Written By: - Date published: 6:31 pm, November 28th, 2007 - 67 comments
Categories: polls -
Tags: polls
The Herald reports that:
National has hit its highest rating – 51.3 per cent – in a Herald-DigiPoll survey.
Translated to votes, that would mean the party could govern alone.
Labour has slipped to 13.2 points behind National, increasing last month’s 12.4-point gap…
Helen Clark is still well ahead of Mr Key as preferred prime minister, favoured by 48.7 per cent (down 2.1) to Mr Key’s 36.7 per cent (down 0.6).
Translated to seats, and assuming party leaders with seats retain them, the poll results would give National 65 seats in a 121-seat Parliament, Labour 49, the Maori Party 4 and Jim Anderton’s Progressives, Rodney Hide’s Act and Peter Dunne’s United Future would have one each.
Facebook
Twitter
Scoopit!
But have you spotted the mistakes? Take a look at the sample, add the total and then ask a statistician for a comment.
300 sample size in AKL ?
people who said they would vote green Greens =2
Total no of extra people needed on total sample size to get green over 5% = 14
The obvious thing wrong with this poll is the greens only on 3.5%, down from 8.3% to 0.9% in Auckland. This has got to be complete rubbish.
The other weird thing is the Nats on 52% party vote, but Helen leading preferred PM by 12%. On one of the other recent polls Key and Helen were almost equal on preferred PM, however Labour was only a few percent behind on party vote.
Does anyone have any insights into how the different polls are carried out? i.e. the questions and the order they are asked?
Add up the numbers, 0.6% missing of the vote and also “decided” voters, hmmm wheres the undecided??? Has the Herald (another boo boo as not an independent or respectable source) decided that the “don’t knows” don’t count? and yeah Luke, wonder what the questions were? Remember if I was doing a poll for a political party, I can dress the stats to suit my purpose i.e. make the sponsor hear what they want to hear. The Herald presented a poll from another source stating the gap was about 5% and two days later in their own poll becomes 13%. Go figure
Sounds like they use the same people who count marchers
Sounds like they use the same people who count marchers
most likely. “excuse me sir, last month I asked you who would you vote for, you said (whoever), same again sir?”
Perhaps, since you folks were so keen on John Armstrong’s review of the Key DVD, you would like to hear his opinions on this poll result?
Hint: he doesn’t think it’s good for Labour.
http://www.radionz.co.nz/audio/national/aft/the_panel_-_part1
Lamprey – you do realise that the Herald doesn’t run the poll itself (this helps prevent it being hacked in favour of Teh Party)
I really think Labour have missed an opportunity by not bringing in a pre-election poll ban at the same time as they pushed through the EFB. As so many polls are being published using flawed methodology and with questionable objectivity I think a real case could be made that they become at least as anti-democratic as any anonymous, big-budget attack-ad campaign. The idea certainly isn’t without precedence overseas.
Also, it would make election night much more exciting.
I don’t know if National would be too happy if this “govern alone” idea takes hold in the media. Problems:
1) It’s based on both NZ First and Greens being out of Parliament. This is not going to happen.
2) A general, vague anti-government sentiment will suit National as long as people are focussed on step one: getting rid of Labour. Once that becomes presented (falsely) as inevitable, then the “antis” will start to focus on step two: What *kind* of government will replace it? National really don’t want people to think about that, until their votes are safely locked up in the ballot box.
For example, on closer inspection, it turns out that the caricature of Aunty Helen’s Sisterhood Nanny State will remain unaltered by National. Section 59? No change. Smoking in bars? No change. Civil Unions? No change. And so on. It’s your basic conservative approach to social progress, throughout history: oppose, then keep.
The Kiwiblog crowd are too thick to have worked this out yet. But it’s going to be very funny watching it dawn on them s-l-o-w-l-y …
Actually the greens have lost all credibility because a lot of the sort of people usually associated with them have been exposed as wannabe terrorists, or are busily protesting on behalf of the wannabe terrorists. Most Kiwis are just glad the cops lock up people who wanna kill people, and now that the evidence has leaked all over the place, you lefties can’t pull that ol’ “civil liberties” “suppression of the truth on a legal technicality” game.
Double understandering- stats to suit my purpose i.e. make the sponsor hear what they want to hear.
Try reading the whole statement not bits.
Phil
Are these the same ‘wannabe terrorists’ against whom the Solicitor-General advised that no terrorism charges should be laid?
And obviously you’re a fan of trial by media. I personally prefer 800 years of civil liberties, starting with the premise that I’m innocent until someone (in a court of law, with proper representation and judged by a panel of my peers) finds me guilty.
“Does anyone have any insights into how the different polls are carried out? i.e. the questions and the order they are asked?”
