Joyce wrong on Collins poll

On Q&A yesterday Steven Joyce made a questionable claim in relation to a Colmar poll about the possible influence of Judith Collins on voting intentions.

Politicians are frequently poor interpreters of poll results. Many journalists are poor at reporting polls too, due to ignorance or due to the pressures of making an interesting or headline grabbing story out of a few numbers. Politicians may also be ignorant of poll interpretation, or they may be deliberately misrepresenting polls to try and score political points.

Joyce is usually very well informed about issues he has prepared for. He may have been taking advantage of interviewer and public ignorance of polls, unless he just got it wrong.

The Q&A Colmar questions and results were:

Do you personally think Judith Collins should remain a minister?

  • Yes 42%
  • No 42%
  • Don’t know 17%

Do you think her behaviour has been damaging to National’s level of public support, or do you think it will make no difference?

  • Yes, it has been damaging 50%
  • No it won’t make a difference 42%
  • Don’t know 9%

On balance, how well do you think Prime Minister John Key has handled issues with Judith Collins and Maurice Williamson? Would you say…

  • Well 46% (very well 11%, quite well 35%)
  • Not well 42% (not that well 29%, not at all well 18%)
  • Don’t know 11%

Which of these statements best describes how these issues will influence your vote in the upcoming election?

  • These issues will be a factor in your decision 23%
  • These issues will not have much influence 75%

(Sample size 500 eligible voters. The maximum sampling error is approximately ±4.4%-points at the 95% confidence level.)

Discussing the poll on Q&A Joyce said:

It’s the Labour-Greens voters that say, ironically, it would change their vote. I’m not sure where they would change them to.

Colmar pollster Andrew Robertson has commented on this.

Mr Joyce had clearly seen the report because he cited results that were in the body of it – results that had not yet been discussed by the Q+A panel.

Unfortunately, Mr Joyce either misread or misunderstood the results.

The question did not ask eligible voters if they’d change their vote. It asked whether these issues would be a factor in their voting decision. That’s a very different question. One is fairly blunt, and would need to be understood in the context what party people have changed their vote from and to. The other allows people to consider how important these issues are in relation to others.

When asked if these issues would be a factor in their own voting decision, most say the issues won’t have much influence.

Note that the question did not ask people if they would change their vote, it asked whether these issues would be one of the issues they would consider in their decision. There are many other issues, of course, such as education, jobs, housing, child poverty, crime, and the list goes on.

So Joyce was wrong. The poll didn’t ask anything about changing vote. And even if it caused a Labour voter to switch their party vote it could be to Greens, NZ First, Mana, Internet Party.

Or it could make it more likely they will vote, or more likely they won’t vote. These possibilities can all affect the outcome of an election.

The detailed report shows a breakdown of National and Labour & Green supporters.

Colmar Collins poll

In an election where a 2-3% swing could easily decide the outcome 23% of eligible voters is a significant number.

Even 8% of National Party supporters is notable. Analysis of the last election has shown that a significant number of potential National voters decided not to vote, which is a possible explanation for the drop-off in support for them from pre-election polls compared to the election itself.

If an issue like Collins/Oravida caused some national leaning voters to not vote or to switch to NZ First or Labour, and it encouraged more people to get out and vote, for Labour or Green or Mana or Internet Party or NZ First, it could have a major effect on the outcome.

What polling can’t do know is measure how much of an influence the Collins/Oravida issue will have in four months time.

The economy is expected to be a major decider, and associated with this jobs and perceptions of poverty.

Leaders’ personalities can also influence voters, and many people take little notice of politics until the campaign proper begins – this will be in August.

And the last election showed that a late and seemingly trivial issue can have a major effect. The Key/Banks cup of tea reshuffled a lot of tea leaves.

Colins/Oravida will have had some effect on an accumulation of voter perceptions but it’s impossible to tell whether it would decide the election.

It should be noted that there may be more yet to happen with Collins. She has obviously been under pressure and has acted irrationally. She could yet cause more problems for National, even to the extent of resigning.

