
The Foundation for Economic Growth is a group of likeminded individuals who have decided to act rather than accept New Zealand's continuing poor economic performance. The Foundation is not affiliated with any political party. Add Your Comments here. 

Most recent update: Sep 1st, 2014  15:12:56


Newsletters
:
Current Newsletters
:
1 September 2014
Thought for the Day
When I look at the various polls of people's intention to vote I am struck by how stable they are.
We must remember that these polls are only on a base of 500 to 1000 participants and that the ' standard error of the mean' usually amounts to 3% or 4%.
So National varying from 46% to 54% by different polls is within the standard error which means any variation could be just statistical error and nothing to do with the 'true' mean showing any change.
This begs the question of just what the true mean is. If we imagine we are measuring the mean weight of the population then we might approach this problem by weighing everyone and dividing by the number of people in NZ. This would take a long time.
So we take a sample and measure that. If we measure 100 people then we might consider that perhaps such a small number would not give the same answer as the true mean. If we took another different sample of 100 people we would get a different answer for the mean. All we could say is that the answer is probably about these sample answers.
If we only measured 10 people we would get bigger variations but if we measured 1000 people in each group the answers would tend to be more similar.
Statistics is not telling us the truth. It is giving us a good idea of where the truth is. Statistics can also tell us the limits about how far away from our sample the true answer is. The bigger the sample the closer the true answer will be.
So with a standard error of 4% and an answer of 49% we can say that very probably the true answer is between 45% and 53%. It gives no indication of where in this range the true mean sits. It could be 52% or 46%  we cannot say.
We also note that the question is someone's assessment of their likely action in the future  not something real that can be objectively measured. This also contributes to differences in polls.
So over all with all the polling I think we can say that the polls seem remarkably settled. The only trend  if it can be called that  is that Labour used to be in the mid thirties and is now in the mid twenties. We can also note that it is very hard to learn anything much from parties who get results below 5%.
This morning's DomPost caught my attention with its front page headline, 'National caught deep in the mire'. The opening statement was, 'The snowball effect of the Dirty Politics scandal is threatening to bury National.'
My attention was then diverted to a group of four likely punters who presumably represented the average voter. My immediate (biased) reaction was that they had obviously been selected to support the headline and the main theme of the article.
I was amazed to learn that all four were not thinking of changing their vote. What they all seemed to think was that it was irrelevant grubby electioneering tactics that showed Collins in a bad light but they were basically just ignoring it. National was doing fine!
Far from National being 'caught in the mire' or 'buried' the Dirty Politics has made no difference to the way people intend to vote.
I also noticed this tendency for reporters to lie about what they see with the TV leaders debate. I watched it and thought that Key seemed a bit preoccupied (we can now see why) and that Cunliffe tried to overtalk too much. They both did OK as far as I could see given their differing circumstances.
However, the reporters talking and writing about the event thought that Cunliffe had notched up a good win and Key was in trouble. This was in total variance with the poll conducted at the end of the show which scored Key at 69% winning and Cunliffe at 31% losing.
Journalists are becoming even more unreliable than the statistics they report.
Sep 1, 2014, 10:57


