Poll wrap: Labor maintains 51-49 Newspoll lead, plus many seat polls

The election will be held in four days, on May 18. There will be 151 House seats in the new parliament, up from 150 now. There are 47 NSW seats, 38 Victorian seats, 30 Queensland seats, 16 WA seats, 10 SA seats, five Tasmanian seats, three ACT seats and two NT seats.

Owing to a favourable redistribution for Labor in Victoria, the creation of a third seat in the ACT and Kerryn Phelps' win at the October 2018 Wentworth byelection, the Coalition notionally holds 73 of the 151 seats based on 2016 results, Labor 71 and there are six crossbenchers. Corangamite is on zero margin after a redistribution. To win a majority, either the Coalition or Labor must win 76 or more seats.

Read more: Labor benefits from completed draft boundaries, plus South Australian and Tasmanian final results

The Coalition won the 2016 election with 76 of the 150 seats to 69 for Labor, but Labor gained 14 seats. As a result of these gains and some Coalition retirements, Labor should do better in seat terms than implied by the pendulum, owing to the "sophomore surge" effect, where new members usually do better in swing terms than the overall swing in a state or region.

Assuming no net gains or losses to the crossbench, analyst Kevin Bonham estimates Labor could lose the national two party vote 50.3-49.7, and still have a better chance of winning more seats than the Coalition. On the pendulum, Labor needs a 0.6% swing, or a 50.2-49.8 two party win, to have more seats than the Coalition. These are estimates for Labor winning more seats than the Coalition, not a majority of seats.

This week's Newspoll, conducted May 9-12 from a sample of 1,644, gave Labor a 51-49 lead, unchanged since last week. Primary votes were 39% Coalition (up one), 37% Labor (up one), 9% Greens (steady), 4% One Nation (down one) and 4% for Clive Palmer's United Australia Party (UAP) (steady).

44% were satisfied with Scott Morrison's performance (steady) and 44% were dissatisfied (down one), for a net zero approval. Bill Shorten's net approval was up eight points to -10, his best net approval since March 2015. Morrison led Shorten as better PM by 45-38 (46-35 last week). Bonham says Morrison's better PM lead is weak given voting intentions.

A key reason for the Coalition's competitiveness at the election is Morrison's relative popularity. After his honeymoon wore off, Malcolm Turnbull was usually unpopular, while Tony Abbott became unpopular shortly after becoming PM. Australia has not had a PM with enduring popularity since Kevin Rudd in 2008-09.

Morrison is conservative enough that most of the hard right like him, while they hated Turnbull. However, he has not done anything that the vast majority of voters disliked, such as Abbott's knighting of Prince Philip or the 2014 budget. In this way, Morrison is similar to John Howard.

Overall, the national polling is consistent with a narrow Labor victory on Saturday. But Labor supporters should not be complacent, as the polls could be understating the Coalition's vote, or there could be a shift to the Coalition as late deciders make up their minds. On the other hand, a blowout Labor win cannot be ruled out either.

I expected an Essential poll, but Essential appears to have delayed their poll until the final days. I will write a final poll wrap article early on Election Day.

Morgan poll: 52-48 to Labor

A Morgan poll, conducted in "May", gave Labor a 52-48 lead. I am not sure if this is Morgan's face-to-face poll, or another poll they are conducting. If it is the face-to-face, it implies that this poll is an average of May 4-5 and 11-12, and that the May 11-12 poll was about 53-47 to Labor, a two-point gain for Labor since last week. Primary votes in this Morgan poll were 38.5% Coalition, 35.5% Labor, 10% Greens, 4% One Nation and 3.5% UAP.

Newspoll state breakdowns

Newspoll has released state breakdowns for all its five polls conducted in April to May. Nationally, Labor leads 51-49 (50.4-49.6 to Coalition in 2016). In NSW, the Coalition leads by 51-49 (50.1-49.9 to Labor). In Victoria, Labor leads by 54-46 (51.8-48.2). In Queensland, there is a 50-50 tie (54.1-45.9 to Coalition in 2016). In SA, Labor leads 52-48 (52.3-47.7). In WA, the Coalition leads 52-48 (54.7-45.3). Newspoll includes the ACT in its NSW breakdowns.

The problem I have with these data is that the first two polls in this sample were from early April, soon after the NSW March 23 election, which the Coalition won. This election result probably assisted the federal Coalition in NSW, but it may not carry through at the federal election. In 2016, the Coalition won NSW by 50.5-49.5, virtually the same as the 50.4-49.6 national margin. I am sceptical of the swing in NSW being very different from the national swing.

