Māori electorate polls

For the third election in a row, Curia has done polls in the Māori electorates for Whakaata Māori (Māori TV). The results can all be found here. In four of the electorates TPM leads, in two Labour leads and in one, Greens lead.

I thought it would be useful to caution that electorate polls, and especially Māori electorate polls, should not be seen as predictive of the election outcome as party vote polls.

There are six reasons why these polls are less predictive.

1. They have a smaller sample of 500, and hence a larger margin of error of 4.5%. A gap between candidates of less than 7% is not statistically significant at 95% confidence.
2. People are less certain about their electorate vote. Most people have a fairly good idea of parties they will vote for, or not vote for. But they keep their options open on candidates until closer to the election. There is always a much higher level of undecideds with electorate vote than party vote.
3. Māori electorates are harder to poll than general electorates. As it is not cost effective to call random households hoping that 1 in six may have a Māori adult, you target areas more likely to have Māori respondents. This makes the poll less random.
4. Māori households tend to have higher levels of poverty, and those with the highest levels of deprivation can be harder to contact.
5. In most seats the electorate contest is primarily between a centre right candidate and a centre left candidate, making the decision more binary. In Māori electorates all the leading candidates tend to be centre left, so respondents won’t be as guided by their party vote choice.
6. Most electorate polls are done a month or two before the actual election, so can’t be seen as predictive as a poll taken the week before the election.

So the results from the polls are very very interesting, and can tell us a lot about how party and candidates are doing. But unless there is a huge gap between candidates, they should not be seen as particularly predictive of the outcome.