Looking at the methodology gives me an inkling as to why that might be.
The intention with political polls is to survey a group of people that accurately reflects the population of the country. Any poll undertaken in New Zealand would aim to get roughly the same proportion of men, women, Pakeha, Maori, workers, retirees, business owners etcetera as the general population.
Yesterday, I did a quick Google search on the methodologies used by five of New Zealand's main political pollsters. All five relied on phone surveys. One only surveyed landlines, one targeted a mixture of landlines and mobile phones, the other three did not specify the types of phones.
It is very unusual for anyone under the age of 35 to have a landline phone unless they either live with their folks or run a business from home, and this is becoming more common in the older generations as well as mobile calling and data prices come down.
It's also common for young people to ignore a phone call from a number they don't recognise, knowing that if the call is important, the caller will leave a message.
I often wonder if the reliance on phone surveys is partly why polls seem to be less effective these days.
I guess we'll find out on September 23.