Steven Joyce may have been trying to deflect from this.

For more details and discussions on this:

Polls – One News Colmar and 3 News/Reid Research

Two new polls out tonight.

One News/Colmar 3 News/Reid Research
National 47% (-4) 45.9% (+1.4)
Labour 31% (-3) 31.2% (-2.3)
Greens 11% (+3) 11.2% (-1.2)
NZ First 7% (+3.9) 4.9% (-0.8)
Maori Party 0.9% (-0.2) 1.5% (+0.5)
Conservative Party 2.3% (+1.0) 1.9% (-0.2)
Mana Party  0% (nc) 1.1% (+0.8)
ACT Party  0.3% (-0.1) 1.1% (+1.1)
UnitedFuture  0.1% (-0.2) 0.1% (+0.1)
Internet Party 0.4% (+0.4)


ONE News Colmar Brunton Poll 22-26 March 2014, margin of error 3.1% (at 50%)
Don’t know 13%, Refused 5% out of 1003

Click to access 140330_ONE_News_Colmar_Brunton_Poll_report_22-26_Mar_2014_prelim.pdf

This 3 News Reid Research poll was carried out among a random sample of 1000 eligible New Zealand voters from March 18 to March 26, 2014. The maximum sampling error for a simple random sample of 1000 eligible voters is +/- 3.1 percent at the 95 percent confidence level.


Daily Blog poll denial – “trying to manipulate” opinion

The Daily Blog fires a double barreled blast at the One News/Colmar poll with two posts calling foul – after they claimed their own self selecting online poll might be predicting the future. It predicted the opposite of what has happened.

Martyn Bradbury claims We have nothing to fear but TVNZ Polls.

Oh the National Party that is overseeing the largest erosion of civil liberties this country has seen since the 1951 waterfront lockout while contributing to the inequality that is dooming so many has enough support in the latest TVNZ Poll to govern alone? 


So despite this being landline poll, despite 13% don’t knows, despite it being held over summer when most Green Party supporters are out in the sunshine, John Key is sooooooo popular he can govern alone is he?


These opinion polls are not reflecting public opinion, they are trying to manipulate it.

Don’t believe the hype!

I wonder if this reaction has anything to do with the poll saying the opposite of what Christ Trotter and Bradbury suggested last week – that a Daily Blog poll showing Greens ahead of Labour was a sign of changing support. See Canaries In A Coal Mine: Has The Daily Blog Poll anticipated Labour’s Collapse?

This poll showed an apparent collapse in Green support from 14 to 8% and Labour holding steady.

And Frank Macskasy is also in denial in Latest TV1-Colmar Brunton Poll – Back To The Future IV?

It was a shocker of a poll on Sunday evening (23 February); the TV1-Colmar Brunton poll had National soaring to stratospheric heights.

It was a shocker for those hoping for and predicting something quite different.

There is no figure given for Undecideds/Refused to Say, which kind of makes the stats a bit dodgy. 

The “Don’t Know/Refused to say” was a whopping 13%!

That’s a sizeable chunk of voters who could yet decide the election outcome.

But how credible is a polling figure of 51% for any political party?

The answer? Not very.

It’s credible within a 3.1% margin of error at 95% confidence – that’s polling 101.

The highest Party Vote for any political party since the introduction of MMP in 1996, was 47.31%, achieved by National in the 2011 election.

So is 51% a credible indicator for National’s re-election chances?

Again, not very.

That’s different. It’s many months until the election, many things could change before then. But it’s a strong indicator that National is in a strong position right now.

In a February 2011 TV1-Colmar Brunton poll, National stood at… 51%. In fact, the 2011 Poll is a remarkable mirror of the current Colmar results; 

So it’s credible poll result now then, it’s a result that National has achieved before.

It is further worth noting that the actual election night result on Saturday 26 November 2011 was as follows;

  • National:  47.31%
  • Labour: 27.48%
  • Greens: 11.06%
  • NZ First: 6.59%

No other Party breached the 5% threshold.

