Hard News by Russell Brown

Read Post

Hard News: Polls: news you can own

59 Responses

First ←Older Page 1 2 3 Newer→ Last

  • tussock,

    Though after the last election, polling of non-voters suggested more National voters stayed home because National "was going to win" than did those who might have voted for opposed blocks.

    So declaring a victor before polling really does make everyone less likely to vote. And more so for National matches their polling results quite well.

    Or maybe Labour just GOTV better every time. Greens similarly poll well with younger intending voters, only to have young people ever less likely to actually vote on the day. Maybe the Greens need to join the Labour program and help young people get to the polling station, at least in the seats with the widest total margin between intent and result for them.

    Winston polls better with old folk, and they still do vote very strongly, so their comparative proportion rises on the day (not to mention the climbing proportion of old folk in society over time, and how they don't have to skip work to vote). Supporters dying out indeed. If anything it's the religious brigade who are slowly disappearing.

    Since Nov 2006 • 611 posts Report

  • Russell Brown, in reply to Craig Ranapia,

    And, frankly, FUCK “COMMERCIAL SENSITIVITY”. You’re been presented as expert opinion, you don’t get to pretend you’re a disinterested party in how your industry is perceived.

    Is there any part of his maths you question? Because that post is really just maths. And it's on the UMR website.

    Auckland • Since Nov 2006 • 22850 posts Report

  • Russell Brown, in reply to Pete George,

    ‘Don’t know’ in the Colmar poll were not 18%, they were 13% (up 3).

    My mistake. It's the don't know + refuse to say figure, which is up 5%, but I accidentally left out out the second part. I've corrected the wording.

    Auckland • Since Nov 2006 • 22850 posts Report

  • Jason Kemp,

    What I get from all of this is that news media take the surveys and then make up whatever they want to editorialise on. And that is regardless of what a more careful analysis might show.

    In the US there is a huge volume of polls around the election and FiveThirtyEight was able to use those to calibrate some more meaningful insights. I don’t think we get real analysis here for the most part.

    From having seen other election cycles in NZ I know that what happens is that the two main parties will get closer (as the election approaches.)

    Also ( MMP notwithstanding) it is very hard for one grouping to make it 3 elections in a row and so to some extent that pushes the momentum over to a Labour led grouping.

    It is - however Labours election to lose. There does still seem to be very little coming from Labour that will win votes and they need to fix that promptly.

    I would think going in as underdog is a good thing as it may get more people to the polling booths on the day. But in my view I hear a lot more sense coming from the Greens than I do from Labour at this point.

    Also Winstone - he is a Rasputin / Rumplestiltskin character who sleeps between elections and only really wakes up during the election campaign.s so he can do some more dog whistling.

    As for conservatives, internet party and ACT. We don’t need any more grandstanding idiots in parliament. I don’t see that any of them have real support bases at all.

    The internet party has some policies that sound interesting but nothing that the Greens or Labour couldn’t co-opt.

    Auckland • Since Nov 2006 • 368 posts Report

  • Craig Ranapia, in reply to Russell Brown,

    Is there any part of his maths you question? Because that post is really just maths.

    No, because that kind of maths is well above my pay grade. But, yeah, I would like pollsters commenting on political polling (and how they are reported) to be more forthcoming about any potential or actual conflicts of interest they may have in their client lists. If it's a good enough bar for David Farrar, it's good enough for everyone else.

    North Shore, Auckland • Since Nov 2006 • 12370 posts Report

  • Ianmac, in reply to Sofie Bribiesca,

    Yes Sofie. Gower asks the question which shows he wants the answer to match his intent and even if the interviewee gives a conflicting answer Gower ploughs ahead with his version of what was said even when blatantly untrue. Then reports his untrue version on the 6pm News.
    Ought to be a law against it. They shoot mad dogs don't they?

    Bleneim • Since Aug 2008 • 135 posts Report

  • Russell Brown, in reply to Craig Ranapia,

    No, because that kind of maths is well above my pay grade. But, yeah, I would like pollsters commenting on political polling (and how they are reported) to be more forthcoming about any potential or actual conflicts of interest they may have in their client lists. If it’s a good enough bar for David Farrar, it’s good enough for everyone else.

