New Zealand politics – Page 3

US election minus four weeks

Biden extends his national lead after the first debate and Trump’s coronavirus. Meanwhile, NZ Labour retains a large lead.

Guest post by Adrian Beaumont, who joins us from time to time to provide commentary on elections internationally. Adrian is an honorary associate at the University of Melbourne. His work on electoral matters for The Conversation can be found here, and his own website is here.

This is an updated version of an article I had published for The Conversation on Wednesday.

With four weeks left until the November 3 election, the FiveThirtyEight aggregate of US national polls gives Joe Biden a 9.5% lead over Donald Trump (51.7% to 42.2%). Biden’s lead has increased 1.9% since last week’s article.

In the key states, Biden leads by 7.7% in Michigan, 6.9% in Wisconsin, 6.9% in Pennsylvania, 4.6% in Florida and 4.4% in Arizona. If Biden wins the states that voted for Hillary Clinton in 2016, plus Michigan, Wisconsin and Pennsylvania, he wins the election with at least 278 of the 538 Electoral Votes.

Pennsylvania and Wisconsin are tied as the “tipping-point” states that could potentially put either Trump or Biden over the 270 EVs required to win. Pennsylvania has been polling closer to Wisconsin and Michigan than in the recent past. The current difference between Pennsylvania and the national vote is 2.6% in favour of Trump. At the moment, that gap isn’t much help to Trump.

There are five states where Biden is either just ahead or just behind: North Carolina, Georgia, Texas, Iowa and Ohio. If Biden won all of them, he would win a blowout victory with over 400 EVs.

In the FiveThirtyEight forecast, Trump still has a 16% chance to win, though only a 7% chance to win the popular vote. Trump’s chances have declined 5% since last week. Still, a 18% chance is the probability of rolling a six on a six-sided die.

Trump’s ratings with all polls in the FiveThirtyEight aggregate are 43.4% approve, 53.3% disapprove (net -9.9%). With polls of likely or registered voters, Trump’s ratings are 43.8% approve, 53.2% disapprove (net -9.4%). His net approval has declined about one point since last week. RealClearPolitics averages have Biden at +6 net favourable, while Trump is at -12.

The FiveThirtyEight Classic Senate forecast gives Democrats a 72% chance to win, up 4% since last week. The most likely outcome is a narrow 51 to 49 Democratic majority, unchanged from last week. The forecast gives Democrats an 80% chance of holding between 48 and 55 seats after the election.

Trump’s coronavirus

Perhaps there would have been some public sympathy for Trump had his coronavirus appeared to be bad luck. But it is likely Trump and other prominent Republicans’ coronavirus infections occurred at a September 26 event to announce Amy Coney Barrett as Trump’s nominee to replace the late Ruth Bader Ginsburg on the Supreme Court.

Pictures in the linked article show people sitting close together, without face masks. This created an impression of reckless conduct by Trump and other Republicans in ignoring medical advice.

In a CNN poll taken after Trump’s coronavirus, 60% disapproved and 37% approved of Trump’s handling of coronavirus; his -23 net approval is a record low on that issue. By 63-33, voters thought Trump had acted irresponsibly.

An additional problem for Trump is that coronavirus is back in the headlines. As Trump is perceived to have been poor on this issue, that helps Biden. New US daily cases have plateaued between 30,000 and 50,000.

US employment growth slows

The September US jobs report was the last before the November 3 election. 661,000 jobs were created, and the unemployment rate dropped 0.5% to 7.9%. This was the first month with fewer than a million jobs added since the April nadir. The unemployment rate has almost halved from April’s 14.7%. But the gain in September was mainly attributable to a 0.3% slide in the participation rate, to 61.4%.

Trump may have undermined his advantage on the economy by withdrawing from negotiations with Democratic House Speaker Nancy Pelosi over a new stimulus bill. An article by analyst Nate Silver says stimulus spending was very popular: in a September Siena poll for The New York Times, voters supported a $US 2 trillion stimulus by a 72-23 margin.

NZ Colmar Brunton poll: Labour leads National by 47-32

With nine days left until the October 17 NZ election, a Colmar Brunton poll gave Labour 47% (steady since last week), National 32% (down one), the right-wing ACT 8% (steady), the Greens 6% (down one) and NZ First 2% (up one). Jacinda Ardern led Judith Collins as better PM by 50-23 (54-23 last week). This poll was conducted October 3-7 from a sample of 1,007.

