Why does one poll count and another vanish?
- Grant McLachlan
- 20 hours ago
- 6 min read
When TOP recently cracked five percent, the same poll became a story about National’s lead — or no story at all — depending on who held the microphone.
In May, The Opportunity Party did what it had failed to manage in nearly a decade of trying — it polled above the five percent threshold. The Roy Morgan poll put Qiulae Wong’s rebranded party on six percent, a record high that on those numbers would deliver seven seats and a first entry into Parliament. By any reasonable measure, that is a significant story.
The country’s main newsrooms barely registered it. The broadcasters who run their own polls said nothing; the outlets that did engage waited days, and several used the result less to report a number than to argue about the party.
The question worth asking is not whether a single poll is correct. It is why our media decide a minor party exists only once it is balanced on the lip of the five percent cliff — and why those same outlets treat the polls they pay for as news and everyone else’s as background hum.
Contents
The poll that wasn’t news
The Opportunity Party has never reached five percent. Its best result was 2.4 percent in 2017 and 2.2 percent in 2023.
For months the press pack rated its chances at near-zero, on the reasoning that no new party has entered under MMP without a recognised former MP at the helm — a line the NZ Herald itself ran in February. Then a poll broke the pattern, and the broadcasters who commission quarterly polls of their own — 1News and RNZ — did not report it at all.
It did not vanish entirely. The NZ Herald noted the five percent crossing in a headlines bulletin two days later, and the Otago Daily Times weighed the result in an editorial. But the outlets with their own polls to protect said nothing, and the 1News interview that ran days earlier shows why: it was pegged to the network’s own Verian poll of 3.3 percent, and framed Wong’s task as convincing voters their ballot would not be wasted.
The instinct was to ignore the number, not to treat a six percent reading for a party written off as hopeless as the news it plainly was.
The cliff decides who is real
Coverage tracks the threshold, not the policy.
A party safely above five percent is a player; a party beneath it is a curiosity, or a wasted vote. The mechanics reward the framing. Under the Sainte-Laguë formula a party on 5.1 percent collects around six seats while a party on 4.9 percent collects none, its votes redistributed to those who cleared the bar. I have argued for more than a decade that the five percent threshold is the catalyst for abuse. Reporting that treats the cliff as the only thing worth covering makes the cliff more dangerous still.
Our media decide that a party exists only once it is standing on the lip of the cliff.
What Jordan Williams understood
None of this is new.
On 30 July 2020, Taxpayers’ Union director Jordan Williams teased a poll on Twitter and urged followers to watch the six o’clock bulletin; the result was talked up as Act hitting five percent. The 1News Colmar Brunton poll, when it screened, said that Act was at 5 percent, rounded up from the 4.8 percent stated in its press release. It was the first time since 2004 that Act had been near the 5 percent threshold and that threshold being mentioned in the media broke the psychological barrier and created momentum.
Two days earlier, National’s internal Curia numbers had been leaked while Colmar Brunton was still in the field.
Releasing one poll while a rival is polling can nudge the next — and a party seen to be near five is suddenly a party worth covering.
Williams grasped the lesson precisely: in New Zealand, the media take you seriously only when the number tells them to.
Whose poll counts
Here is the tell.
The broadcasters’ own polls — 1News–Verian, RNZ–Reid Research — are reported as events in themselves, with graphics, panels and analysis. Polls they did not pay for are downplayed or ignored, however sound the method. The irony is sharp, because the number everyone now waits on, the Curia–Taxpayers’ Union poll, comes from a firm that is no longer a member of the Research Association of New Zealand after complaints and the resignation of its principal, David Farrar. Farrar founded the Taxpayers’ Union and polls for National.
The poll being treated as the benchmark has the clearest partisan lineage of the lot.
One poll, two headlines
Nothing shows the trick more cleanly than how Newstalk ZB handled the Roy Morgan result.
Senior political correspondent Barry Soper analysed it as a story about the National-led Government’s lead — National up five points to 30.5 percent — with the Opportunity Party nowhere in the billing. Across the same network, his colleague and wife Heather du Plessis-Allan had already made Opportunity a story in its own right, and after the poll cast the party as the dark horse of the election.
Same numbers, same newsroom, same household — and two opposite headlines. The poll did not decide the story; the storyteller did.
Even sympathetic coverage tends to manage a minor party as much as report it.
In a Herald column a fortnight before the poll, du Plessis-Allan called Opportunity’s chances its best yet — then insisted it was not a centre party but a left-wing one whose land-tax-funded universal income was classic redistribution, and reasoned that, with the left likely to lose anyway, a vote for it would cost little.
The endorsement and the containment arrive together: the party is worth watching, but a vote for it is still a punt.
The home-team scoreboard
If that instinct seems too cynical, watch how the same newsrooms covered the NZ Media Awards last month.
The morning after, Stuff led with “Stuff wins top media award for third year running”. NZME told NZ Herald readers it was “celebrating multiple top honours”. RNZ reported that “Sam Sherwood wins three top Media Awards”.
Three outlets, one night, three different “top” awards — each the one the outlet in question happened to win.
Stuff’s “top media award” was Digital News Provider of the Year; the night’s actual Newspaper of the Year went to an NZME title.
There is no neutral scoreboard. Each masthead reports the slice of reality in which it is the winner — and treats a rival’s trophy, like a rival’s poll, as someone else’s news.
The rogue-poll trap
Politics runs on momentum, and one poll can move the next. The danger now is obvious.
Commentators dismiss the Roy Morgan result as a rogue poll; the dismissal hardens into received wisdom; the next poll, softened by that narrative, duly lands lower; and the prophecy fulfils itself.
Even the serious coverage leans this way: the Otago Daily Times opened its editorial with the reminder that one poll does not make an election — true enough, though the same caution is rarely extended to the broadcasters’ own polls.
All eyes are on the Curia–Taxpayers’ Union poll to settle the question, which is exactly the problem.
A single contested number, from whichever stable, should not be allowed to decide whether more than 150,000 voters are permitted to matter.
Kill the cliff
The remedy is not to police which polls get airtime. It is to remove the reason the games are played at all.
Scrap the five percent threshold and regulate polling, and the cliff disappears: a party on four percent simply wins the seats four percent earns. The Electoral (Integrity of Polling, Media, and Elections) Amendment Bill would do precisely that. Without the cliff, no leaked number, no teased bulletin and no convenient rogue-poll line can erase a party between editions.
A democracy worth the name does not let the newsroom — or the next poll — decide who is allowed to exist.
