Inside Story

The by-election that won’t tell us much but means a lot

By-elections sometimes reflect important political trends, and sometimes they don’t, writes Peter Brent. So why are we watching Canning so closely?

Peter Brent 28 August 2015 1141 words

Potential backfire? Mobile ads funded by the CFMEU for the union campaign against the Coalition in Canning. Rebecca Le Ma/AAP Image


It was Saturday 26 March 1988 and Coalition supporters were cock-a-hoop. Just a week earlier twelve years of NSW Labor government had been swept away, big time, and now the federal electorate of Port Adelaide had swung to the Liberals by a whopping 11.1 per cent at a by-election caused by the resignation of Labor’s Mick Young.

Port Adelaide stayed Labor – just – but that kind of double-digit swing is (or was) rare. Political observers vividly recalled the 14 per cent slide in Bass in 1975 that foretold the Whitlam government’s election demolition later that year.

Just a month before Labor’s knock in Port Adelaide, the Liberals had taken Adelaide in a by-election with an 8.4 per cent swing. For the five-year-old, twice re-elected (most recently, just nine months earlier) Hawke government, all was doom, gloom and introspection. Opposition leader John Howard felt obliged to warn his party room against complacency, insisting the next election was not a foregone conclusion, but his future looked bright.

It turned out, however, that Labor was not even halfway through its thirteen-year reign. Howard was replaced as leader barely a year later, and at the next election in 1990 Adelaide returned to Labor and Port Adelaide became safe ALP territory again.

Two potent political fallacies were on show in this short-lived frenzy: that state election results point to federal ones, and that by-elections portend general elections. Both myths persist to this day.

The history of the political class’s obsession with by-elections would make a fascinating study. Presumably we get this bizarre sub-genre from the Brits, who pore over by-election results at least as much as we do.

For every big by-election result that correctly points to the demise of a government (Bass in 1975, Canberra in 1995), another (Port Adelaide in 1988, Ryan in 2001) doesn’t. And here’s a statistic: only once since federation has a federal Labor government achieved a positive swing at a by-election: in Fremantle in 1994. Yet two years later Australian voters sent the Keating government out the back door in no uncertain terms.

Applying by-election swings to the pendulum to produce jaw-dropping seat outcomes is a favourite journalists’ pastime – watch out for it from 20 September – but it is an utter waste of time. As we know, by-elections usually move against sitting governments. Various factors contribute to the result, such as which side held the seat (and is losing its MP’s personal vote) and the manner of that member’s departure. (The 1994 Fremantle vote was propelled by the political persona of the Labor candidate, former premier Carmen Lawrence.)

But the main reason by-elections are unreliable indicators is that so little is at stake for the voter. They are not deciding who should form government, and so they can vote about other things. Local issues and, for once, the quality of candidates can play substantial parts. Delivering an arrogant government a wake-up call is always tempting.

And in three weeks’ time – here we go again – it’s the turn of Canning in Western Australia. The stakes are low for voters but high for the major parties. There is no rational reason for this to be so, but it’s a self-fulfilling conviction within the political class. Both leaders have acknowledged it.

And ACTU secretary Ged Kearney has assured the Australian newspaper that “we will be using our resources and our campaigning power in Canning to influence the result.” All rather brave considering so much of the outcome will be in the lap of the gods.

The late Liberal MP Don Randall had a very high personal vote. These things are impossible to precisely measure, but the pattern of House of Representatives and Senate votes in 2013 suggests that if the sitting member had been some other average Liberal, the current margin might be about 6 or 7 per cent instead of 12. And a 6 or 7 per cent by-election swing would be quite routine.


Polling so far has suggested the seat may be up for grabs. A Newspoll found the Liberals ahead 51–49 after preferences, while ReachTel puts it at 50–50. But the Canning result will barely affect the situation in parliament. It will marginally influence the outcome in that seat at the next election, because whoever takes it on 19 September will enjoy an advantage in personal vote and financial resources. But it’s still only one seat.

A large proportion of inhabitants of the political bubble believe in electoral “momentum” – that a by-election victory for one side will not just point to the next election outcome (dodgy enough) but also somehow influence it. This illusion is wrapped up in the commentator conceit that their words can make things happen: that if they pronounce that a government or opposition is travelling well, it will be so.

So the parties and leaders care about Canning because they have to. It is important because it is believed to be important. Some have scribbled that Tony Abbott is finished if the Liberals lose. While chatter has negligible influence on electoral outcomes, leadership speculation in the media can be self-fulfilling. That would certainly be a “scalp” for Bill Shorten, but it is probably in the opposition leader’s interests for Abbott to remain prime minister until the next election.

In this Seinfeldian election about nothing, voters can latch onto esoteric motivations. If they believe they can effect change at the top of the Liberal Party, then that might influence them. Some who desperately want to see a Liberal victory in 2016 may vote Labor because replacing Abbott as prime minister would make Coalition survival more likely.

Others might wish to send a message to Bill Shorten and Labor. The Liberals would love to turn the vote into a referendum on a mining tax. The union movement, meanwhile, seems to have inhaled its own press about its self-proclaimed responsibility for last year’s election outcome in Victoria; if high-profile union involvement in the campaign becomes an issue in itself that could backfire.

Western Australia remains the government’s brightest (or less dim) spot in national polling, but the polling swing there is if anything worse than in the rest of the country. The most recent Newspoll quarterly had Labor ahead 53–47 nationally, a 6.5 per cent swing, and 50–50 in the west, an 8 per cent change from the 2013 result.

By-elections are like massive political surveys with tiny margins of error. Their downside is that the question is not “how would you vote if a general election were held today?” and so we don’t know how to interpret the results.

Canning won’t really tell us anything about the next election. But because it’s being so heavily scrutinised it will mean a lot. •