If this was the election result, Labour would win 60 of the 120 seats, one short of a majority. National would win 41, ACT 11 and the Greens 8. However, the Greens will be anxious about clearing the 5% threshold, as NZ polls have tended to overstate them. If the Greens miss the threshold, Labour would win a majority on this poll, as it has a 47-40 lead over the combined vote for National and ACT. My personal website has more discussion of last week’s poll, including personal ratings of Ardern and Collins.

US election minus five weeks

Two strong polls for Biden in Pennsylvania make an Electoral College/popular vote split less likely, as Trump loses the first debate.

Five weeks before the November 3 election, FiveThirtyEight’s national aggregate gives Joe Biden a 7.6% lead over Donald Trump (50.5% to 42.9%). This is a slight improvement for Biden since last week, when he led by 7.3%. In the key states, Biden leads by 7.0% in Michigan, 6.9% in Wisconsin, 5.6% in Pennsylvania, 3.7% in Arizona and 2.0% in Florida.

The best polling news for Biden was two Pennsylvania polls from FiveThirtyEight A+ rated pollsters. Both the Siena poll for The New York Times and the ABC/Washington Post poll gave Biden a nine-point lead in Pennsylvania.

If Biden wins states carried by Hillary Clinton in 2016, plus Michigan, Wisconsin and Pennsylvania, he wins the Electoral College with at least 278 Electoral Votes (270 are required). The strong polls for Biden in Pennsylvania have moved it to be 2.0% more favourable for Trump than nationally; last week it was 2.7% better for Trump.

There are several states that are not near the Electoral College “tipping-point”, but where Biden is currently narrowly ahead or behind. Since last week, Biden has taken the lead in Ohio and Iowa, and is tied in Georgia. He trails by just two points in Texas. With Biden also narrowly ahead in North Carolina, it could be a blowout victory.

Biden is doing best relative to Clinton in the Midwestern and northeastern states where there are many who voted for Barack Obama in 2012, but Trump in 2016. In 2016, both Trump and Clinton were unpopular candidates, but this year Biden is currently at a net +3 favourability, while Trump is at -13 in RealClearPolitics averages. Biden appears to have some appeal to whites without a university education, who swung to Trump in 2016.

The FiveThirtyEight forecast still gives Trump a 21% chance to win the Electoral College, down 1% since last week. Trump wins the popular vote just 11% of the time. An article by Nate Silver says that, if the election were held today, Trump would have only a 9% chance to win.

Trump’s ratings with all polls in the FiveThirtyEight aggregate are 43.7% approve, 52.9% disapprove (net -9.2%). With polls of likely or registered voters, Trump’s ratings are 44.2% approve, 52.7% disapprove (net -8.5%). His net approval has improved about one point since last week.

The FiveThirtyEight Classic Senate forecast gives Democrats a 68% chance to win, up 1% since last week. The most likely outcome is a narrow 51 to 49 Democratic majority, unchanged from last week. The forecast gives Democrats an 80% chance of holding between 47 and 55 seats after the election.

Biden wins first presidential debate

The first presidential debate between Biden and Trump occurred Tuesday. A CBS News post-debate scientific poll gave Biden a narrow 48-41 victory, while a CNN poll gave him a far more emphatic 60-28 win. Trump needed a clear win to change the current polling. There will be two more presidential debates on October 15 and 22, and a vice presidential debate on October 7.

The major headlines from the debate were that it was a shouting match, and Trump’s refusal to denounce white supremacists. I have said before that the US economy’s fast recovery from the April coronavirus lows is Trump’s best asset for re-election, but he did nothing during the debate to tell a positive story about the economy.

Concerning the Supreme Court fight over the replacement for the late Ruth Bader Ginsburg, a Morning Consult poll found that a record 62% supported the Affordable Care Act (Obamacare), while 24% were opposed. In March, this was 55-29 support. There is clear danger for Trump and Republicans in appointing a judge who will overturn Obamacare.

New Zealand poll: Labour short of majority

A new Colmar Brunton New Zealand poll has Labour on 47%, National 33%, ACT 8% and the Greens 7%. If repeated at the October 17 election, Labour would win 59 of the 120 seats, two short of a majority. You can read more at my personal website.

Colmar Brunton: NZ Labour leads National by 48-31

Three-and-a-half weeks before the October 17 New Zealand election, the first media poll since late July has Labour still well ahead.

