Polls Jan 2011

[Ab]

Barbecue time.

The latest New Zealand Roy Morgan Poll shows support for John Key’s National-led Government has risen to 59.5% (up 6%). Support for Prime Minister Key’s National Party is 55% (up 6.5%), the Maori Party 2.5% (unchanged), ACT NZ 1.5% (unchanged) with United Future 0.5% (down 0.5%).

Support for Opposition Parties has fallen to 40.5% (down 6%) with Labour Party 29% (down 6%), Greens 7%, (unchanged), New Zealand First 4.5% (up 1.5%) and Progressive Party 0% (down 0.5%).

[fixed_truth]

Maybe Goff is just into online panel polls which actually account for the mood of the swing vote.

In which case, yes,turn the heat up.

[CCS]

Lol. Any poll that puts NZ First at 8.9% is clearly a bullshit poll.

[Redneck]

I'd like to say you were right, I really would

[CCS]

Originally Posted by Redneck

I'd like to say you were right, I really would

You don't have doubts?

[GM]

Sample of 1718, so the margin of error = ~2.4% (50% result in poll), however: "The Horizon online poll differs from many others by asking "undecided" voters which way they are leaning". Without an 'undecided' option I'd add a dash of scepticism. What fixed_truth said about the poll "actually account[s] for the mood of the swing vote" is made up by the polling company, there's no real statistical backing to that at all.

[fixed_truth]

Originally Posted by GM

What fixed_truth said about the poll "actually account[s] for the mood of the swing vote" is made up by the polling company, there's no real statistical backing to that at all.

Could you explain what you mean by this? The Horizon poll asks the undecideds to give their preference (which reduces that group by about half) rather than only reporting the decided vote share (which most other polls do).

[GM]

Response bias

[fixed_truth]

Polling companies are politically neutral so I'm not sure how asking people who are undecided which way they are leaning would result in a 'response bias'.

[Lightspeed]

I'm not sure how useful it is knowing the political leanings of the undecided. It probably tells us something, but who knows what?

[fixed_truth]

It gives a clearer picture of how an election held today would actually turn out. But yeah, these people probably change their mind from month to month.

[GM]

Originally Posted by fixed_truth

Polling companies are politically neutral so I'm not sure how asking people who are undecided which way they are leaning would result in a 'response bias'.

Sorry I started to write something about but I also wanted to go to bed and tiredness won out.

The mood of the swing vote in the questionnaire is polled separately to the polling question by the looks of it and the results are aggregated. The results reflect a double question in other words, it's also possible that there's no neutral response to the first question, any questionnaire that forces opinion is going to introduce a small amount randomness to it because people who have no opinion on the subject will simply choose the easiest response - whatever that might be.

It's odd I think that the company has chosen to do it this way and I suspect they're just looking for an angle to present their data in the public arena, hence why I said that this method is made up by the market research company and presented as a legitimate polling result.

[GM]

I also suspect some sampling bias too - online poll filled with old people.

[fixed_truth]

Originally Posted by GM

The mood of the swing vote in the questionnaire is polled separately to the polling question by the looks of it and the results are aggregated. The results reflect a double question in other words, it's also possible that there's no neutral response to the first question, any questionnaire that forces opinion is going to introduce a small amount randomness to it because people who have no opinion on the subject will simply choose the easiest response - whatever that might be.

I don't think that your interpretation of how the poll works is accurate. People who are undecided are asked their party preference but they can still remain undecided. Reading DPF on this apparently all public polling companies do this but other companies only report decided voters, while as I said Horizon includes undecided and those who prefer not to say.

So there's no conspiracy theory here about a polling company making up research to get propaganda for their political cause.

Originally Posted by GM

I also suspect some sampling bias too - online poll filled with old people.

Yes I think sampling bias is a factor (along with not only reporting decided votes) in it's results being different than other polls.

[GM]

I never said anything about a conspiracy, its simply how market research firms decide to use data to cut a point of difference, and in this instance they've decided to use a method that is different to everyone else - a method that isn't explained very well at all I might add and one that causes more confusion than clarifying the voting situation (as all polling surveys should do).

Just look at the results and run a your common sense filter past it - 8% for the NZ first party?

[GM]

Originally Posted by fixed_truth

So there's no conspiracy theory here about a polling company making up research to get propaganda for their political cause.

I'd just like to add that this is an incredibly douchey thing to say when I'm trying to explain to you my reasons for suggesting their method might be dodgy. That and I never said anything about political cause or making up research, jesus h christ you can be fucking twat sometimes.

[spigalau]

So forcing an undecided voter to pick a side, wouldn't they (the undecided) normally not bother to vote as it ain't compulsory in NZ to do.

[fixed_truth]

Originally Posted by GM

Just look at the results and run a your common sense filter past it - 8% for the NZ first party?

The method (ie including the level of undecided voters) seems to be praised by most political commentators. It's most likely it's the online panel factor that exaggerates NZ First's share.

I would like to see the other major polls include undecideds in order to give a more accurate result.

Originally Posted by GM

I'd just like to add that this is an incredibly douchey thing to say when I'm trying to explain to you my reasons for suggesting their method might be dodgy. That and I never said anything about political cause or making up research, jesus h christ you can be fucking twat sometimes.

Got your period bub?

When you said that there's no real statistical backing to the polling and brought up response bias and made up that people being are forced to choose the easiest response - I took from that you were suggesting the company were angling for particular answers. My bad precious.

[fixed_truth]

Originally Posted by spigalau

So forcing an undecided voter to pick a side, wouldn't they (the undecided) normally not bother to vote as it ain't compulsory in NZ to do.

They're not forced to, they're "probed". According to Horizon this reduces the undecided group by about half. Also if people say they don't vote then they aren't counted.

