New Zealand First Leader Rt Hon Winston Peters
12 August 2019
POLS111 Guest Lecture
Two months ago, on 9 June 2019, two political polls came out simultaneously on the 6 o’clock TV news.
One was published by TVNZ, the other by Newshub.
While it is clear that in the era of a twenty-four hour news cycle the 6 o’clock news is a relic of the past, media organizations seem reluctant to adapt.
Normally, the errant assumptions of such polls would go unchallenged by commentators.
If the Labour Party has a lead, speculation would form around its ability to assemble a coalition, and vice versa for National.
And the electoral chances of minor parties would invariably be rubbished.
But in this case, scrutiny of their conclusions could not be avoided.
Indeed, the two polls were diametrically opposed.
The contrasting results fell well outside the margins of error. One seriously favoured the Labour Party, the other the National Party.
Furthermore, one suspects they both underestimated the power of smaller parties, as polls often do.
This led to a flurry of analysis and so-called ‘soul searching’ on the part of polling companies, in this case Colmar Brunton and Reid Research.
No one could recall two polls coming out at the exact same time, which painted such vividly different pictures of the New Zealand electorate.
As is often the case with broken systems, only after their defects become impossible to ignore, do the chattering classes begin to notice.
Since the 1980s, media organizations have sought to reveal the voting intentions of the general public through polls.
There is a common adage – which happens to be correct - that “the only poll that counts is the one on election day.”
Despite that, media organizations pride themselves on their accuracy in predicting the vote, years out from elections.
Although such claims are patently absurd, the deception continues.
In those June polls, both TVNZ and Newshub claimed their polls to be superior – with Tova O’Brien of Newshub saying that Newshub has “most accurate political poll” in the country.
Ms O’Brien’s self-confidence is impressive, but obviously misplaced.
Her hubris, and that of a mixed herd of others, demonstrates deeper problems in political polling which will be very hard to fix, and are only getting worse.
This is not to say that inaccurate polling is a recent development.
Even three years before the MMP era began, defective polling has steadily become the norm.
Indeed, on election night 1993, Prime Minister Jim Bolger summed up what many of us feel about political polls in three words – “bugger the pollsters”.
It only got worse.
What Mr Bolger could not have anticipated is just how bad things have become in the last few years – not only due to the quality of polls, but also their quantity.
As the coffers of media organizations dwindle in the digital age, their ability to poll has diminished.
One used to be able to expect a steady trickle of polls between election cycles – with an average derived from them painting a more accurate picture – albeit still critically flawed.
Now, each release of a poll constitutes a media event – and receives coverage far greater than its actual worth.
And as we saw in June, it is hard to express any kind of confidence in them.
This is not a problem unique to New Zealand. In three short years, we have a myriad of other examples.
Those told us that:
- The United Kingdom would remain in the European Union;
- Hilary Clinton would win the United States presidential election;
- Theresa May would increase her majority in the United Kingdom;
- The Labor Party would win power in Australia; and
- The BJP would lose their majority in India’s Parliament
None of those came to pass – the voters belied polls at every given chance.
For anyone with a political brain, it was plain to see that those outcomes did not match prevailing public opinion.
In the case of Brexit and Trump, they did not consider an overwhelming desire for change.
But the poll data did not match those self-evident truths.
Here in New Zealand, we are not immune from this deluge of incompetence.
There are two critical deficiencies which explain this:
technological disruption, and bias – both wilful and unconscious.
The most obvious example of technological disruption is the rapid transition from landline phones to mobile phones.
When polls were first conducted in the 1970s, an outfit called Heylen actually conducted them on-foot.
People would go house-to-house, and talk to people directly about what they thought of the Government.
Then, in the 1980s, telephone polls began to appear.
In the twentieth century, virtually everyone had a landline. The numbers for those landlines were catalogued, and the polling companies could consult those lists to conduct their polls.
It is not hard to see why the Heylen model was abandoned, when a company could seek voters’ views from the comfort of a call centre.
Then came the cell phone. Gradually adopted over the course of the twenty first century, cell phones have quickly supplanted landlines.
That trend was only accelerated by the advent of the smartphone in the last decade.
Now, smartphones are our society’s main mode of contact.
Most people – particularly the young and poor, don’t bother having landlines at all.
But smartphones aren’t catalogued in a similar way, and pollsters struggle to utilize them.
Confronted with this problem, pollsters had a few options.
They could revert back to the Heylen model and interview people in person – a method which was championed by some in the aftermath of June’s disastrous polls.