A New Zealand Colmar Brunton poll, conducted September 17-21 from a sample of 1,008, gave Labour 48% (down five since late July), National 31% (down one), the right-wing ACT 7% (up two), the Greens 6% (up one) and NZ First 2% (steady). Labour PM Jacinda Ardern led Opposition Leader Judith Collins by 54-18 as better PM (54-20 in July).

Under New Zealand’s proportional representation system, parties must either win a single-member seat or clear the 5% threshold to be represented in parliament. NZ First, which is currently a Labour coalition partner, is unlikely to return to parliament.

Even though the Greens have 6% in this poll, they should be anxious about clearing the 5% threshold. In the past, NZ polls have overstated the Greens. ACT is far better placed as their leader holds the single-member seat of Epsom.

If these poll results were the outcome on October 17, Labour would win 62 of the 120 seats, just enough for a majority. National would win 41 seats, ACT nine and the Greens eight.

There have been two other relatively recent NZ polls. A privately conducted late August to early September UMR poll gave Labour 53%, National 29%, ACT 6.2%, NZ First 3.9% and the Greens 3.2%. A Morgan poll, conducted in August, gave Labour 48%, National 28.5%, the Greens 11.5%, ACT 6% and NZ First 2.5%.

A concern for Labour is that their vote could continue to fall. If Labour’s election vote is under 45%, they may need the Greens to clear the 5% threshold to form a Labour/Green government.

Time running out for Trump

Seven weeks before the November 3 election, Biden still leads Trump by about five points in the Electoral College tipping-point states.

In FiveThirtyEight’s national aggregate of Donald Trump vs Joe Biden polls, Biden leads Trump by 6.9% (50.3% to 43.4%). This is an improvement for Trump from three weeks ago, when he trailed by 8.2%. In the key states, Biden leads by 7.5% in Michigan, 6.8% in Wisconsin, 5.0% in Arizona, 4.8% in Pennsylvania and 2.3% in Florida.

In my article three weeks ago, the difference in Trump’s favour between the Electoral College tipping-point state and the national vote had widened to three points, but this difference has fallen back to two points, with Arizona and Pennsylvania currently two points more favourable to Trump than national polls.

If Biden wins all the states carried by Hillary Clinton in 2016, plus Michigan, Wisconsin and Arizona, he gets exactly 269 Electoral Votes, one short of the 270 required for a majority. Maine and Nebraska award one EV to the winner of each of their Congressional Districts, and two to the statewide winner. All other states award their EVs winner-takes-all.

Under this scenario, Biden would need one of either Nebraska’s or Maine’s second CDs for the 270 EVs required to win the Electoral College. Nebraska’s second is a more likely win for Biden as it is an urban district.

The US economy has rebounded strongly from the coronavirus nadir in April. Owing to this, the FiveThirtyEight forecast expects some narrowing as the election approaches. Every day that passes without evidence of narrowing in the tipping-point states is bad news for Trump. Biden’s chances of winning in the forecast have increased from a low of 67% on August 31 to 76% now.

While Trump has improved slightly in national polls, some state polls have been very good for Biden. Recently, Biden has had leads of 16 points in Minnesota, 21 points in Maine, 10 in Wisconsin and 10 in Arizona.

Trump’s ratings with all polls in the FiveThirtyEight aggregate are currently 43.1% approve, 52.8% disapprove (net -9.7%). With polls of likely or registered voters, his ratings are 43.8% approve, 52.9% disapprove (net -9.1%). In the last three weeks, Trump has gained about two points on net approval, continuing a recovery from July lows.

The RealClearPolitics Senate map has 47 expected Democratic seats, 47 Republican seats and six toss-ups. If toss-ups are assigned to the current leader, Democrats lead by 51-49, unchanged from three weeks ago.

Coronavirus and the US economy

The US has just passed the grim milestone of over 200,000 deaths attributable to coronavirus. However, daily new cases have dropped into the 30,000 to 40,000 range from a peak of over 70,000 in July. Lower media attention on the coronavirus crisis assists Trump.

In the US August jobs report, 1.4 million jobs were created and the unemployment rate fell 1.8% to 8.4%. The unemployment rate has greatly improved from its April nadir of 14.7%. The headline jobs gained or lost are from the establishment survey, while the household survey is used for the unemployment rate. In August, the household survey numbers were much better than the establishment survey, with almost 3.8 million jobs added.

