Polls and elections

France – so far so good for Macron

The first round of presidential elections has shown a lead for Macron (27.8 percent) over Le Pen (23.2 percent). Mélenchon from the left-centre has 22.0 percent and Zemmour from the populist hard right has 7.1 percent. If these were Australian figures with preferential voting, one would add up the votes on a left-right basis and predict a final election outcome of 49.8 percent for Macron (27.8 + 22.0), 30.3 percent for Le Pen (23.2 + 7.1), and 19.9 percent for others – easy enough to get Macron over the line.

But a run-off system gives the same results as preferential voting only if people who vote for non-leading candidates come out for the final round of voting. Macron’s main challenge is to get Mélenchon’s supporters out for that round and it won’t be easy.

The Guardian has a projection on its website estimating how the second round will go, taking into account that while almost all of Macron’s and Le Pen’s supporters will turn up, many of the others won’t. By their estimates Macron will have a final vote of 53 percent, or 53:47 in Australian TPP terms. The Economist has a similar projection.

Wikipedia has a neat map of first-round voting, revealing a pattern common in many countries. Macron and Mélanchon have done well in Paris and other large cities, while Le Pen has done well in the northeast.

The Economist has a non-paywalled pre-election site Fractured France, including an innovative political metric, a graph showing the relationship between Macron/Le Pen support and voters’ distance from a train station. Macron dominates in regions within about 20 km of a station, while Le Pen dominates in the rest of France. (An Australian refinement could be to add the presence or otherwise of railway station car parks as another factor.)

Voting polls – mixed messages, Coalition still well behind

William Bowe’s BludgerTrack shows the gap between Labor and the Coalition is starting to close. At this stage the two-party-preferred (TPP) for Labor averaged across all polls is 55:45, and for Newspoll is 53:47.

Regaining majority government is a huge task for the Coalition. That doesn’t mean it’s easy for Labor, however. Even though most journalists are conditioned to think of federal politics only in terms of winner-take-all outcomes, and see a minority government as something not-quite-right – a “hung parliament” – the reality is that this election could see Labor having to secure the support of a number of independents if it is to form government. That’s nothing odd: in fact majority government is an exception in most mainland European democracies. As Ross Gittins writes in his regular column “Both sides want us to believe this would be a bad thing, leading to instability. That’s the reverse of the truth. Minority governments are so common at state level that their presence goes unremarked.”

The combined Labor + Coalition vote has been on a downward trend for 70 years, and is now at a point that provides openings for independents and small parties in an ever-increasing number of electorates.

Newspoll – false analogies with 2019

Coalition hopefuls and Labor pessimists remind us that at this stage in 2019, six weeks out from the election, Newspoll showed a 52:48 lead for Labor, but the election result was actually the other way around: 52:48 in the Coalition’s favour. Newspoll claims to have now rectified that serious bias.

There is a widespread view that in 2019 Labor’s vote support fell during the campaign, thanks to Morrison’s mastery of the dark arts of marketing spin, but in fact Newspoll and other polls showed no change in the TPP situation over the election period. Newspoll had a bias, but there is no reason to believe it was inaccurate in its trend estimates. That’s an important distinction.

Newspoll’s current figures are on William Bowe’s Pollbludger site. In round numbers, in comparison with the 2019 election, the Coalition’s primary vote is down by 4 percentage points and Labor’s vote up by 4 percentage points. There has been some recent tightening, but the figures are not off trend.

We can draw out a little more from Newspoll figures if we consider the vote shift since 2019 in terms of “right-populist” and “centrist” groupings:

This latest Newspoll therefore shows the “right-populist” vote at 43 percent (36+3+4 percent) and the “centrist” vote at 47 percent (37+10). That’s 5 percent down for the “right-populist”, and 3 percent up for the “centrists”. Within the “right-populist” group the Coalition has lost, while One Nation plus UAP has held at 6-7 percent. The sustained strength of the populist far right is missed by most media.

There has to be some explanation why the “right-populist” vote is down by 5 percent while the “centrist” vote is up by only 3 percent. That remaining 2 percent is explained by a rise in the “other” vote, which was 8 percent in 2019 and is 10 percent now. That’s all a bit too neat, and there is a large margin of error in estimating support for small parties, but Newspoll has been showing a rise in the “other” vote for the last 9 months. And, because of problems to do with ease of recall and name recognition, polls have a bias against independents. It’s fairly safe to assume that independents will do well in the coming election.

That’s why, while the return of a majority Coalition is highly unlikely, it would be a brave person who could make a prediction between a majority Labor government and a minority Labor government. Because of the dynamics of preferential voting, very small differences in primary votes can tip seats.

Morgan – a surge for independents and small parties?

Morgan has a poll reporting Labor’s support at 36.0 percent (the same low figure as Newspoll) and the Coalition’s at 32.5 percent (one of its worst). Yet the TPP figure, based on respondent-allocated preferences, is 57:43 in Labor’s favour.

A plausible explanation for the low combined Labor + Coalition vote (68.5 percent) and for Labor’s high TPP figure is high support for small parties and independents whose preferences will flow heavily away from the Coalition.

The poll is based on surveys from Monday April 4 to Monday April 11.

Other polls – generally bad news for the Coalition but treat them with caution

Resolve Strategic has published a poll of state-by-state estimates, pointing to big losses for the Coalition. It is published in the Sydney Morning Herald – Voters deserting Coalition in Queensland and WA: Resolve survey – but that’s behind a paywall. William Bowe has a summary on his PollBludger site however.

In this election there is a flood of polling, particularly at a state and electorate level. This is understandable as more independents and candidates from minor parties look like having a strong influence, and in view of the big state differences revealed in recent elections. But these polls should be treated with caution, because the smaller the population surveyed, the greater the sampling error, and when it comes to distribution of preferences tiny differences in primary votes can have a big difference in results.