Colmar Brunton has been running the 1 NEWS poll for more than 20 years. We are New Zealand’s most transparent pollsters when it comes to our results and the methods we employ.
Over recent General Elections the 1 NEWS Colmar Brunton poll has consistently been close to the final election result.
For full details of our most recent poll and our poll method you can download the full report below. The full report is released on this website 48 hours after the results are aired by TVNZ.
A full description of the method is included in our report.
You will have to wait and see…
We don’t want to give anyone the opportunity to sway the results before we conduct a poll, so we do not publicly announce our polling dates.
Eligible New Zealand voters.
Each poll is taken from a nationally representative sample of around 1,000 eligible New Zealand voters.
We now use a mix of telephone and online sampling. 50% of the sample we collect by calling randomly generate mobile numbers. The remaining 50% we collect online – with New Zealanders who have signed up for market research studies.
For mobile, the random nature of the calling together with our strong response rates delivers a robust sample. For online we put in place careful interviewing targets to ensure our sample is nationally representative. Weighting is then applied to the entire sample as a final check that our 1000 respondents are representative of eligible voters in New Zealand.
Not anymore. Up until 2017 the poll was conducted completely by phoning randomly generated landline numbers. However, access and use of landlines has declined notably in recent years making it more challenging to collect nationally representative samples. In response to this shift we switched the poll to mobile and landline calling in 2018 (after running several tests ahead of the 2017 election). Following further testing at the 2020 election, we made the decision to remove landline calling and use a mix of mobile calling and online interviews.
Our testing at the 2020 election showed this combined approach was highly effective at reaching a nationally representative sample, while also improving our accuracy in measuring voter intention at the 2020 election.
The maximum sampling error is approximately ± 3.1%-points at the 95% confidence level. This is the sampling error for a result around 50%. Results higher and lower than 50% have a smaller sampling error. For example, results around 10% and 5% have sampling errors of approximately ±1.9%-points and 1.4%-points respectively at the 95% confidence level.
Undecided voters, non-voters, and those who refuse to answer are excluded from the data on party support. Party support results are based only on those who claim to be ‘quite likely’ or ‘very likely’ to vote.
It is important to consider the margin of error when reading poll results. The maximum sampling error for the 1 NEWS Colmar Brunton poll party support results is approximately ± 3.1%-points at the 95% confidence level. This is the sampling error for a result around 50%.
Sometimes poll results are compared to actual election outcomes to gauge their accuracy. It’s important to remember that the poll results provide a snapshot in time of where the electorate’s preferences lie – they are not a prediction. People may change their minds in response to events.
That said, the 1 NEWS Colmar Brunton poll has a strong record in reflecting the outcome of elections over the last 25 years. In addition, at the 2020 election the polls correctly reflected voters decisions on both the 2020 cannabis and end of life referenda. We have continued to evolve the polling method to make it as accurate as possible.