- Look up ‘Textor/Crosby’. They conducted questionnaires to provide the Nats with stats that support their views, and which show them which messages appeal to peoples prejudices to sell messages like anti-Maori and anti-PC so they didn’t have to focus on proper policy during an election.
Perhaps, since you folks were so keen on John Armstrong’s review of the Key DVD, you would like to hear his opinions on this poll result?
Hint: he doesn’t think it’s good for Labour.
well no shit Sherlock, biggest understatment of the year. As for his DVD review, I’m not interested in personal views from media as one eyed.
“I’m not interested in personal views from media as one eyed.”
- Then you’re surely not interested in ‘Amateur in New Zealand – Meet John Key, because I’ve yet to find anything since seeing it that’s more biased.
Oh, what spectacular re-writing of even very recent history, Gruela. The S-G didn’t say that terrorism charges should not be laid. He said that there were extremely serious activities going on, that the Police should be praised for their investigation, that they were right to ask that charges be laid, but that the anti-terrorism law was so badly drafted that he could not use it to bring anti-terrorism charges against the accused.
Hardly an exoneration for the alleged terrorists, is it?
Bye bye Winston. Bye bye Greens.
And Gobsmacked thinks the right’s getting it slowly. Heh.
“Bye bye Winston. Bye bye Greens.”
See 1999, 2005 …
Both redbus, I have to practice what I preach. I wouldn’t be surprised if John was told to tow the party line as some staff have most likely been told (a certain business editor was given a slight push for his more ‘leftie” views) and this is a bit of a smoke screen to please certain elements of readership.
Back to me, sorry, the Herald is just too much opinionated rubbish. To be honest, if the results were the other way I still wouldn’t read into it too much. As for the DVD, was rated M for moron
Okay, Insolent, no exoneration for those caught up in the police action. My bad. But the charges laid against most of them will be a lot less serious than was first implied by the police.
And I still say that it was disgraceful and frightening that basically the entire police case against them was leaked to the media. This cuts against every intent of ‘impartial justice’. Whoever was responsible should be immediately kicked out of the force.
Gruela
“And I still say that it was disgraceful and frightening that basically the entire police case against them was leaked to the media.”
Did it ever cross your mind that it may have been leaked by their side, to prevent a fair trial taking place?
Also, I notice that you are quick on the “ban it” trigger for anything media related. Have you had some history there or something?
Double
Brilliant. I have to admit, the thought never crossed my mind, (and it should have.) Dur on me.
I suppose I’ll have to find out when the first leak was made, and if the defendants were still in jail at the time. (It could have been one of their lawyers, but I’d be more inclined to believe it was from within the police. Also, the police had more to gain from the leaks, since they were coming in for so much stick about the way the raids were conducted.)
And I’m not big on banning the media. Au contraire, mon amie, I want there to be a much freer market amongst the NZ media, rather than have it dominated by a couple of big overseas companies like it is now. And if regulation is needed to attain this, then so be it.
Of course the charges brought against the accused will be much less serious than the Police implied, Gruela. The Police alleged that the people involved were engaged in terrorist activity. The Police assumed–wrongly as it turned out–that the law Parliament passed would be sufficiently robust to lay charges. Instead the law was not robust: so shoddy, in fact, that the S-G refused to lay charges. It had nothing to do with the evidence or the substance of the Police allegations.
I would be very careful about claiming that the Police leaked the affidavit, Gruela. The affidavit I have seen was annotated by one of the accused. If you read the documents, it is pretty clear who leaked them. It’s less clear why that person leaked them. But if you’re planning to overthrow the government through armed insurrection, and assassinate politicians, then you may not be using the same logic process as the rest of the population.
I must say, Gruela, of all the Standard’s left-wing commenters, you do make the most effort to engage in a discussion, rather than insult your opponents.
People should be cautious about interpreting these polls from a stats point of view. Firstly, the sample population is never reported ie how many people had to be phoned in order to get 914 decided voters. Secondly, the true denominator (all people agreeing to participate in the poll) is unknown, as only decided voters are reported. So a measure of undecided voters in never included.
As to why polls produce different results, that comes down to the uncertainly factor. The polls typically report their estimated standard error ( /- 3%, bound to be rounded). Standard error is an estimate of the standard deviation of error – or a measure of the likely spread of error. It is not a measure of uncertainty about the results likely to be obtained from repeated polls. To obtain that measure, confidence intervals should be used ie the 95% confidence interval. Basically take the standard error and multiply it by 1.96. So the uncertainty around the National party poll result is 51.3 /- (3 X 1.96) or 51.3 /- 5.9%. Thus the spread of results if this poll was repeated 95 times (hence the 95% confidence interval) is 45.4% to 57.2%. 95% is an arbitrary value, and there is no reason a 99% confidence interval cannot be constructed, but the confidence interval would be wider as a consequence.
