A Newspoll taken over the weekend shows the Turnbull government on 49 per cent after preferences for the second fortnight in a row. Anthony Albanese has shot to 40 per cent as preferred Labor leader against Bill Shorten’s 29. Other surveys are showing similar voting intentions.
On those national figures, the next election isn’t looking such a sure thing for Labor anymore.
And yet the commentariat, almost to a person, is excitedly reporting exactly the opposite, simply because of a modest by-election swing in a seat called Longman. In the process, they’ve airbrushed the negligible swing in Braddon on the same day.
Also gone missing is the fact that Bennelong saw a bigger swing to Labor last December, yet was largely perceived as a disappointment for the opposition.
Inevitably, some people are plotting the Longman result on the pendulum and listing the seats that would fall to the same swing. But at a modest 3.7 per cent it doesn’t make for great drama. You could take any of the dozens of 53–47 national polls over the past year and a half and apply that 3.4 per cent swing (from the Coalition’s 50.4 per cent at the last election) to get something similar.
What, or who, is driving this interpretation of Saturday’s results?
There are, of course, the furious partisans who roam social media in permanent campaign mode, insisting that the Albanese leadership story was made up, Lindsay MP Emma Husar is the victim of a beat-up and it’s about time journos publicly acknowledged the beauty and magnificence that is Shorten and the Labor Party.
Another group, Malcolm Turnbull’s enemies inside the Coalition movement, are also availing themselves of this weekend’s opportunity.
And there are journos themselves, some of them feeling silly for having contemplated Liberal victories on Saturday. It’s not really their fault — that’s what the public polling and the “leaked” internal polling told them was possible, even likely. So a big mea culpa, Bill, sorry we doubted you; and now let’s overcompensate and lay it on with a trowel. (Remember this when we’re talking about your leadership again in a month.)
Some commentary comes close to justifying itself simply by the fact the political class believes this is important, therefore it behoves us all to play along. Certainly, while by-elections are pretty meaningless in themselves, they do take on a certain mysticism — a ridiculous amount of fuss about one in Canning in 2015 played a role in Tony Abbott’s demise, for example — but that can’t be a reason for commentators jumping on board.
Common to most of the above is the apparent conviction that if the political class, especially the media, believes something to be true, that will make it so. Momentum and all that. But momentum is illusory in electoral politics.
Before Saturday we all thought Labor would probably win the next election anyway, didn’t we? Back when the by-elections were announced, few gave the Liberal National Party any chance of taking Longman. But those subsequent polls encouraged an expectation that has now been dashed.
It’s true that the 9 per cent collapse in the LNP’s primary vote is noteworthy. But most of it went to One Nation and returned to the LNP in preferences, which doesn’t do Labor much good. And it’s right to point out that One Nation preferenced the LNP this time, and if we imagine it had done the same in 2016 (leaving aside the fact that Wyatt Roy would have retained the seat) and then slot in some estimates then we find a larger swing to Labor on Saturday. Perhaps 1 per cent bigger.
On these pages on Monday Tim Colebatch made the excellent point that the record of by-election swings to oppositions in opposition-held seats is much more modest, an average of 1.5 per cent. But there’s a reason for that, illustrated most clearly by the biggest outlier in Tim’s table, a 13.4 per cent swing to John Gorton’s Liberal government in the Australian Capital Territory in 1970, after the death of a very popular Labor member, Jim Fraser.
By-elections are usually caused by members retiring or dying, which sees their personal vote disappear. On average, this has exacerbated the swing against governments in seats they hold, and softened it in opposition-held seats.
That wasn’t the case in Longman and Braddon (and Fremantle and Mayo). In fact, it was the opposite: these MPs had been newly elected in 2016, which should (all else being equal) generate a personal vote — that is, a “sophomore surge.” In Braddon it was what I call a “single” surge, because former Liberal member Brett Whiteley recontested. In Longman it was a “double” (Wyatt Roy’s 2016 personal vote disappeared, and Susan Lamb generated a new one) — although former state MP Trevor Ruthenberg was already known in parts of the electorate, which eats into that “double” a little.
Remember Liberal Jackie Kelly’s 5 per cent swing in Lindsay in 1996? Sophomore surge contributed to that.
Anyway, by-elections should never be seen as tests for general elections, because people aren’t voting about who will form government. That 1970 swing in Jim Fraser’s seat didn’t do the Coalition much good at the next election.
Yes, you can find big by-election results that portend changes of government: Bass (14.3 per cent) in 1975 and Canberra (16.1) in 1995 come to mind. But the Hawke government survived big swings in its third term (11.1 per cent in Port Adelaide; 8.4 per cent in Adelaide). John Howard came close to losing the election after Lindsay and Ryan (9.7 per cent), in March 2001, gave him a wake-up call.
The only by-election swing to a Labor government in federal history took place in Fremantle in 1994; two years later the Coalition was elected in a landslide. And we’re making a fuss about 3.7 per cent in Longman?
Meanwhile, according to the national polls, Malcolm Turnbull might actually have been doing something right politically over the last few months. He needs to be careful about jumping at by-election shadows. •