Two Very Differing Polls, and TV1's Possible Biasses

pn410

When two polls taken within a week of each other come up with quite different results it's understandable that many people will question the value of polls, especially when the pollsters were wrong about the recent Australian election results.

Some said the difference was due to the (slightly) different poll dates when political interest was high due to National's budget leaks. Others said this would make no difference.
Right-wing blogger David Farrar says both results can't be right, but does not say which is right; left-wing blogger Pete George says it's all to do with margins of error, and others say it's due to the different ways in which the two polls were taken and analysised.

Such as the different numbers approached on landline and cellphone;  the "scaling" to compensate for the population that was not polled, and up to 15% of those asked who refused to answer or were undecided.  

And that's before adding the unstated number who did not answer their phones! (Click here and here for links to the main arguments.)

Possible Bias
But even more questionable —and important because it hints at bias by part of our national media whose job is to report and comment fairly, not to influence how we see the news —was that TV1 News only reported one of the polls, the Colmar Brunton poll they'd  commissioned.  

Add to this their heading "National back in front while Labour takes a dive (my emphasis) in latest Colmar Brunton poll", and the comment by their political commentator Jessica Mutch McKay that the Colmar Brunton poll result "reflects poorly for government". As possible causes she instanced the "Wellbeing budget not playing well for government" and a string of supposed Labour negatives — the "backtracking" on the Capital Gains Tax, the teacher strikes, Bridges Budget leak, and the "failure" of KiwiBuild.

What's happened  to the TV1 News impartiality? The two polls released their results on the same day, their results were very different (see table below) but TV1 News on air and Mutch McKay's comments only mentioned the poll where results favoured National — the one where Labour support had supposedly "dived."

The Party Poll Results
The table below shows just how much the results differed.

Would vote:	Colmar Brunton 4-8 June		Newshub/Reid 29 May - 7 June
	Percent	Change	Percent	Change
Labour	42	-6	50.8	+3.3
National	44	+4	37.4	-4.2
Greens	6	0	6.2	+1.1
NZ First	5	+1	1.28	-0.1

 Other parties 1% or less.

Labour down 6% or up 3.3%, both can't be correct. 
  
Mutch McKay also overstated the support for the smaller parties whose support National may need at the next election. She says the poll result "shows there's an appetite for these less traditional parties".  

I doubt it showed anything of the sort, even allowing for polling method and margin of error.  

ACT, New Conservatives and TOP (The Opportunities Party) received 1% each,  a long, long way off the 5% parliamentary theshold, unless National gives them an electorate, as it has for ACT. 

Though more encouraging exposure and support from the media would also help!

Preferred Prime Minister
The two polls were close on "preferred PM": 

Percentages --	Simon Bridges	Judith Collins	Jacinda Ardern
Colmar Brunton	5	6	45
Newshub/Reid	4.4	7.1	49

Winston Peters had 5% support in the Colmar Brunton poll. He wasn't mentioned by Newhub/Reid. 

Note that neither poll totals 100%. Presumably the other 35-44% —more than one-third of the people interviewed! — had no opinion or refused to answer. Other than showing Ardern's lead and the Bridges-Collins contest no further comment is needed.


"Performance"
The polls' questions on "performance" also differed. Colmar Brunton asked about  "economic performance". 38% thought government was performing "poorly." Newshub/Reid asked about "government performance". 72% thought them performing "well."   

The questions were different but related. One would have expected some coincidence,  but the results were polls apart. Am I imagining that the pollsters' choice and wording of questions suggest bias?


Budget leaks 
 Newshub/Reid also asked whether National was right in seeking out and releasing details of the Budget before Budget day. 32.6% answered "yes", 55.4% "no" and 12% were "undecided. It would have been interesting to know which party the yes, no and undecided would vote for, especially when (or if) their party differed from their "expected" answer.

Cannabis
On support for legalising recreational cannabis, the polls also differed. Colmar Brunton had 34% approving compared with 41.7% for Newshub/Reid. 

While Greens MP Golriz Ghahramin commented on the influence of the media in shaping opinion, my thoughts are that there will be no meaningful poll on cannabis use until government spells out its intended referendum. Opinions on its recreational and medical use are likely to differ. A poll by the Christian lobby group Family First in April showed under 20% support for recreational use, and 53% support for medical use.

So there we have it for another round of poll releases. Take from them what you will. Watch out for media bias. And wait impatiently for the next round of polls on which to speculate.

-- ACW