Saturday’s Newspoll, published in The Australian newspaper, would have made sombre reading for SA Best candidates across the state. From a high of 32 per cent in December – the poll that kickstarted Xenophon’s momentum and prompted an influx of hopeful candidates that saw him dramatically increase the scope of his campaign – his statewide vote crashed back to 21 per cent.

The notion that his support was in “freefall” was duly echoed across all media throughout the weekend.

But the spark of hope for Xenophon came in an intriguing paragraph in the article accompanying the poll: “The drop in statewide support for SA Best, from 32 to 21 per cent, appears to be linked to the party not running candidates in all 47 lower house seats.”

“However, if SA Best did secure 21 per cent of the statewide vote on March 17, this would still represent an average of 27 per cent voter support in each of the 36 seats in which it is fielding candidates,” it read.

This prompted plenty of social media confusion as to the methodology behind the discrepancy.

Latest Newspoll method is wrong. @SABEST_Party still a strong third, with capacity to win House of Assembly seats. @Nick_Xenophon has to get a boost from being in three-way leaders’ debate. @mjowen “freefall” claim is just barracking.

— Michael Atkinson (@MickAtko) March 3, 2018

Further, a response from a YouGov-Galaxy pollster – the company behind Newspoll – suggested there had been an “adjustment” to account for SA Best contesting fewer seats than its major party rivals.

The final poll people expect to be reflective of the election result. Of course when nominations have closed and certain parties are not running in every seat is the right time to make that adjustment. There’s no great conspiracy here…

— Campbell White (@DrCampbellWhite) March 3, 2018

Galaxy managing director David Briggs told InDaily today: “Our estimate of the vote across the whole state is 21 per cent… however we recognise they’re only standing candidates in 36 seats, so they have to achieve 27 per cent in each [to reach that statewide figure].”

We estimate SA Best will achieve 21 per cent across the state – and that allows for 11 seats in which they record zero

Yet, a basic reading of a 21 per cent statewide vote would suggest that the party only garners 21 per cent in any given seat – rather than the vote being shoehorned into the number of seats it’s contesting.

But Briggs insists “we allow for that”.

“We look at the geographic nature of the sample.”

“When all the votes are counted across the state, we estimate SA Best will achieve 21 per cent across the state – and that allows for 11 seats in which they record zero,” he said.

“We have to make an allowance for the assumption that if someone says they’re going to vote SA Best, we have to assume everyone won’t have that option – in 11 seats they won’t… we’re saying they’re likely on average to achieve 21.”

The distinction is significant, because SA Best is only shooting for enough seats to hold the balance of power: if it achieves an average of 21 per cent across all the seats it contests, it could struggle to win any at all. But an average of 27 per cent across the 36 seats would likely see it finish in second place in a range of contests, and potentially snare several on preferences.

Briggs says that the December poll at which SA Best garnered a 32 per cent primary – higher than both major parties – was predicated on an assumption it was running in all 47 seats.

“Unless we know absolutely for certain what the situation is going to be, we don’t impose any restrictions [on the poll],” he said.

So, if SA Best was contesting all 47 seats, how would it have fared in Saturday’s poll?

“Probably it would be closer to 27 than 21,” says Briggs.

“The more direct comparison is [a slip from] 32 to 27.”

But Newspoll’s intent is not to track individual seats, rather to provide an accurate reflection of the statewide vote, rather than a party’s popularity per se.

“It’s not misleading – it’s accurately reflecting the statewide vote,” Briggs says.

“It’s certainly comparing like with like, but the most recent poll factors in that SA Best is not fielding candidates in all seats and that might be incorporated within the overall result.”

Our assessment is that they will achieve the 27 per cent based on these numbers

Having said that, Briggs says the raw numbers of the poll across the entire state were closer to 21 than 27.

“Our poll indicated that it’s very close to 21 per cent, however using some analysis based on region, that number declined slightly to the number in our poll,” he said.

“We do some modelling that involves some fairly detailed analysis of the numbers – what we’re saying is the statewide vote is 21 per cent and to achieve that they need 27 per cent [in each seat].”

But does that mean they will achieve 27 per cent in each seat?

Because if the statewide result of 21 per cent was not a reflection of the higher vote across SA Best’s 36 prospective electorates, it could mean Xenophon’s vote is even lower again – perhaps more like 16 per cent.