At 34% current polling (by Colmar Brunton), this is still 6.52 percentage points above the 2011 election night results. Not a bad starting point to go into an election.

But 51% for National? Not in the realm of possibility. That is the polling they started from in February 2011 – and still they finished at 47.31%.

It is in the realm of possibility.

One thing that Macskasy fails to mention is that Labour got 33% in the February 2011 poll, so they dropped more than National, to 27.28%. Going by Frank’s logic 34% shouldn’t be a credible result for Labour either.

In any case this election is quite different to 2011.

  • National look like benefiting from an improving economy.
  • Labour have dumped two leaders since the last election and are struggling with their third, and have so far failed to show any sign of rebuilding, or retiring dead wood MPs and re-neweing talent.
  • Greens held their 2001 surge in support until recently but are starting to frighten some people, plus have had negative news with Norman’s association with Dotcom revealed, his confrontation with Colin Craig and Metiria Turei’s jacket and castle publicity.

I think it’s very likely National will drop below the 50% line, under MMP we have never had a single party with a majority and that’s most likely to continue.

But Bradbury and Macskasy sound like they are flailing about and lashing out in denial of a series of unfavourable poll results.

And they might be a little disappointed that their own very dubious blog poll was not the grand soothsayer they hoped.

It’s more than a bit ironic that Bradbury accuses One News and Colmar Brunton of deliberately manipulating opinion when that’s what he has just blatantly tried to do – unless he really believed in his own bull.





Three polls

Three polls – Fairfax and One News taken over identical periods (19 – 23 October) and Roy Morgan over a much longer time (14 – 27 October).

Fairfax/IPSOS One News/Colmar Roy Morgan
National 50.2 45 42
Labour 33.6 34 35.5
Green Party 10.7 13 11
NZ First 2.3 3.9 4.5
Conservative 0.7 1.6 2.5
Maori Party 0.7 1.4 1.5
Mana 0.7 0.3 0.5
ACT 0.1 0.5 1.0
United Future 0.1 0.1 0.5
Other 0.9 0.0 1.0


Period: 19 – 23 October
Sample size: 1030
Margin of error: +/- 3.1%
Don’t know or don’t intend to vote: 21.9%
Question: “Which way would you vote if an election were held today?”


Period: 19 – 23 October
Sample size: 1014
Margin of error: ± 3.1% – points at the 95% confidence level.
Undecided: 11%
Question: “If a general election was held today, how likely would you be to vote?”
NOTE: Those claiming they would be ‘quite likely’ or ‘very likely’ to vote have been included in the party support analysis.

Roy Morgan

Period: 14 – 27 October
Sample size: 847
Margin of error: not given but > 3.2%
Undecided: 3.5%
Question: “If a New Zealand Election were held today which party would receive your party vote?”

Two polls, similar and different

Today two polls were published, Fairfax/IPSOS and OneNews/Colmar. Some results are quite similar, some are notably different, especially for National (50 and 45). Other differences are within the margins of error.

Fairfax/IPSOS OneNews/Colmar
National 50.2 45
Labour 33.6 34
Green Party 10.7 13
NZ First 2.3 3.9
Conservative 0.7 1.6
Maori Party 0.7 1.4
Mana 0.7 0.3
ACT 0.1 0.5
United Future 0.1 0.1
Other 0.9 0.0

Particularly being a year out from the election these polls are indicative of current party support only and vary from poll to poll and between polls.


Period: 19 – 23 October
Sample size: 1030
Margin of error: +/- 3.1%
Don’t know or don’t intend to vote: 21.9%
Question: “Which way would you vote if an election were held today?”


Period: 19 – 23 October
Sample size: 1014
Margin of error: ± 3.1% – points at the 95% confidence level.
Undecided: 11%
Question: “If a general election was held today, how likely would you be to vote?”
NOTE: Those claiming they would be ‘quite likely’ or ‘very likely’ to vote have been included in the party support analysis.

