The Dim-Post

July 24, 2011

Deep question

Filed under: media,polls — danylmc @ 8:12 am

Why does the SST keep running stories based on Horizon polls? Is it because the staff there don’t know anything about polls? Or is it because they think that their readers don’t know anything about polls and the staff don’t really care if the stories they publish are nonsense?

About these ads

51 Comments »

  1. SST paid good money for the Horizon annual subscription. That is all..

    What else do you want from your weekly lifestyle/food/real estate/mish mash of bad opinion and vacant blandness, actual news?

    Comment by andy (the other one) — July 24, 2011 @ 8:26 am

  2. Why do they include the Progressives, when there will be no Progressives after November? The Horizon poll is cr*p.

    Comment by Inventory2 — July 24, 2011 @ 8:30 am

  3. They also include NZ First as a credible coalition option. Crazy.

    Comment by Simon Poole — July 24, 2011 @ 9:08 am

  4. …John Hartevelt said the deals were understandable but were also the “kind of stuff that makes voters sniffy about MMP”. He said such accommodations could appear anti-democratic, and in the case of Ohariu the “stitch-up is the smelliest”.

    Ohariu? A dirty deal being done to elect one MP to an electorate seat is going to make voters sniffy about MMP? Well, if they’re particularly stupid perhaps, John – but what’s your excuse?

    Comment by Psycho Milt — July 24, 2011 @ 9:13 am

  5. It is cheap. That is the only reason they use this rubbish poll.

    Comment by Barnsley Bill — July 24, 2011 @ 10:33 am

  6. the staff don’t really care if the stories they publish are nonsense

    Given that the story doesn’t even state the support for Labour and National, this would seem to be the case.

    Comment by bradluen — July 24, 2011 @ 10:36 am

  7. Isn’t it also a bit weird that they are talking about Labour forming a coalition with the Progressives when Jim Anderton is retiring after election?

    Comment by Amy — July 24, 2011 @ 11:34 am

  8. “They also include NZ First as a credible coalition option. Crazy.”

    Well, yes, crazy as they are polling below 5%, but Peters has yet to delcare his intentions about running and yet his party polls better than many of the other minor parties. He is campaigning at a grassroots level out of the eye of the mass media:

    http://mauistreet.blogspot.com/2011/04/winston-peters-targets-maori-vote.html

    If and when he declares, his support will jump. I can imagine him attacking Don Brash and John Banks and the whole ACT thing – and by association John Key and National – in a way Labour have failed to capitalise on. This week’s political deals are a gift to Peters.

    I wouldn’t write him off yet.

    Comment by Dotty — July 24, 2011 @ 11:58 am

  9. Danyl, are you suggesting that Horizon polls are nonsense?

    Feel free to argue that, of course, but let’s have some evidence.

    I’m aware of 3 differences between Horizon polls and the other main opinion polls in this country. These differences could well explain the difference in results between Horizon than the others, and not in a way that necessarily makes Horizon less reliable:

    (1) Horizon asks undecided voters to say who they think they are most likely to vote for, whereas the other polls discount them as non-voters

    (2) Horizon’s norming criteria include income level, whereas (according to Chris Trotter) the norming criteria for the other polls do not

    (3) Horizon poll a group of people who have agreed to answer by internet, then norm the results based on demographic information in an attempt to cancel out any bias stemming from this, whereas the other polls use the people they get by cold-calling, minus those who refuse to participate, then norm the results based on demographic information to cancel out any bias stemming from this.

    are you asserting that these differences, or others that I am unaware of, make the Horizon polls rubbish?

    Comment by Kahikatea — July 24, 2011 @ 12:09 pm

  10. Kahikatea: item three strikes me as deeply problematic. I would be very surprised if internet access is as common as owning a landline (the latest stats I could find were from 2009 in which Statistics NZ reported 75% of households had internet access, vs 91% with a landline at home). You can cold-call more or less randomly. But you can’t randomly contact via internet (unless you spam, which is illegal and at the required scale would probably result in a prosecution in NZ). It seems to me that Horizon’s pool of respondents is more of a self-selecting pool and from a smaller group to start with.

