Inside Story

Current affairs & culture from Australia and beyond

925 words

More polls, but still no election

8 April 2019

Do today’s figures tell us anything at all about how the parties are faring?


Not the election campaign: prime minister Scott Morrison, with foreign minister Marise Payne (right) and the Liberal candidate for Macquarie, Sarah Richards, talks to journalists at North Richmond, Sydney, yesterday. Joel Carrett/AAP Image

Not the election campaign: prime minister Scott Morrison, with foreign minister Marise Payne (right) and the Liberal candidate for Macquarie, Sarah Richards, talks to journalists at North Richmond, Sydney, yesterday. Joel Carrett/AAP Image

One day, soon, this seemingly endless 2019 pre-campaign period will be over, but in the meantime we have two fresh opinion polls.

One is the latest Newspoll in the Australian. Newspoll sits front and centre of the political class’s obsessions, fetishised by players and addicts alike, and has even been a fashionable prime ministerial assassination accessory. It’s now a contest in itself, with parties “winning” or “losing” Newspoll. The other, less closely watched one is Ipsos in the Nine papers.

Newspoll has the Coalition on 38 per cent, Labor 37, Greens 9 and One Nation 6, washing through to 52–48 after preferences. Ipsos puts it at 37, 34, 13 and 5, with 53–47 two-party-preferred.

The big difference is in Greens primary support, of which Ipsos is a serial overestimator. During and before the 2016 campaign, for example, the pollster had the minor party on 13 or 14 per cent; on election day it received 10.2 per cent of the vote. The other polls, including Newspoll, had much more realistic measurements, as this cheat sheet shows.

(Like most minor parties, the Greens tend to get inflated opinion poll figures when they’re included in the initial “Which of these would you vote for?” question. If they’re not included, they’re understated. My guess, without being privy to the pollsters’ sausage-producing procedures, is that while they all include the Greens, most of them — but not Ipsos — adjust their result downwards in some way.)

If we take three or four points off Ipsos’s Greens vote and give 80 per cent of it to Labor and 20 per cent to the Coalition (which is about how they would flow in preferences anyway), we get… very similar numbers to Newspoll’s.

Both pollsters agree that Labor is on about 52 or 53 per cent.

The Australian, always a strong believer in the power of positive thinking — and of the media’s (particularly its own) ability to influence political “momentum” — is pushing that 52–48 (“It’s Game On: Budget Bounce for Morrison”) for all it’s worth, while the Age and its siblings see 53–47 as no bounce at all.

Speaking of that archaic annual ritual inflicted with preposterous fanfare on a long-suffering nation last week (and already a distant memory to most), both pollsters agree the budget was well received.

What these and other polls remind us is that while Labor has been ahead in the surveys for more than two years now, the lead has never been huge. Nothing like, for example, the heights reached by the Coalition under Tony Abbott against the Gillard government over much of 2011 to 2013.

Last time the Coalition lost office federally, in 2007, the Howard government’s final pre-campaign Newspoll had Labor on 58 per cent and the Coalition on 42, having averaged 56.5 to 43.5 since Kevin Rudd had become opposition leader the previous December. Six weeks later, Labor’s actual vote came in at 52.7 per cent. A similar improvement for the Coalition next month would see it retain office.

But eleven and a half years ago Labor support was artificially inflated by Rudd’s stratospheric personal popularity. Bill Shorten certainly doesn’t suffer from that affliction.

The surveys will become more useful once this election is called. Most of the pollsters will change their initial question from “If an election were held today, which of the following would you vote for?” to something like “Which of the following will you vote for on May X?”

As we get closer to election day, people’s answers will become more accurate. Changes in sentiment picked up by polls during campaigns are not usually a result of actual “events” but come about because respondents are concentrating their minds. The question becomes less and less hypothetical.

At some point in the campaign, One Nation support will drop significantly in most polls. That’s because it will become clear that the party isn’t running in all, or probably even most, of the 151 electorates. (Psephologist Ben Raue has the current tally at twenty-one.) As much as they can, most pollsters will include them as an option for respondents only in those electorates.

Allocating One Nation preferences has been a bit of a nightmare for pollsters. Should they follow the flow at the 2016 election, when the party uncharacteristically preferenced Labor more than the Coalition? Or should the assumed flow be more favourable to the Coalition? (Correct answer: the latter.)

If One Nation goes down to, say, 2 per cent, most of the other 3 or 4 per cent will shift to the major parties, and most of that to the Coalition. The size of that “most” will affect the two-party-preferred figures a little.

Most observers expect Labor to win off a low primary vote (a problem for both sides), as in the above survey results. Labor almost did it in 2016, but it benefited from a kind of underdog status then; at the very least, no one seriously expected it to win, and it performed better than generally expected. It enters this election as heavy favourite, and that’ll produce a different dynamic, with more policy scrutiny.

And, as I like to say, polls come and go, but for the pollsters there’s only one that really counts: the one before election day. That’s the one against which their “performances” are judged. Which is why Newspoll at least, and probably others, gives it extra resources. The betting markets will also have a better idea on election eve as well. And so will the rest of us.

And of course we’ll all be terribly smart and wise about what happened and why at around 7.30pm eastern standard time. If we ever get there. •

Read next

924 words

Landscape with figures

5 April 2019

Cinema | Bill Nighy delivers a characteristically ambiguous performance in Sometimes Always Never


Cunning blend: Bill Nighy in Sometimes Always Never.

Cunning blend: Bill Nighy in Sometimes Always Never.