Confidence intervals provide some estimation as to how poll results might differ, all things being equal. However, they will not estimate why polls might provide consistently different answers. As other commentators have suggested, the answer lies in the method – differences in sampling techniques and bias in constructing questions being the most probable explanations.
The other main issue with poll results is how the data gets cut up. The overall sample has a standard error of 3%, but once you start dicing up the data into cities, the standard error changes (as the ample is smaller) and thus the confidence intervals around these estimates will be much wider (and possibly even uninformative as they may get so wide). And of course, not all analyses are reported, so we don’t know what the other results are. Some will be interesting, others not. But some might simply be interesting because the data has been diced up some much the result is random error (due to small sample size), rather than reliable estimate. Within stats, this is known as data dredging and is widely discouraged. Not so in marketing, though.
Gob – Fair enough point about the “conservative approach to social progress” i.e. oppose, then keep. It can be observed on occasion.
As a social liberal (as well as an economic one) I’m quite happy with that for the most part.
You may wish to note that National did not actually oppose any of those three examples you gave…
Civil Unions: Conscience Vote – although I admit that many opposed as individuals, presumably being socially conservative or believing their electorate to be so. See http://en.wikipedia.org/wiki/Civil_unions_in_New_Zealand
Smoking in Bars: Conscience Vote. See http://en.wikipedia.org/wiki/Smokefree_Environments_Amendment_Bill
Section 59: Was this a conscience vote too? I recall National voted for it after their amendment to make it slightly better drafted by specifically allowing police discretion was accepted. See http://en.wikipedia.org/wiki/Child_Discipline_Act
Frankly, this is hilarious. The poll (and presumably all the others from the last 18 months) are the result of bias, or use improper methodology, or will be rectified when people realise. Is it just inconceivable to you that people do not like this government?
Oh, the ingratitude!
After the TVNZ Colmar-Brunton then Herald poll is traditionally the least favourable to Labour. However, I think Labour supporters have to accept that the Herald running a partisan anti-government campaign is hurting Labour support in Auckland. To that extent, the Herald’s poll is a self-fufilling outcome of the papers recent overt (and more damagingly, constant covert editorial slant) move to the hard right.The challenge for Labour is to come up with some sort of counter to a paper that clearly sees itself now as a right wing change agent.
Tom Semmens:
Quite right. The Herald is an enemy of the state. The Government should either change its tax status, make it illegal to campaign against the Government in election year, or, at least, have the editors and senior management shot. The socialist revolution is more important than liberty and freedom. The Herald’s actions of late are a suitable justification for suspending democracy and making Helen Clark the prime minister for life.
People should be cautious about interpreting these polls from a stats point of view. Firstly, the sample population is never reported ie how many people had to be phoned in order to get 914 decided voters. Secondly, the true denominator (all people agreeing to participate in the poll) is unknown, as only decided voters are reported. So a measure of undecided voters in never included.
Good stuff, exactly. Might have an answer to that this morning
Within stats, this is known as data dredging and is widely discouraged. Not so in marketing, though.
Andrew, your the man.
Thank you TomS, agree with that
Prick, you’ll have to work harder on your sarcasm. That was weak.
Frankly, this is hilarious. The poll (and presumably all the others from the last 18 months) are the result of bias, or use improper methodology, or will be rectified when people realise.
To me, sounds like Andrew could be a statistician, as my wife is. Any graphic in a newspaper is most likely to be presented statistically wrong. Example is using a table and along the x axis have 10, 15,20,30,40,60,80. which is used in conjuction with anm article to make an impact. With stats, the art of lying is made easy. Otherwords you can manipulate to suit your needs and when presenting to an uneducated auidence (in terms of stats) well what do you think?
“…Teh…”
- does anybody understand why he does this? Is it his thing – like how the characters on ‘Friends’ had a thing. Theirs were funny, ‘teh’ is just lame.
Isn’t it funny that IP will comment on other people and how he’s judged them to be more constructive and engaging, then spurts out the usual tory sort of giggerish in the form of
“The Herald’s actions of late are a suitable justification for suspending democracy and making Helen Clark the prime minister for life.”
I guess he’ll understand that of all the right-wing commentators, no-one would consider him the one who makes the most effort to engage in discussion!
On a relevant note, does anyone have a link for any of the NZ political meta-polls? I’ve seen them mentioned every now & then and wouldn’t mind checking them out…
The fact that the Green vote in Auckland is so out of kilter with what it normally is in these polls must throw the value of the poll into doubt.. Usually the Greens poll higher in Auckland than nationwide, in this poll they get 0.9%
Good post Andrew Jull
What I take from this is
Andrew Jull is not DPF