Pages

Sunday, July 26, 2020

David Farrar: Latest poll


Newshub have released a Reid Research poll which shows Labour 36% ahead of National.
Now Reid Research is a very good polling company, but even the best polling company will get results outside the normal margin of error. When you say there is 95% confidence that the poll results are within 3% of the actual, that means there is a 5% chance they are not (putting aside non sampling errors).
I do not believe this poll is even close to reflecting the actual situation. There is no way there is a 36% gap between the parties. The numerous polls I have seen and done have it way way closer.
It is worth also noting what a Stuff survey has found:

Prior to the leadership change, the number of people responding to the survey who said they would probably vote National stood at 23.7 per cent support. After Collins took over, that number jumped to 40.2 per cent. Likewise, when people were asked about which leader they felt closest to, only 10 per cent named former leader Todd Muller, while 52 per cent named Prime Minister Jacinda Ardern. But those numbers changed markedly after the leadership change, with 31.7 naming Collins, while Ardern dropped to 47.2 per cent.
Now the Stuff/Massey survey is not a scientific random poll, it is a self selecting survey. So you can’t say the topline data reflects the NZ electorate.
But the huge change pre and post Judith is significant. There has been a galvanising effect for National. And again that doesn’t mesh with a poll showing National dropping into the mid 20s.
The other thing worth noting is that no party has ever got over 50% in an election in the past 69 years. The last time was in 1951.
Again Reid Research is a very good polling company, but all pollsters can get “outlier” results. Last election Colmar Brunton (also a very good company) had two polls which showed Labour ahead, which didn’t mesh with any other poll, and then six days later another poll which showed the gap changed from 4% to -9%.
There is a good article at Five Thirty Eight about “outlier” polls and how they are part of the business. They say:
So what’s my advice to you as news consumers when you encounter a poll that looks like an outlier?
To a first approximation, the best advice is to toss it into the average. Definitely do not assume that it’s the new normal. 
The trouble in NZ is we have relatively few public polls. So my best advice is just to wait for the next couple of public polls, especially ones done entirely after the last week which was an especially messy week for both major parties.
I’m just sad political betting is illegal in NZ as I’d love to bet against Labour getting over 50% at the election, and make lots of money. I do miss the days of iPredict where I made (off memory) over $10,000 from the political markets there.
David Farrar runs Curia Market Research, a specialist opinion polling and research agency, and the popular Kiwiblog. He previously worked in the Parliament for eight years, serving two National Party Prime Ministers and three Opposition Leaders.

1 comment:

Unknown said...

I did however notice that a number of parties were missing from that poll. I wonder why that is and as to why they were not listed