It is probably fortunate for Biden that the September jobs report, to be released in early October, will be the last voters see before the election. The October report will be released November 6, three days after the election.

I believe Trump should focus on the surging economy in the lead-up to the election, and ignore other issues like the Kenosha violence and culture war issues.

With the election drawing closer, I will do these reports every week from now on.

NZ: poll drought continues

A month out from the October 17 New Zealand election, there have not been any media-commissioned polls since late July. The only recently reported poll is a privately-commissioned UMR poll, conducted August 25 to September 2. That gave Labour 53%, National 29%, ACT 6.2%, NZ First 3.9% and the Greens just 3.2%.

The Greens’ low figure in this poll is very different from a Morgan NZ poll, conducted during August, which had the Greens on 11.5%. That poll gave Labour 48% and National 28.5%. Unless a party wins a single-member seat, the threshold required to enter parliament is 5%.

Affairs of state

One finely crafted electoral news item for every state (and territory) that is or might ever conceivably have been part of our great nation.

A bone for every dog in the federation kennel:

New South Wales

Gladys Berejiklian has backed a move for the Liberal Party to desist from endorsing or financially supporting candidates in local government elections, reportedly to distance the state government from adverse findings arising from Independent Commission Against Corruption investigations into a number of councils. Many in the party are displeased with the idea, including a source cited by Linda Silmalis of the Daily Telegraph, who predicted “world war three” because many MPs relied on councillors to organise their numbers at preselections.

Victoria

The second biggest story in the politics of Victoria over the past fortnight has been the expose of the activities of Liberal Party operator Marcus Bastiaan by the Nine newspaper-and-television news complex, a neat counterpoint to its similar revelations involving Labor powerbroker Adem Somyurek in June. The revelations have been embarrassing or worse for federal MPs Michael Sukkar and Kevin Andrews, with the former appearing to have directed the latter’s electorate office staff to spend work time on party factional activities.

Together with then state party president Michael Kroger, Bastiaan was instrumental in establishing a conservative ascendancy with help from Bastiaan’s recruitment of members from Mormon churches and the Indian community. Having installed ally Nick Demiris as campaign director, Bastiaan’s fingerprints were on the party’s stridently conservative campaign at the 2018 state election, which yielded the loss of 11 lower house Coalition seats. Religious conservatives led by Karina Okotel, now a federal party vice-president, then split from the Bastiaan network, complaining their numbers had been used to buttress more secular conservatives.

The Age’s report noted that “in the days leading up to the publication of this investigation, News Corporation mastheads have run stories attacking factional opponents of Mr Bastiaan and Mr Sukkar”. Presumably related to this was a report on Okotel’s own party activities in The Australian last weekend, which was long on emotive adjectives but short on tangible allegations of wrongdoing, beyond her having formed an alliance with factional moderates after the split.

Continue reading “Affairs of state”

Electoral College may save Trump

Just over two months before the November 3 election, the gap between Biden’s national lead and the likeliest tipping-point states widens to about three points.

In the FiveThirtyEight poll aggregate, Donald Trump’s ratings with all polls are 42.2% approve, 54.3% disapprove (net -12.1%). With polls of registered or likely voters, Trump’s ratings are 43.3% approve, 53.4% disapprove (net -10.1%). Since my article three weeks ago, Trump’s net approval has improved about one point, continuing a recovery from July lows.

In FiveThirtyEight’s national aggregate of Joe Biden vs Trump polls, Biden’s lead has slightly increased to a 50.6% to 42.2% margin, from a 50.0% to 42.5% margin three weeks ago. In the key states, Biden leads by 7.6% in Michigan, 5.9% in Wisconsin, 5.3% in Pennsylvania, 5.1% in Florida and 3.7% in Arizona. FiveThirtyEight adjusts state polls to the current national vote trends.

On current polling, Pennsylvania and Florida are the most likely tipping-point state. If Biden wins either of those states, and all other states more favourable for him, he wins the Electoral College. Trump wins by winning Pennsylvania, Florida and more pro-Trump states.

The problem for Biden is that the gap between the national vote and the vote in the tipping-point state has widened from 1.5% three weeks ago to 3.1%. That makes the scenario where Trump loses the popular vote by up to five points, but sneaks a win in the Electoral College more realistic. In 2016, Trump lost the popular vote to Hillary Clinton by 2.1%, but won the tipping-point state by 0.8% – a 2.9% difference.