    It's not quite the same thing. White wrote up his review for his blog on the SayIt website, where he talks about all kinds of things, including UMR's own polls on various issues.

    I suppose Mediawatch could have noted that one of their many clients is the Labour Party when they asked him about the blog post. But it's not as if he's doing political commentary in the media, or he's a prominent activist in a major political party. And this is just a simple review of historical numbers. I don't see how there's a conflict of interest.

    Auckland • Since Nov 2006 • 22850 posts Report

  • ThoughtSpur,

    I strongly disagree that the way the poll is conducted doesn’t skew the result or impose bias on the result. A phone poll via land line is almost laughable.

    “A survey in 2012 by Canstar Blue found that a third of Generation Y in New Zealand have abandoned landlines; 14 per cent of Generation X and 8 per cent of Baby Boomers.”

    Those percentages will have increased since 2012 (as is the international trend) – so, if you want to know what the elderly, reactionary and laggard think – go right ahead and give them a tinkle.

    If younger voters are going to influence the next election then you’ll need some more sophisticated, or at least meaningful methods of inquiry.

    TV3’s Reid poll coverage was a cringe-inducing joke. Patrick Gower was beside himself (in a state of ecstasy) calling the election – months out and before a single shot has been fired in anger.

    http://www.stuff.co.nz/the-press/christchurch-life/8761615/Landline-phones-heading-for-extinction

    Auckland • Since Aug 2007 • 15 posts Report

  • Rob Stowell,

    I'm curious if anyone else sees a consistent media fascination with ACT and equally consistent tendency to ignore Mana. They seem to have similar levels of support - Mana have a safer seat than ACT, since ACT's is courtesy of another party, and it seems possible that Mana could pick up another seat (Waiariki). Both are tied to one side of the aisle - no coalition options.
    Is it because ACT has a past life as a more significant party; because it's wealthy, and the party of the wealthy; because it's so nutty and liable to implode in various ways that make good stories?

    Whakaraupo • Since Nov 2006 • 2120 posts Report

  • bmk,

    What I find interesting is that the polls in the US and Australia when aggregated prove very accurate but in NZ this isn't the case at all.

    One factor may be the lack of exit polls. Pollsters can use the demographic data obtained from exit polls to help weight their polling via demographics.

    I'm not sure whether NZ polling companies do exit polls and don't publish the data for this case - or if this is legal. As a keen political observer I do wish exit polls were allowed here and I really don't understand why they aren't.

    Since Jun 2010 • 327 posts Report

  • bmk, in reply to BenWilson,

    But these are actually questions that possibly could be statistically answered. I’m curious whether polling companies would attempt that, considering that they do get competitive advantage out of having better results. By better I mean “results that more accurately predict what customers really want to know – what the actual election results will be”.

    Agreed. They already weight by other factors. You would think if they statistically observed that say there were consistently a 3% discrepancy between those who said they would vote National and those who actually did then you would weight for it.

    In the same way as (in the US anyway) they weight for ethnicity etc and match it up for typical turnout figures.

    Since Jun 2010 • 327 posts Report

  • BenWilson, in reply to Jason Kemp,

    Also ( MMP notwithstanding) it is very hard for one grouping to make it 3 elections in a row and so to some extent that pushes the momentum over to a Labour led grouping.

    Quite the opposite. It's very hard for them NOT to make 3 elections. That's the median and the mode across NZ history since the 1930s. Mode="most common value". In other words, Labour will actually be bucking the trend if they manage to win this time. In the MMP period, 3 terms is the only number of terms (we can't count this one, it might not be finished) that parties have managed.

    A phone poll via land line is almost laughable

    I can't get why they are still so ubiquitous. It's not like the cost of the damned phone call to cell phones is a really significant part of a survey cost.

    Auckland • Since Nov 2006 • 10657 posts Report

  • bmk, in reply to ThoughtSpur,

    A phone poll via land line is almost laughable.

