Surprise surprise, The Herald's 'poll of polls' completely biased in favour of the left
November 16, 2026

Summarised by Centrist
The NZ Herald's pre-election Poll of Polls substantially inflated Opportunity's support compared with both the election result and the underlying surveys, according to pollster David Farrar.
Writing on Kiwiblog, Farrar compared the Herald's projections with preliminary election results and found its biggest error was Opportunity, which was forecast to receive 9% of the party vote but secured just 6%.
Farrar had previously questioned why the Herald placed Opportunity ahead of ACT when individual published polls consistently showed ACT leading. He said the Herald justified its projections through assumptions about political momentum.
The difference is that the underlying polls were considerably closer to the election result than the Herald's model.
An average of the final polls from five major pollsters put Opportunity at 6.5%, against its actual 6%. NZ First averaged 10% and received 9.8%, while ACT averaged 10.3% and secured 10.9%. National was the more notable exception, outperforming its 27.2% polling average to receive 29.2%.
The Herald's model, by contrast, had Opportunity at 9%, ACT at 9.5% and National at 27.7%.
It also projected Te Pāti Māori at 1.5%, compared with its preliminary result of 0.6%. However, the model underestimated Labour's support by 2.6 percentage points and overestimated NZ First by 1.4 points, meaning its errors were not uniformly favourable to the left.
Farrar warned that overstating Opportunity's popularity could have influenced voters, although there is no established evidence that the model changed voting decisions.
The central question is why a model intended to improve on individual polls produced a substantially less accurate picture of Opportunity's support than the polls themselves.
The official election results are due on November 27.

Community Discussion
Comment Policy