The Poll Bludger has incorporated the Newspoll breakdowns into BludgerTrack. Labor leads by 51.7-48.3, a 2.0% swing to Labor since 2016. There are swings to Labor in all states except NSW, where the Coalition has a 1.6% swing. BludgerTrack currently gives Labor 79 of the 151 seats, to 66 Coalition and six crossbenchers. Bonham also gives Labor the same seat count.

Queensland Galaxy: 51-49 to federal Coalition

A Queensland Galaxy poll, conducted May 8-9 from a sample of 848, gave the federal Coalition a 51-49 lead, a three-point gain for the Coalition since the last Queensland Galaxy poll in February, which was probably a pro-Labor outlier. Primary votes were 38% Coalition (up three), 33% Labor (down one), 9% Greens (down one), 9% One Nation (up one) and 5% UAP (steady).

Queensland is a conservative state, and this poll represents a 3% swing to Labor since the 2016 election.

Seat polls of Herbert, Lindsay, Corangamite, Bass, Boothby, Kooyong and Higgins

In late April, The Poll Bludger wrote that seat polls at state elections and federal byelections since the 2016 federal election have continued to be inaccurate, and somewhat biased to the Coalition. In polls conducted during the final fortnight of election campaigns, the Coalition's primary vote was on average overstated by 1.9% with an error of 9.5%. Labor's primary vote was understated by 0.5% with an error of 6.5%. An error of 4-5% is expected for polls with a sample about 500.

Newspoll conducted seat polls of Herbert, Lindsay, Corangamite and Bass from May 9-11 with samples of 500-580 per seat. In Herbert, the LNP led Labor by 52-48 (Labor barely won it in 2016). In Lindsay, the Liberals led Labor by 52-48 (51.1-48.9 to Labor in 2016). In Corangamite, Labor led by 51-49 (no margin after redistribution). In Bass, Labor led by 52-48 (55.4-44.6 in 2016).

Herbert and Lindsay were previously polled by Newspoll on April 20. There has been a two-point swing to the LNP in Herbert since that poll, and a three-point swing to the Liberals in Lindsay. Bass and Corangamite were previously polled by ReachTEL in the first week of the campaign. Comparing these Newspolls to those earlier ReachTEL polls gives a three-point swing to Labor in Corangamite, and a six-point swing to Labor in Bass.

Read more: Poll wrap: Palmer's party has good support in Newspoll seat polls, but is it realistic?

Primary votes in Herbert were 35% LNP (up four since April 20), 30% Labor (up one), 13% Katter's Australian Party (up three), 7% Greens (up two), 7% One Nation (down two) and 7% UAP (down seven). In Lindsay, primary votes were 44% Liberals (up three since April 20), 39% Labor (down one), 6% UAP (down one) and 4% Greens (steady). In Corangamite, primary votes were 42% Liberals, 37% Labor, 10% Greens and 4% UAP. In Bass, primary votes were 40% Liberals, 39% Labor, 10% Greens, 4% UAP and 2% Nationals.

A YouGov Galaxy poll for The Advertiser, conducted May 9 from a sample of 522, gave the Liberals a 53-47 lead (52.7-47.3 in 2016). Primary votes were 47% Liberals, 37% Labor, 9% Greens and 3% UAP. Centre Alliance (formerly Nick Xenophon Team) is not contesting Boothby, so Labor's primary vote is up 10% and the Liberals up 5%.

The Guardian reported a Greens-commissioned poll of Kooyong gave Liberal Josh Frydenberg a 52-48 lead over the Greens' Julian Burnside (62.8-37.2 vs Labor in 2016). Primary votes, after excluding 8% undecided, were 45% Frydenberg, 23% Burnside, 17% for Labor's Jana Stewart and 10% for climate-focused independent Oliver Yates. This poll should be treated with extra scepticism as it was commissioned by the Greens, who are using it to argue for a vote for Burnside to oust Frydenberg. The sample was very large for a seat poll at 1,741.

A Greens-commissioned seat poll of Higgins was also reported in The Guardian. Primary votes were 36% Liberal (51.6% in 2016), 30% Labor (16.5%) and 29% Greens (24.2%). If this poll is even roughly accurate, whichever of Labor and the Greens finishes second will win Higgins on the other's preferences. However, it's a Greens-commissioned poll that sampled just 400 voters.

Adrian Beaumont, Honorary Associate, School of Mathematics and Statistics, University of Melbourne

This article is republished from The Conversation under a Creative Commons license. Read the original article.