    This is only a factor if the people who don't own a land line vote differently from those who are in the same demographic and do. So say half the people between 18-30 don't have a land line if the other half vote the same way then it doesn't really matter. It just makes it harder for the polling company to get their pool of 18-30 year olds so they are more likely to weight the answers they do get - this can lead to higher variance from poll to poll. From what I can remember from (I think it was on 538?) in the US they found that polls that included mobiles weren't any more accurate than land line only polls.

    I get tired with some people on the left saying "but polls only count land lines" as if that will magically provide them with victory. I remember this delusion at the last election too.

    On your fundamental point I do agree though that pollsters will need to change and adapt their methods for the future; there will come a time when polling only land lines will be too unrepresentative.

    Since Jun 2010 • 327 posts Report

  • Kumara Republic, in reply to BenWilson,

    I can’t get why they are still so ubiquitous. It’s not like the cost of the damned phone call to cell phones is a really significant part of a survey cost.

    I read somewhere that the biggest complexity of cellphone polling is that a lot of people have multiple cellphones - like one for home and one for work - that aren't officially registered like landlines.

    The southernmost capital … • Since Nov 2006 • 5446 posts Report

  • Pete George, in reply to bmk,

    Roy Morgan include cellphones.

    Yesterday's polls plus the last Roy Morgan poll for the four biggest parties:

    National - RM 45.5, CB 47, RR 45.9
    Labour - RM 31.5, CB 31, RR 31.2
    Green Party - RM 14, CB 11, RR 11.3
    NZ First - RM 3.5, CB 7, RR 4.9

    The polling periods are different, especially for Roy Morgan which was earlier, but it's Labour supporters complaining the most about polls (not coincidentally because they don't like their results) but there seems to be a lot of consistency for their lack of traction.

    Dunedin • Since Dec 2011 • 139 posts Report

  • Pete George,

    A pollster on cellphones in polling:
    "The company that I work for has no policy on “…refusing to call cell phones.In fact, they do randomly dial cell phones for telephone surveys."

    "At present my view is that, in New Zealand, non-response is a far far bigger source of error than non-coverage."

    It's worth reading the whole post.
    http://grumpollie.wordpress.com/2014/03/13/rob-salmonds-post-on-cell-phone-polling/

    Dunedin • Since Dec 2011 • 139 posts Report

  • bmk, in reply to Pete George,

    Roy Morgan include cellphones.

    Exactly. I was aware of this but as the results you list above show there isn't really a significant difference. I think the non-polling cell phones isn't a big deal.

    I think there is some flaw in the methodology that the pollsters use which means National and Greens seem to be typically overstated while NZF understated. But I don't believe it's a result of cellphones.

    And still this difference is only around 3% I believe not as significant as some people like to imagine. I think the adjusted tracking poll talked about above is the best of the kind we have available in NZ. That's the aggregated poll I would predict would be most likely to call an election correctly.

    Since Jun 2010 • 327 posts Report

  • Pete George,

    Assuming that National and Green variances last election will be repeated this election are risky. There's no way of knowing if polling companies have changed their weightings, and each election has different circumstances.

    I think National last support late in the last election because of the cup of tea fiasco, because some people didn't want them to rule alone (they'd been polling over 50%) and because some people didn't vote because they knew National would 'win' easily.

    I could swing the same way for different reasons this time, but it could just as easily swing the other way. For example people may be motivated to vote against a potential Labour+Green+NZFirst+ Mana+Internet Party coalition. That could change as people evaluate possibilities closer to the election. 13% is a lot of undecided voters.

    Dunedin • Since Dec 2011 • 139 posts Report

  • bmk, in reply to Pete George,

    Assuming that National and Green variances last election will be repeated this election are risky.

    This is true. And if it were only 2011 that followed this pattern then I wouldn't be so sure. But it's also based on 2008 where the same thing. Most polls had National 2-3% higher than their final result. While the Greens I believe have always polled higher than their actual election tally.

    I also can't see people breaking to National unless National drop from their current point. I think there is a natural ceiling for parties and for National I think it's about 48%. In fact the polls showing over 50% support for National probably hurt them more than it helps them as some people won't like the idea of a one-party majority.