One News Colmar poll

“Which political party would you vote for?”
  • National  46% (-3)
  • Labour 33% (no change)
  • Greens 14% (+5)
  • NZ First 3.3% (-0.3)
  • Maori 1.6% (+0.3)
  • Act 0.7% (+0.4%)
  • Conservative 0.6% (-1.1)
  • Mana 0.2% (-1.0)
  • United Future 0.2% (-0.5)
  • Undecided 10%

National appear to have taken a bit of a hit, the Henry inquiry debacle won’t have helped them.

Labour holding steady on 33 is being hailed as success, which is both good and sad.

Greens are picking up, Russel Norman has been strong holding National to account without petty politicking.

Preferred Prime Minister

“Now thinking about all current MPs of any party, which one would you personally prefer to be Prime Minister?”
  • John Key 41% (-1)
  • David Shearer 13% (+1)
  • Winston Peters 4%
  • Russel Norman 3% (+1)
  • Helen Clark 2%
    David Cunliffe 2%
    Jacinda Ardern 1%
    Metiria Turei 0.9%
    Grant Robertson 0.6%
    Hone Harawira 0.4%
    Steven Joyce 0.3%
    Shane Jones 0.2%
    Tariana Turia 0.2%
    Judith Collins 0.2%
    John Banks 0.2%
    Bill English 0.1%
    Annette King 0.1%
    David Parker 0.1%
    Other 2%
    Don’t know 24%
    None 4%
    Refused 1%

Details here (PDF).

Four polls – don’t rely on a single one

Roy Morgan’s fortnightly poll proves why you shouldn’t read too much into a single poll.

One News
Colmar Brunton
3 News
Reid Research
 Polling period 17-23 May 17-23 May “days after
May 16”
13-26 May
National               49      47.1 49.1            41.0
Labour                33             33.1 31.9         35.0
Greens        9   12.0 11.2         12.0
NZ First             4               2.2 3.2          4.5
Conservative         2           1.5 1.6          2.5
Maori         1       2.2 ?          2.0
Mana           1          0.5 ?         0.5
Act      –            0.2 ?          0.5
United Future          1           0.4 ?           0.5
Other          1.5

There’s a big difference between the first three and the Roy Morgan poll. The only obvious difference is Roy Morgan polled over a much longer period, and unlike all the others started before the budget.

The best lessons from this are:

  • Only one poll every three years counts
  • You can’t judge a lot off a single poll.

And if you look at the Roy Morgan trends it shows there are big variations from poll to poll.


Three polls

On Sunday One News and 3 News released polls, and today Fairfax have reported on their latest poll. All polling was done over a similar period, just after the budget.

One News
Colmar Brunton
3 News
Reid Research
17-23 May 17-23 May “days after
May 16”
National                      49                   47.1 49.1
Labour                     33                    33.1 31.9
Greens                   9             12.0 11.2
NZ First                     4                 2.2 3.2
Conservative                      2                     1.5 1.6
Maori                    1                 2.2 ?
Mana                         1                     0.5 ?
Act                      –                  0.2 ?
United Future                      1                     0.4 ?
  • One News/Colmar look to be rounded to the nearest whole number
  • Fairfax/Ipsos is from media reports that don’t give all the details
    – “UnitedFuture, ACT, Mana and the Maori Party 2.1% between them”

As usual this can’t be projected into a possible election result, we are mid term with a lot to happen before an election in November next year. And there are often late movements in support leading up to an election.

National will be happy with consistently high support hovering around them being able to rule alone (but voters have never yet obliged on that under MMP). It’s the economy and less risky stewardship of Finance, and that’s likely to be a significant factor in the election.

Labour should be worried that after major policy releases their support is drifting back. Fairfax lay much of the blame on David Shearer (see below) but associating closely with Greens must also be impacting.

Greens will be disappointed they are not gaining support.

Labour+Greens combined are struggling to compete with National:

  • Labour+Green – 42, 45.1, 43.1
  • National – 49, 47.1, 49.1

A Roy Morgan poll is also due this week but will cover a slightly later time period.

