Polls are great. I love ‘em and I love the occasional surprise we get from them.

They're really useful in providing those snapshots of public opinion at any given moment.

And you'll be aware the political parties use them extensively, not just to see how they're tracking but for all sorts of nefarious stuff like testing out our reactions to policy ideas.

On top of that there's also a lot of research into the way leading figures are perceived, all part of the process of moulding and shaping them for maximum ballot-box efficiency.

But the thing about polls that we rarely take much notice of, and should, is the sampling process.

It's vital, if you want accurate answers to your questions, to be scrupulous about the sampling process.

Let me explain.

Say you're doing a political poll; you want to be sure your respondents match the demographics of the population as a whole.

So, you end up having to interview so many men, so many women, and a set number of people in each age group to match the voting population.

For example, the 1News poll uses roughly half and half randomly selected people on mobiles and volunteers from a panel; these are people who've done interviews in the past and have agreed to take a phone call from time to time.

Like most of the others, a sample of about a thousand people are called up over a two or three-day period.

Thing is, no matter how careful you are errors start to creep in. For a poll using a thousand people, even carried out very cautiously, your results are only good to within plus or minus three points.

So reporting numbers within that margin —in this example say, how many people are going to vote NZ Loyal— well it's pretty much pointless.

What's the point of all this?

The point is that I suspect polls influence the way people vote, possibly even whether people bother to vote.

And I'm going to suggest that this is such an important election that you can't leave it to the polls to make up your mind. Yet at the same time, we're all voting tactically. We kinda want that extra bit of information.

For what it's worth, and again I'm not trying to influence your vote, I've made up my mind on the issues—on the principles, and the issues.

I'll vote on Monday and try to forget about it until 14th.

Let's just hope we're collectively a little less grumpy once this is all over.

Polls are great. I love ‘em and I love the occasional surprise we get from them. 
They're really useful in providing those snapshots of public opinion at any given moment. 
And you'll be aware the political parties use them extensively, not just to see how they're tracking but for all sorts of nefarious stuff like testing out our reactions to policy ideas. 
On top of that there's also a lot of research into the way leading figures are perceived, all part of the process of moulding and shaping them for maximum ballot-box efficiency. 
But the thing about polls that we rarely take much notice of, and should, is the sampling process. 
It's vital, if you want accurate answers to your questions, to be scrupulous about the sampling process. 
Let me explain. 
Say you're doing a political poll; you want to be sure your respondents match the demographics of the population as a whole. 
So, you end up having to interview so many men, so many women, and a set number of people in each age group to match the voting population. 
For example, the 1News poll uses roughly half and half randomly selected people on mobiles and volunteers from a panel; these are people who've done interviews in the past and have agreed to take a phone call from time to time. 
Like most of the others, a sample of about a thousand people are called up over a two or three-day period. 
Thing is, no matter how careful you are errors start to creep in. For a poll using a thousand people, even carried out very cautiously, your results are only good to within plus or minus three points. 
So reporting numbers within that margin —in this example say, how many people are going to vote NZ Loyal— well it's pretty much pointless. 
What's the point of all this? 
The point is that I suspect polls influence the way people vote, possibly even whether people bother to vote. 
And I'm going to suggest that this is such an important election that you can't leave it to the polls to make up your mind. Yet at the same time, we're all voting tactically. We kinda want that extra bit of information. 
For what it's worth, and again I'm not trying to influence your vote, I've made up my mind on the issues—on the principles, and the issues. 
I'll vote on Monday and try to forget about it until 14th. 
Let's just hope we're collectively a little less grumpy once this is all over.

A fresh and intelligent start to your day - catch the very latest international and domestic news developments, sport, entertainment and business on Early Edition with Ryan Bridge, on Newstalk ZB.

A fresh and intelligent start to your day - catch the very latest international and domestic news developments, sport, entertainment and business on Early Edition with Ryan Bridge, on Newstalk ZB.