Uncategorized

Nation's most influential pollster can't explain election disaster

You've probably never heard of David Briggs. But you've very likely heard of Newspoll. That's the opinion poll that Malcolm Turnbull formalised as the benchmark of prime ministerial performance.

Remember? Tony Abbott had to go when he "lost" 30 in a row. Then, eventually, Turnbull himself had to go after he "lost" 38 of them.

Briggs is the man behind the poll. So that made him, in effect, the arbiter of whether Australia's leaders were seen to be succeeding or failing. He was the spokesman for the jury, as well as the judge, in the courtroom of Australian politics. All it took to finish the process was the executioners in the party caucuses to deliver the punishment.

But Briggs is much more than Newspoll, published by Rupert Murdoch's The Australian. He's also the man behind the YouGov Galaxy polls published by Murdoch's tabloid papers. And he's the man behind the exit poll conducted for Nine at the federal election.

Advertisement

Beyond his polling for media outlets, Briggs this year was also the pollster for the Labor Party. Taken together, this made him by far the most influential and important pollster in the land, the narrator of the Australian political story.

He also got the election result wrong. And because he got it wrong, all his clients got it wrong. That included the Labor Party which, to the end, thought it was cruising to victory two weekends ago.

So when I phoned Briggs last week and asked him how he was doing, I wasn't surprised when he replied "shithouse". He made no attempt to gild the lily. "It's very sad," he said.

He was also as bewildered as everyone else as to how his polls had consistently pointed to a Labor win. "Our final poll of the 2016 election campaign was the most accurate there has ever been," he said, when measured against the actual result.

"We used exactly the same methodology for this election that we used in 2016. Since I started at Newspoll in 1985, there hasn't been this style of disaster. Australia has been well served." Till now. He had already begun a post-mortem examination of the Newspoll poll data.

Briggs points out this election had some unique features. Clive Palmer's $60 million ad campaign, for instance, which was more than double the sum spent by the two main parties combined. We might ask what effect that had, Briggs says, but "it still doesn't explain why we were overstating Labor's vote by 3 to 4 per cent."

How did one pollster reach such a position of dominance? Through the quality and consistency of his results. Other polling companies always cast a nervous eye at his results as they published their own. Clients went to Briggs because of his reputation.

The long-time Labor pollster John Utting was quick to ask whether Briggs was guilty of a conflict of interest. He pointed out that Briggs' company, Galaxy YouGov Research, presented itself as an "honest broker and dispassionate observer" while at the same time it was "intimately involved in Labor's campaign." This, said Utting, "beggars belief."

Briggs has two responses. First: "It's not a conflict of interest. All our clients want the same thing. They all want accurate information. We were doing our best for all our clients." And second, he discounts Utting as a fair-minded critic: "It's sour grapes from someone who lost the contract" as the quantitative pollster for Labor.

Loading

But if this were only the story of the failure of Briggs' polling constellation, the problem would be easier to isolate. In fact, all the major published polls were wrong, and wrong in the same way.

The Essential poll published by The Guardian and the Ipsos poll published by The Sydney Morning Herald, The Age and The Australian Financial Review, consistently pointed to the same outcome – a Labor victory.

There were variations in the detail. For instance, the Ipsos finding of a very low Labor primary vote turned out to be exactly and uniquely right. But in the election-deciding measure, the two-party preferred share of the vote, all the pollsters indicated Labor would win with 51 or 52 per cent. In the event, it was the Coalition that won 51.6 per cent. Labor lost with 48.4 on the count so far.

This was an industry failure. Like Briggs, the Ipsos pollster, Jess Elgood, is baffled. "We treated and presented our data identically to the 2016 election," she says.

As for the reason for the clustering of the various polls around the same – wrong – conclusion, she points out that Ipsos has been unafraid to publish "outlier" results, even though it has been criticised for doing so. She has no explanation for the clustering in the final polls: "I think it's far too early to say." Ipsos, like all the others, is doing its own introspection.

All the pollsters could take shelter under the defence of margin of error. All polls are just estimates of a larger reality, and all are published with the note that they have margins of error of 2 to 3 per cent. But all the pollsters reject this as a cop-out. All recognise that they have a duty to do better.

"For me," says Elgood, "this is a lesson in caution." That should be the lesson for the country at large. The betting markets are often cited as a better indicator. They were spectacularly wrong-footed in this election too. The theory that punters are smarter than pollsters is now a dead letter.

But other indicators gave contrary signals. Ipsos conducted mid-campaign focus groups for this newspaper that pointed to a lack of appetite for change of government, and the main front-page news report began: "Uncommitted voters know about Labor's plans for new taxes but have heard almost nothing about the promised benefits, in a sign of potential trouble for Opposition Leader Bill Shorten's election prospects."

Our reporters on the road, testing opinion the old-fashioned way, also reported the lack of appetite for change of government, the scepticism about Labor and Bill Shorten. Opinion polls conducted in individual key seats also proved to be more useful indicators than national averages, although polling a meaningful number is a big and expensive exercise. The political parties poll around 20 each to map their battles.

But because of our long conditioning to the pseudo-scientific infallibility of the opinion polls, the country allowed alternative data points to be pushed aside. The pollsters need to address their problems. And the rest of us need to recover our common sense.

Peter Hartcher is the political editor.

Click Here: