Three new polls showing a narrowing of the Opposition’s lead this week may reflect hesitation about voting Labor, writes statistician ADRIAN BEAUMONT.

THE final Resolve poll for Nine newspapers, conducted May 12-17 from a sample of 2049, gave Labor just a 52-48 lead by 2019 election preferences, a two-point gain for the Coalition since last fortnight’s Resolve. By respondent preferences, Labor’s lead was narrower at 51-49, a three-point gain for the Coalition.

Primary votes were 34 per cent Coalition (up one), 31 per cent Labor (down three), 14 per cent Greens (down one), 6 per cent One Nation (up one), 4 per cent UAP (down one), 6 per cent independents (up two) and 4 per cent others (steady). 86 per cent said they were now committed to their first preference (up 10), while 14 per cent were not yet committed (down 10).

Fifty per cent thought Scott Morrison was doing a bad job and 43 per cent a good job for a net approval of -7, up two points. Anthony Albanese gained three points for a net approval of -8. Morrison led as preferred PM by 40-36 (39-33 previously).

Labor and Albanese led the Liberals and Morrison by 32-30 on keeping the cost of living low (34-28 previously). On economic management, the Liberals led by 40-30 (42-27 last time).

The poll supplemented its usual online sample of about 1400 for campaign polls with several hundred respondents interviewed by telephone.

In the three polls so far this week, Resolve has had the most dramatic narrowing. Essential has generally had better results for the Coalition than other polls, and Labor’s lead after preferences has been as low as one point twice this year. The narrowing in Morgan was not all it seemed.

I don’t think the Coalition’s campaign launch on Sunday and their housing policy is responsible, as the fieldwork for these polls began well before then. With Morrison’s ratings still well in negative territory, the narrowing may reflect hesitation about voting Labor.

I expect more polls from Newspoll, Ipsos and perhaps a final Morgan poll by Friday night.

Essential: 48-46 to Labor with undecided included

The final Essential poll, conducted May 11-16 from a sample of 1600, gave Labor a 48-46 lead with undecided included (49-45 last fortnight). Primary votes were 36 per cent Coalition (steady), 35 per cent Labor (steady), 9 per cent Greens (down one), 4 per cent One Nation (up one), 3 per cent UAP (down one), 6 per cent Others (up one) and 7 per cent undecided (up one).

With undecided excluded, the two party would be 51-49 to Labor. Analyst Kevin Bonham estimated 51.6-48.4 to Labor by 2019 preference flows.

Forty-nine per cent disapproved of Morrison’s performance (up one since April) and 43 per cent approved (down one), for a net approval of -6, down two points. Albanese’s net approval was up one point to +1. Morrison led as better PM by 40-37 (40-36 previously).

Thirty-four per cent said the government deserved to be re-elected (up one since last fortnight), and 49 per cent said it was time to give someone else a go (up three).

Morgan poll: Labor’s lead narrows to 53-47, but …

A national Morgan poll, conducted May 9-15 from a sample of 1366, gave Labor a 53-47 lead, a 1.5-point gain for the Coalition since the previous week’s poll. Primary votes were 34 per cent Labor (down 1.5), 34 per cent Coalition (steady), 13 per centGreens (steady), 4 per cent One Nation (steady), 1 per cent UAP (steady), 9 per cent independents (up 0.5) and 5 per cent others (up one).

This two party result is based on 2019 preference flows. Until last week, Morgan was using respondent preferences, which were better for Labor. Bonham gets a Labor lead of 53.9-46.1 from Morgan’s primaries, implying Morgan miscalculated the 2019 flows.

It’s likely Morgan’s high independent vote is because they continue to ask for independents in all seats, even though most seats don’t have viable independents. Resolve was the other pollster that used to have high independent votes, but dropped the independent option in its last poll in most seats, leading to a surge for the Greens.

It’s not mentioned in the poll report, but Labor’s two party estimate using respondent preferences was actually up 0.5 points from the previous week to a 56.5-43.5 lead for Labor.

Seat polls galore, but mainly in NSW and WA

As I’ve said before, seat polls have been unreliable at past elections. The polls listed here are relatively poor for Labor in WA, but strong in NSW, which most of these polls focused on. There are two potential NSW losses for Labor: Eden-Monaro and Hunter, but far more for the Coalition.

Data for the seat polls would mostly have been collected last week, before any national narrowing began.

Polls were good for “teal” independents in Wentworth and Goldstein, and for Labor in inner Brisbane seats, but none of these polls surveyed a regional Queensland seat.

A note on seat margins: in Australia, the margin is the winning party’s two party percentage minus 50 pre cent, not the difference between the two leading candidates. For example, Parramatta is Labor held by a 3.5 per cent margin, meaning that Labor won it by 53.5-46.5 at the 2019 election, a 7.0 per cent difference.

Utting research robopolls of four WA seats were conducted May 12-13 from samples of 400 per seat for the WA Sunday Times. Compared to March polls of the same four seats, these are much better for the Coalition. Labor would still gain Swan and Pearce.

In Swan (Lib, 3.2 per cent margin), Labor’s March lead is down from 59-41 to 53-47. In Pearce (Lib, 5.2 per cent), Labor’s lead reduced from 55-45 to 52-48. In Hasluck (Lib, 5.9 per cent), the Liberals lead by 55-45 after trailing 52-48 previously. And in Tangney (Lib, 9.5 per cent), the Liberals have a 54-46 lead after a 50-50 tie last time.

WA has nearly always been much more pro-Coalition at federal elections than the country overall. These polls suggest that it has reverted to type. Labor’s national poll leads may reflect swings to Labor in the eastern states since the campaign began.

Polls for the Industry Association were reported by Sky News on Sunday. They surveyed seven NSW seats from samples of 800 per seat. Fieldwork dates and pollster used were not mentioned.

In Robertson (Lib, 4.2 per cent), Labor led by 58-42. In Reid (Lib, 3.2 per cent), Labor led by 53-47. In Parramatta (Lab, 3.5 per cent), Labor led by 54-46. In Gilmore (Lab, 2.6 per cent), Labor led by 56-44. In Shortland (Lab, 4.5 per cent), Labor led by 57-43. In Hunter (Lab, 3.0 per cent), Labor led by 51-49. And in Lindsay (Lib, 5.0 per cent), the Liberals led by 57-43.

The report also said that similar polling earlier in the campaign showed losses for the Coalition in Banks (Lib, 6.3 per cent) and Bennelong (Lib, 6.9 per cent).

The weighted share in a Compass poll of North Sydney (Lib, 9.3 per cent), conducted in the week of May 6 from a sample of 507, gave the Liberals 40.5 per cent, Labor 21.6 per cent, an independent 13.6 per cent and the Greens 12.9 per cent. Analyst Kevin Bonham estimated this would be 50.5-49.5 to Labor.

A Redbridge poll of North Sydney for Climate 200, conducted May 3-14 from a sample of 1267, gave the Liberals 35.5 per cent, the independent (Kylea Tink) 24.8 per cent and Labor 18.9 per cent. Bonham said respondent preferences gave Tink a 54.5-45.5 lead over the Liberals.

A Redbridge poll for independent Allegra Spender in Wentworth (Lib, 1.3 per cent vs independent Kerryn Phelps in 2019), reported by The Guardian, gave the Liberals 36 per cent, Spender 33.3 per cent, Labor 11.7 per cent, the Greens 6.2 per cent and UAP 5.3 per cent. Spender would win from these primary votes.

The Poll Bludger reported Tuesday that a Laidlaw poll of Fowler (Lab, 14.0 per cent), conducted three weeks ago from a sample of 618, had Labor’s Kristina Keneally leading independent Dai Le by 45-38 after preferences with 17 per cent undecided.

In the Victorian seat of Goldstein (Lib, 7.8 per cent), Samantha Maiden reported Saturday that a uComms poll for the left-wing activist GetUp! with a sample of 831 gave independent Zoe Daniel a 59-41 lead over Liberal incumbent Tim Wilson.

uComms has had very strong results in its polls for “teal” independents. The Poll Bludger is sceptical as they have not altered their weighting since the 2019 election, when not weighting by education is thought to have caused the poll failure.

YouGov polls for Labor in May of Brisbane (Qld, LNP, 4.9 per cent), Ryan (Qld, LNP, 6.0 per cent), Bennelong and Higgins (Vic, Lib, 2.6 per cent) from samples of 400 per seat had Scott Morrison’s disapproval rating at 57 per cent in Bennelong, 58 per cent in Ryan, 62 per cent in Brisbane and 65 per cent in Higgins according to The Guardian.

Maiden reported Tuesday that uComms polls for GetUp! in Gilmore, Ryan, Eden-Monaro (NSW, Lab, 0.9 per cent), Page (NSW, Nat, 9.5 per cent) and Macquarie (NSW, Lab, 0.2 per cent) gave Labor a 57-43 lead in Gilmore, 55-45 in Ryan and 56-44 in Macquarie. But the Nationals led by 51-49 in Page and the Liberals by 51-49 in Eden-Monaro.

Tasmanian upper house: Labor loses Huon

Tasmanian upper house elections occurred in Huon, McIntyre and Elwick on May 7. The last two were decided by large margins, but in Huon preferences were distributed Tuesday after the postal reception deadline.

Primary votes were 25.1 per cent Labor, 23.7 per cent for conservative independent Dean Harriss, 22.7 per cent Liberals, 20.9 per cent Greens and 7.8 per cent Local Party. After preferences, Harriss defeated Labor by 52.6-47.4. This means Labor drops from five seats to four in the 15-member upper house.

Adrian Beaumont, Honorary Associate, School of Mathematics and Statistics, The University of Melbourne This article is republished from The Conversation under a Creative Commons license.