That's true, TV3 gets Reid Research to do their polling, and TV1 use Colmar Brunton. Both discard the 'don't knows', and the "not saying" results.
But if you look at both firms, they have different methodologies.
Colmar Brunton also use over 1,000 voters who can be called via landlines, the same as Reid Research. Reid Research say that they ensure a representative cross-section by age, sex and geography. Sounds good, doesn't it?
But Colmar Brunton go a step further. They allow for those weightings, but also weight for ethnicity. That's probably quite important. They also weight on household size. In the latest Colmar Brunton poll, the undecided and no answer rate was 10%, they at least released the data.
They also say that because they discard the refusals and undecided to come up with the 100% total, that these polls can only be relied on to reflect voting trends, they cannot be used to predict election results.
But of course that's exactly what the TV channels do with the data. So we're hearing this morning from TV1 reporting on TV3's poll "So what are Labour doing wrong?" and other daft comments. The difference is that polling has started, this shouldn't be a lightweight game being played by the news media now. Why aren't they spending detailed time on policy differences? - but that would be a lot of work.
http://www.colmarbrunton.co.nz/wp-co...rt-2-6-Sep.pdf
Don't forget that the other weakness in both polling methods is that they only call landlines to make their contacts. Many households that are struggling, have long ago given up their landline and gone to cellular, rather than pay for two systems. All of these people are then off the grid of the pollsters.
In any case, I think I'd trust the Colmar Brunton Poll to get closer to the actual voting trend, and before the TV channels get too much time to sway voters with their latest take, a new TV1 poll will be out in a day or so.