[GM]

Originally Posted by fixed_truth

The method (ie including the level of undecided voters) seems to be praised by most political commentators. It's most likely it's the online panel factor that exaggerates NZ First's share.

I would like to see the other major polls include undecideds in order to give a more accurate result.

Got your period bub?

When you said that there's no real statistical backing to the polling and brought up response bias and made up that people being are forced to choose the easiest response - I took from that you were suggesting the company were angling for particular answers. My bad precious.

Without seeing the structure of the survey the reasons I gave are the most likely for skewed results especially if the market research company is capable at weighting (any statistician worth their salt can do this), no statistical backing simply means they've decided to zig while everyone else zagged reducing the rigor that results like these require when doing analysis.

But carry on in your ignorance spouting whatever bullshit you decide to tell everyone...

[fixed_truth]

You're the idiot who said were were talking about a "questionnaire that forces opinion".
Most polling companies probe undecided people to see which way they are leaning.

This poll is only different in that it includes the level of undecided voters in it's results and that it is an online poll.

Including undecided vote = good
Online poll = Not so good (atm)

It's that simple.

[CCS]

fixed_truth is the last word in polls and statistics!

[GM]

Originally Posted by fixed_truth

You're the idiot who said were were talking about a "questionnaire that forces opinion".
Most polling companies probe undecided people to see which way they are leaning.

This poll is only different in that it includes the level of undecided voters in it's results and that it is an online poll.

Including undecided vote = good
Online poll = Not so good (atm)

It's that simple.

No. I said: "it's also possible that there's no neutral response to the first question.

Fuck you really are a nitwit, why are you quoting me out of context and why do you continue to believe that there is some sort argument that you must win here - I work in this field and am simply giving my thoughts on why the survey is presenting data that doesn't quite fit with common sense.

[fixed_truth]

Originally Posted by GM

No. I said: "it's also possible that there's no neutral response to the first question.

Fuck you really are a nitwit, why are you quoting me out of context and why do you continue to believe that there is some sort argument that you must win here - I work in this field and am simply giving my thoughts on why the survey is presenting data that doesn't quite fit with common sense.

Your "thoughts" about the polls method are based on a premise that isn't true. People aren't "forced" into choosing "the easiest response". The polls method is essentially the same as all the others. It's just reported differently. Why can't you grasp this?

The survey is presenting data that doesn't quite fit with common sense because it's taken from an online panel. So add as many post icons as you like and call me mean names - your "thoughts" on the problems of there being no neutral response don't apply to the poll we are discussing.

[GM]

lol, you're just pathetic now. Ok, so since you think you have all the answers, tell me this: why didn't they weight the tables before running the results if there's uncertainty around sample bias?

Also, do you have the survey questions there with you and what are they?

And just to finish off, are the 'undecided that were probed and changed their mind' aggregated with those that had an answer straight off?

My thoughts are based on the information I have to hand, not some made up bullshit that you think is right.

Retard.

[fixed_truth]

Originally Posted by GM

My thoughts are based on the information I have to hand, not some made up bullshit that you think is right.

Like you had information to hand that the questionnaire forces people with no opinion to vote? ie

Originally Posted by GM

any questionnaire that forces opinion is going to introduce a small amount randomness to it because people who have no opinion on the subject will simply choose the easiest response - whatever that might be.

So now you're just ignoring my points and plain lying. Horizon's home page is here if you're interested. I can't be bothered with your bullshit anymore.

https://www.horizonpoll.co.nz/page/8...tailed-results

This result is based on respondents who did not rule out voting and who were eligible to vote. It combines the results for those who have already decided how they will vote, with the results from those who are not sure, but have a preference for one party or another.

[Draco T Bastard]

Originally Posted by fixed_truth

Polling companies are politically neutral

BS

so I'm not sure how asking people who are undecided which way they are leaning would result in a 'response bias'.

Depends upon how the question is asked.

[MrTTTT]

I don't like Draco T Bastard's political opinion.

[GM]

Originally Posted by Draco T Bastard

BS


Depends upon how the question is asked.

Sanity prevails, thought I'd taken some crazy pills.

[fixed_truth]

Originally Posted by Draco T Bastard

BS

Depends upon how the question is asked.

You're right. NZ's polling companies aren't capable of applying professional standards and techniques for preventing response bias in their polls.

I mean this question for example is obviously leading:

Q24. If a New Zealand general election were held tomorrow, which party would you vote for, or which do you have a preference towards at the moment?

/conspiracy theory

[MrTTTT]

because it uses the words 'New Zealand' it makes me want to vote for New Zealand First.

[GM]

Originally Posted by fixed_truth

You're right. NZ's polling companies aren't capable of applying professional standards and techniques for preventing response bias in their polls.

I mean this question for example is obviously leading:


/conspiracy theory

lol. you still don't understand what response bias is and how it may affect the results:

Q24. If a New Zealand general election were held tomorrow, which party would you vote for, or which do you have a preference towards at the moment?

...is a double question, it both asks which one you'd vote for OR which do you have a preference for. It should be 2 questions. Also, they should specify that it is a party vote (MMP).

Modelling for political leaning is quite difficult and there are a number of ways of doing it, some better than others. In this instance there may (and I stress may) be some issue with response bias as I've said in my previous posts.

[Golden Teapot]

Yep - not great questions.

But the important thing is that National will win again and that's all that everyone needs to know. And, specifically this means that the lefties may as well stay home and not vote at all.

[spigalau]

Join our HorizonPoll nationwide panel right now and we will
* make sure you have some influence – by letting decision makers know what you think
* put you in regular draws for prizes!
* put you in our March 31, 2011, draw for $1000 cash and your own iPad! (Wi-Fi, 16GB) valued at $799!

^ and that is why, one should never trust or believe a Horizon Poll.