    Since Jun 2010 • 327 posts Report

  • James Green,

    In my academic (and non-political) research, we've also found non-response to be an enormous issue. It's clear that people are increasingly sick of answering cold calls. The political pollsters clearly have some tricks up their sleeves, as their reponse rate is higher than we've been able to get, but in our experience, upward of 50% aren't even interested in talking.

    I'd also recommend Andrew's best estimate of the actual methods of the five public polls.
    http://grumpollie.wordpress.com/nz-public-poll-methods-grid/

    Limerick, Ireland • Since Nov 2006 • 703 posts Report

  • Russell Brown, in reply to James Green,

    It’s clear that people are increasingly sick of answering cold calls.

    I blame brand perception surveys. Being asked to quantify my feelings from 1-7 on 20 minutes' worth of repetitive, barely distinguishable questions is not at all my idea of a good time. The dropout rate during those surveys must be huge.

    Auckland • Since Nov 2006 • 22850 posts Report

  • James Green,

    I’d also point out that political polling has grown up around trying to predict two horse races, with relatively little interest in those dwelling in single digits. In contrast, in New Zealand, we have a very great interest in the results of our small parties.

    A second important issue in play here is that in certain respects MMP makes political polling “easier”, but probably less accurate. In an FPP system you need to be able to predict the outcome of each race. This then means that you’d need to know which races were so blue/red that they effectively didn’t need polling, but would need to be able to poll individual seats to understand their outcome to forecast an overall result. In contrast, because MMP is proportional, there is no great need to understand voter patterns in anything less than the whole country.

    So to contrast polling in NZ with say Nate Silver’s 538 forecasts, Silver has far, far more polls to work with, as the pollsters try to understand each senate race and electoral college seat and congressional district. And not just more polls, but much better resolution. This brings me back to my initial point, that in New Zealand we are really interestd in the fate of the small parties, but our National polls have quite low resolution, as well as being less numerous.

    Limerick, Ireland • Since Nov 2006 • 703 posts Report

  • James Green, in reply to Russell Brown,

    I blame brand perception surveys. Being asked to quantify my feelings from 1-7 on 20 minutes’ worth of repetitive, barely distinguishable questions is not at all my idea of a good time. The dropout rate during those surveys must be huge.

    They are certainly part of the problem, along with sales calls, donation soliciting etc. There are periods of the day where it is almost not worth answering the phone. The other issue, that you highlight there, is that the response of some to people not taking part is to make the surveys longer and continuing to try to bother more people until you achieve your desired sample size, both of which jade more people and make response worse.

    And then there are all the other surveys, for students, for your car repairer, for the community organisations you belong to. Many of which are too long, too ambiguous, or don't involve any actual sampling, and are sent to every person they can... </rant>

    Limerick, Ireland • Since Nov 2006 • 703 posts Report

  • James Green, in reply to Russell Brown,

    Being asked to quantify my feelings from 1-7 on 20 minutes’ worth of repetitive, barely distinguishable questions is not at all my idea of a good time.

    Though presumably at least one person suggested that "Spark" sounded "dynamic" and "innovative" ...

    Limerick, Ireland • Since Nov 2006 • 703 posts Report

  • Hebe, in reply to Russell Brown,

    At the least, the increase in undecided voters -- some of whom may be turning away from labour and Cunliffe, remember -- seems worthy of comment.

    Did you riff off Bomber for this post? I inadvertently set him on to this today.

    Beloved and I clocked it last night as the only interesting trend line in the polls: don't knows being up 5 per cent to 18 per cent in a month. That's a humungous leap. Admittedly I'm slightly obsessed with Don't knows/won't vote since the methodology surrounding them was changed a few years ago. Now most political journos don't seem to recognise them as at all interesting. In MMP, those voters are crucial.

    And don't knows are quite different creatures to won'ts. Yet they are no longer differentiated by pollsters. Major fail. I bet the parties try to figure that.

    Don't knows can only be good for NZ First.

    Christchurch • Since May 2011 • 2899 posts Report

First ←Older Page 1 2 3 Newer→ Last

Post your response…

This topic is closed.