    Comment by Stephen — July 24, 2011 @ 12:37 pm

  11. I reckon it’s all a cunning right-wing plot to persuade 4.9% to vote for Winston Peters.

    Labour should respond by promising to cut super. Lose enough votes to Winston, and Goff’s in.

    Comment by sammy — July 24, 2011 @ 12:39 pm

  12. Kahikatea: Well, I’d think (3) is the killer. Rule number 1 of any statistical work is “never use a self-selecting sample”.

    Sociologists would laugh at this methodology. And I think that says everything we need to know about Horizon’s polling.

    Comment by Idiot/Savant — July 24, 2011 @ 12:39 pm

  13. Kahikatea: No 1 is incorrect. UMR, Colmar Brunton, Roy Morgan and Digipoll all probe the undecided voters as to any minor preference. This is near standard methodology in any political poll.

    Also you can make judgements on pollsters not just by methodology, but by results. You can see with all the above companies how close they got the 2008 election result. And the average of all the public pollsters was deemed by an academic study to be very very close to the final result.

    While the only result one can judge Horizon on is the Auckland Mayoralty where their last poll said Brown was 31% ahead and he in fact won by 14%.

    I’ve blogged on this in more detail at http://www.kiwiblog.co.nz/2011/05/horizon_polls.html.

    Comment by dpf — July 24, 2011 @ 1:22 pm

  14. Internet polls are not necessarily horrible, but phone polls very usually do better:

    http://www.fivethirtyeight.com/2010/05/telephone-polls-closer-to-mark-in-uk.html

    The British firm YouGov, the Internet pollster with the best record (except in the 2010 UK general, but everyone missed that one), claims to actively recruit most of its panel, i.e. not just rely on a “sign up here!” button on their website, in a way that they hope reduces self-selection bias. Haven’t heard any similar claim from Horizon.

    Comment by bradluen — July 24, 2011 @ 1:36 pm

  15. Ohariu? A dirty deal being done to elect one MP to an electorate seat is going to make voters sniffy about MMP?

    I dunno if I’d put it the same way as Hartevelt but it still irritates me. Voters shouldn’t be having their options for electorate representatives reduced because of a much bigger political strategy. If that’s acceptable then we may as well just do away with electorates entirely. But in this case, two candidates, including one whom a majority of voters might’ve actually wanted to represent them more than Dunne, are effectively being withdrawn because of petty squabbles between parties over electorate-winning karma.

    If it’s kept then a possible improvement to MMP would be to replace the FPP-style electorate voting with something like STV, just to fix vote-splitting issues.

    Comment by izogi — July 24, 2011 @ 1:43 pm

  16. “You can cold-call more or less randomly. ”

    sure, but cold-calling still doesn’t select people randomly – a voter who lives in a house with 5 registered voters has only 20% of the chance of being selected as a voter who lives in a house where he or she is the only eligible voter.

    Comment by Kahikatea — July 24, 2011 @ 2:05 pm

  17. I dunno if I’d put it the same way as Hartevelt but it still irritates me.

    It would annoy the hell out of me if I lived in the Ohariu electorate, too. But the current shenanigans there are a product of the FPP system used for electorate votes, nothing to do with MMP.

    Comment by Psycho Milt — July 24, 2011 @ 2:25 pm

  18. Horizon has no screening for multiple entries. Try joining their poll with more than one email address. It has no checking mechanism at all. the results are being skewed by individuals with multiple identities.

    Comment by Barnsley Bill — July 24, 2011 @ 2:27 pm

  19. The part of the SST article that’s about the poll is a straight rewrite of this Horizon release:

    http://www.horizonpoll.co.nz/page/141/less-than-1-

    Note that the article on Horizon’s website has a link to the spreadsheet of poll results. So does the version of the press release on Scoop.

    This implies that the reporters at the Sunday Star-Times are too fucking lazy to open an Excel spreadsheet to get the actual results of the poll they’re writing about.

    For the record, it’s National 37.7%, Labour 27.5%, out of 91.7% expressing a preference.

    Comment by bradluen — July 24, 2011 @ 2:28 pm

  20. @Psycho Milt, yeah, no argument that it’s a consequence of the electorate voting system. But (being nitpicky) it is to do with MMP because that’s how MMP is defined to handle electorate voting. Possibly the fact that parties don’t usually need to rely on electorates for their overall influence makes these kinds of silly trade-off games that penalise electorate voters even more likely under MMP. Hopefully this gets sorted when there’s a review, if MMP’s kept. It shouldn’t be hard to fix.

