How did the Herald Poll of Polls go?
I blogged some time ago that the Herald Poll of Polls was producing nonsense results as it said Opportunity was ahead of ACT, despite every single poll showing the reverse. They justified it on the basis of assumptions about momentum.
The results shows their model was not high quality.
| Result | Herald Poll of Polls | Variance | |
| National | 29.2% | 27.7% | 1.5% |
| Labour | 29.4% | 26.8% | 2.6% |
| Greens | 13.3% | 13.6% | 0.3% |
| ACT | 10.9% | 9.5% | 1.4% |
| NZF | 9.8% | 11.2% | 1.4% |
| Opp | 6.0% | 9.0% | 3.0% |
| TPM | 0.6% | 1.5% | 0.9% |
As you can see they had Opportunity 50% higher than what they actually got. They also had ACT and NZ First both significantly lower, and TPM significantly higher.
I’m a huge fan of poll of polls models in theory. But one that actually contradicts the actual polls is not a good model. And it misled potentially hundreds of thousands of New Zealanders by insisting Opportunity was 50% higher in support than they actually were. This could well have had a non-trivial impact on how people voted.
Perhaps Opportunity should declare the cost of the Herald Poll of Polls as an election expense on their return?