Tue Jun 13

Resolve first national poll to have 'no' ahead in Voice referendum, but Essential has 'yes' far ahead

Written by Adrian Beaumont, Election Analyst (Psephologist) at The Conversation; and Honorary Associate, School of Mathematics and Statistics, The University of Melbourne

A federal Resolve poll ^[1] for Nine newspapers, conducted last week from a sample of 1,606, had “no” ahead by 51-49 in a forced choice question on the Indigenous Voice to parliament referendum to be held later this year.

This is the first “no” lead in any national Voice poll. In the past two months, there has been a rapid movement to “no” in Resolve polls. “Yes” was ahead by 58-42 in April, but this dropped to a 53-47 lead in May, and now “no” has taken the lead.

Party breakdowns suggest a large drop in support since April from Labor voters, with their support down 12 points to a 63-37 split in favour of “yes”.

State breakdowns were based on aggregate data from May and June with an overall sample of 3,216. As May and June results were averaged, “yes” led overall by 51-49. But these breakdowns suggest the Voice would not get the required majority in four of six states, as it was trailing narrowly in South Australia and Western Australia, and much further behind in Queensland.

Only 30% of voters said they could confidently explain the Voice to someone else. Not understanding the detail is likely a key driver of the swing to “no”. Initial preferences were 42% “yes” (down two since May ^[2]), 40% “no” (up one) and 18% undecided (steady).

Essential poll: ‘yes’ leads by 60-40

An Essential poll ^[3], conducted last week from a sample of 1,123, gave “yes” to the Voice a 60-40 lead, a one-point gain for “yes” since May. In April ^[4], Essential and Resolve were only two points apart, with Resolve at 58-42 “yes” and Essential 60-40. Now there is an 11-point gap.

I believe Resolve and Newspoll are more credible as they have been tested and performed well at the federal 2022, Victorian 2022 ^[5] and New South Wales 2023 elections ^[6]. In my opinion, Resolve had the best final poll at the federal election, while Newspoll performed best at both the Victorian and NSW elections.

In contrast, Essential greatly overstated both major parties’ support at the federal election, and did not produce a voting intentions poll close to either the Victorian or NSW elections.

It’s possible that Essential’s skew to “yes” is caused by a bias in their samples towards better educated respondents. Seats with a higher educational attainment are where the Greens and teal independents performed best at the last federal election.

Relative to other polls, Essential has had much ^[8] higher Greens votes, with their latest poll putting the Greens on 16% (up one since last fortnight), compared to 12% in Resolve ^[9] and 12% in last week’s Newspoll ^[10]. For this parliamentary term, the Greens support has usually been about 14% in Essential compared to 10-11% in other polls.

Last week’s Newspoll gave “yes” an overall 46-43 lead with 11% undecided. Education breakdowns ^[11] had university-educated people favouring “yes” 56-35, but “no” was leading with both TAFE/college (by 48-43) and no tertiary education (by 45-41).

The skew to “yes” in Essential may be explained by having too many university educated people in their samples. Better-educated people are more likely to have decided how to vote on the Voice a long time ago, explaining the stable referendum vote in Essential.

A struggle ahead for the ‘yes’ campaign

In early May, I wrote that just one of 25 Labor-initiated referendums had succeeded, and that early poll leads for proposals often collapse by the referendum date. Furthermore, Labor-initated referendums have lost much more heavily when held as a midterm referendum than when held with general elections.

The Resolve poll and last week’s Newspoll (46-43 to “yes”) implied that support for the Voice is slumping. Given the history of Labor-initiated referendums, if these polls are correct, it is likely that the Voice will be defeated. Supporters of the Voice will hope that Essential is right, but Essential’s track record is worse than either Resolve or Newspoll.

Fadden byelection: July 15

A byelection for the federal Queensland seat of Fadden will occur on July 15 after the resignation of former Liberal minister Stuart Robert. The LNP won Fadden by a 60.6-39.4 margin against Labor at the 2022 federal election.

Fadden has only been won by Labor once, in 1983 on the election of the Hawke government, and has been more pro-LNP than Queensland overall since 1984 according to the ABC guide ^[13].

In my first look at this byelection, I thought Labor might not contest given how unlikely it is they would win. However, Labor will contest with their 2022 candidate. Nominations for the byelection close on June 22 and will be declared the next day. The LNP should retain easily.

WA state poll: 61-39 to Labor

The Poll Bludger ^[15] reported on June 3 that an Utting research poll for The West Australian, conducted May 31 from a sample of 800, gave Labor a 61-39 lead from primary votes of 52% Labor, 28% Liberals, 5% Nationals and 8% Greens. This poll was taken two days after WA Premier Mark McGowan announced his resignation ^[16].

While Labor is well down from the record landslide margin of 69.7-30.3 at the March 2021 WA election, this would still be a massive win for Labor if repeated at an election.

New Labor Premier Roger Cook had ratings of 42% approve, 26% disapprove, while Liberal leader Libby Mettram was at 31% approve, 33% disapprove. Cook had a 50-24 lead as preferred premier over Mettram. McGowan’s ratings were 68% approve, 19% disapprove.

There will be a byelection ^[17] in McGowan’s former seat of Rockingham on July 29. At the 2021 WA election, McGowan won Rockingham ^[18] by an 87.7-12.3 margin over the Liberals, from a primary vote of 82.7%. There is likely to be a big swing against Labor at the byelection, but they should still hold Rockingham easily.