    (Oops, I accidentally logged into my izogi WordPress account with my previous post.)

    Comment by MikeM — July 24, 2011 @ 2:38 pm

  21. Psycho Milt wrote: “It would annoy the hell out of me if I lived in the Ohariu electorate, too. But the current shenanigans there are a product of the FPP system used for electorate votes, nothing to do with MMP”

    Actually, I think they’re largely a consequence of the so-called ‘one-seat rule’, which means that a party with less than 5% support can be eligible for list seats if it wins one electorate.

    Comment by Kahikatea — July 24, 2011 @ 2:50 pm

  22. “It has no checking mechanism at all. the results are being skewed by individuals with multiple identities.”

    That should favour smaller parties (should they be involved in the practice), it’s a lot easier to get a significant change to say Act or Mana with a few extra votes compared to the larger parties.

    Comment by Pete George — July 24, 2011 @ 2:58 pm

  23. Barnsley is quite right. Between the two of us we manage to submit about twenty surveys to Horizon each time round. Just doing our bit to ensure cheap junk is never worth anything more than cheap junk.

    The dumbarses don’t even know I live in Australia.

    Comment by Adolf Fiinkensein — July 24, 2011 @ 3:20 pm

  24. Actually, I think they’re largely a consequence of the so-called ‘one-seat rule’..

    I doubt anyone in National seriously imagines Peter Dunne might bring additional MPs into Parliament with him. In this instance, they’re happy to accommodate the one UF MP, Dunne, because the alternative is Labour gets the seat and National gets zip.

    Comment by Psycho Milt — July 24, 2011 @ 3:34 pm

  25. Danyl, you should probably blame this on the editors as the story has no byline at all – probably because it’s a straight copy of the media release as bradluen suggests above and also because none of the journalists want to put their name to the results.

    Comment by Dave Guerin — July 24, 2011 @ 3:52 pm

  26. Regarding the comments made by people above about the deals being made by National:

    At least the deals are made more or less openly. And frankly it is RealPolitik. It makes sense to have such strategic deals for all of the parties involved. It is simply Labour’s arrogance that denies them the ability to do the same with the Greens and so on. They did an implicit deal with Jim Anderton in Wigram in the past, and how his heir apparent is running for Labour. Unfortunately the Maori Party’s inability to come to an agreement with the Mana Party could hurt them.

    So for example, if Labour were to do a deal with the Greens, then the Greens could possibly win a seat, such as Coromandel.

    Comment by David in Chch — July 24, 2011 @ 4:43 pm

  27. then the Greens could possibly win a seat, such as Coromandel.

    Sandra Goudie won the seat for National with 22,653 votes. The Labour and Green candiates combined managed just over 12,000.
    The window of opportunity for the Greens to nab that seat has well and truly passed.

    Looking down the electorate list, there really isn’t anywhere that stands out as being prime Green territory. Maybe they could arrange a deal in Nelson? Nick Smith won the seat, but Labour and the Greens combined got more party votes. Smith’s retirement (whenever that might be) could open up the seat for competition. It would also require Maryan Street being pushed out of the way.

    Comment by Phil — July 24, 2011 @ 5:19 pm

  28. “So for example, if Labour were to do a deal with the Greens, then the Greens could possibly win a seat, such as Coromandel”

    I think at this point the Greens would rather maximise their party vote than win an electorate seat. At least, why bother if it requires an endorsement from Labour – I think winning a seat without an endorsement from Labour would be a good source of mana.

    and if they did want to win an electorate, there are probably about 15 that are more promising than Coromandel.

    Comment by Kahikatea — July 24, 2011 @ 5:23 pm

  29. @Phil, I think Rongotai, Dunedin North and Wellington Central are the most promising seats for the Green Party to win.

    Comment by Kahikatea — July 24, 2011 @ 5:25 pm

  30. The story is all too typical of the SST which combines a pleasing (to me) layout with dire content across all sections of the paper, including especially dim-witted and mean-spirited editorials and column writers whose only contact with the world seems to occur through the media.

