Australia Decides: Polling Special Edition

The PM and Opposition Leader during their fractious Mothers Day debate

With less than 10 days to go to the election, we figured this would be a good point to review the polling and what it is telling us, consider some of its weaknesses and look at the betting markets for an alternative indication of voter sentiment.

Newspoll

The latest Newspoll, published in the Australian Newspaper on 8 May, shows a widening of the ALP's lead in the national vote.

According to Newspoll and based on preference flows from recent State and Federal elections, the ALP extended its lead by one to an eight point lead: 54% to 46%. Newspoll has Mr Morrison on a virtual tie with Mr Albanese for preferred Prime Minister, with 44% to 42% respectively, while 14% were uncomitted.

Ipsos

The latest Ipsos poll, published in the Australian Financial Review on May 8 is telling us a similar story.

Most notably, Ipsos' methodology provides the person being polled an opportunity to allocate their preference, which is particularly important where they are planning a minor party primary vote.

A few things to note from the Ipsos poll, extracted above:

• Between the 2019 election and now, the Liberal/National primary vote has fallen 12.4% to 29% (though 7% of voters surveyed responded they did not know who they would vote for).

• Since 2019, Labor's primary has grown only 1.7% to 35%.

• For two-party preferred vote "with a stated preference" (highlighted in orange) Labor is on 50% - though 15% remain undecided.

• Consistent with Newspoll, the preferred Prime Minister question gives Mr Morrison only a 5% advantage - with nearly a quarter of voters uncommitted.

Resolve Political Monitor

The Resolve Poll published in the Sydney Morning Herald and the Age on 30 April, indicates what appears a tighter contest, as the poll does not attempt to allocate preferences.

Perhaps most startling about the poll's trend is a fall in Liberal and National party support over the last 12 months: dropping 7 points since June 2021; and 8 points since the May 2019 election.

The poll gives the Greens 15% of the vote. This is highly questionable on the basis of previous election outcomes, and is likely to be inaccurate and an outlyer result.

To further increase uncertainty, the poll asks voters how firm they are in their vote. With 24% uncommitted, Resolve tells us there is a trend toward Labor among voters, but also tells us that this election is wide open.

Can we rely on the polls?

When asked to comment on published opinion polls, the cliché political response generally comes that the only poll that counts is the one held on election day.

With the spectacular failures of published opinion polls in 2019, it is useful to ask a few questions:

• What is the impact of published polling?

• Why do they get it wrong? and What are some of the problems facing published polling?

• Where else should you look for a reliable indication of the likely electoral outcome?

What is the impact of published polling?

During election campaigns momentum is everything.

A good published result from a reputable national polling outfit can put a kick in the step of candidates, and their campaigns and can build positive sentiment among voters. For example, creating a ‘social proof’ which might open the minds of voters to supporting a party or candidate.

Political parties also make resource allocation and messaging decisions based on their own confidential, private polling. Following the 2019 election, party officials on both sides have said that a clear and winning swing to the Liberal and National parties had been identified in the lead up to election day. While national published polls missed the surge to the Coalition, private polling was giving the major parties what they needed to make decision about resources. In those last 10 to 14 days of the 2019 campaign the Liberals moved their resources to capitalise on momentum, while the ALP shifted to defence.

Why do the published polls get it wrong?

The value of social research, including public opinion polling, comes down to a reliable methodology, which in turn depends on accurate and sizeable data collection and representative sampling. Where pollsters seek to make further inferences and extrapolations, such as for example 2-party preferred allocations, greater strain is placed on the methodology. Pollsters generally use a mix of online surveys and computer assisted telephone interviewing. A long time issue among pollsters is the question of how you get to a representative sample and the big question of whether it is best to use landlines or mobile numbers.

Mobiles versus landlines, and other sampling difficulties

There has been a long standing debate among pollsters on the impact of landline versus mobile phone survey responses.

On one hand calls to landlines may have a higher level of acceptance, however this is at least partly because of the older demographic still using landline phone numbers. According to the Australian Communications and Media Authority, to June 2020, 60% of Australians had no landline for voice calls.

The same survey found 83% of 25 to 34 year olds and 76% of 35 to 44 year olds, were mobile-only. In other words, any poll where more than 50% of responses are provided via landline is likely to be skewed towards voters over the age of 55.

In addition, simply gaining access to phone numbers has become problematic with many people choosing to keep their mobile and landlines numbers off published lists.

Some pollsters have turned to using brokers to overcome these issues and gain access to survey participants.

Considering these constraints, and then adding degrees of sampling difficulty across gender, urban, regional and rural populations, then perhaps it is a legitimate question to ask whether the time of accurate and useful political opinion polling has passed?

Do the betting markets provide an alternative reliable indication of the likely electoral outcome?

Australia has well-entrenched and mature electoral betting markets. Outfits like Sportsbet, Betfair and others provide endless entertainment for political tragics.

A review of Sportsbet’s market (with 10 days to go to the poll) has:

1. Labor majority as the favourite at $1.62 (versus $2.40 at three weeks ago).

2. Labor minority at $3.95 (versus $3.75 at three weeks ago).

3. Liberal/National Majority at $5.00 (versus $2.95 at three weeks ago).

4. Liberal/National minority at $6.50 (versus $5 at three weeks ago).

While we might expect betting odds to tighten as the election nears, on these numbers the betting markets are increasingly favouring Labor to form the next Australian Government.

For educational purposes only, you can find the market here.

Some of the markets for highly contested seats, versus three weeks ago, are set out in the following table:

Some interesting reflections from the betting markets:

• Goldstein and Kooyong in Victoria have markedly tightened over the last three weeks. In Goldstein the Teal Independent is now favourite, while in Kooyong, the market is showing effectively a dead-heat.

• In North Sydney the Liberals appear ahead, while in Wentworth the Teal Independent is now favourite.

• Must win seats for the ALP like Chisholm, Longman, Swan, Pearce, Bass, Braddon and Boothby have either remained steady or strengthened in the ALP's favour.

Analysis of the published opinion polls and comparison with the betting markets reminds us of one thing. Elections are won by the party that gains more seats, not more votes.

While Newspoll, Ipsos and Resolve each give a sense that the election is there for the winning by the ALP, looking at the seat-by-seat picture, the result is far less clear and electing the magic number of 76 Members of Parliament remains uncertain. For example, should the challenges of the Teal Independents fall flat, if Labor falls short in several of its target seats, or if the LNP manage to pick up Gilmore and/or Parramatta, the result is, all of a sudden, wide open.

Anyone can win from here.