Or they could have found a way to communicate with New Zealanders digitally, with accuracy – something no one seems to have achieved.
But instead of changing their methodology to reflect that New Zealanders have increasingly cut the cord, pollsters are largely stuck in the Twentieth Century.
They still conduct much of their research over landlines, or a mix of landlines and mobile phones.
These technological hurdles are an unavoidable by-product of disruption, and nobody can blame the polling companies for that.
But what they can be blamed for are the biases, unconscious and wilful, which have tainted their results.
The most pernicious example of this are pollsters’ habits of favouring certain parties when offering respondents ones to choose from.
My party, New Zealand First, is vocal about this disturbing trend because we are often on the receiving end of it.
A poll typically follows a standard procedure.
Responses from 1000 individuals are picked, with demographic variables weighted to represent the general population – often with mixed results.
Those individuals are asked who they would vote for if a general election were to be held the following day – with viable parties listed as options.
Time and time again, substantive evidence exists that pollsters list every party in Parliament, apart from New Zealand First.
It goes without saying that this leads to an underrepresentation of New Zealand First in the results of such polls.
This leads to shock when we exceed expectations and form a significant force in the House of Representatives – above and beyond their projected numbers.
Whilst this might be a satisfying exhibition of their inability to do their jobs, it is harmful to New Zealand’s political culture.
We are too often pegged to calculations based on false data, tainted by human error and technological disruption.
The question that naturally arises out of this conundrum is just how can polling companies return to more accurate data?
Unfortunately, the only conceivable option is a return to the person-to-person polls of the 1970s, as companies’ failed attempts to use the internet become painfully clear.
That would be expensive, and considering that media organizations can barely make ends meet as things are, it is inconceivable that they would spend money on something so ephemeral.
As students of political science, polling between electoral cycles is likely to have played some part in your assessment of Governments and their policies.
If that’s the case, here’s why you’re wrong.
The idiom that a ‘week is a long time in politics’ was coined by British Prime Minister Harold Wilson in the 1960s, presumably when he was having a particularly tough week. But it is true.
The fleeting nature of media cycles and public interest betray the fact that the importance of short term popularity is often overstated.
It distracts from what really matters – Governments carrying out legislative change, which imprints their vision on the lives of ordinary New Zealanders.
Public opinion is undoubtedly important – especially when re-election is considered.
But the actual business of Government, and indeed Opposition, is far more important than individual polls – or even the sum of their parts.
To make matters worse, as those polls become less common and less reliable, a hole has formed for the vacuous type of commentary that moulded around them.
Rather than focusing on policy positions and economic metrics, editorials are little better than gossip columns, and have come to dominate our newspapers, websites, and television stations.
Such stories are somehow even worse than those about flawed polls. Rather than using data, they often contain journalist’s evaluation of ‘perceptions’ of the public.
This is dangerous territory.
Those perceptions are little more than conjecture, formed by those in the press gallery who lack the funding or ability to seek and deliver real news.
To make matters worse, they are often absorbed in a political bubble which the general public does not care for.
This creates a harmful cycle, where the garbage that some journalists produce disillusions the public, which causes their readership to fall, which leads to smaller budgets and even worse journalism.
And the demise of polls has a large part to play in that.
With all factors considered, it is clear that the quality of political polls has degraded to the point where they are practically worthless.
June’s polls were just the tip of the iceberg.
And the media’s response to this degradation is to double down and start hypothesizing about scenarios which don’t exist.
You might consider this to be an overstatement, or a reflection of my personal relationship with the Press Gallery – one that can mildly be characterized as adversarial.
But the value of the Press is undeniable. Without them, no Government can truly be held to account – a salient task, when the current Leader of the Opposition is doing such a poor job of it.
You are all old enough to remember political polls in the last three years, and those in both Australia and the UK before that.
Prime Minister Turnbull called a snap election in Australia based on the polls. When the election was over, he had lost 14 seats.
Prime Minister May in the UK did likewise – the result was 13 less seats and a compromised majority.
Remember Brexit, remember Trump, and remember the British, Australian, and Indian elections. The polls did not reflect reality, and there is little doubt that the same will happen here.
Let there be a lesson. Too many in the commentariat, in spite of overwhelming evidence, are always drinking their own kool-aid. What we all need to do spike the kool-aid with experience and reality – for the good of the fourth estate, and New Zealand as a whole.
In short, make sure you leave this university with a Political Science degree based on reality and facts – not dreaming.