    Comment by Tinakori — July 24, 2011 @ 5:40 pm

  31. Tinakori, I’ve inferred from your previous comments that we probably have rather different beliefs on most topics, but in this I am 100% of your mind.

    Comment by Stephen — July 24, 2011 @ 6:18 pm

  32. From Metiria Turei’s Twitter:

    “All reports of the Greens doing a deal over Ohariu are WRONG. C’mon some media, check your sources, come to us direct.”

    NZ media get it in their mind that something is so, construct a story around it, pull in facts to support their story. Underresourced or merely incompetent, the results are similar.

    Comment by George D — July 24, 2011 @ 7:30 pm

  33. This can go a couple of ways. Colmar etc. may predict the election better a week or less out, but it’s distinctly possible that come the election, Horizon’s polls over the last few months will have been far better predictors than Colmar Brunton polls which have had National at ~55%.

    Comment by Graeme Edgeler — July 24, 2011 @ 7:43 pm

  34. GE for a person of some intelligence you show a lamentable failure to understand the basis of polling.

    They measure a persons intentions on the day of polling. For you to suggest that in the event National’s share of the party vote declines from it’s present level of 55% to something below 50% cum election day, that somehow demonstrates a flaw in Colmar-Brunton’s accuracy is bizarre.

    I think you are a lawyer? Do I take it you would expect a trial judge to ignore all evidence relating to events which occurred less than three months before the date of the crime? That’s what you’re asking the pollsters to do.

    Comment by Adolf Fiinkensein — July 24, 2011 @ 8:25 pm

  35. David: At least the deals are made more or less openly. And frankly it is RealPolitik. It makes sense to have such strategic deals for all of the parties involved.

    Absolutely, it makes perfect sense for the parties because that’s where all the incentives are. I just think the fact that these incentives still exist (for removing voters’ options and for exploiting overhang and for letting in some small parties but keeping out others with equal proportions) is a fault in MMP. It should be addressed if it possibly can be. The electoral system’s meant to represent voters as fairly as possible, and it shouldn’t provide opportunities or incentives for parties to use tactics to skew that representation in their favour.

    Comment by MikeM — July 24, 2011 @ 9:17 pm

  36. Adolf – that’s not what I’m saying at all.

    Comment by Graeme Edgeler — July 25, 2011 @ 2:02 am

  37. I believe that this is one of the things that needs to change as a result of the MMP review, assuming the referendum chooses to keep MMP.

    Parties that win a seat but less than 5% shouldn’t be able to coat-tail in additional MPs – those MPs that win seats should effectively be counted as independent MPs (and I’m thinking not get Party Leader budgets) until their party gets better than 5%. That simple change would rule out all of these shenanigans.

    Of course it’s hard for me to propose that without looking self-interested, as the Greens are the only party that does better than 5% and doesn’t tend to win electorate seats; such a rule-change disadvantages all the other smaller parties other than us. Nevertheless I believe it’s the right thing to do.

    Comment by James Shaw — July 25, 2011 @ 6:29 pm

  38. James,

    If you are going to do that, then why not look at lowering the representational threshold also?

    Comment by Andrew Geddis — July 25, 2011 @ 7:19 pm

  39. I think 5% has proven to be a bit high, 4% might be a better balance.

    And to resolve the single seat qualification change that to require two or more electorate MPs to bypass the threshold requirement and to get party rights and funding.

    Comment by Pete George — July 25, 2011 @ 7:34 pm

  40. or raising it if you really want to get rid of act/mana/progressives.

    Comment by will — July 25, 2011 @ 7:38 pm

  41. I don’t want to get rid of act/mana/progressives, I’d prefer as wide a representation of parties (or independent MPs) as possible with minimal distortions.

    I doubt I’d ever support Act or Mana but if they get enough votes for seats their voters deserve to be represented as much as anyone.

    Comment by Pete George — July 25, 2011 @ 7:43 pm

  42. Surely the simplest solution is a 1% threshold and a 100 seat parliament.
    If you’re good enough to get 1 in every 100 people to tick your box, good on you.

    Best of all, it gets rid of 20 of the less useful assholes cluttering up the back benches on either side of the house.

