This ain’t exactly real

From Matthew Beveridge on Twitter, who also has a post on reporting standards.

In a poll of 1000 people, if we make the (generous) assumption that it’s a uniform random sample, these minor-party results mean 8 people for The Opportunities Party, 7 for the Māori Party, 4 for ACT and 4 for UnitedFuture.  That’s not a very impressive lead.

A 95% confidence interval (the equivalent of the usual margin of error statement, using my cheatsheet)  for TOP would be 0.3% to 1.6%; for the Māori Party 0.3% to 1.4%, and for ACT and UnitedFuture 0.1 to 1.0%.  There’s a lot of overlap.   A 95% confidence interval for the ratio of TOP to ACT support goes from 0.63 to 7.5, showing that more-sophisticated analysis confirming the conclusion: ACT could be more popular than TOP, or much less popular, but we don’t have enough information to be sure. Obviously, the same thing will be true only more so for the TOP/Māori comparison.

But there’s a more important problem: for at least three of these parties, the party vote is relatively unimportant. The Māori Party [correction] has won one  list seat in its history, and neither ACT and UnitedFuture has for the past two elections.  What matters most for these parties is their support in their key electorates.  It’s hard to poll for electorate support, both because it’s hard to sample from electorates and because you have to distinguish party vote and electorate vote intentions reliably. Judging from past election campaigns, we’ll probably get some polling of the Māori electorates, but it’s unlikely we’ll get any useful polling of Ōhāriu and Epsom.  However, I would feel pretty safe in predicting that ACT will end up with a seat, and much less confident that TOP will.