FiveThirtyEight now has a model forecasting the election result, and that model currently gives Biden a 70% chance to win, down from about 72-73% a few days ago. Biden has received virtually no bounce from the Democratic national convention last week, while Trump could get some bounce from the Republican convention that concluded Friday in Australia.

Why has the national vote to tipping-point state gap widened recently? The Midwestern states have a higher percentage of non-University educated whites than nationally. Trump’s general behaviour has offended better-educated voters, and they are likely to vote for Biden.

Whites without a university education may have moved slightly back to Trump because new US coronavirus cases are slowing and the economy is improving. There are currently 30,000 to 50,000 daily new coronavirus cases, down from a peak of over 78,000 in late July. There are over 1,000 daily deaths on most days, but this is down from the April peak of well over 2,000.

On the economy, there is a clear downward trend in new jobless claims since their peak in April, and also a down trend in continuing jobless claims. As I wrote last time, the sharp increase in real disposable income in April owing to a stimulus explained Trump’s better economy ratings.

If the jobs situation continues to improve, and there is no resurgence in coronavirus, Trump could win another term in the same way he won his first term – by exploiting the greater share of whites without university education in the electoral battlegrounds than nationally. The swing to the right among such voters explained the Coalition’s win in Australia and the Conservative triumph in the UK last year. Will they be enough to re-elect Trump?

In the RealClearPolitics Senate map, Republicans currently lead Democrats by 46 seats to 44, with ten toss-ups. If toss-ups are assigned to the current leader, Democrats lead by 51 to 49, unchanged from three weeks ago.

NZ: Ardern has large leads over Collins on economy and coronavirus

Owing to the recent outbreak of coronavirus in Auckland, the New Zealand election has been moved from September 19 to October 17. Sadly, there have been no voting intention polls since late July, before this outbreak.

The only recent poll is a Horizon Research poll, conducted in late August, that did not ask voting intentions. Labour PM Jacinda Ardern led National leader Judith Collins by 54% to 26% on managing the economic recovery from coronavirus, and by 64% to 18% on managing coronavirus. Both leads were down from Ardern’s leads over Todd Muller in July.

Reid Research: NZ Labour with massive poll lead

Eight weeks before the September 19 New Zealand election, a new poll gives Labour a 61% to 25% lead over the opposition National.

There have been three leaders of the conservative National party in the last two months. On May 22, the unpopular Simon Bridges was ousted in a leadership spill, and replaced by Todd Muller. However on July 14, Muller resigned for health reasons, and was replaced by Judith Collins.

A Reid Research poll, conducted July 16-24 – after the second leadership change – gave Labour a thumping 60.9% (up 4.4% since the last Reid Research poll in early May). National was on 25.1% (down 5.5%).

New Zealand seats are allocated proportionally to parties that either win at least 5% of the national vote or a single-member seat. The Greens, with 5.7% in this poll, appear the only other party likely to clear the 5% threshold, although the right-wing ACT, with 3.3%, will also qualify if their leader holds Epsom. NZ First, which is currently a Labour coalition partner, has just 2.0%, and is unlikely to re-enter parliament.

After polls in May showed a blowout Labour lead, there was a much better poll for National in late June, when Muller was leader. That Colmar Brunton poll had Labour’s lead dropping to 50% to 38%, from 59% to 29% in May. I believe the big differences between the June Colmar Brunton poll and the July Reid Research poll are most likely caused by the coronavirus response, not by National leadership chaos.

In June, there was a mess-up in New Zealand’s quarantine system, in which two women from Britain were allowed to leave quarantine on compassionate grounds without being tested for coronavirus; they later tested positive. Alarm over this incident, which could have revived coronavirus in New Zealand, likely contributed to Labour’s poll drop.

Since this incident, quarantine has been strictly enforced. While there are 21 active coronavirus cases in New Zealand, these are returned overseas travellers, and there has been no recent news that would indicate community spread.

Globally, there have been over 16 million coronavirus cases and over 650,000 deaths. Up until a month ago, Australia looked good, but the hundreds of new cases in Victoria every day have damaged Australia. So it’s not as if New Zealanders need to look far afield to see what happens in countries that do not handle coronavirus well.

In my opinion, this election is primarily about handling of coronavirus. If there are no community transmissions in New Zealand before the September 19 election, Labour is likely to win easily. If the virus comes back, the election will likely be more competitive.

Jacinda Ardern’s performance recorded 85.3% approved of in this poll, and just 8.2% disapproved, for a net approval of +77.1. Collins was at 39.5% approve and 30.8% disapprove (net +8.7).