    Comment by Phil — July 25, 2011 @ 8:30 pm

  43. James,

    If you are going to do that, then why not look at lowering the representational threshold also?

    Because James’ party consistently polls above 5%, so only other voters are disenfranchised by the threshold. Why would he care about people who aren’t him?

    I think 5% has proven to be a bit high, 4% might be a better balance.

    Closely followed by:

    …if they get enough votes for seats their voters deserve to be represented as much as anyone.

    So, which is it? “Enough votes for a seat” is 1/120th of the vote, or 0.833%. But you think 4% would be a better “balance” (a balance between what isn’t clear), so in fact you don’t think their voters deserve to be represented as much as anyone.

    Comment by Psycho Milt — July 25, 2011 @ 8:38 pm

  44. I think 5% has proven to be a bit high, 4% might be a better balance.

    Maybe. A risk with lowering the threshold is that it increases the number of smaller parties that can take advantage of an incentive to seriously exploit overhang, which I think could happen sooner or later.

    There’s already an incentive for existing big parties to exploit overhang of smaller parties, which bothers me about MMP. No idea how to fix it. If they wanted to, National or Labour or any party reliably getting more than the threshold could split themselves into 2 parties, tell supporters to give the electorate vote to Party A and the party vote to Party B. Party A gets a ridiculous amount of overhang, undermining all proportionality, and joins in a natural coalition with Party B.

    Some voters might punish a party for trying such tactics so obviously, but it probably won’t be so painfully obvious. In essence it’s what happens when a party like The Maori Party focuses on electorate seats only then tells all its supporters to give their party vote to Labour, which from memory has happened in the past. (My memory’s very sucky for these types of things, though.)

    Comment by MikeM — July 25, 2011 @ 9:14 pm

  45. Mike M wrote “A risk with lowering the threshold is that it increases the number of smaller parties that can take advantage of an incentive to seriously exploit overhang, which I think could happen sooner or later.”

    I really don’t think it does increase that risk. For a small party to ‘exploit’ overhang in that way requires it to win electorates, so I think it’s progression to doing this would consist of it proving its ability to win electorates, not getting into Parliament exclusively by winning list seats.

    Comment by Kahikatea — July 25, 2011 @ 9:20 pm

  46. The Maori Party focuses on electorate seats only then tells all its supporters to give their party vote to Labour, which from memory has happened in the past.

    I don’t know if they tell them how to vote, or the voters are just smarter and get the most out of their two votes they can.

    Comment by Pete George — July 25, 2011 @ 9:41 pm

  47. The Maori Party has always campaigned for two ticks. The Greens explicitly encouraged tactical vote-splitting in the Maori seats. Don’t recall Labour doing so, and don’t see the incentive for them to do so given their frosty relations with Turia.

    http://waatea.blogspot.com/2008/06/turei-anti-turia-on-vote-splitting.html

    ***

    Don’t remember who it was, but after the 1993 referendum, somebody in National proposed the grand rort of splitting National into two parties, then running NationalA on the list and NationalB in the electorates. Bolger vetoed the idea.

    Comment by bradluen — July 26, 2011 @ 8:43 am

  48. “Don’t remember who it was, but after the 1993 referendum, somebody in National proposed the grand rort of splitting National into two parties, then running NationalA on the list and NationalB in the electorates. Bolger vetoed the idea.”

    Maurice Williamson?

    Comment by Kahikatea — July 26, 2011 @ 9:45 am

  49. It’s the Cold War all over again. :-P

    @Pete, You’re right and I guess my intended point (which I didn’t make well) was that smart voters really shouldn’t be able to step outside the proportionality of MMP either, given that one of its big selling points is that it’s proportional.

    Comment by MikeM — July 26, 2011 @ 10:08 am

  50. Williamson sounds right, though I can’t find a reference.

    Rob Salmond has a good post at Pundit on the Horizon Poll and the tendency of some to only hear what they want to hear.

    Comment by bradluen — July 26, 2011 @ 1:39 pm


RSS feed for comments on this post. TrackBack URI

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

The Rubric Theme. Create a free website or blog at WordPress.com.

Follow

Get every new post delivered to your Inbox.

Join 419 other followers

%d bloggers like this: