Labor enters the federal election campaign in a stronger position than it was just six weeks ago, with polling released today suggesting the government could even be returned in majority.
Research from pollster YouGov has found that if an election were held today, Labor would end up with a median estimate of 75 seats in the next parliament, with a lower estimate of 69 and an upper estimate of 80.
That central estimate is just one short of 76, a majority in parliament.
The Coalition is behind on 60, with a lower estimate of 55 and an upper estimate of 68.
The estimate is the result of the organisation’s latest MRP model, in which 38,629 surveys were conducted, with the results modelled across all 150 electorates to produce an estimate of the vote share in each seat. It will not predict every single seat accurately.
It is a marked change from YouGov’s last MRP model published six weeks ago, which found at that time that the Coalition were the favourite to win the election. Labor is now ahead in 10 of the seats where it was behind in the February modelling.
The seats that could change hands
This latest research finds the Coalition would gain six seats: Bennelong, Gilmore, Werriwa and Robertson in NSW, Lyons in Tasmania and the seat of Aston, which it lost in a by-election. Labor would gain Griffith and Brisbane, which are currently held by The Greens. It also suggests Labor could gain Deakin in Victoria from the Liberals, which would be a surprising result.
It is just one model and should not be taken as a prediction of the outcome of the election, but it is consistent with other quantitative research that has recorded a tightening in the race in recent weeks.
The model suggests there is still significant uncertainty in the election outcome. The surveys were conducted between February 27 and March 26, almost entirely before the federal budget was announced, along with several new policies from the major parties.
YouGov estimates a 38 per cent chance of a Labor majority government were an election held today, and a 61 per cent chance of a hung parliament.
Nationally, the poll found Labor had 50.2 per cent support in two-party preferred terms, with the Coalition on 49.8 per cent. That would represent a swing back toward Labor of around 1.3 per cent since YouGov’s February poll.
That result is consistent with the shift recorded toward the government in the past month across the polling industry as a whole.
YouGov’s director of public data Paul Smith puts at least part of the shift away from the Coalition down to Peter Dutton scoring own goals with “very unpopular” policies.
“Dutton needed to win working-class votes in outer suburban Sydney and Melbourne but his Trump-style policies of ending work from home and cutting 40,000 public servants have seen his support dramatically impacted,” he said.
Crossbenchers see position improve in latest model
The model now estimates a larger crossbench than was predicted six weeks ago, with a central estimate of 15 seats. It suggests the Greens are ahead in two of its four seats, and incumbent independents are generally performing well.
It has independent candidates Alex Dyson ahead in Wannon in regional Victoria, and Caz Heise ahead in Cowper on the NSW mid-north coast, with Nicolette Boele very close to overtaking the Liberals in the Sydney seat of Bradfield.
This poll, like YouGov’s last MRP, was conducted with a “generic ballot”, where respondents were asked who they would vote for from a list of parties without being told the names of candidates running in their seat.
This is standard practice for many pollsters at this point in the election cycle, where the full list of candidates is unknown, but it introduces considerable uncertainty in predicting the results in seats with strong independent or minor party challenges.
In general, MRP models have a better record of accuracy when estimating seat shares for established parties than for independents. MRP models also struggle where there are significant local factors specific to a particular seat.
Figures show election result is highly sensitive
MRP models are a snapshot in time and will not correctly predict the winner in every single seat, but they give us a better idea of geographic trends and patterns in voter behaviour.
This model continues to record the largest swings against the government in outer metropolitan seats.
As well as indicating a general tightening in the contest, the significant shift in the estimates suggests that the outcome of the election is sensitive to small shifts in voter behaviour.
There are 14 seats that YouGov predicts are “toss-ups” or “leaning” towards one party, and there are 24 seats in the model where the leading candidate has less than 52 per cent of the two-candidate-preferred share of the vote.
There are a very large number of soft and swing voters right now who could change their minds during the campaign.
It demonstrates that the quality of the major parties’ campaigns could have a significant impact on how many seats they win.
“Events can shift outcomes very quickly,” Paul Smith says.