But Briggs is adamant that “we believe we’ve factored that into our calculation”.

“What we’re trying to do is to come up with an estimate which we believe is the best possible statewide assessment of their total level of support,” he said.

“There’s an awful lot of stuff that goes in there in terms of the number crunching [but] our assessment is that they will achieve the 27 per cent based on these numbers.

“It’s a statewide poll – some latitude is required… we rely on our expertise in understanding the numbers and analysis of the data [but] ultimately the number we predict is our best estimate for the current statewide vote.”

Respondents were asked the same questions and given the same options across both published polls.

The emerging Australian Conservatives garnered six per cent in the Newspoll, but are only running in 33 seats.

Briggs says this means if they were running in all 47 seats their figure would “probably [be] more than six”.

Perhaps the question, after all this, is what can be read into the statewide poll?

Not much, according to one senior major party insider.

“They’re almost worthless, they really are… that’s not meant as a criticism [but] it’s just for journalists to talk about,” the source said.

“It provides no insight into it, I don’t think… we don’t compete on statewide, we compete on 47 individual seats.”

To that end, perhaps the snapshot polls in various seats – such as those published weekly by The Advertiser, also conducted by Galaxy – provide a better clue as to the actual election outcome.

The series continued today with polls in three seats held by independent MPs in the spotlight, with Geoff Brock on track for re-election in Frome, Frances Bedford looking solid in Florey but Duncan McFetridge struggling in Morphett, where the Liberals have built an imposing primary vote of 39 per cent.

Seat polls are frequently a load of rubbish

SA Best is only contesting one of those three seats – Morphett, where its primary has plunged to 17 per cent, behind both major parties.

But Tasmanian-based psephologist Dr Kevin Bonham says these polls too can be problematic.

“[The outcome] is very difficult to model, because you’ve got seat polls coming out but seat polls are frequently a load of rubbish,” he said.

“The real problem with seat polls is doing the demographic scaling for them – it’s a mess [and] we’ve seen consistent seat poll failures in Australia for years because of this.”

Bonham says reading too much into the statewide Newspoll is “a bit tricky”, because it’s hard to get a read on whether, for example, voters “would be aware the party’s not running in those seats”.

He said secrecy surrounding the methodology doesn’t help the interpretation of polls.

“It’s a bit of a problem that there isn’t a full and clear public explanation of how they’re doing this, and people are left to ponder the mysteries of exactly what’s going on,” he said.

“There’s a lot of mystery here, because of the lack of written documentation about exactly what was done… this is a problem with polls generally in Australia.”

“We don’t get a lot of documentation about how they’re doing things [so] it’s very difficult to track the statewide figures and apply them to seats.”

Bonham says “even if we assume that [SA Best] are averaging 27 per cent in the seats they’re running in, that still – all things being equal – wouldn’t win them very many seats.”

He said the equation appeared to be “a few, not several” seats the fledgling party could hope to pick up on March 17.

“It’s not double figures… it’s a few to several.”

He said there was a similar situation with statewide polling inflating One Nation’s chances at the recent Queensland election, when Pauline Hanson’s party fell wildly short of expectations.

But retiring Labor stalwart Michael Atkinson is still wary, at least publicly, tweeting that SA Best “could pick up half a dozen seats on a statewide primary vote of 21 per cent”.

“Its vote wouldn’t be uniform across the 47 state districts and the higher than average vote would put it in the winning zone,” he wrote.

In reality, much will come down to preferences rather than polls, making the Greens’ decision to preference against SA Best in 15 seats – as revealed by InDaily last week – a potential game-changer.

Loading…

Taking too long?

Reload document

|

Open in new tab

Download

In a statement today, Xenophon hit out at the decision, urging Greens voters to “ignore” the party’s How To Vote cards “in those seats where the Greens are preferencing the Liberals and Labor ahead of SA Best”.

“How could the Greens Party do this dirty deal when they well know SA Best – unlike the Liberal Party – is against oil and gas drilling in the Great Australian Bight, voted with the Greens to disallow water regulations that were undermining the Murray Darling Basin flows, particularly into SA, and we are strongly in favour of dispatchable, reliable and renewable energy,” he said.