Vernon Small (Fairfax): National leaving Labour in its wake

Halfway through its second term, Prime Minister John Key’s National is riding high on 49.1 per cent support, up 4.5 percentage points since February, and would be able to govern alone.

Over the same period Labour had shed 4.4 per cent to 31.9 per cent, with respondents pointing to Mr Shearer as weak and negative, Ipsos pollster Duncan Stuart said.

It marks a sharp reversal from our February poll when the Left and Right were neck and neck.

The survey, taken in the days after the May 16 Budget, suggests an improving economic mood has lifted National’s poll ratings.

Tracey Watkins (Fairfax): Shearer’s invisible cloak thinning

If the results of today’s Fairfax Media-Ipsos poll are a precursor to the next election, the news is all bad for Labour – and not just because the poll has it shedding support, though that is bad enough.

But because it reverses a trend that had Labour slowly clawing into contention.

What changed? To lean on a cliche, the economy, stupid.

And with optimism on the rebound, National’s message at the election in 2014 looks like an increasingly potent one – we’ve taken our medicine, done the hard yards, and we’re starting to reap the gains. Why put that all at risk?

But something else may also be changing. Mr Shearer may be morphing from Mr Invisible to something worse in voters’ eyes. Mr Negative.

The overall poll results and trends conflict with Patrick Gower’s take at 3 News – Poll: Labour, Greens close gap on Nationalthe 3 News results showed National and Labour+Greens closer than the others but the movements from their previous poll were within margins of error.

Democracy demands no National veto of Parental Leave bill

If the Parental Leave bill passes through Parliament as expected National should reconsider their threat to veto it. With both majority Parliamentary support and strong poll support National don’t have any democratic justification for opposing it.

Increasing paid parental leave from 14 weeks to 26 weeks is supported by…

  • a majority in Parliament
  • a clear majority of the public
  • John Key says National was not opposed to the idea in principle

…so it looks like a no-brainer to do it.

I understand the need for fiscal restraint, but Governments are always making decisions on allocating their budget for necessary expenditure. There is a strong social case for increasing paid parental leave, and there is a double dose of democratic support – Parliament and people.

A One News/Colmar poll confirms strong public support for the bill:

“Do you support extending paid parental leave to 26 weeks?”

Yes 62%No 34%Unsure 4%

Don’t support extended leave: National Party supporters 53%

So there is nearly 50% support even from National voters. Labour and Green supporters and younger people overwhelmingly support it.

John Key says:

“There’ll be a time I’m sure one day when paid parental leave will be expanded but it has to be when we can afford it and not when we’re running up bill on the credit card.”

That day should be the day this bill is passed by Parliament.

The problem is the cost, National claim that will be $150m a year. This is disputed by Labour, but remarkably:

Officials are currently working out the actual annual cost to the taxpayer of 26 weeks leave.

The cost should have been worked out long ago – like at least approximately when the bill was drafted, or at least as soon as possible after the bill was drawn from the Member’s ballot.

And Labour don’t help their case when they are bombing the MRP share float, which will cost the country money – coincidentally by $100-200 million by some estimates.

But National should allow what Parliament and the people want.

If they veto the Parental Leave bill, especially at this stage of their second term, they risk a voter backlash, and that would be justified.

Democracy demands no National veto of the Parental Leave bill.

Two contrasting polls

One News Colmar poll:

National 43 (-6)
Labour 36 (+3)
Green 13 (+2)
NZ First 3 (-1)
Conservative 2 (+1)
Maori Party 1
United Future 1 (+1)

This is similar to the recent Roy Morgan poll.

3 News Reid Research goes against the trend in stark contrast:

National 49
Labour 30
Green 13
NZ First 3
Conservative 2
Maori party 1

Confidence in both Key and Shearer down a bit.

Contrasting result confusion may be an appropriate result- the polls were taken before the NZ Power announcement so the political climate may have changed somewhat and are now largely irrelevant.