

  2. Lanthanide 2

    One interesting point is that wasted votes, eg those that voted for a party that doesn’t make the 5% threshold, strengthen the largest bloc in parliament.

    At the last couple of elections, that has been National. But in this upcoming election, it’s likely that Labour + Greens > National.

    So if Winston doesn’t get back in, and any extra wasted vote from Maori P, UF, Act, Conservatives and Mana would favour Labour.

    The difference can actually be quite dramatic: a government of Labour + Greens + Mana + Maori (assuming 1 and 3 seats) if NZFirst get 5%, vs a government of Labour + Greens if NZFirst get 4.9%.

    • Colonial Viper 2.1

      Which would be why National is stalling on dropping the threshold to 3% or 4%.

      • Stalling? I thought they were pretending nobody even brought up the idea, along with that whole thing about removing the electorate exemption.

        Honestly though I think their self-interest plays well here, as taken as a whole the recommendations were terrible and shouldn’t be implemented, even though some individual ones were OK.

  3. Olwyn 3

    When Labour shows that it intends to take genuine, unequivocal steps toward turning our Ponzi economy into a real economy, when it firmly promises to stop the draconian treatment of beneficiaries and the undermining of the unions, when it guarantees we will not be spied on under their watch; when it promises to either tax the rich to achieve these things, or at least demand their positive participation; when they show such commitment that I can rule out their forming a grand coalition with, or giving confidence and supply to, National, then I may be able to take an interest in whether or not the polls are flat lining.

    • Murray Olsen 3.1

      I think we’ll be waiting a while, Olwyn. You mention all the things that make it more than just a problem that can be fixed by a telegenic leader. Just as well I never took much interest in polls, and luckily Mana and Greens exist.

    • Colonial Viper 3.2

      +1 Olwyn

  4. McFlock 4

    Pretty much agree with that assessment, James.

    re: the greens vote, my recollection (not just from 2011, for the “it was just coz of the Rena” crowd) is that they tend to campaign very well, then lose percentages during the term (probably due to larger-party exposure). I reckon over the next few elections they’ll stabilise at the ~20% mark. As someone said yesterday, if Bread Stale is predicting 10% for the greens, that’s a good omen for them 🙂

  5. Enough is Enough 5

    Key cannot govern next year without Winston.

    Ket is popular. Probably the most popular PM ever. He will retain strong populaity into the next election. If his 5 years of making a shambles of this country hasn’t dented his popularity yet, I can’t see anything damaging him in the next 15 months.

    That means his only potenetial partner needs to be taken out.

    We should be using our collective energy now to highlight how racist and out of touch Winston is. An ideal scenario is Winston polling 4.9%. How do we make it happen?

    • If Winston gets 4.9%, he deserves at least 4.9% of the house. Let’s not sink to their level please, if Labour can’t govern with just the Greens, but they can govern, then they should make the best of that, and hope to grow into a two-party majority.

  6. Pete 6

    I do think Christchurch will swing away from National next year. People aren’t very happy there. Auckland just got a big spending commitment, though. If Labour can keep to that and ensure they won’t need to sell their assets, I think they’ll make some headway.

  7. Yes 7

    Are you guys serious. The man ban is a killer…men will now flock to Maori, mana, nzf and oh..wait for it national.

    Redistribute the polls..national up 5.. Nzf up 1 Maori up 1 and mana up 1.

    Will that leaves the gays and greens holding up only 25% of the voting population.

    Vote Labour and be told what to do. At least Helen Clarke had balls

  8. Andrew 8

    These charts are interesting.

    Rolling averages are usually used to smooth out trend lines when you have very small per wave sample sizes (eg, 50-200 per wave) but for this poll you’ve got somewhere around 900 per wave.

    What’s your logic for using a rolling average on such a large sample?

    Given the frequency of this poll and the sample sizes, the overall trends are still fairly clear. (Estimates at n=900 are usually pretty reliable.)

    Aren’t you removing contextual effects, which is one of the key advantages of the Roy Morgan poll? Because it’s so regular and has a large sample size, it should be very useful for seeing the influence of recent political events. Isn’t this advantage being removed because you’re using a rolling average?

    • lprent 8.1

      Can’t answer for the James, but I’d say that it makes quite a bit of difference to the display. In the end what you are interested in with polls are trends, not individual data points. It usually takes several months for political events to go from them happening to making a voting intention effect in voting population. So anyone interested in politics looks at the trends rather than the points..

      Why? There is a lot of jitter in any poll (for a number of reasons). That in the Morgan poll is something in the order of +-3% from the observed variations (and their statistical variation is in the same order) – which is pretty easy to observe in the graph’s points. You can literally see a poll in one two week period for a party to drop by 3% and then jump up by 6%.

      So it makes sense to project political samples as a rolling average to look at the underlying trend.. It gets rid of the unusable jitter

      • Andrew 8.1.1

        “It usually takes several months for political events to go from them happening to making a voting intention effect in voting population.”

        That’s not really true though. The impact on voting intention depends on the specific event. Did you see how quickly John Key became preferred PM?

        Something I posted on DimPost:

        “Some survey methods should theoretically produce more stable results and less noise than others. When I talk about stability, I’m not talking about results that don’t change. Results should of course change to reflect ‘true changes’ in public sentiment. An unstable or volatile poll will fluctuate more around whatever the true value is, whereas a stable poll will fluctuate less around that value and less within the given margin of sampling error.

        I appreciate that it’s hard to know what the ‘true value’ is, especially for a political poll when you only get that value once every three years. Political polls are the most visible and public polls, but actually they are in the minority relative to all surveys carried out in NZ. There are many other surveys, that have been carried out over many years, and there is a large body of international research in the area of survey methodology, which suggests there are some good and not-so-good ways to carry out a survey. Someone recently said to me “Surveys are easy to do, but are hard (and expensive) to do well.” Really – anyone at all can bang a survey together and claim that the results are robust.

        I do understand that most people are not at all interested in survey methods, so views about which surveys are right/wrong will always be based on more obvious things such as frequency, sample size, online/landline/cell phone sampling, whether the results are consistent with our views based on events at the time, which blogs we read, and which hairs on the back of our necks stand up.”

        I’ve put together a matrix detailing NZ public poll methods, if anyone is interested. There is quite a lot of information available about how companies carry out their polls:
        https://grumpollie.wordpress.com/nz-public-poll-methods-grid/

        The following two websites are also useful if anyone is interested in what makes a good/bad survey:

        AAPOR: http://www.aapor.org/Poll_andamp_Survey_FAQ/5511.htm

        UK Magenta book (Chapter 5 in particular): http://www.civilservice.gov.uk/wp-content/uploads/2011/09/the_complete_magenta_book_2007_edition2.pdf

      • Andrew 8.1.2

        “In the end what you are interested in with polls are trends, not individual data points.”

        Actually I don’t believe this to be true either. If the poll is reasonably non-volatile, those changes between data points can provide insight into what the public really care about. Very useful for political strategists.

Links to post

The server will be getting hardware changes this evening starting at 10pm NZDT.
The site will be off line for some hours.