Miscellany: issues polling, drug law reform, Eden-Monaro wash-up, NZ poll

Concern about the state of the economy pushes climate change down the issue agenda batting order; evidence of a trend in favour of legalisation of cannabis; and New Zealand Labour still on track for a landslide in September.

Beneath this post is the latest offering from Adrian Beaumont on the polling picture in the United States ahead of the November presidential election. Closer to home, a few items of poll-related news:

• Pollster JWS Research has published results of its occasional True Issues survey, in which respondents are prompted to identify the five most important issues from a list of 20. The key changes since the last survey in February are a 17% increase for the economy and finances to 52% and an 11% drop in environment and climate change to 31%. The result for health issues has in fact changed little over recent surveys, although it has gained the top spot in the latest survey with a three point increase to 56%, overtaking cost of living which is down six to 53%. Interestingly, defence, security and terrorism is up six to 26%, which I take to reflect growing nervousness about China. Various other questions on COVID-19 are also featured, including findings that satisfaction with federal and state government performance is at record highs, with both scoring 19% for very good and 39% for good. The report notes that strongest results for state governments were recorded in Western Australia (83% combined very good and good) and the weakest were in Victoria (57%), although this is going off small sub-samples. The poll was conducted July 1 to 5 from a sample of 1000, just as the breakout in Victoria was beginning to gather pace.

• The Australian Institute of Health and Welfare has published the National Drug Strategy Household Survey 2019, in which 22,274 respondents were surveyed by Roy Morgan between April and September 2019 about their use of and attitudes towards illegal drugs. On the latter count, it found a plurality in favour of legalising cannabis for the first time, with 41% supportive and 37% opposed, with support having risen from 21% since 2007. It also found 57% support for allowing pill testing with 27% opposed.

• Kevin Bonham offers an interesting look at the unweighted data on voting intention that Essential Research effectively provides in its otherwise voting intention-less poll results, by way of identifying the size of the subsamples in its survey question breakdowns (for example, in the latest polls you can see from the “base” rows in the tables breaking down responses by voting intention that the sample included 299 Labor voters, 420 for the Coalition and 108 for the Greens). Notwithstanding the lack of weighting, the results paint an intuitively plausible picture of collapsing government support at the time of the bushfires, a reset when COVID-19 first reared its head, and an ongoing surge in Coalition support on the back of its support packages and the largely successful efforts to suppress the virus. These movements are considerably more variable than anything recorded by Newspoll, which has maintained the unnatural stability that was its hallmark before the 2019 election, despite its methodological overhaul.

Some wash-up from the Eden-Monaro by-election:

• John Black, former Labor Senator and now executive director of Australian Development Strategies, offered an ecological analysis of voting patterns in the Eden-Monaro by-election in The Australian on Monday. This pointed to a strong age-related effect in which older areas swung Labor and younger areas swung Liberal. Labor-swinging areas were also low-income with large accommodation and food industry workforces, while Liberal-swinging areas were white-collar and with high levels of employment in public administration. None of this would surprise students of the electorate and the result, given the Liberal swing in Queanbeyan and the Labor swing along the coast.

• Counting in the by-election is nearly complete, with today being the last day that postal votes received will be entered in the count. The latest results are continuing to be updated as they come through on my live results page. With probably a couple of dozen postals to be entered in the count, Labor holds a lead of 764. Of remaining interest will be the distribution of preferences, presumably to be conducted early next week, which will offer some insight into exactly how many Nationals and Shooters preferences flowed to Labor – contentious subjects both on the conservative side of politics.

Meanwhile across the pond:

• Roy Morgan published a New Zealand voting intention poll this week that was shortly overtaken by events, with the conservative opposition National Party experiencing its second leadership change in two months earlier in the week. The poll had Labor down two points from the previous poll in May to 54.5%, National up half a point to 27%, the Greens up two to 9%, Act New Zealand up 1.5% to a new peak of 5%, and New Zealand First apparently headed towards extinction with a one point drop to 1.5%. The poll was conducted by phone from a sample of 879, but all we are told of the field work period is that it was conducted during June.

• Concurrent with the New Zealand election on September 19 will be a non-binding referendum on cannabis legalisation. Poll results on this question are all over the shop: one poll last month, by Colmar Brunton, had 40% for and 49% against, while another, by Horizon Research, had 56% for and 43% against.